Volume 1 Number 4 October 2010 - Scientific Research Publishing · 2011. 9. 7. · used to study the temperature dependent conductivity by the ITC 502S Oxford temperature controller

Materials Sciences and Applications, 2010, 1, 177-258 Published Online October 2010 in SciRes (http://www.SciRP.org/journal/msa/)

Copyright © 2010 SciRes. MSA

TABLE OF CONTENTS

Volume 1 Number 4 October 2010

Electrical Conductivity, Magnetoconductivity and Dielectric Behaviour of (Mg,Ni)-Ferrite below Room Temperature

S. Ghatak, A. K. Meikap, M. Sinha, S. K. Pradhan………………………………………………………………………………177

Synthesis and Electrical Characterization of BaTiO3 Thin Films on Si(100)

V. R. Chinchamalatpure, S. A. Ghosh, G. N. Chaudhari……………………..……………………………………………………187

The Effect of the pH of Ammonum Nitrate Solution on the Susceptability of Mild Steel to Stress Corrosion Cracking (SCC) and General Corrosion

F. S. Mohammed, A. G. Elramady, S. E. A. Yahya………………………………………………………………………………191

Effect of Heat Treatment on Mechanical Properties of Al-1.5Cu-9.5Zn-3Mg Rapidly Solidified Alloy

E. S. Gouda, E. M. Ahmed, N. L. Tawfik………………………………………………………………………………………199

Anti-Corrosion Performance of Cr+6-Free Passivating Layers Applied on Electrogalvanized

C. R. Tomachuk, A. R. D. Sarli, C. I. Elsner…………………..…………………………………………………………………202

Effect of Mn Doping on Solvothermal Synthesis of CdS Nanowires

Z. Jindal, N. K. Verma……………………………………………………………………………………………………………210

Nanoindentation Study of Al356-Al2O3 Nanocomposite Prepared by Ball Milling

Y. Mazaheri, F. Karimzadeh, M. H. Enayati…………………………………………………………………..…………………216

Molar Binding Energy of Zigzag and Armchair Single-Walled Boron Nitride Nanotubes

L. Chkhartishvili, I. Murusidze…………………………………………………………………………………………………222

Effect of Solution Concentration on the Electrospray/Electrospinning Transition and on the Crystalline Phase of PVDF

L. M. M. Costa, R. E. S. Bretas, R. Gregorio Jr.…………………………………………………………………………………246

Polypyrrole Coated PET Fabrics for Thermal Applications

A. C. Sparavigna, L. Florio, J. Avloni, A. Henn………………………………………………………………..………………252

Materials Sciences and Applications (MSA)

Journal Information

SUBSCRIPTIONS

The Materials Sciences and Applications (Online at Scientific Research Publishing, www.SciRP.org) is published quarterly by

Scientific Research Publishing, Inc., USA.

Subscription rates: Print: $50 per issue.

To subscribe, please contact Journals Subscriptions Department, E-mail: [email protected]

SERVICES

Advertisements

Advertisement Sales Department, E-mail: [email protected]

Reprints (minimum quantity 100 copies)

Reprints Co-ordinator, Scientific Research Publishing, Inc., USA.

E-mail: [email protected]

COPYRIGHT

Copyright©2010 Scientific Research Publishing, Inc.

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by

any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except as described below, without the

permission in writing of the Publisher.

Copying of articles is not permitted except for personal and internal use, to the extent permitted by national copyright law, or under

the terms of a license issued by the national Reproduction Rights Organization.

Requests for permission for other kinds of copying, such as copying for general distribution, for advertising or promotional purposes,

for creating new collective works or for resale, and other enquiries should be addressed to the Publisher.

Statements and opinions expressed in the articles and communications are those of the individual contributors and not the statements

and opinion of Scientific Research Publishing, Inc. We assumes no responsibility or liability for any damage or injury to persons or

property arising out of the use of any materials, instructions, methods or ideas contained herein. We expressly disclaim any implied

warranties of merchantability or fitness for a particular purpose. If expert assistance is required, the services of a competent

professional person should be sought.

PRODUCTION INFORMATION

For manuscripts that have been accepted for publication, please contact:

E-mail: [email protected]

Materials Sciences and Applications, 2010, 1, 177-186 doi:10.4236/msa.2010.14028 Published Online October 2010 (http://www.SciRP.org/journal/msa)


177


Somenath Ghatak1, Ajit Kumar Meikap1, Manika Sinha2, Swapan Kumar Pradhan2

1Department of Physics, National Institute of Technology, Deemed University, Durgapur India; 2Department of Physics, University of Burdwan, Burdwan, India. Email: [email protected] Received March 19th, 2010; June 17th, 2010; October 9th, 2010.

ABSTRACT

We report a comprehensive study of electrical transport properties of stoichiometric (Mg,Ni)-ferrite in the temperature range 77 ≤ T ≤ 300K, applying magnetic field upto 1T in the frequency range 20 Hz-1 MHz. After ball milling of MgO, NiO and -Fe2O3 and annealing at 1473K, a (Mg,Ni)-ferrite phase is obtained. The temperature dependency of dc re-sistivity indicates the prevalence of a simple hopping type charge transport in all the investigated samples. The activa-tion energy decreases by annealing the samples by 1473K. The dc magnetoresistivity of the samples is positive, which has been explained by using wave function shrinkage model. The frequency dependence of conductivity has been de-scribed by power law and the frequency exponent ‘s’ is found to be anomalous temperature dependent for ball milling and annealing samples. The real part of the dielectric permittivity at a fixed frequency was found to follow the power law /(f,T) Tn. The magnitude of the temperature exponent ‘n’ strongly depends on milling time and also on annealing temperature. The dielectric permittivity increases with milling and also with annealing. An analysis of the complex im-pedance by an ideal equivalent circuit indicates that the grain boundary contribution is dominating over the grain con-tribution in conduction process. Keywords: Ferrites, Chemical Synthesis, X-Ray Scattering, Transport Properties

1. Introduction

Small ferri-magnetic oxides, technically known as fer-rites have attracted considerable attention not only from a fundamental scientific interest but also from a practical point of view for growing applications in the magnetic, electronic and microwave fields [1-7]. Simultaneous presence of magnetic and dielectric nature of ferrites is vastly exploited in a variety of applications at different frequencies. The special feature of these materials is that the properties can be tailored over wide ranges by appro-priate substitution of various ions in the chemical for-mula unit and control of processing procedures. Ferrites are extensively used in magnetic recording, information storage, colour imaging, bio-processing, magnetic refrig-eration and in magneto optical devices [5-7]. Ferrites also have great promise for atomic engineering of materials with functional magnetic properties. The formation of corrosion product on the out of core surfaces in pressur-ized heavy water reactors (PHWRs) are major problem.

Ferrite having spinal structure such as magnetic and nickel etc play a major role to prevent such problem. Thus at-tempts are being made to study the various ferrites to evaluate the impact of substitution of the divalent metal ions to modify the properties of these oxides.

Spinals are characterized by a very compact oxygen array with cations in tetrahedral (A) and octahedral (B) coordination and may be described by the IV(A1-iBi) VI(B2-IAi)O4 structural formula, where IV and VI repre-sent tetrahedrally and octahedrally coordinated sites, A and B are cations with variable valency and i the inver-sion parameter. Normal spinal are those with i = 0, in-verse spinals those with i = 1.

Different synthetic roots are employed in preparation of ferrites [8-10]. High energy ball milling is a very suit-able solid state processing technique for the preparation of nanocrystalline ferrite powder exhibiting new and un-usual properties [11-14]. The objectives of the present work are 1) to prepare the Mg-Ni ferrite by ball milling



178

the stoichiometric mixture of MgO, NiO and α-Fe2O3 and 2) to study the anomalous transport properties of ballmilled and post-annealed samples below room tempera-ture.

2. Experimental

MgO, NiO and α-Fe2O3 powders were taken in 0.25: 0.25: 0.5 mol% respectively and were hand-ground by an agate mortar pestle in a doubly distilled acetone medium for more than 5 h. The dried homogeneous powder mixture was then termed as unmilled stoichiometric homogene-ous powder mixture. A part of this mixture was ball milled at room temperature in air in a planetary ball mill (Model P5, M/S Fritsch, GmbH, Germany) with hard-ened chrome steel vial of volume 80 ml using 30 hard-ened chrome steel ball of 10 mm diameter, at ball to powder mass ratio 40:1 up to 20 h. Some of the selected ball milled samples (8 h and 20 h) were post annealed at 1473K each for 1h duration in a programmable furnace. The X-ray diffraction (XRD) patterns of the unmilled, ball milled and post annealed powders were recorded (step size = 0.020 2, counting time = 5 sec, angular range = 15-800 2) using Ni-filtered CuKα radiation from a highly stabilized and automated Philips generator (PW1830) operated at 40 KV and 20 mA. The generator is coupled with a Philips X-ray powder diffractometer con-sisting of a PW 3710 mpd controller, PW1050/37 gonio- meter and a proportional counter. The Rietveld’s analysis based on structure and microstructure refinement of XRD data [15-19] is adopted in the present case for micro-structure characterization and phase transformation ki-netics of ferrite phase in the course of milling and post annealing the ball-milled powder mixture.

The electrical conductivity of the samples was meas-ured by a standard four probe method by using 81/2-digit Agilent 3458 multimeter and 6514 Keithley Electrometer. The ac measurement was carried out with a 4284A Agilent Impedance analyzer up to the frequency 1 MHz at different temperatures. Liquid nitrogen cryostat was used to study the temperature dependent conductivity by the ITC 502S Oxford temperature controller. To measure the ac response, samples were prepared as 1 cm dia pel-lets by pressing the powder under a hydraulic pressure of 500 MPa. The density of the pressed pellets were in the range 3.74 g/cc to 5.89 g/cc. Fine copper wires were used as the connecting wire and silver paint was used as coat-ing materials. The capacitance (CP) and the dissipation factor (D) were measured at various frequencies and temperatures. The real part of ac conductivity and real and imaginary part of dielectric permittivity have been calculated using the relations /(f) = 2fo//(f), /(f) = CPd/0A and //(f) = /(f)D respectively, where 0 = 8.854 × 10-12 F/m, A and d are the area and thickness of the

sample respectively. CP is the capacitance measured in farad; f is the frequency in Hz. The magnetoconductivity was measured in the same manner varying the transverse magnetic field B 1T by using an electromagnet.

3. Results and Discussion

Figure 1 shows the recorded XRD patterns of unmilled (0 h) and ball milled mixture of MgO, NiO and α-Fe2O3 powders for different durations of milling. The powder pattern of unmilled (0 h) mixture contains only the indi-vidual reflections of MgO, NiO and -Fe2O3 phases. The intensity ratios of individual reflections are in accordance with the stoichiometric composition of the mixture. After 3 h of milling, the particle size of all phases reduces con-siderably which is evident from the broadened reflections of all phases. There is no clear evidence of ferrite phase formation in 3 h ball milled sample as all intense reflec-tions of ferrite phase are overlapped with broadened re-flections of starting phases. But the intensity ratios of the starting phases were changed after 3 h of milling and a careful observation of the reflection at 2θ = 35.42º clearly reveals the fact that the (110) reflection of α-Fe2O3 phase stands out as the most strongest reflection. This signifi-cant change in intensity indicates the formation of the ferrite phase which has its most intense (311) peak at 2θ = 35.715º. Intensities of MgO reflections show quite a large value in ball milled samples up to 20 h milling and MgO phase stands out as the major phase in the course of milling. This increment of MgO phase may be attributed to the formation of MgO-NiO solid solution [20] as both

20 30 40 50 60 70 80

8000

10000

12000

14000

(111

)

(440

)

(400

)

(222

)(3

11)

(220

)

(Mg Ni)Fe2O

4

(220

)

(200

)

MgOFe

2O

3

NiO(1

04)

(012

)

(101

)

(2

20)

(1

0 1

0)

(3

00)

(214

)

(018

)

(1

16)

(0

24)

(006

)(1

13)

(110

)

(104

)

(012

)

0h

8h

20h

Inte

nsity

(arb

.unit)

2(degree)

Figure 1. X-ray diffraction patterns of unmilled and ball milled stoichiometric mixture of MgO, NiO and -Fe2O3 powders.



179

the phases have same crystal structure (MgO; cubic, space group Fm 3 m (ICDD PDF # 87-0653) and NiO; cubic, space group Fm 3 m (ICDD PDF # 040835)) and radii of Mg+2 (0.72 Å) and Ni+2 (0.69 Å) ions are very close.

The mole fraction, lattice parameter and particle size of all phases present in the unmilled and ball milled sam-ples with increasing milling time are given in Table 1. The mole fraction of -Fe2O3 phase decreases rapidly with increasing milling time and that of NiO phase de-creases slowly. The NiFe2O4 phase is noticed to form after 3 h milling and its content increases with increasing milling time. The continuous increase of the content of MgO phase up to 20 h of milling above its starting value (0.25 mole fractions) indicates that MgO phase is not contributing in ferrite phase formation, furthermore, a part of NiO phase is diffused into MgO matrix. Therefore only a small percentage of Ni+2 ions participate in ferrite formation and the formed ferrite phase is eventually a Ni-ferrite phase. It is clearly evident from the Table 1 that the ferrite phase has formed initially with a low value of lattice parameter (0.833 nm) and then saturates at a value 0.840 nm at higher milling times. The lattice parameter of MgO decreases and that of NiO increases up to 8 h of milling and after that both the lattice pa-rameter values saturate at higher milling time. The con-traction of MgO lattice is due to the substitution of larger Mg+2 ions by smaller Ni+2 ions in the MgO lattice. Simi-larly the small increase of lattice parameter of NiO phase is due to the replacement of small amount of smaller Ni+2 ions by larger Mg+2 ions. After 8 h milling no further solid solution is formed because most part of MgO and NiO were used up in the formation of both MgO solid solution and ferrite phases. It is also evident from the Table 1 that the lattice parameter of the -Fe2O3 phase did not change appreciably with increasing milling time, indicates that both MgO and NiO phases did not diffuse into -Fe2O3 lattice. On other hand, all starting phases are showing a decrease in their particle size with in-creasing milling time. The particle size of -Fe2O3 phase decreases sharply from ~161 nm to a value ~17 nm within 3 h of milling time and remains almost unchanged in higher milling time. NiO phase also shows a consid-erable decrease in its particle size (from ~46 nm to ~20 nm) within 3 h of milling and further milling has a very slow decreasing effect on its particle size. The MgO phase initially has a low value of particle size (~25 nm) in comparison to the other two starting phases and de-crease in particle size of MgO phase is very small with increasing milling time. The ferrite phase formed with a very small particle size (~4 nm) and with increasing mill-ing time the size decreases very slowly and finally attains a value ~3 nm after 20 h of milling.

Figure 2 shows the XRD patterns of ball milled sam-

ples annealed at temperature 1473K. It seems that the (Mg,Ni)-ferrite phase is formed completely after this heat-treatment. However, a critical Rietveld analysis re-veals the presence of a very small amount of NiO phase along with the ferrite phase (Table 1). It indicates that almost a stoichiometric (Mg,Ni)-ferrite phase has been obtained at 1473K. The Rietveld analysis also reveals that ~0.92 mol fraction inverse spinel ferrite phase is formed both in 8 h and 20 h ball milled samples. This indicates that the amount of ferrite phase formation is independent of milling time. By measuring particle size we actually measure the coherently diffracting zone of a grain. The particle or crystallites re separated from each other by grain boundaries and the grain boundaries are nothing but bulk crystal imperfections in a crystal. The size of the crystallites in the ball milled samples is in the nanometer range. As can be seen from the experiment, annealing the sample increases the size of the particles. Heat energy helps to annihilate the deformations in the crystals. As a result of grain boundaries started to vanish during annealing and the small crystallites agglomerate together to form larger particles due to intra-grain diffu-sion. The experimentally observed patterns (I0) of the annealed samples are fitted with theoretically simulated patterns (Ic) as shown in Figure 3. The accuracy of fit-ting is shown by the fitting residual I0-Ic, plotted at the bottom of respective patterns.

The dc resistivity of different (Mg,Ni)-ferrite samples was measured as a function of temperature. It is observed

20 30 40 50 60 70 80

4000

6000

8000

(104

)

(012

)

(222

)(1

01)

(533

)

(440

)

(422

)

(511

)

(400

)

(311

)

(220

)

(111

)

NiO

Fe2(Mg,Ni)O4

8h

20h

Inte

nsity

(ar

b. u

nit)

2(degree)

Figure 2. X-ray powder diffraction patterns of ball-milled mixtures of MgO, NiO and -Fe2O3 powders annealed at temperature 1473K.



180

Table 1. Microstructure parameters of unmilled and ball milled (Mg,Ni)-ferrite revealed by Rietveld’s X-ray powder struc-ture refinement analysis.

Lattice parameter Sample Phase Present Mole fraction

a (nm) c (nm) Particle size (nm)

MgO 0.2549 0.4212 25.12

NiO 0.1703 0.4177 46.72 MNF-0 h

(0 h ballmilled)

α- Fe2O3 0.5747 0.5035 1.3753 161.47

MgO 0.3715 0.4197 20.94

NiO 0.1102 0.4178 20.25

α- Fe2O3 0.4149 0.5038 1.3758 17.03

MNF-3 h

(3 h ballmilled)

NiFe2O4 0.1033 0.8330 4.44

MgO 0.3871 0.4193 15.88

NiO 0.1103 0.4180 15.61

α-Fe2O3 0.3565 0.5041 1.3761 16.08

MNF-8 h

(8 h ballmilled)

NiFe2O4 0.1460 0.8404 4.99

MgO 0.4589 0.4194 12.78

NiO 0.0402 0.4180 12.05

α- Fe2O3 0.1514 0.5050 1.3733 15.72

MNF-20 h

(20 h ballmilled)

NiFe2O4 0.3495 0.8401 2.71

Fe2(Mg,Ni)O4 0.9232 0.8357 495.49 MNF-8 h-1473K

(8 h ballmilled &

annealed at 1473K) NiO 0.0768 0.4195 217.67

Fe2(Mg,Ni)O4 0.9208 0.8353 574.44 MNF-20 h-1473K

(20 h ballmilled &

annealed at 1473K) NiO 0.0792 0.4194 211.41

that resistivity decrease with increasing temperature, which suggests the semi conducting behavior of the sam-ples. Generally in ferrite phase the conduction mecha-nism arises due to exchange of electrons between the ions of the same elements present in more than one val-ance state, more are randomly distributed over crystallo-graphic lattices site. It is well known that in spinal ferrite Ni and Mg ions prefer to occupy ‘B’ and ‘A’ sites re-spectively. The conductivity in (Mg,Ni)-ferrite may be due to electron hopping between Fe+3 Fe+2 and hole hopping between Ni+2 Ni+3 at octahedral site. The decrease in resistivity with increase in temperature is due to the increase in drift mobility of the charge carriers. The temperature dependence of resistivity found to fol-low the Arrhenius equation,

0 exp a

B

ET

K T

(1)

where (0) is the resistivity at infinite temperature, Ea is

the activation energy, KB is the Boltzmann constant. Ac-cording to the Figure 4 the linear variation of ln[(T)] with 1/T indicates the prevalence of a simple hopping type charge transport in all the investigated samples. The values of Ea are obtained from the slopes of the different straight lines curves in the Figure 4 (0.29 eV for MNF-8 h, 0.13 eV for MNF-8 h-1473K, 0.36 eV for MNF-20 h, 0.12 eV for MNF-20 h-1473K). Hence the activation energy increases by increasing milling time due to de-crease of particle size. However, it is also seen that the activation energy decreases by annealing the samples by 1473K. This decrease of activation energy may be due to the increase of particle size where the metal core in-creases by vanishing the grain boundaries by annealing the samples.

The magnetic field dependent resistivity of the sam-ples has been measured under the influence of magnetic field of strength < 1T. The variation of magnetoresistiv-ity with magnetic field at T = 300K for different samples



181

20 30 40 50 60 70 80

4000

6000

14000

16000

18000

20000

NiO

Fe2(Mg,Ni)O4

_.

Ic

I0

I0-I

c

I0-I

c

8h

20h

Inte

nsity

(ar

b.un

it)

2(degree)

Figure 3. Observed (.) and calculated (-) X-ray powder dif-fraction patterns of post annealed ballmilled powder mix-ture of MgO, NiO and -Fe2O3 revealed by Rietveld powder structure refinement analysis. Peak positions of phases pre-sent are shown at base line as small markers (I). is shown in Figure 5. It is observed that the room tem-perature magnetoresistivity of all samples is positive. The magnitude of the maximum percentage change of resistivities [((B,T) − (0,T))/(0,T) × 100] in the presence of magnetic field of 0.8 T at 300K were ob-served about 57.4% for MNF-0 h, 24.1% for MNF-20 h and 3.4% for MNF-20 h-1473K. It is observed that the magnetoresistivity of the investigated samples decreases with increasing the milling time and also by annealing. The measured magnetoresistivity data could be explained by simple phenomenological model that consists of two simultaneously acting hopping processes, namely the wave function shrinkage model [21,22] and the forward interference model [23-25]. The wave function shrinkage model corroborates the fact that by applying a magnetic field the wave functions of electrons are contracted and reduces the average hopping length. This corresponds to a positive magnetoresistivity (negative magnetoconduc-tivity) i.e., resistivity increases with increasing magnetic field. On the other hand, the forward interference model takes into account the effect of forward interference among random paths in the hopping process between two sites spaced at a distance equal to the optimum hopping distance and the theory predicts the negative magnetore-sistivity (positive magnetoconductivity). For the sample having small localization length, the average hopping length Rhop = (3/8)(TMott/T)1/4Lloc is small and the wave

Figure 4. Temperature dependence of dc resistivity of dif-ferent (Mg,Ni)-ferrite samples.

Figure 5. Variation of dc magnetoconductivity with mag-netic field at temperature T = 300K of different samples. function shrinkage effect is dominated. But this effect is not evident in samples having large localization length, where the quantum interference effect [26,27] is domi-nated. Therefore, the sign and magnitude of the magne-toresistivity changes due to competition of the two (wave function shrinkage and quantum interference) types of contributions. As the magnetoresistivity ratio of the in-vestigated samples increases with increasing magnetic field at a temperature 300K, we assume that the contribu-tion due to wave function shrinkage model predominated over the quantum interference model. So, we analyzed our measured data in the light of the wave function shrinkage model. According to this model, for a small magnetic field, the magnetoconductivity ratio can be expressed by the following relationship [21]

3/42 42

1 2

,ln

0,loc MottB T e L T

t BT T

(2)

Where, t1 = 5/2016 and Lloc is the localization length. Figure 5 shows a linear variation in the plot of ln [(B,T)/(0,T)] versus B2 for different samples. The



182

points are the experimental data while the solid lines represent the best fits obtained on the basis of the wave function shrinkage model. It is evident from the Figure 5 that the experimental data can be well described by the theory as indicated in (2). It is observed from the fitting that the slope of the curves are 0.59 for MNF-0 h, 0.19 for MNF-20 h and 0.07 for MNF-20 h-1473K samples. The slope of unmilled MNF-0 h sample is much greater than the ballmilled and annealed samples. This is because in unmilled sample, the individual phases of oxides of Fe, Ni and Mg contribute the magnetoresistivity. However in ballmilled and annealed samples, ferrite phases exist and by milling, the particle size decreases and more disorder presents in the sample. Since, in sample with higher dis-order, electronic wave functions are more localized within smaller regions resulting smaller localization length. Therefore, the lowering of slope arises due to reduction of localization length.

The ac conductivity of Mg-Ni ferrite samples are in-vestigated in the frequency range 20 Hz to 1 MHz and in the temperature range 77 T 300K. The measured data showed that the variation of conductivity with frequency at a particular temperature is prominent at higher fre-quencies, whereas at low frequencies it is almost inde-pendent with frequencies, this could be attributed to the dc contribution. A general feature of amorphous semi-conductors or disordered systems is that the frequency dependent conductivity ac(f) obeys a power law with frequency. The total conductivity /(f) at a particular temperature over a wide range of frequencies can be ex-pressed as [28-30]

/ ( ) ( ) sdc ac dcf f f (3)

where dc is the dc conductivity, is the temperature dependent constant and the frequency exponent s < 1. The value of ac(f) has been determined upon subtraction of the dc contribution from the total frequency dependent conductivity /(f). Figure 6 shows the linear variation of ln[ac(f)] with ln(f) at different temperatures for the sam-ple MNF-20 h-1473K. Similar behavior was observed for all other samples. The value of ‘s’ at each temperature has been calculated from the slope of ln[ac(f)] versus ln(f) plot for each temperature. The trend of change in ‘s’ with temperature is shown in Figure 7 for different sam-ples. The temperature dependency of ‘s’ of disordered systems has been explained by two physical processes such as correlated barrier hopping (CBH) [30] and quan-tum mechanical tunneling (QMT) like electron tunneling (ET) [31], small polaron tunneling (SPT) [30] and large polaron tunneling (LPT) [29]. As the nature of tempera-ture dependency of ‘s’ for different conduction processes are different, the exact nature of charge transport may be obtained experimentally from the temperature variation

Figure 6. Frequency dependent ac conductivity at different temperatures of MNF-20 h-1473K sample. of the frequency exponent ‘s’. According to the corre-lated barrier hopping model ‘s’ increases with the de-crease in temperature. From the trend of change in ‘s’ with temperature for unmilled sample (MNF-0 h), it is presumed that the correlated barrier hopping is suitable. According to this model, the charge carrier hops between the sites over the potential barrier separating them and the frequency exponent ‘s’ is given by the expression [30].

0

61

1ln

B

H B

k Ts

W k T

(4)

where WH is the effective barrier height and o is the characteristic relaxation time. According to (4), for large values of WH/kBT, the variation of ‘s’ with frequency is so small that it is effectively independent of frequency [32]. On the other hand, the linear variation of ln[ac(f)] vs ln(f) in Figure 6 supports that ‘s’ is independent of frequency in our investigated samples. Therefore, we fitted experimental data with (4) as function of tempera-ture alone with WH and o as fitting parameters. In Fig-ure 7 the points represent the experimental data whereas solid lines are the theoretical best fit values obtained from (4) for MNF-0 h sample and the best fitted values of the parameters WH and o (at a fixed frequency of f = 10 KHz) are 1.16 eV and 5.97 × 10-13 S. The value of WH are, as expected, higher than the activation energy meas-ured from dc contribution and the values of the characteristic relaxation time o are comparable with those that would be expected for typical inverse phonon fre-quency. Therefore, it may be concluded that the ac con-ductivity of MNF-0 h sample can be described by CBH model. But for MNF-20 h sample, ‘s’ has to increase first upto T < 150K and then decrease with further increasing of temperature (T > 150K). Similar trend was observed in annealed (MNF-20 h-1473K) sample with weaker temperature dependency. Hence, the anomalous behavior



183

Figure 7. Variation of frequency exponent ‘s’ with temperature of different samples.

of ‘s’ with temperature (T < 150K) for MNF-20 h and MNF-20 h-1473K cannot be understood completely with CBH model and indicates the another mechanism of transport for carriers in these investigated systems. How- ever, according to small polaron tunneling ‘s’ only in-creases with increasing temperature. We try to fit the experimental data for T < 150K with the small polaron tunneling (SPT) theory [30], but the fit yields the un-physical values of the parameters. So the temperature dependence of ‘s’ is in conflict with SPT theory. To have a clear concept of this, we plotted the variation of /(f) with temperature for MNF-20 h-1473K sample in the Figure 8. At a particular frequency the real part of com-plex conductivity increases with temperature and is found to follow a power law /(f) Tn, which are shown as the solid lines in Figure 8. The values of n have been calculated from the power law fitting and found to be strongly frequency dependent. With increasing frequency from 1 KHz to 1 MHz the values of ‘n’ decreases from 10.1 to 7.3 for MNF-20 h and 9.5 to 7.2 for MNF-20 h-1473K. According to the CBH model [30] the ac con-ductivity /(f) is expressed as /(f) T2R

6 Tn with n = 2 + (1-s)ln(1/o) for broad band limit and /(f) R

6 Tn with n = (1-s)ln(1/o) for narrow band limit, where R = e2/o[WH – kBTln(1/o)]. We have calculated theoretically the values of ‘n’, taking o = 4.34 × 10-14 s and the value of s = 0.74 for 300K and 0.79 for 77K for MNF-20 h sample. With increasing frequency from 1 KHz to 1 MHz, the calculated values of ‘n’ varies 7.73 to 5.93 for 300K and 6.63 to 5.17 for 77K for broad band limit and 5.7 to 3.9 for 300K and 4.6 to 3.2 for 77K for narrow band limit. The experimental values did not match with the theoretical values, this indicates that both the broad band and narrow band are not suitable for ex-

Figure 8. AC conductivity as a function of temperature at different frequencies of MNF-20 h-1473K sample. plaining the temperature dependency of ac conductivity. Therefore, this observation cannot be understood com-pletely in terms of the existing theory of charge transport. Anyway, more studies are necessary to formulate the true mechanism and this experimental result may add impetus to the theoretical community to rethink this issue.

The variation of real part of dielectric permittivity /(f,T) with temperature is shown in Figure 9 for differ-ent samples at f = 1 MHz. In the /-T plot there is no sharp peak till the temperature is raised to 300K, which is the maximum temperature employed in our investigation. At a particular frequency the real part of dielectric per-mittivity increases with temperature and is found to fol-low a power law /(f) Tn, which are shown as the solid lines in Figure 9. The values of n have been calculated from the power law fitting and found that its value is strongly dependent on milling time and also on annealing temperature. Generally the ferrite exhibits interfacial polarization due to structural inhomogeneities and exis-tence of free charges [33]. It is thought that the hopping electrons at low frequencies may be trapped by the in-homogeneities. The increase of /(f) with temperature at a particular frequency is due to the drop in the resistance of the ferrite with increasing temperature. The low resis-tance promotes electron hopping, hence resulting in a larger polarizability or larger /(f). The frequency de-pendence of real part of the dielectric permittivity /(f) have also been studied for different samples and shown in Figure 10 for T = 300K and in Figure 11 for MNF- 20 h-1473K sample at different yet constant temperatures. A weak variation is noticed in the dielectric pemittivity at lower temperature, although a large variation of the same is observable at higher temperature for all the samples.



184

Figure 9. Thermal variation of dielectric constant of differ-ent samples at 1 MHz frequency.

Figure 10. Variation of dielectric constant as function of frequency at T = 300K for different samples.

Figure 11. Variation of dielectric constant as function of frequency at different temperatures of MNF-20 h-1473K sample.

At a fixed temperature, the dielectric pemittivity /(f) increase sharply with decreasing frequency and this sharp increase shifts to lower frequencies as the temperature is reduced. Such sudden increase of real part of the dielec-tric constant /(f) at low frequency can be attributed to the presence of large degree of dispersion due to charge transfer within the interfacial diffusion layer present be-tween the electrodes. The magnitude of the dielectric dispersion is temperature dependent. At lower tempera-ture, the freezing of the electric dipoles through the re-laxation process is easier. So there exists decay in po-larization with respect to the applied electric field, which is evidenced by the sharp decrease in /(f) at lower fre-quency region. When the temperature is high, the rate of polarization formed is quick and thus the relaxation oc-curs in high frequency. Due to this, the position of the sharp increase shifts towards higher frequency by in-creasing temperature. Therefore, the frequency behavior of /(f) is due to inhomogeneous nature (containing dif-ferent permittivity and conductivity regions) of the sam-ples, where the charge carriers are blocked by poorly conducting region. The effective dielectric permittivity of such inhomogeneous systems is given by Maxwell Wagner capacitor model [34-35]. The complex imped-ance of such systems can also be modeled by an ideal equivalent circuit consisting of resistance and capaci-tance due to grain and interfacial grain boundary contri-bution and it can be expressed as

/ / /

0

1Z Z iZ

i C (5)

/

2 21 1

g gb

g g gb gb

R RZ

R C R C

(6)

2 2/ /

2 21 1

g g gb gb

g g gb gb

R C R CZ

R C R C

(7)

where sub indexes ‘g’ and ‘gb’ refer to the grain and in-terfacial grain boundary respectively, R = resistance, C = capacitance, = 2f and C0 = free space capacitance. The real part of the complex impedance have been cal-culated from the experimental data for real (/) and imaginary (//) part of the dielectric permittivity by using the relation Z/(f) = //(f)/[Co(/(f)2 + //(f)2)] for different samples and analyzed by (6). Figure 12 shows the fre-quency dependence of the real part of the complex im-pedance at room temperatures for different samples. The points are the experimental data and the solid lines are the theoretical values obtained from (6). The grain and grain boundary resistances and capacitances have been extracted from this analysis at room temperature, whose values lie within the range 25 to 87 K for Rg, 0.5



185

Figure 12. The real part of the complex impedance versus frequency at T = 300K of different samples. to 1.02 M for Rgb, 0.63 to 1.56 nF for Cg and 0.43 to 0.92 nF for Cgb for different samples. It is observed from the Figure 12 that the experimental data are reasonably well fitted with the theory. It is observed that the grain capacitance (Cb) are comparable with the grain boundary capacitance (Cgb) for the investigated samples. However, the resistance due to interfacial grain boundary is much larger in compare to the grain resistance. This implies that the grain boundary contribution dominates over the grain contribution. Since the ferrites are semiconductors, the conduction process can be explained by hopping mechanism, where the carrier mobility is dominated by a factor that increases with temperature exponentially. This temperature dependent factor is controlled by thermal activation in order to overcome the potential barrier be-tween the sites by hopping.

4. Conclusions

The above experimental observations suggest the following facts: 1) A Ni-ferrite phase and MgO-NiO solid solution is obtained in ball milling the powder mixtures of MgO, NiO and -Fe2O3. 2) Ni-ferrite phase is ob-tained in the ball milling process is a non-stoichiometric phase with a number of cation vacancies. 3) Particle size of Ni-ferrite phase reduces to ~3 nm within 20 h of mill-ing. 4) After annealing at 1473K, ~0.92 mol fraction of (Mg,Ni)-ferrite phase is obtained. The dc resistivity de-creases with increasing temperature and the same follows a hopping type charge transport. The magnetoresistivity is positive and its magnitude reduces with increasing the milling time and also by annealing and it can be ex-plained by the wave function shrinkage model. The real part of the complex ac conductivity was found to follow

the power law /(f) T pf s. The magnitude of the tem-

perature exponent ‘p’ strongly depends on frequency and its value decreases with increasing frequency. A detailed analysis of the temperature dependence of the universal dielectric response parameter ‘s’ revealed that the corre-lated barrier hopping is the dominating charge transport mechanism for only unmilled sample, however, anoma-lous temperature dependency has been observed for ball milling and annealing samples, which cannot be ex-plained in terms of existing theory of charge transport. Anyway, more studies are necessary to formulate the true mechanism and this experimental result may add impetus to the theoretical community to think about this issue. At a particular frequency the real part of the dielectric per-mittivity was found to follow the relation /(f,T) Tn. The magnitude of the temperature exponent ‘n’ strongly depends on milling time and also on annealing tempera-ture. The frequency dependent real part of the dielectric permittivity shows large degree of dispersion at low fre-quency, but rapid polarization at high frequencies, which can be interpreted by Maxwell-Wagner capacitor model. The complex impedance of such systems can also be modelled by an ideal equivalent circuit consisting of re-sistance and capacitance due to grain and interfacial grain boundary contribution. The details analysis of this indicates that the grain and grain boundary capacitances are comparable with each other; however, the resistance due to interfacial grain boundary is much larger in com-pare to the grain resistance.

5. Acknowledgements

This work has been carried out under Grant nos. F.27-1/ 2002. TS.V dated 19.03.2002 and F.28-1/2003.TS.V dated 31-03.2003 sanctioned by the MHRD, Government of India. The authors gratefully acknowledge the principal assistance received from the above organization during this work.

REFERENCES

[1] I. Anton., I. D. Dabata and L. Vekas, “Application Orien-tated Researches on Magnetic Fluids,” Journal of Mag-netism and Magnetic Materials, Vol. 85, No. 1-3, 1990, pp. 219-226.

[2] R. D. McMickael, R. D. Shull, L. J. Swartzendruber, L. H. Bennett and R. E. Watson, “Magnetocaloric Effect in Su-perparamagnets,” Journal of Magnetism and Magnetic Materials, Vol. 111, No. 1-2, 1992, pp. 29-33.

[3] D. L. Leslie-Pelecky and R. D. Rieke,” Magnetic Proper-ties of Nanostructures Materials,” Chemistry of Materials, Vol. 8, No. 8, 1996, pp. 1770-1783.

[4] T. Hirai, J. Kobayashi and I. Koasawa, “Preparation of Acicular Ferrite Fine particles Using an Emulsion Liquid



186

Membrane System,” Langmuir, Vol. 15, No. 19, 1999, pp. 6291-6298.

[5] R. H. Kodama, “Magnetic Nanoparticles,” Journal of Magnetism and Magnetic Materials, Vol. 200, No. 1-3, 1999, pp. 359-372.

[6] K. V. P. M. Shafi, Y. Koltypin, A. Gedanken, R. Pro-zorov, J. Balogh, J. Lendvai and I. Felner, “Sonochemical Preparation of Nanosized Amorphous NiFe2O4 Particles,” The Journal of Physical Chemistry B, Vol. 101, No. 33, 1996, pp. 6409-6414.

[7] D. Niznansky, M. Drillon and J. L. Renspinger, “Prepara-tion of Magnetic Nanoparticles (γ-Fe2O3) in the Silica Matrix,” IEEE Transaction on Magnetics, Vol. 30, No. 2, 1994, pp. 821- 823.

[8] J. M. Yang, W. J. Tsuo and F. S. J. Yen, “Preparation of Ultrafine Nickel Ferrite Powder Using Ni and Fe Tar-trates,” Journal of Solid State Chemistry, Vol. 145, No. 1, 1999, pp. 50-57.

[9] Y. Shi, J. Ding, X. Liu and J. Wang, “NiFe2O4 Ultrafine Particles Prepared by Co-Precipitation/Mechanical Al-loying,” Journal of Magnetism and Magnetic Materials, Vol. 205, No. 2-3, 1999, pp. 249-254.

[10] P. Druska, U. Steinike and V. Sepelak, “Surface Structure of Mechanically Activated and of Mechanosynthesized Zinc Ferrite,” Journal of Solid State Chemistry, Vol. 146, No. 1, 1999, pp. 13-21.

[11] V. Sepelak, K. Tkacova, V. V. Boldyrev and U. Steinike, “Crystal Srtucture Refinement of the Mechanically Acti-vated Spinel-Ferrite,” Materials Science Forum, Vol. 228-231, 1996, pp. 783-788.

[12] V. Sepelak, A. Yu, U. Rogachev, D. Steinike, C. Uecker, S. Wibmann and K. D. Becker, “Structure of Nanocrys-talline Spinel Ferrite Produced by High Energy Ballmilling Method,” Acta Crystallographica A, Suppl. Issue, Vol. A 52, 1996, p. C367

[13] V. Sepelak, A. Yu, U. Rogachev, D. Steinike, C. Uecker, F. Krumcich, S .Wibmann and K. D. Becker, “The Syn-thesis and Structure of Nanocrystalline Spinel Ferrite Pro-duced by High Energy Ball-Milling Method,” Materials Science Forum, Vol. 235-238, 1997, pp. 139-144.

[14] V. Sepelak, U. Steinike, D. C. Uecker, S. Wibmann and K. D. Becker, “Structural Disorder in Mechanosynthe-sized by Zinc Ferrite,” Journal of Solid State Chemistry, Vol. 135, No. 1, 1998, pp. 52-58.

[15] H. M. Rietveld, “Line Profile of Neutron Powder Diffrac-tion Peaks for Structure Refinement,” Acta Crystal-lographica, Vol. 22, 1967, pp. 151-152.

[16] H. M. Rietveld, “A Profile Refinement Method for Nu-clear and Magnetic Structures,” Journal of Applied Crys-tallography, Vol. 2, 1969, pp. 65-71.

[17] R. A. Young, “The Rietveld Method,” Oxford University Press, Oxford, 1996.

[18] L. Lutterotti, P. Scardi and P. Maistrelli, “LSI-a Com-puter Program for Simultaneous Refinement of Material Structure and Microstructure,” Journal of Applied Crys-tallography, Vol. 25, No. 3, 1992, pp. 459-462.

[19] L. Lutterotti, “MAUD Version 2.046.” http://www.ing. unitn. it/~luttero/maud

[20] W. D. Kingery, H. K. Bowen and D. R. Uhlmann, “In-troduction to Ceramics,” 2nd Edition, John Wiley and Sons, New York, 1976.

[21] B. I. Shklovskii, “Positive Magnetoresistance in the Vari-able Range Hopping Conduction Regimes,” Soviet Phys-ics Semiconductors, Vol. 17, 1983, p. 1311.

[22] B. I. Shklovskii and A. L. Efros, “Electronic Properties of Doped Semiconductors,” Springer, Berlin, 1994.

[23] V. L. Nguyen, B. Z. Spivak and B. I. Shklovskii, “Tunnel Hops in Disordered Systems,” Soviet Physics-JETP, Vol. 62, 1985, p. 1021.

[24] U. Sivan, O. Entin-Wohiman and O. Imry, “Orbital Mag-netoconduction in the Variable-Range-Hopping Regime,” Physical Review Letters, Vol. 60, No. 15, 1988, pp. 1566-1569.

[25] R. Rosenbaum, A. Milner, S. Hannens, T. Murphy, E. Palm and B. Brandt, “Magnetoresistance of an Insulating Amorphous Nickel-Silicon Film in Large Magnetic Fields,” Physica B, Vol. 294-295, 2001, pp. 340-346.

[26] M. S. Fuhrer, W. Holmes, P. L. Richards, P. Delaney, S. G. Louie and A. Zettl, “Nonlinear Transport and Local-ization in Single Walled Carbon Nanotubes,” Synthetic Metals, Vol. 103, No. 1-3,1999, pp. 2529-2532.

[27] Y. Yosida and I. Oguro, “Variable Range Hopping Con-duction in Bulk Samples Composed of Single Walled Carbon Nanotubes,” Journal of Applied Physics, Vol. 86, No. 2, 1999, pp. 999-1003.

[28] N. F. Mott and E. Davis, “Electronic Process in Noncrys-talline Materials,” 2nd Edition, Clarendon, Clarendo, 1997.

[29] A. R. Long, “Frequency-Dependent Lossin Amorphous Semi-Conductors,” Advances in Physics, Vol. 31, No. 5, 1982, pp. 553-637.

[30] S. R. Elliott,” A.c. Conduction in Amorphous Chalco-genide and Pnictide Semiconductors,” Advances in Phys-ics, Vol. 36, No. 2, 1987, pp. 135-217.

[31] A. L. Efros, “On the Theory of a.c. Conduction in Amor-phous Semiconductors and Chalcogenide Glasses,” Phi-losophical Magazine B, Vol. 43, No. 5, 1981, pp. 829-838.

[32] M. Ghosh, A. Barman, S. K. De and S. Chatterjee, “Transport Properties of Hcl Doped Polyaniline and Poly-aniline-Methyl Cellulose Dispersion,” Journal of Applied Physics, Vol. 84, No. 2, 1998, pp. 806-811.

[33] B. G. Soares, M. E. Leyva, G. M. O. Barra and D. Khast-gir, “Dielectric Behaviour of Polyaniline Synthesized by Different Techniques,” European Polymer Journal, Vol. 42, No. 3, 2006, pp. 676-686.

[34] S. S. Suryavanshi, S. R. Patil, S. A. Patil and S. R. Sa-want, “d.c. Conductivity and dielectric Behaviour of Ti4+ Substituted Mg-Zn Ferrites,” Journal of the Less-Com-mon Metals, Vol. 168, No. 2, 1991, pp. 169-174..

[35] J. C. Maxwell, “A Treatise on Electricity and Magnet-ism,” Vol. 1, Oxford University Press, Oxford, 1988.



187


V. R. Chinchamalatpure, S. A. Ghosh, G. N. Chaudhari

Nanotechnology Research Laboratory, Shri Shivaji Science College, Amravati, India. Email: [email protected] Received May 10th, 2010; June 13th, 2010; July 29th, 2010.

ABSTRACT

BaTiO3 thin film has been deposited on Si(100) substrate using sol-gel process and deposited by using spin – coating technique. The BaTiO3/Si(100) structures were studied by structural and electrical characteristics. The X-ray diffrac-tion of BaTiO3/Si(100) shows that the diffraction peaks become increasing sharp with increasing calcination tempera-tures indicating the enhance crystallinity of the films. Scanning electron microscopy of BaTiO3 thin films shows the crack free and uniform nature. The capacitance-voltage measurement of BaTiO3 thin film deposited on Si(100) an-nealed at 600 shows large frequency dispersion in the accumulation region. The current-voltage measurement of BaTiO3/Si shows the ideality factor was approaches to unity at 600. Keywords: Sol-Gel Technique, BaTiO3 Thin Film, C-V, I-V

1. Introduction

Barium Titanate (BaTiO3), a well-known dielectric mate-rial has been used as an insulating material to fabricate MIS structures. BaTiO3 exhibits several advantages, properties such as high charge storage capacity, good insulating property, low leakage current density and high dielectric breakdown strength. It is also a potential mate-rial for active microwave tunable device because of its variable dielectric constant under external electric field. It has also been shown that BaTiO3 works as an excellent buffer layer for YBa2Cu3O7-δ high-Tc superconductor on various substrates, in particular Si and Al2O3 [1]. Re-cently, J S Lee et al. [2] fabricated n-MOSFET structures sequentially with the CMOS process and investigated the insulator characteristics, programming and resistance behaviour of BaTiO3 films. BaTiO3 are transparent in the visible and infrared [3] and possess strong electro-optic coefficients making them attractive for active and passive optical components.

BaTiO3 thin film can be grown using pulsed laser de-position [4], chemical vapor deposition (CVD) [5], metal organic chemical vapour deposition [6-7], polymeric precursor method [8-11], sol-gel synthesis [12] and mo-lecular beam epitaxy (MBE) [1,13]. MBE can be used to grow complex oxide thin film substrate with atomic layer accuracy [14]. High quality BaTiO3 thin films are gener-

ally grown on lattice matched substrate such as MgO and SrTiO3. Integration of BaTiO3 thin film with silicon processing may enable the properties of ferroelectric to be utilized in combination with CMOS in multi material integration. Ion beam – assisted deposition of various template layers such as MgO and Yttria – stabilized zir-conia has been used to develope thin film of super con-ducting YBa2Cu3O7 as well as oriented layers of (Pb, Ba) TiO3 [15,16].

In recent years, the sol-gel technique has gained inter-est in the area of processing of thin films because of the several advantages it offers, such as easier composition control, better homogeneity, low processing temperature and low equipment cost [17]. For the present investiga-tions BaTiO3 thin films have been deposited on Si(100) substrates using sol-gel technique through organic pre-cursor route.

2. Experimental Details

(100) oriented Si wafers were well rinsed with warm acetone and methanol followed by etching in HNO3/HF (1:1) for 1 minute. BaTiO3 thin films were deposited on the polish side of the Si using sol-gel process.

Barium acetate (Sigma Aldrich, UK) and Titanium butoxide were used as a starting material. Barium acetate was dissolved in glacial acetate acid and reflux in a re-flex condenser at a temperature of about 120 for six



188

hours to obtain a clear solution, after obtaining a clear solution, Titanium butoxide was added in a proper mole ratio and the solution was stirred to get homogenous precursor. The viscosity of the solution was varied by the addition of 2-methoxy ethanol. The precursor solution was coated on Si(100) wafers by spin coating technique. The spin coating was done at the rate of 3000 rpm. The film was annealed at 300 and 600 for 30 minutes.

The focus of the present work is to study of BaTiO3 thin films prepared by the sol gel technique and depos-ited by using spin – coating technique. The BaTiO3/ Si(100) structures were studied by structural and electri-cal characteristics. The crystalline nature and the phase formation of the deposited thin film were confirmed by power X-ray diffraction analysis using Cukα radiation (α = 1.5418 A.U.) surface morphology of the film were ob-served using scanning electron microscopy.

The C-V characteristics of the MIS diodes were meas-ured at 1 MHz frequency at room temperature in the dark using a lock in amplifier (EG & G Model 5204). The ohmic contact was made on the back side of Si(100) by successive deposition of Au-Ge (2:1) alloy under a vac-uum of 10-6 mbar and annealed at 400 under argon atmosphere for 2 minutes.


The crystalline quality of BaTiO3 precursor were pre-pared by sol-gel technique and deposited by using spin – coating technique. The Figure 1 shows the XRD pattern of BaTiO3 thin film on Si(100) substrate with different annealing temperatures. As deposited BaTiO3 thin film and annealed at 300 shows amorphous in nature. The crystalline behaviors of the deposited BaTiO3 thin film annealed at 600 shows sharp peaks indicating the en-hance the crystallinity of the films. The structure of the BaTiO3 thin film was found to be tetragonal.

Figure 2 show the FTIR spectra of BaTiO3 thin films annealed at 600. The intense band at the edge of detec-tion (260 cm-1) is assigned to Ti-O mode in developed network [13]. The Figure shows that the broad and strong band observed at 450 cm-1, characteristics of Ti-O bond.

Figure 3 (C1 & C2) shows the SEM micrograph of BaTiO3 thin films annealed at 300 and 600. The BaTiO3 thin films deposited on Si(100) annealed at 300 shows the agglomeration of the films shown in Figure 3 (C1). The BaTiO3 thin films deposited on Si(100) an-nealed at 600 shows the films became crack free and uniform nature.

The electrical properties of BaTiO3/Si(100) were stud-ied by using 1 MHz capacitance-voltage characteristics. Figures 4(a) and (b) shows room temperature C-V char-acteristics for the sample annealed at 300 and 600 of BaTiO3/Si(100). Figure 4(a) shows that the capacitance

Figure 1. XRD pattern of BaTiO3 thin film on Si (100) sub-strate annealed at 600.

Figure 2. FTIR spectra of BaTiO3 thin films annealed at 600. remains unchanged with in that bias range. Figure 4(b) shows BaTiO3 thin film deposited on Si(100) and an-nealed at 600 showed large frequency dispersion in the accumulation region. The capacitance was changed from the inversion value to a value above flat band capacitance at 1 MHz. The electrical properties of BaTiO3/Si(100) structure improved with annealing temp. at 600 shown in Figure 4(b). The result shows that the deposited Ba-TiO3 insulator is good enough to use as gate insulator of MIS FET.

For 300 annealed temperature, the C-V curve shows virtually flat which is an evidence of little modulation of the surface potential by the applied voltage, However when the film is subjected at 600 annealing tempera-ture the C-V plot shows a dramatic change in the 1 MHz freq. C-V behavior and considerable reduction in the frequency dispersion suggest a much closer approach to



189

Figure 3. SEM of BaTiO3 thin films annealed at (C1) 300 and (C2) 600.

Figure 4. Capacitance-Voltage characteristics of BaTiO3/Si (100) annealed at (a) 300; (b) 600.

surface accumulation, i.e., usually observed in the metal insulator semiconductor structure. Hence, the electrical properties of BaTiO3/Si(100) MIS structure improved with annealing temperature at 600.

Figure 5 shows Current-Voltage (I-V) Characteristics of a typical diode with 90 Ao thick BaTiO3 film on Si (100). The room temp. Current-voltage characteristics were obtained to access the quality of deposited thin films. The forward bias current – voltage characteristics were measured and fitted to the equation

/ 1avoJ J e nKT (1)

where J is the current density in amps/cm2 and n is ideal-ity factor of the diode. When n is close to unity, the re-sults of thermonic emission theory are valid such that,

** 2 / KToJ A T e (2)

where A** is the modified Richardson constant in amp/cm2 and Φ – Barrier height.

The saturation current density Js is given by

2* exps

q BnJ A T

kT

(4)

The parameter n is given by 1

(ln *)1

kT An

V q V

(5)

where is the height of Schottky barrier, is the Schottky barrier lowering and A* is the effective Rich- ardson constant.

For the annealing temperature at 300 and 600 of BaTiO3/Si shows the ideality factor was found to be varying

Figure 5. I – V Characteristics of a Typical diode with 90 Ao thick BaTiO3 film on Si (100) with annealed at (A) 300 and (B) 600.



190

between 1.2 to 1.1. The value of n = 1 is consistent with current injection through a shottkey barrier, While large value indicates conduction by generation recombination. A Poor ideality factor can have variety of interpretation among them, the effect of Tunneling can be excluded because of low carrier density in substrates. The ideality factor decreases with increasing annealing temp. of Ba-TiO3/Si as shown in Figure 5.

These results show that the electrical properties of Ba-TiO3/Si, MIS structure improved at higher annealing temp.

4. Conclusions

BaTiO3 thin film has been synthesized using sol-gel process and deposited on Si (100) by using spin – coating technique. The BaTiO3/Si(100) structures were studied by structural and electrical characteristics. The XRD and SEM of BaTiO3/Si(100) indicates the enhance crystallin-ity of the films with annealing temperature at 600. The C-V measurement of BaTiO3 thin film deposited on Si(100) annealed at 600 shows large frequency disper-sion in the accumulation region. The current-voltage measurement of BaTiO3/Si shows the ideality factor was approaches to unity at 600.

REFERENCES [1] S. Witanachi, S. Patel, D. T. Shaw and H. S. Kwok, “Ef-

fect of buffer layers on low‐temperature growth of mir-ror‐like superconducting thin films on sapphire,” Ap-plied Physics Letters, Vol. 55, No. 3, 1989, pp. 295-297.

[2] J. S. Lee, Y. H. Bae and Y. H. Lee, “A novel Ba-TiO3/SiO2-based voltage-programmable link applicable to on-chip programmable devices,” Semiconductor Sci-ence and Technolology, Vol. 15, No. 3, 2000, pp. 267-269

[3] Petraru, J. Schubert, M. Schmid and C. Buchal, “Ferro-electric BaTiO3 thin-film optical waveguide modulators,” Applied Physics Letters, Vol. 81, No. 8, 2002, pp. 1375-1377.

[4] T. García, P. Bartolo-Pérez, E. de Posada, J. L. Peña and M. Villagrán-Muniz, “Studies of pulsed laser deposition processes of BaTiO3 thin films,” Surface and Coatings Technology, Vol. 201, No. 6, 2006, pp. 3621-3624 .

[5] T. T. W. Kim, Y. S. Yoon, S. S. Yom and C. O. Kim, “Ferroelectric BaTiO3 films with a high-magnitude di-electric constant grown on p-Si by low-pressure metalor-ganic chemical vapor deposition,” Applied Surface Sci-ence, Vol. 90, No. 1, 1995, pp. 75-80.

[6] X. D. Wu, A. Inam, M. S. Hegde, B. Wilkens, C. C.

Chang, S. Miura, S. Matsubara, Y. Miyasaga and N. Shohata, “High critical currents in epitaxial YBa2Cu3O7−x thin films on silicon with buffer layers,” Applied Physics Letters, Vol. 54,No. 8, 1989, pp. 754-756.

[7] D. K. Fork, F. A. Ponce, J. C. Tramontana and T. H. Ge-balle, “Epitaxial MgO on Si(001) for Y‐Ba‐Cu‐O thin‐film growth by pulsed laser deposition,” Applied Physics Letters, Vol. 58, No. 20, 1991, pp. 2294-2296.

[8] V. Agrawal and M. Liu, “Preparations of Barium cerate- based thin films using a modified pechni process,” Jour-nal of Materials Science, Vol. 32, No. 3, 1997, pp. 619- 625

[9] V. Bouquet, S. M. Zanetti, C. R. Foschini, E. R. Leite, E. Longo and J. A. Varela, In: N. P. Bansal, K. V. Logan and J. P. Singh, Eds., Innovative Processing and Synthe-sis of Ceramic, Glasses, and Composites, Ceramic Transactions, The American Ceramic Society Inc., Wes-terville, Vol. 85, 1998, p. 333.

[10] S. M. ZanettiL, E. R. Eite, E. Longo and J. A. Varela, “Preparation and characterization of SrBi2Nb2O9 thin films made by polymeric precursors,” Journal of Materi-als Research, Vol. 13, No. 10, 1998, pp. 2932-2935.

[11] S. M. Zanetti, E. Longo, J. A. Varela and E. R. Leite, “Microstructure and phase evolution of SrTiO3 thin films on Si prepared by the use of polymeric precursors,” Ma-terials Letters, Vol. 31, No. 3-6, 1997, pp. 173-178

[12] L. Beckers, Y. Schubert, W. Zander, J. Ziesmann, A Eckau, P. Leinenbach, C. Buchal, “Structural and optical characterization of epitaxial waveguiding BaTiO3 thin films on MgO,” Journal of Applied Physics, Vol. 83, No. 6, 1998, pp. 3305-3310.

[13] S. Kwak, K. Zhang, E. P. Boyd, A. Erbin, B. J. Wilkens, “Metalorganic chemical vapor deposition of BaTiO3 thin films,” Journal of Applied Physics, Vol. 69, No. 2, 1991, pp. 767-772.

[14] M. E. Naggar, D. A. Boyd and D. G. Goodwin, “Charac-terization of highly-oriented ferroelectric PbxBa1-xTiO3,” Journal of Material Research, Vol. 20, No. 11, 2005, pp. 2969-2976.

[15] L. A. Wills, B. W. Wessels, D. S. Richeson and T. J. Marks, “Epitaxial growth of BaTiO3 thin films by or-ganometallic chemical vapor deposition,” Applied Phys-ics Letters, Vol. 60, No. 1, 1992, pp. 41-43.

[16] Lemoine, B. Gilbert, B. Michanx, J. P. Pirad and A. J. Leclux, “Synthesis of barium titanate by the sol-gel proc-ess,” Journal of Non-Crystalline Solids, Vol. 175, No. 1, 1994, pp. 1-13.

[17] M. Burgos and M. Langlet, “The sol-gel transformation of TIPT coatings: a FTIR study,” Thin Solid Films, Vol. 349, No. 1-2, 1999, pp. 19-23.



191


Fathia S. Mohammed1, Alyaa G. Elramady2, Salheddin E. Abu Yahya2

1Department of Chemical Engineering, American University of Sharjah, Sharjah, United Arab Emirates; 2The Petroleum Institute, Abu Dhabi, United Arab Emirates. Email: [email protected] Received April 18th, 2010; August 6th, 2010; August 16th, 2010.

ABSTRACT

This work investigates the relative aggressiveness of nitrate solutions at different pH values on mild steel towards stress corrosion cracking (SCC) and general corrosion. Electrochemical behavior and stress corrosion cracking susceptibility measurements were carried out in 52 Wt% ammonium nitrate solutions at 368° K and various pH values ranging from 0.77 to 9.64. Constant load stress corrosion test at 90% yield stress was conducted. Tested specimens were prepared and examined using the scanning electron microscope (SEM). The potentiodynamic polarization curves for different pH values again emphasized the validity of the gravimetric measurements and hence the mechanism of cracking was at-tributed to the stress that assisted the dissolution process. Keywords: Stress Corrosion Cracking, Ammonium Nitrate Solution, Mild Steel, Constant Load Test, Effect of PH

1. Introduction

Stress corrosion cracking (SCC) can lead to rapid and catastrophic failure in many different metals and alloys. This phenomenon occurs under conditions where a com-ponent is exposed to a mildly corrosive environment while under applied and/or residual tensile stress. Hence, metal parts with severe SCC can appear bright and shiny whilst being filled with microscopic cracks [1]. These factors along with the rapid progress of SCC make it common for SCC to go undetected prior to failure. Steel/ nitrate interaction is an issue in nitrogenous fertilizer plants, waste heat recovery boilers (WHRBs) in power generating plant and nuclear wastes [2].

The effect of pH and types of nitrate solution has been investigated [3], which concluded that the order of de-creasing aggressiveness of nitrate solutions corresponded to the order of increasing (initial) pH for a given chemi-cal strength, i.e., NH4

+, Ca++, K+, Na+. The aggressiveness of ammonium nitrate when com-

pared to other nitrates was attributed to its lower pH. Parkins [4] reported that the marked decrease in potency

at initial pH values in the region number 4 compared to either slightly higher or lower values depends probably upon pH changes in the solution during the test. The above work was concerned with the changes in pH of the bulk of solution and not with the pH at the crack tip re-gion, which undoubtedly more acidic. Steel has been characterized as being very susceptible to SCC at near- neutral pH [5].

In this work a comprehensive study on the effect of the pH of ammonium nitrate solution on the susceptibility of mild steel to stress corrosion cracking and general corro-sion was carried out. The results indicate that in some pH values the general and localized corrosion were the cause of failure. In other pH values the stress corrosion crack-ing were the cause of failures. The severity was con-firmed by the calculation of crack growth rate, morphol-ogy of the fracture surface by SEM and by polarization work.

2. Experimental Approach

2.1. Material

The work was carried out on mild steel of the following



192

composition (wt %): C 0.070, Mn 0.300, Si 0.093, S 0.044, P 0.019. The

Material was supplied in the form of 19 mm diameter rods. The corroding solution was prepared by using am-monium nitrate.

2.2. Specimen Preparation

2.2.1. Electrochemical Measurement The steel rods were hot-rolled at 1200° K to 4 mm thick strips. This was reheated to 1200° K in the furnace for 900 s, and then allowed to cool to room temperature. Most of the oxide film was removed by pickling in 30% HCl solution, and the surface was finally cleaned for cold- rolling by mechanical abrasion. The strips were reduced to 0.5 mm thick by cold-rolling.

Samples of 20 mm by 13 mm were prepared. A 3 mm holes were drilled at one end to suspend the samples, then the specimens were degreased with ether, annealed at 1200° K for 3.6 ks. The specimens were attached to the holder and the whole assembly coated apart from an area of 100 mm2 on one face.

2.2.2. Stress Corrosion Measurement The steel rod was hot-rolled at 1200° k and swaged cold to approximately 10 mm diameter. It was then annealed at 1200° k for 900 s, furnace cooled to 850° k, followed by air cooling to room temperature. The specimens were machined from the rod as shown in Figure 1. They have a gauge length of 15.8 mm and a gauge diameter of 3.2 mm.

2.2.3. General Corrosion Testing Samples of 40 mm by 15 mm were prepared. A 3 mm holes were drilled at one end to suspend the samples, then the specimen were degreased with ether, annealed at 1200 k for 3.6 ks.

2.3. Apparatus

2.3.1. Electrochemical Measurements For electrochemical measurements on unstressed speci-mens, a glass cell comprising two compartments was designed. The main compartment contained the working electrode and the platinum counter electrode. The refer-ence compartment contained a saturated calomel elec-trode. The complete cell is shown in Figure 2. The two compartments were connected by a salt bridge with a Luggin capillary. The glass joints that carried the work-ing and the counter electrodes also had a screw cap joint for the thermometer. There was another two openings in the main compartment, one for water condenser, and the other for gas and solution inlet, when working with de-aerated system. The reference compartment has a thermometer gas inlet and liquid inlet together with the

Figure 1. Stress corrosion test specimen (dimensions in mm).

Figure 2. Electrochemical polarization cell (schematic). 1) Main compartment; 2) Specimen/working electrode; 3) Platinum electrode; 4) Luggin capillary; 5) Thermometer; 6) Condenser and gas outlet; 7) & 13) Solution inlet; 8) & 14) Gas inlet; 9) Salt bridge; 10) Gas Outlet; 11) Reference compartment; 12) Saturated calomel reference electrode. saturated calomel electrode in one joint. The cell capacity is 0.4 dm3 of test solution. Only the main compartment of the cell was immersed in an oil bath controlling the re-quired temperature, the reference compartment was held at room temperature.

2.3.2. Stress Corrosion Measurement The majority of the work was conducted using the con-stant load method. The tensile properties of the material were measured in triplicate using Instron Tensile Testing Machine. In all the constant load tests the load applied was 90% of the predetermined yield stress.

For electrochemical measurements on stressed speci-mens, a glass cell consisting of two compartments was used; the main compartment contained the stress corro-sion specimen and the platinum counter electrode. The reference compartment contained the saturated calomel electrode, similar to the reference compartment described before. The two compartments were connected by a salt



193

bridge with a Luggin capillary. The capacity of the cell is 0.25 dm3, and the details are shown in Figure 3.

2.3.3. General Corrosion Measurement A flat-bottomed one-liter glass vessel with two necks was used. One neck held the water condenser; the speci-mens were suspended by a glass hook from the lower end of the condenser. A thermometer, inserted through the second neck.


To validate the results, all measurements were conducted at least three times under each specific environment.

3.1. Stress Corrosion Life and Corrosion Potential

The entire stress corrosion test carried out under a con-stant load of 90% of the yield stress (206.5 MN m-2). A series of stress corrosion tests were carried out to deter-mine the stress corrosion life in 52 Wt% NH4NO3 solu-tion at 368° k and various pH values ranging from 0.77 to 9.64. The corrosion potential was also recorded during the tests.

Figures 4, 5 and 6 show the changes in the corrosion potential during the stress corrosion test at different pH values. In solution of pH 2.78 and above jumps in the corrosion potential to more negative values were ob-served before failure occurs. At pH 0.77, no oscillations were observed.

Figure 7 shows the whole range of potential change, the initial and the final pH of the solution, and the stress corrosion life for some of the above tests. For lower pH values, the pH was more basic at the end of the test, while at high pH values (5.7 and above), the pH became more acidic at the end of the test. While results indicate that the critical range for cracking (i.e., where the stress corrosion life is minimum) is between pH 3.0 and pH 7.5, cracking occurs even at pH 9.6 but after very long period of time.

Staehle [6] reported that with respect to pH, a change of one unit of pH changes the solubility of oxide by three orders of magnitude for three valent ions such as Fe3+ and by two orders of magnitude for two valent ion such as Fe2+. It was reported that nuclear waste are all alkaline, with pH in the range 11-14. Even under these highly alkaline conditions, the presence of certain constituents, such as ni-trates can make the carbon steel susceptible to SCC [7]. Other researchers indicated that elevated pH are long considered corrosion inhibitors and did not have a domi-nant effect on the susceptibility to SCC in the range 10- 13.5 [8,9].

Figure 8 summarizes the relation between the pH of

Figure 3. Stress corrosion test cell (schematic). 1) Main compartment; 2) Specimen/working electrode; 3) Platinum electrode; 4) Luggin capillary; 5) Thermometer; 6) P.T.F.E nut; 7) Rubber washers; 8) Stainless steel nut; 9) Top shackle; 10) To condenser and gas outlet; 11, 18) Solution inlets; 12, 17) Gas inlets; 13) Salt bridge outlet; 14) Gas outlet; 15) Reference compartment; 16) Saturated calomel reference electrode.

Figure 4. Effect of pH on the corrosion potential/time behavior of mild steel during stress corrosion testing in 52 Wt% NH4NO3 at 368° K. the solution and the corrosion potential at different time during the test. The graph shows no straight forward sys-tematic correlation between the pH of the solution and the potential.

3.2. Metallographic Examination

Selected specimens were prepared for examination using the scanning electron microscope and the ordinary met-allographic microscope. The results are in Table 1.

Figures 9(a) and 9(b) show a fracture surface of a specimen which was stress corroded in a solution of pH



194

Figure 5. Effect of pH on the corrosion potential/time behavior of mild steel during stress corrosion testing in 52 Wt% NH4NO3 solutions at 368° K.

Figure 6. Effect of pH on the corrosion potential/time behavior of mild steel during stress corrosion testing in 52 Wt% NH4NO3 at 368° K. (pH 9.64). 0.77. The first shows an area where localized attack has occurred, whilst the second shows the ordinary structure.

The crack growth rate was very small (3 nms-1) with big reduction of the specimen diameter (660 μm) which indicates the high general and intergranular corrosion occurs during the test.

Very heavy attack was observed when specimens were broken in a solution of pH 4.2.

Figure 10 Illustrates part of the fracture surface. It is clear that that cracking occurs and the cracking rate was very high (71 nms-1), with very small reduction in the diameter (8 μm) which indicates the severity of such en-vironment.

Figure 11 shows a specimen which was stress corroded in a solution of pH 9.64.

Figure 7. The relationship between stress corrosion life, corrosion potential variations and solution pH changes for mild steel in 52 Wt% NH4NO3 solutions at 368° K.

Figure 8. Effect of pH on the corrosion potential of mild steel under applied stress in 52 Wt% NH4NO3 solutions at 368° K.

(a) (b)

Figure 9. Fracture surface morphologies of a stress corro-sion specimen after failure in NH4NO3 solution of pH 0.77 at 368°K (X 950). (a) localized intergranular attack and cracking; (b) Ductile failure.

In summary the results of the present work indicate the following facts: 1) a presence of stress corrosion cracking at high pH values (pH 9.64, Figures 6, 11 and Table 1). 2)



195

Table 1. The Effect of pH on the stress corrosion life and morphology of attack in 52 Wt% NH4NO3 solution at 368° K.

Type of Attack and Number of Locations pH

Average Stress Corrosion Life

(Ks) A B C

Maximum Depth of Penetration in

Grains

Reduction in Diameter (μm)

Crack Growth Rate (nm/sec)

0.77 62.64 3 5 660 3.2

2.78 56.5 7 25 - 17.7

4.2 27.3 6 1 44 8 71

8.8 108 1 7 - 2.6

9.64 970.2 2 - - -

A: Fine cracks; B: Wide cracks; C: Cracks visible to the naked eye

Figure 10. Fracture surface morphology of a specimen after stress corrosion failure in 52 Wt% NH4NO3 solution of pH 4.2 at 368° K showing the high degree of intergranular at-tack (X 250). The high potency of the solution at pH 2.78 compared to lower or higher pH values (Figure 4). 3) In a solution of pH 0.77, the SCC was associated with high rate of general attack (Figure 4, Figure 9 and Table 1).

Naris Sridhar et al. [10] reported that intergranular stress corrosion cracking (IG SCC) has generally not been observed when the pH greater than 11.0. According to potential pH or pourbaix diagram [11] dangerous zones

Figure 11. An illustration of catastrophic crack formation in a stress corrosion specimen after failure in 52 Wt% NH4NO3 solutions of pH 9.64 at 368° K. where SCC caused by nitrate solution is between pH 2.2 and 5.2, with corresponding potential between 0-I000 mVSCE.

Parkins et al. [12] reported that changes in the pH during the test show more relation to the results than the initial pH values. They showed that the time to failure correlated more significantly with the final pH of the solution than with initial values.

Relating the results shown in Figures 4, 5 and 6 to the stages of SCC indicate the followings: There are no sudden jumps to more active potentials

before failure at a very low pH value (0.77), and hence no indications of fast propagation period [13]. This suggests that at this pH value, general corro-sion and not stress corrosion is the predominant factor.



196

In the pH range 2.0 – 8.0, oscillations in potential are evident, occurring over a period representing 20 – 25% of the total life. This suggests that variations in pH above and below the natural pH of 4.2 do not affect the percentage of the total time taken up by the fast propagation period [13].

At relatively high pH (9.64), the duration of the fast propagation period is relatively unchanged, but it occupies only about 1% of the total life. This can be attributed to the very low rate of attack at such high pH value.

3.3. Corrosion Rate and Corrosion Potential Measurement on Unstressed Specimens

Figure 12 shows the effect of pH of the solution on both the general corrosion rate and the stress corrosion life. This figure indicates that the increase in the stress corro-sion life at lower pH values was associated with high general corrosion, while at higher pH values (greater than 7.5) it was associated with low general corrosion.

The variation of the corrosion potential with time for unstressed specimens was measured for period of ~70 ks at different pH values, ranging from 1.05 to 9.13 (Figure 13 and 14). The potential changed to the more noble di-rection as the test proceeded over almost the whole range of the investigated pH except at pH 9.13. At this pH the potential at the beginning was less noble, and after ap-proximately 30 ks more noble values were observed.

From Figures 12, 13 and 14 the following are observed: High general corrosion at lower pH values is ac-

companied by a more active potential. The non dependence of pH on the corrosion rate

between pH values of 4.2 and 6.9 is associated with unchanged corrosion potential.

The decrease in the corrosion rate with increasing pH in the range 6.0 – 8.5 is characterized by erratic changes in the corrosion potential, e.g. it is com-paratively noble at pH 7 and yet more active at higher values.

The above behavior probably reflected the difference of the solubility of the corrosion product in solutions of different pH values.

Figure 15 shows the effect of stress on the maximum corrosion potential during testing in 52 Wt% NH4NO3 at 368° K at different pH values. The stress appears to cause a shift in the corrosion potential to more active values in the range of the critical pH.

3.4. Crack Growth Rate

From the microscopic examination and the results of stress corrosion life in different pH values reported in Table 1, the crack growth rate was calculated by dividing

Figure 12. Effect of pH on the corrosion rate and stress corrosion life of mild steel in 52 Wt% NH4NO3 solutions at 368° K.

Figure 13. Effect of pH on the corrosion potential/time be-havior of mild steel in 52 Wt% NH4NO3 solutions at 368° K.

Figure 14. Effect of pH on the corrosion potential/time be-havior of mild steel in 52 Wt% NH4NO3 solutions at 368° K.



197

Figure 15. Effect of pH on the maximum corrosion potential value attained for mild steel in 52 Wt% NH4NO3 solutions at 368° K. the maximum measured crack depth by the total time to failure. The obtained value gives an estimate of the rate since it does not take into account the time to initiate cracks. The growth rate of any observed cracking is as-sumed to be constant throughout the exposure period. These estimates provide a simple semi quantitative di-agnostic to classify the SCC propensity. The following points regarding the crack growth rate are concluded from Table 1. In the solution of pH 0.77 the crack growth was only

3.2 nms-1 but there was a big reduction (18%) in the diameter which clearly indicates that general cor-rosion was dominating.

The crack growth rate at pH 2.78 is 17.7 nms-1 while it is 71 nms-1 at pH 4.2; thus it is obvious that the latter environment is more prone to SCC.

In the solution of pH 8.8 the crack growth rate was 2.6 nms-1 and this is probably because of the longer initiation period.

3.5. Potentiodynamic Polarization

Figure 16 shows the potentiodynamic polarization curves for different pH values of 52 Wt% NH4NO3 solution at 368 K ranging from 2.04 to 8.35.

The scanning started approximately 200 mv more negative than the open circuit potential (OPC) in the noble direction to more than +1500 mv using sweep rate of 0.33 mVs-1.

Several distinct characteristics for solution of different pH are as below: For pH 2.04 1) An active dissolution regime be-

tween −500 mVSCE and −150 mVSCE, 2) First pas-sive plateau at a current density of 2 × 103 Am-2 between potential -150mVSCE and zero mVSCE, 3)

Figure 16. Effect of pH on the potentiodynamic anodic po-larization behavior of mild steel in 52 Wt% NH4NO3 solu-tion at 368 K (sweep rate 0.33 mVs-1).

Broad active to passive transition peak starting at zero mV with corresponding current density of 0.85 × 103 Am-2, 4) Second dissolution regime between + 500 mV and 600 mV, 5) Second passive plateau at current density of 1.3 × 103 Am-2 between 600 mV and 1300 mVSCE, 6) A transpassive regime starting at 1300 mVSCE.

For pH 4.2 1) An active dissolution regime between −500 mVSCE and −350 mVSCE, 2) First active-passive transition starting at −350 mV with a corre-sponding current density of 1 × 103 Am-2, 3) An active dissolution regime between −300 mVSCE to −150 mVSCE, 4) A second active-passive transition starting at −150 mVSCE with a corresponding current density of 3 × 103 Am-2, 5) Transition peak at po-tential 0 mV with a corresponding current density of 0.6 × 103 Am-2, 6) A passive plateau at a current density of 0.4 × 103 Am-2 between 500 mVSCE and 1400 mVSCE, 7) The initiation of the transpassive regime at 1450 mVSCE.

For pH 6.97 1) An active dissolution regime be-tween −500 mVSCE and −150 mVSCE, 2) A passive plateau at a current density of 3.5 × 103 Am-2 be-tween −100 mVSCE and +500 mVSCE, 3) Sudden decrease in the current density to 0.2 × 103 Am-2 at −520 mVSCE, 4) passive plateau at current density of 0.2 Am-2 between 550 mVSCE and 1200 mVSCE, 5) The initiation of transpassive regime at 1200 mVSCE.

For pH 8.35 1) A slow dissolution from −250 mVSCE to 800 mVSCE to a maximum current density of 0.1 × 103 Am-2, 2) Another dissolution behavior from 800 mVSCE to 1200 mVSCE to a maximum current den-sity of 0.9 × 103 Am-2, 3) Passive plateau at a cur-rent density of 1 × 103 Am-2 between 1200 mVSCE



198

and 1600 mVSCE, 4) The initiation of transpassive regime at 1600 mVSCE.

From above it is clear that the potentiodynamic po-larization curves again emphasize the validity of the gra-vimetric measurement and show the influence of pH on the anodic dissolution characteristics of mild steel. These results were found to be in accord with previous research [14].

4. Conclusions

At very low pH values, stress corrosion cracking is asso-ciated with very high rate of general corrosion. In the region of pH 2.0 to 4.2 the stress corrosion life is relatively unchanged and general corrosion rate decreases with increasing of the pH level. Between pH 4.2 and 6.0, the corrosion rate and stress corrosion life are almost constant. In the region of pH 6.0 to 7.5, the stress corrosion life increases slightly and the corrosion rate decreases. Above pH 7.5, there is a noticeable increase in the stress corro-sion life while the general corrosion rate shows a marked decrease. Experimental observation suggests that an oxide film of critical physical properties is formed at immersion. This film suffers localized breakdown at the grain boundary. These limited grain boundary micro fissures will only propagate if aided by stress which facilitates the continuing action of the corrosion process. Subsequently, the precipitation of a layer of stifling corrosion product re-occurs and the cyclic process is repeated until failure. The local dissolution rate at the crack tip is accelerated with test time, which would be attributed to the continu-ous increase in the stress concentration. This reflects the interaction of stress and anodic dissolution during the SCC process.

REFERENCES [1] ASM International, “Metals Handbook (Desk Edition),”

Chapter 32 (Failure Analysis), American Society for Metals, 1997.

[2] C. Sean Brossia, et al., “A Study of Stress Corrosion Cracking and Localized Corrosion of Carbon Steel in Ni-trate Based Nuclear Waste,” NACE Conference, Georgia, 2009.

[3] R. N. Parkins and R. Usher, “The Effect of Nitrate Solu-tion in Producing Stress Corrosion Cracking in Mild Steel,” Proceeding of the First International Congress on Metallic Corrosion, London, 1961, p. 289.

[4] R. N. Parkins, “Environmental Aspect of Stress Corrosion Cracking in Low Strength Ferrite Steels,” Proceeding of the International Conference on Stress Corrosion Crack-ing and Hydrogen Embrittlement of Iron Base Alloys, Firming, 1973.

[5] Y. Z. Wang, R. W. Revie and M. T. Shahatas, “Early Stages of Stress Corrosion Cracks Development of X65 Pipeline Steel in Near Neutral pH Solution,” Materials for Resource Recovery and Transport, L. Collins, Ed., The Metallurgical Society of CIM, Montreal, 1998, p. 71.

[6] R. W. Stachle, “Framework for Predicting Stress Corro-sion Cracking,” Proceedings of Environmentally Assisted Cracking; Predictive Methods for Risk Assessment and Evaluation of Materials, Equipments and Structure, Or-lando, 2000.

[7] K. D. Boomer, J. beavers, et al., “A Study of Corrosion and Stress Corrosion Cracking of Carbon Steel Nuclear Waste Storage tanks,” Material Science and Technology Conference and Exhibition, Michigan, 2007.

[8] F. Gui, C. S. Brossia, et al., “On the Anodic Polarization Behavior of Carbon Steel in Hanford Nuclear Wastes,” Corrosion 2007, NACE, Houston, 2007.

[9] C. S. Brossia, C. Scott, et al., “Inhibition of Stress Corro-sion Cracking of Carbon Steel Storage Tanks at Han-ford,” Corrosion 2007, NACE, Houston, 2007.

[10] Narasi Sridhar, et al., “Proceeding of Environmentally Assisted Cracking Predictive Methods for Risk Assess-ment and Evaluation of Materials, Equipments and Strac-ture”,Orlando,STP 1401,241(2000).

[11] R. G. J. Leferink and W. M. M. Huijbregts, “Nitrate Stress Corrosion Cracking in Waste heat Recovery Boil-ers,” Anti – Corrosion Methods and Materials, Vol. 49, No. 2, 2002, pp. 118-126.

[12] R. N. parkins, et al., “Stress Corrosion Test Method,” British Corrosion Journal, Vol. 7, 1972, p. 154.

[13] F. S. Mohammed, “Stages of Corrosion Cracking of Mild Steel in Nitrate Solution,” Third International Material Conference, College of Engineering, Australia, 2008.

[14] Corrosion Source, Last Revised December 2005. http:// www.corrosionsource.com/handbook/testing/scc.htm.



199


El Said Gouda1,2, Emad Makboul Ahmed1, Nabih Lotfi Tawfik1

1Metal Physics Lab., Solid State Physics Department, National Research Center, Dokki, Egypt;

2Physics Department, Faculty of Science, Jazan University, Jazan, K.S.A. Email: [email protected] Received April 29th, 2010; June 17th, 2010; June 18th, 2010.

ABSTRACT

Ribbons with the composition Al-1.5Cu-9.5Zn-3Mg were prepared by melt spinning technique. Microhardness and ten-sile strength were measured. The melt spun hardness and ultimate tensile strength values were as high as 291 HV and 660 MN/m2, respectively. Hardness values are relaxed to lower values on prolonged thermal annealing to around 50%. X-ray diffraction lines corresponding to Cu, Zn and Mg were disappeared for the as melt spun ribbons, which indicates a complete solubility of these element in Al matrix. On prolonged thermal annealing these alloying elements were pre-cipitated. Keywords: Microhardness, Rapid Solidification, UTS, Tensile Strength

1. Introduction

The study of the material strength is an important subject because it is the first characteristic comes in mind when used in industrial applications specially that subjected to shock loading. Steel is a good example for the most strength materials, but its high density restricts its uses. Aluminum alloys are increasingly employed in many important manufacturing areas, such as the automobile industry, aeronautics and the military [1]. Currently, it offers the greatest potential for cost effective weight savings in automotive body structures and closures. With a density of only 33% of that of steel and a greater strength to weight ratio, there is the possibility for a weight savings of 40% to 50%. Also, Mg alloys are very attractive materials for producing lightweight compo-nents for automobiles because they have densities that are 66% of Al alloys and 22% of steel. With their lower density and moderate strength, Al-Mg alloys are well suited for a number of applications, ranging from steer-ing wheels and instrument panels to engine and transmis-sion components. The mechanical properties of the Al-Mg plastically processed alloys depend on the content of magnesium in the alloy. With an increase of magne-sium from 0.5 to 5% the properties increase; this rise is

greater when magnesium increases from 3 to 6% [2]. There are many studies characterize the strength and mechanical properties of Al-based and Mg-based alloys with different elements [3-8]. The present paper aims to characterize hardness and tensile strength of the quater-nary alloy Al-1.5Cu-9.5Zn-3Mg as an example for a high strength material.

2. Experimental

Al-1.5Cu-9.5Zn-3Mg alloy was prepared from 99.75 wt.% pure Al, 99.9 wt.% pure Cu, Zn and Al-10 wt.%Mg master alloy. The required quantities were weighted out and melted in electrical resistance furnace then thermally agitated to ensure the homogenization. The molten alloy was casted into graphite moulds to produce rods of 25 mm length and 4 mm diameter. A stream of the molten alloy at 850 was ejected by argon gas at a gauge pres-sure of 1.5 atm., from a silica tube with 0.5 μm orifice diameter. The melt jet fell on a copper wheel of the melt spinning apparatus fixed at 2950 r/m. The estimated cooling rate was about 105 K/s. The resulting alloys are in ribbons form of about 50 μm thickness and width 2 mm. X-ray diffraction analyses was performed to iden-tify the structure of the ribbons using a 1390 Philips X-ray Diffractometer with Cu radiation. The ribbons



200

were tensile tested with a gauge length of 5 cm at strain rate of 1.66 × 10-5 s-1, at room temperature. Measure-ments of hardness were done with the specimen placed against a glass slide with a load of 150 gm and indenta-tion time of 10 s. Tests were done for the as melt-spun ribbons and for fully aged ribbons.


3.1. X-Ray Diffraction

The cooling rates of the melt spinning process exceed 105

/s was high enough to retain the high concentrations of the alloying elements in solid solution with the Al- 1.5Cu-9.5Zn-3Mg alloy. This was confirmed by the presence of Al reflections only in the X-ray diffraction pat-tern of the as melt-spun ribbons as illustrated in Figure 1. For aged ribbon, the Al-Cu compound and the alloying elements start to precipitate and crystal growth starts to be exist, so additional X-ray diffraction peaks were formed for the aged ribbons as illustrated in Figure 2. The additional diffraction lines were corresponding to pure Mg, Zn and Al2Cu phases.

3.2. Tensile Test

Samples for the tensile test were chosen such that, a minimum variation in width and thickness has been ob-tained. The stander deviation in cross sectional area for each sample is ± 5%. Samples with gauge length 5 cm were tested. Figure 3 shows the load-elongation curves for as melt-spun and annealed ribbons. Each curve can be divided into two regions. The first region is a linear and ends at strain ratio ε/εf about 80% and 30%, for the as melt-spun and annealed samples, respectively, εf is the fracture strain. The second region is slightly curved due to yielding near the end of the test. Slope of the straight line in the first region represents Young's modulus of the as melt-spun sample. The ultimate tensile strength (UTS) for the as melt-spun sample was 660 MN/m2. This value decreased by annealing at 300 for 5 h to 442 MN/m2. Also, toughness, which is expressed as the area under the load-elongation curve until fracture, was calculated. It was 3.98 MN/m2 for the as melt-spun samples and 3.39 MN/m2 for the annealed sample.

3.3. Microhardness

Hardness of the as melt spun Al-1.5Cu-9.5Zn-3Mg alloy is 291 MN/m2 and decreases to 145 MN/m2 by thermal ageing at 220 for 5 h. This decrease takes place in two stages as illustrated in Figure 4. An initial fast stage fol-lows by slow stage in which a slight decrease in hardness can be observed. The as-cast rod with the same composi-tion gives the value of 141 MN/m2. By comparing the two values, it is noticed that, the as melt-spun ribbon

has a much higher value than that of the as cast rods, which agreement with other results [9] for alloy rapidly solidified at different cooling rates. On thermal aging, the

Figure 1. X-ray diffraction pattern of the Al-1.5Cu-9.5Zn- 3Mg melt spun ribbon.

Figure 2. X-ray diffraction pattern of the Al-1.5Cu-9.5Zn- 3Mg annealed ribbon.

(a) (b)

Figure 3. Load versus elongation for Al-1.5Cu-9.5Zn-3Mg melt spun ribbons (a) as melt spun and (b) aged for 5 h at 300.



201

Figure 4. Variation of Vickers hardness ratio H (t)/H (0) at 170 and 220 with aging time.

hardness decreases to reach limiting values of 50% of the as melt-spun values. It was observed that the hardness relaxation during isothermal ageing is much slower than that of resistivity relaxation at the same temperature. This behavior can be explained as that, hardness is closely related to size of precipitates. Also the electrical resistiv-ity is sensitive to point defects which are usually the first properties to recover. On the other hand, the hardness depends more on line imperfection which may require higher temperatures for recovery. So, rapid solidification has a significant effect on increasing the hardness. The higher value of hardness for the as melt spun state can be attributed to the effect of the solute atoms upon the sol-vent lattice and the nature of the lattice forces operative owning to the interaction of different atomic spices.

4. Conclusions

The effects of rapid solidification and thermal heat treatments on the mechanical properties of quaternary Al- 1.5Cu-9.5Zn-3Mg melt spun alloy were studied. A maximum solid solubility value of 9.5 wt.% Zn in α-Al was obtained for the as melt-spun alloy, 3% Mg and 1.5% Cu were also obtained as solid solutions in Al-matrix. The existence of the alloying elements as solid solution in α-Al significantly enhances microhardness, ultimate ten-

sile strength U.T.S values. After aging the values relaxed to lower values as a result of the Al2Cu, Zn and Mg pre-cipitations. Hardness and U.T.S were as high as 291 HV and 660 MN/m2, respectively.

REFERENCES [1] S. J. Maddox, “Review of Fatigue Assessment Procedures

for Welded Aluminum Structures,” International Journal of Fatigue, Vol. 25, No. 12, 2003, pp. 1359-1378.

[2] J. R. Davis, “Aluminum and Aluminum Alloys,” ASM International, Materials Park, OH, 1998.

[3] C. Zhou, X. Yang and G. Luan, “Investigation of Micro-structures and Fatigue Properties of Friction Stir Welded Al–Mg Alloy,” Journal of Materials Chemistry and Phys-ics, Vol. 98, No. 2-3, 2006, pp. 285-290.

[4] B. Grushko E. Kowalska, B. Przepi and M. Surowiec, “An Investigation of the Al-Cu-Cr Phase Diagram: Phase Equilibria at 800-1000,” Journal of Alloys and Com-pounds, Vol. 417, No. 1-2, 2006, pp. 121-126.

[5] D. R. Fang, Z. F. Zhang, S. D. Wu, C. X. Huang, H. Zhang, N. Q. Zhao and J. Li, “Effect of Equal Channel Angular Pressing on Tensile Properties and Fracture Modes of Casting Al-Cu Alloys,” Journal of Materials Science and Engineering A, Vol. 426, No. 1-2, 2006, pp. 305-313.

[6] B. B. Straumal, B. Baretzky, A. A. Mazilkin, F. Phillipp, O. A. Kogtenkova, M. N. Volkov and R. Z. Valiev, “For-mation of Nanograined Structure and Decomposition of Supersaturated Solid Solution during High Pressure Tor-sion of Al-Zn and Al-Mg Alloys,” Journal of Acta Mate-rialia, Vol. 52, No. 15, 2004, pp. 4469-4478.

[7] A. Białobrzeski, and E. Czekaj, “An Attempt to Use Alloy Synthesis in Evaluating the Corrosion Behaviour of Al- and Mg-Based Alloys,” Journal of Materials Processing Technology, Vol. 175, No. 1-3, 2006, pp. 27-32.

[8] E. M. AbdelHady, N. L. Tawfik and E. S. Gouda, “Me-chanical Properties of Some Al-Based Alloys with Heat Treatment,” Journal of Heat Treatment and Surface En-gineering, Vol. 8, No. 1-2, 2007, pp. 39-45.

[9] M. Kamal and E. S. Gouda, “Effect of Cooling Speed on Structure and Properties of Rapidly Solidified Pb-25wt.% Sn Alloy,” Journal of Radiation Effects and Defects in Solids, Vol. 162, No. 9, 2007, pp. 691-696.




Célia Regina Tomachuk1, Alejandro Ramón Di Sarli2, Cecilia Inés Elsner2

1Energy and Nuclear Research Institute, IPEN/CNEN-SP, CCTM, Av. Prof. Lineu Prestes, São Paulo, Brazil; 2CIDEPINT: Research and Development Center in Paint Technology (CICPBA-CONICET), Av.52 s/n entre 121 y 122. CP. B1900AYB, La Plata-Argentina. Email: [email protected], [email protected] Received April 29th, 2010; July 22nd, 2010; August 4th, 2010.

ABSTRACT

Hexavalent chromium-based passivation treatments have been successfully used as promoters of conversion coatings for many years. Their effectiveness is without question although there are many problems with regard to their environ-mental suitability. Hexavalent chromium compounds are carcinogenic and toxic. These problems have lead researchers to evaluate other potential systems, with lower toxicity, to ascertain if they can replace chromates as effective passiva-tors. Researchers have proposed several alternative passivation treatments; these are processes based on molybdates, permanganates, titanates, rare earth metal and Cr3+ (considered to be non-carcinogenic) compounds. In this work, zinc coatings obtained from free-cyanide alkaline bath and submitted to a Cr3+ based passivation treatment with different colors were studied. The corrosion behavior was studied by polarization measurements and mainly by electrochemical impedance spectroscopy in 0.6 N NaCl solution. Morphological observations on the coatings surface were also per-formed. The results indicate that the green-colored Cr3+ passivated coatings have a good corrosion resistance followed by yellow and blue-colored passivation respectively. They could be a less polluting alternative to the traditional chro-mated coatings. Keywords: Zinc, Conversion Treatment, Impedance Spectroscopy, Salt Spray, Corrosion

1. Introduction

Electroplated zinc coating is employed as active galvanic protection for steel. However, as the zinc is an electro-chemically highly reactive metal, its corrosion rate may be also high in indoor but particularly under outdoor ex-posure conditions [1]. For this reason, it is necessary a post treatment in order to increase the lifetime of zinc coatings. In current industrial practice, this treatment consists of immersion in a chemical bath that forms a conversion layer on plated zinc. This latter layer is a di-electric passive layer with high corrosion resistance and is also a better surface for paint adherence. The main problem of traditionally used post treatments is the pres-ence of Cr6+ salts, considered carcinogenic substances which usage is forbidden by European norms [2]. Re-sponding to increasingly more rigorous environmental protection activities, recent years have shown progressive advances in order to reduce the use of environmen-tally-hazardous materials. In line with this purpose, the development of various kinds of chromate-free coated

steel sheets, to be used in food, automotive, appliances, etc. industries, is being extensively explored all over the world. In this sense, the most common transitional alter-native to Cr6+ is Cr3+, which is used since the mid 1970’s [3-9]. According to Fonte et al. [10], the Cr3+ conversion layer formed in a bath containing transition metal ions such us Co2+, Ni2+ and Fe2+ showed higher corrosion re-sistance than those formed in a bath without transition metal ions. This finding was confirmed by Tomachuk et al. [11,12].

Molybdates, tungstates, permanganates and vanadates, including chromium like elements, were the first chemi-cal elements tried as hexavalent chromium substitutes [13-17]. Recently many alternative coatings were devel-oped based on zirconium and titanium salts [18-20], co-balt salts [21,22], organic polymers [23,24] and rare earth salts [25]. However, preparation and corrosion behavior of these coatings is not clear and their practical usage is doubtful.

In order to find an alternative treatment to Cr6+ con-



203

version coating, several treatments that present a good anti-corrosive behavior, a high benefit/cost relation and, mainly, low environmental impact are still to be devel-oped. Usually, the corrosion behavior of coatings is eva-luated using traditional tests such as Salt Spray [26], Kesternich test [27], saturated humidity [28]. However the authors consider important the application of elec-trochemical methods to obtain fast information about the corrosion reactions kinetics.

Among the electrochemical techniques that can be used, the electrochemical impedance spectroscopy (EIS) was selected based on the already obtained results for metal and metal-coated corrosion evaluation [29-32].

The main purpose of the present work was to find an environmentally friendly conversion treatment able to replace satisfactorily those passivating ones based on Cr6+. Electrogalvanised steel covered with Cr6+-free pas-sivating layers were investigated using AC and DC elec-trochemical techniques. Morphological studies of the coatings surface were also performed.

2. Experimental Details

2.1. Samples Preparation

Electrogalvanised steel samples (7.5 × 10 × 0.1 cm) were industrially produced and covered with the following conversion treatments: 1) blue-colored Cr3+ based pas-sivation; 2) yellow-colored Cr3+ based passivation; 3) green-colored Cr3+ based passivation. For each conver-sion layer, an individual commercial conversion bath was formulated and the coating was produced according to the respective supplier recommendations.

2.2. Thickness Measurements

The coatings thickness was measured using the Helmut Fischer equipment DUALSCOPE MP4.

2.3. Morphology

The coatings morphology was determined from scanning electron microscopy (SEM) analyses using a LEICA S440 microscope.

2.4. Electrochemical Behavior

The electrochemical cell consisted of a classic three-electrode arrangement, where the counter electrode was a platinum sheet, the reference one a saturated calomel electrode (SCE) and the working electrode each coated steel sample with a defined area of 7 cm2. All measure-ments were performed at a constant room temperature (22 3) in 0.6 N NaCl solution.

Potentiodynamic polarization experiments were car-ried out using a Solartron 1280 electrochemical system at a swept rate of 1 mV.s-1, over the range ±0.300 V(SCE)

from the open-circuit potential OCP). Before each swept, the electrode in contact with the electrolyte was stabi-lized for several minutes. The corrosion current density (j) and corrosion potential (Ecorr) were obtained from a Tafel slope by extrapolation of the linear portion of anodic and cathodic branches.

Impedance spectra in the frequency range 2.10-2 Hz < f < 4.104 Hz were performed in the potentiostatic mode at the OCP, and as a function of the exposure time in the electrolyte solution, using a Solartron 1260 Frequency Response Analyzer (FRA) coupled to a Solartron 1286 electrochemical interface (EI). The amplitude of the ap-plied AC voltage was 3 mV peak to peak. Each sample’s surface evolution was analyzed until white corrosion products could be seen by the naked eye. The experi-mental spectra were interpreted on the basis of equivalent electrical circuits’ models using the ZView fitting soft-ware by Scribner Associates. All impedance measure-ments were carried out by triplicate in a Faraday cage in order to minimize external interference on the system studied.


The overall coating thickness and description of the sam-ples investigated in this work are reported in Table 1. In it can be seen that these showed similar and uniform thickness; besides, they also exhibited a bright appear-ance throughout their extension. Unfortunately, informa-tion related with the passive layer thickness was not pos-sible to be obtained.

3.1. Morphology

The consideration of the coating morphology after the coating/drying process is very important since the pres-ence of flaws such as pores and/or other defects could be areas were localized corrosion of the treated zinc surface starts from its exposure to a given environment [33]. Therefore, after applying the conversion treatment, the coatings surface morphology was observed up to 1,000X

Table 1. Characteristics of the samples.

Identification Description

Thickness (Zn + conversion

treatment) (µm)

A blue-colored Cr3+

passivation UniFix Zn-3-50 (LABRITS®)

10.8

B yellow-colored Cr3+

passivation UniYellow 3 (LABRITS®)

11.2

C green-colored Cr3+

passivation SurTec S680® 10.4



204

by SEM (Figure 1). All the samples presented surface roughness. Besides, A samples, subjected to blue-colored Cr3+-based passivation, exhibited surface fissures (indi-cated by the red arrows) which reduce its protective properties (Figure 1(a)), while the B samples, subjected to yellow-colored Cr3+-based passivation, exhibited ho-mogenous structure with nodular growth (Figure 1(b)),

(a)

(b)

(c)

Figure 1. Microstructure of the tested coatings. (a) sample A; (b) sample B; (c) sample C.

and C samples, subjected to green-colored Cr3+ passiva-tion, exhibited a gel-like structure (Figure 1(c)). The characteristic cracks of chromate coatings were not present, perhaps due to its thin thickness [34].

3.2. Polarization Curves

Potentiodynamic polarization curves were performed at a swept rate of a 1 mV.s-1 in the range ±0.300 V (SCE) with respect to the OCP. This procedure has been re-peated for all the investigated samples. Figure 2 shows typical potentiodynamic polarization curves for passivated electrogalvanised steel in chloride solution.

Corrosion potential, Ecorr, and corrosion current den-sity, jcorr, values obtained from Figure 2 were reported in Table 2. As it can be seen, the corrosion potential (Ecorr) of A samples was more negative, i.e. less noble, and this means that from the thermodynamic point of view these samples type are more susceptible to be corroded. With regard to B and C samples, both presented similar and more positive corrosion potential values than A samples, indicating that a corrosion resistance improvement took place, probably due to the homogenous morphology of the covering layer showed in Figures 1(b) and 1(c) pro-vided a better barrier resistance.

On the other hand, the corrosion current density (jcorr) of C samples is one order of magnitude less than the corresponding to the other two sample types tested, i.e.,

Figure 2. Polarization curves of the samples tested in 0.6 N NaCl solution, v = 1 mV/s. Table 2. Ecorr and jcorr values of Zn coatings after applying the conversion treatment.

Conversion Treatment Ecorr V(SCE) jcorr A/cm2

A −1.10 0.2

B −1.04 0.4

C −1.04 0.02



205

its corrosion rate is lower.

3.3. Electrochemical Impedance Spectroscopy

EIS measurements carried out in the 0.6 N NaCl solution were discontinued upon the white corrosion products on the surface could be seen by naked eye.

Figure 3 shows a Nyquist representation of the time dependent electrochemical impedance, while Figure 4 illustrates the electrical equivalent circuits able of simu-lating the physicochemical processes taking place at the coated steel surface. It is important to emphasize that experimental impedance data obtained for A and B sam-ples were analyzed on the basis of the electric equivalent circuit depicted in Figure 4(a)., while for the C samples was used the shown in Figure 4(b) [35]. In these figures, Rsol represents the electrolyte resistance between the ref-erence and working (coated steel) electrodes; the first time constant (R1Q1) - where R1 and (Q1 C1) are re-spectively the resistance to the ionic flux and the dielec-tric capacitance of the conversion layer - appears at the higher frequencies. Once the permeating and corro-sion-inducing chemicals (water, oxygen and ionic species) reach electrochemically active areas of the substrate, particularly at the bottom of the coating defects, the me-tallic corrosion becomes measurable so that its associated parameters, the charge transfer resistance, R2, and the electrochemical double layer capacitance, (Q2 C2), can be estimated [3]. Sometimes, the Q2 parameter can be associated to a diffusional process, which not only could be the rate-determining step (rds) of the corrosion reac-tion but also mask part of - or completely its time con-stant. It is important to remark that R2 and C2 values vary directly (R2) and inversely (C2) with the size of the electrochemically active metallic surface.

Distortions observed in these resistive-capacitive con-tributions indicate a deviation from the theoretical mod-els in terms of a time constants distribution due to either lateral penetration of the electrolyte at the metal/coating interface (usually started at the base of intrinsic or artifi-cial coating defects), underlying metallic surface hetero-geneity (topological, chemical composition, surface en-ergy) and/or diffusional processes that could take place along the test. Since all these factors cause the imped-ance/frequency relationship to be non-linear, they are taken into consideration by replacing one or more ca-pacitive components (Ci) of the equivalent circuit trans-fer function by the corresponding constant phase element Qi (CPE), for which the impedance may be expressed as [36,37]:

n

0

jω

YZ

where:

(a)

(b)

Figure 3. Evolution of the A and C samples impedance (Nyquist representation), (a) sample A; (b) sample C.

(a)

(b)

Figure 4. Equivalent circuit models used for fitting the im-pedance data.

Z() impedance of the CPE (Z = Z´ + jZ´´)() j imaginary number (j2 = −1) angular frequency (rad)



206

n CPE power: (n = / constant phase angle of the CPE (rad) Y0 part of the CPE independent of the frequency

(-1) Difficulties in providing an accurate physical descrip-

tion of the occurred processes are sometimes found. In such cases, a standard deviation value (2 < 10-4) be-tween experimental and fitted impedance data may be used as final criterion to define the most probable circuit.

The comparison between simulated and experimental data at different exposure times are omitted for simplicity, however, in all cases, the experimental data were in good agreement with the model predictions.

The more interesting data to discuss are the exposure time dependent resistance R1 of the passivation treatment (giving information on the barrier properties of the con-version layer) coupled in parallel with its Q1 (related to the coating capacitance) and the charge transfer resis-tance R2 (giving information on the kinetic of the corro-sive process). These values, estimated from the fitting analysis of the impedance spectra, are reported in Fig-ures 5 to 7, respectively.

Figure 5 shows the trend of the parameter R1, which was associated to the evolution of the coating barrier properties and consequently with its degradation during exposure time in the aggressive aqueous solution. At zero time, the same and low R1 values for A and B samples suggest poorer barrier properties when compared with the afforded by C samples. Then, it is observed a slight increase of the R1 values until one hour of immersion for C samples and three hours for A and B samples. This was attributed to the blockage of the intrinsic and struc-tural conversion layer defects with the soluble metallic

Figure 5. Values of R1 as a function of immersion time in 0.6 N NaCl solution obtained from impedance data fitting for A, B and C samples.

(a)

(b)

Figure 6. Values of Q1 and its exponent n1 as a function of immersion time in 0.6 N NaCl solution obtained from impedance data fitting for A, B and C samples. corrosion products formed due to the fast permeation of the corrosion inducing chemicals through the thin con-version layer. After that, these values started to decrease probably because of the interfacial corrosion reactions caused an increasing number and/or area of the coating defects and, consequently, of the exposed zinc area at the conversion layer/Zn interface. It is interesting to note that as the R1 values decrease, the correspondingto Q1 in-crease (Figure 6(a)); such a behavior indicates that the involved relaxation process takes place at the same area [38]. At the end of the immersion test, an oscillating be-havior with values less than the initial ones could be ob-served.

Figures 6(a) and 6(b) show values of Q1 and its ex-ponent n1 (see equation expressing the CPE definition) as a function of the immersion time in the 0.6 N NaCl



207

Figure 7. R2 as a function of immersion time in 0.6 N NaCl solution obtained from impedance data fitting for A and B samples. solution. The initial conversion layer capacitance was the lowest for C followed by B samples, result attributed to the more uniform and compact morphology of C samples (Figures 1(b) and 1(c)). During the first hours of immer-sion, the Q1 values of C and B samples increased, being much more evident (two orders of magnitude) the corre-sponding to C samples. These changes, coupled to the n1 values evolution, can be explained assuming that despite the blockage of the fissures and pores of the conversion layer with the corrosion products, these last provides a poor dielectric behavior and, therefore, are unable to in-hibit the corrosion process. For the first hours, A samples exhibited a trend to decrease the Q1 values followed by an increase of approximately one order of magnitude, which is indicative of conversion layer degradation [39]. The decreasing n1 values shown in Figure 6(b) for B and C samples may be interpreted as a trend to change from capacitive to diffusional behavior, or a mix of both. On the other hand, the A samples showed an opposite response [40]. After several hours of exposure, all these changes followed the observed for the Q1 values.

The analysis of R2 (charge transfer resistance) as a function of immersion time is a useful tool for the corro-sion rate evaluation since it gives information about the kinetic of the corrosive process. In such sense, Figure 7 shows that according to the equivalent circuit utilized for fitting the impedance data (Figure 4(b)), C samples did not present the time constant (R2Q2) corresponding to the faradaic process. It means that this type of conversion layer provided barrier properties (Figure 5) high enough as to inhibit the corrosion process throughout the test. On the other hand, the fact that the R2 values were greater for B samples than for A samples means that those showed lower corrosion rates. This was attributed to the fact that

the zinc corrosion products gathered in the conversion layer defects acted as a better partial barrier, but also that such an effect disappeared as the time of exposure elaps- ed.

This analysis of EIS data for the three Cr3+-based con-version treatments showed that a deficient deposition of the conversion layer produces coatings with lower barrier properties and, therefore, lower corrosion protection (as particularly found in the case of A samples).

Summarizing, green-colored Cr3+ passivation exhibited higher protective capacity than yellow-colored Cr3+ and blue-colored Cr3+ passivation layers, which is clearly noted in the polarization curves data (Figure 2).

4. Conclusions

From the results generated during this investigation for three alternative conversion treatments applied on elec-trogalvanised steel, the following conclusions can be made with regard to their corrosion performance in con-tact with a chloride solution at room temperature: the more uniform coating presented lower corro-

sion rate; the electrochemical techniques demonstrated to be

a very useful tool to characterize the corrosion protection provided by different conversion treat-ments;

the EIS data analyses based on equivalent circuit models showed that green-colored Cr3+ conversion treatment (C samples) presented the highest corro-sion protection followed by the yellow-colored Cr3+ conversion treatment (B samples) and blue- colored Cr3+ conversion treatment (A samples), re-spectively. This behavior was in agreement with the results obtained of the polarization curves;

the conversion treatments investigated shown in-teresting results but other experiments need to be performed in order to evaluate alternatives to the traditional and highly effective, but toxic and pol-lutant, Cr6+ based conversion treatment.

In the near future it is likely that stringent legislation will require the total removing of hexavalent chromium as anticorrosive treatment. Consequently, more studies are needed concerning the corrosion protection, ecologi-cal and toxic effects afforded by new alternative treat-ments.

5. Acknowledgements

The authors acknowledge CNPq/PROSUL (Process 490 116/2006-0) of Brazil, CAPES/MINCyT (Process 158/09 of Brazil and BR/08/04 of Argentina), and Comisión de Investigaciones Científicas de la Provincia de Buenos Aires (CIC) and Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) of Argentina by their



208

financial support to this research.

REFERENCES [1] N. Zaki, “Chromate Conversion Coating for Zinc,” Metal

Finishing, Vol. 86, No. 2, February 1988, pp. 75-78.

[2] P. L. Hagans and C. M. Haas, “ASM Handbook,” Surface Engineering, ASM International, Vol. 5, 1994.

[3] F. Deflorian, S. Rossi, L. Fedrizzi and P. L. Bonora, “EIS Study of Organic Coating on Zinc Surface Pretreated with Environmentally Friendly Products,” Progress in Organic Coatings, Vol. 52, No. 4, 2005, pp. 271-279.

[4] R. Berger, U. Bexell, T. M. Grehk and S. E. Hörnström, “A Comparative Study of the Corrosion Protective Prop-erties of Chromium and Chromium Free Passivation Me-thods,” Surface and Coatings Technology, Vol. 202, No. 2, 2007, pp. 391-397.

[5] K. W. Cho, V. S. Rao and H. Kwon, “Microstructure and Electrochemical Characterization of Trivalent Chromium Based Conversion Coating on Zinc,” Electrochimica Acta, Vol. 52, No. 13, 2007, pp. 4449-4456.

[6] Y.-T. Chang, N.-T. Wen, W.-K. Chen, M.-D. Ger, G.-T. Pan and T. C.-K. Yang, “The Effects of Immersion Time on Morphology and Electrochemical Properties of the Cr(III)-Based Conversion Coatings on Zinc Coated Steel Surface,” Corrosion Science., Vol. 50, No. 12, 2008, pp. 3494-3499.

[7] X. Zhang, C. Van den Bos, W. G. Sloof, A. Hovestad, H. Terryn and J. H. W. de Wit, “Comparison of the Mor-phology and Corrosion Performance of Cr(VI)- and Cr(III)-Based Conversion Coatings on Zinc,” Surface and Coatings Technology, Vol. 199, No. 1, 2005, pp. 92-104.

[8] N. T. Wen, F. J. Chen, M. D. Ger, Y. N. Pan and C. S. Lin, “Microstructure of Trivalent Chromium Conversion Coating on Electrogalvanized Steel Plate,” Electro-chemical Solid State Letter, Vol. 11, No. 8, 2008, pp. C47-C50.

[9] N.-T. Wen, C.-S. Lin, C.-Y. Bai and M.-D. Ger, “Struc-tures and Characteristics of Cr(III)-Based Conversion Coatings on Electrogalvanized Steels,” Surface and Coat-ings Technology, Vol. 203, No. 3-4, 2008, pp. 317-323.

[10] B. Da Fonte, Jr. and M. C. Mich, US Patent 4.359.345, 1982.

[11] C. R. Tomachuk, C. I. Elsner, A. R. Di Sarli and O. B. Ferraz, “Morphology and Corrosion Resistance of Cr(III)- Based Conversion Treatments Applied on Electrogalva-nised Steel,” Journal of Coatings Technology and Research, Vol. 7, No. 4, 2010, pp. 493-502.

[12] C. R. Tomachuk, C. I. Elsner, A. R. Di Sarli and O. B. Ferraz, “Corrosion Resistance of Cr(III) Conversion Treatments Applied on Electrogalvanised Steel and Sub-jected to Chloride Containing Media,” Materials Chemis-try and Physics, Vol. 119, No. 1-2, 2009, pp. 19-29.

[13] E. Almeida, T. C. Diamantino, M. O. Figueiredo and C. Sá, “Oxidizing Alternative Species to Chromium VI in Zinc Galvanized Steel Surface Treatment. Part 1. A Morphological and Chemical Study,” Surface and Coat-ings Technology, Vol. 106, No. 1, July 1998, pp. 8-17.

[14] E. Almeida, L. Fedrizzi and T. C. Diamantino, “Oxidizing Alternative Species to Chromium VI in Zinc Galvanized Steel Surface Treatment. Part 2. An Electrochemical Study,” Surface and Coatings Technology, Vol. 105, No. 1-2, June 1998, pp. 97-101.

[15] G. D. Wilcox and D. R. Gabe, “Passivation Studies Using Group VIA Anions. V. Cathodic Treatment of Zinc,” British Corrosion Journal, Vol. 22, No. 4, 1987, pp. 254- 258.

[16] G. D. Wilcox, D. R. Gabe and M. E. Warwick, “The De-velopment of Passivation Coatings by Cathodic Reduc-tion in Sodium Molybdate Solutions,” Corrosion Science, Vol. 28, No. 6, 1988, pp. 577-585, 587.

[17] V. I. Korobov, Y. M. Loshkarev and O. V. Kozhura, “Cathodic Treatment of Galvanic Zinc Coatings in Solu-tions of Molybdates,” Russian Journal of Electrochemis-try, Vol. 34, No. 11, 1998, pp. 1154-1157.

[18] P. D. Deck and D. M. Reichgott, “Characterization of Chromium-Free No-Rinse Prepaint Coatings on Alumin-ium and Galvanized Steel,” Metal Finishing, Vol. 90, No 9, 1992, pp. 29-35.

[19] B. R. W. Hinton, “Corrosion Prevention and Chromates, the End of an Era?” Metal Finishing, Vol. 89, No. 9, 1991, pp. 55-61.

[20] B. R. W. Hinton, “Corrosion Prevention and Chromates, the End of an Era?” Metal Finishing, Vol. 89, 1991, pp. 15-20.

[21] A. Barbucci, M. Delucchi and G. Cerisola, “Study of Chromate-Free Pretreatments and Primers for the Protec-tion of Galvanized Steel Sheets,” Progress in Organic Coatings, Vol. 33, No. 2, 1998, pp. 131-138.

[22] G. D. Wilcox and J. A. Wharton, “A Review of Chromate Free-Passivation Treatments for Zinc and Zinc Alloys,” Transactions of the Institute of Metal Finishing, Vol. 75, No. 4, 1997, pp. B140-B146.

[23] T. F. Child and W. J. van Ooij, “Application of Silane Technology to Prevent Corrosion of Metal and Improve Paint Adhesion,” Transactions of the Institute of Metal Finishing, Vol. 77, No. 2, 1999, pp. 64-70.

[24] S. González, M. A. Gil, J. O. Hernández, V. Fox and R. M. Souto, “Resistance to Corrosion of Galvanized Steel Covered with an Epoxy –Polyamide Primer Coating,” Progress in Organic Coatings, Vol. 41, No. 1-3, March 2001, pp. 167-170.

[25] M. F. Montemor, A. M. Simões, M. G. S. Ferreira and C. B. Breslin, “Composition and Corrosion Behavior of Galvanized Steel Treated with Rare-Earth Salts: The Ef-fect of the Cation,” Progress in Organic Coatings, Vol. 44, No. 2, 2002, pp. 111-120.

[26] “Standard Practice for Operating Salt Spray (Fog) Appa-ratus,” ASTM B 117 Review A07.

[27] “Testing in a Saturated Atmosphere in the Presence of Sulphur Dioxide,” DIN 50018, 1997.

[28] “Atmospheres and their Technical Application Condensa-tion Water Test Atmospheres,” DIN 50017, 1982.

[29] J. R. Mac Donald, “Impedance Spectroscopy Emphasiz-



209

ing Solid State Materials,” John Wiley & Sons, New York, 1987.

[30] F. Mansfeld, “Models for the Impedance Behavior of Protective Coatings and Cases of Localized Corrosion,” Electrochimica Acta, Vol. 38, No. 14, 1993, pp. 1891- 1897.

[31] P. Carbonini, T. Monetta, L. Nicodemo, P. Mastronardi, B. Scatteia and F. Bellucci, “Electrochemical Characteri-zation of Multilayer Organic Coatings,” Progress in Or-ganic Coatings, Vol. 29, No. 1-4, 1996, pp. 13-20.

[32] L. de Rosa, T. Monetta, D. B. Mitton and F. Bellucci, “Monitoring Degradation of Single and Multilayer Or-ganic Coatings. I. Absorption and Transport of Water: Theoretical Analysis and Methods,” Journal of the Elec-trochemical Society, Vol. 145, No. 11, 1998, pp. 3830- 3838.

[33] A. Boukamp, “Equivalent Circuit,” Report CT88/265/128, CT89/214/128, University of Twente, The Netherlands, 1989.

[34] N. M. Martyah and J. E. McCaskie, “Corrosion Behavior of Zinc Chromate Coatings,” Metal Finishing, Vol. 94, No. 2, 1996, pp. 65-67.

[35] C. Delouis, M. Duprat and C. Tournillon, “The Kinetics of Zinc Dissolution in Aerated Sodium Sulphate Solu-tions. A Measurement of the Corrosion Rate by Imped-

ance Techniques,” Corrosion Science, Vol. 29, No. 1, 1989, pp. 13-20.

[36] B. Del Amo, L. Véleva, A. R. Di Sarli and C. I. Elsner, “Performance of Coated Steel Systems Exposed to Dif-ferent Media. Part 1: Painted Galvanized Steel,” Progress in Organic Coatings, Vol. 50, No. 3, 2004, pp. 179-192.

[37] E. P. M van Westing, G. M. Ferrari, F. M. Geenen and J. H. W. de Wit, “In Situ Determination of the Loss of Ad-hesion of Barrier Epoxy Coatings Using Electrochemical Impedance Spectroscopy,” Progress in Organic Coatings, Vol. 23, No. 1, 1993, pp. 89-103.

[38] G. G. Nascimento, O. R. Mattos, J. L. C. Santos and I. C. P. Margarit, “Impedance Measurements on Lacquered Tinplate: Fitting with Equivalent Circuits,” Journal of Applied Electrochemistry., Vol. 29, No. 3, March 1999, pp. 383-392.

[39] F. Mansfeld, “Use of Electrochemical Impedance Spec-troscopy for the Study of Corrosion Protection by Poly-mer Coatings,” Journal of Applied Electrochemistry, Vol. 25, No. 3, March 1995, pp. 187-202.

[40] R. L. Zeller and R. F. Savinell, “Interpretation of AC Impedance Response of Chromated Electrogalvanized Steel,” Corrosion Science, Vol. 26, No. 5, 1986, pp. 389- 399.




Zinki Jindal1,2, N. K. Verma1

1School of Physics & Materials Science, Thapar University, Patiala 147 004, India; 2Department of Physics, Sir Padampat Singhania University, Bhatewar, Udaipur 313 601, India. Email: [email protected] Received May 20th, 2010; July 26th, 2010; September 27th, 2010.

ABSTRACT

High aspect ratio (up to 100) CdS nanowires having average diameter of 15 nm and length varying from 0.5-1.5 μm have been synthesized using solvothermal technique in ethylenediamine as a solvent at 120 and the effect of Mn dop-ing on morphology and optical properties has been studied. X-ray diffraction analysis shows the typical inter-planar spacing and the diffraction peaks corresponding to the hexagonal wurzite phase of CdS. Morphological study has been done through scanning electron microscopy (SEM) and transmission electron microscopy (TEM) and the optical studies have been conducted through absorption spectra and room temperature photoluminescence (PL). Keywords: Nanomaterials, Semiconducting Cadmium Compounds, Growth from Solutions, Nanostructures, Nucleation

1. Introduction

One dimensional (1D) nanostructures are considered to be critical building blocks for nanoscale electronic and optoelectronic devices and have received tremendous attention since the discovery of carbon nanotubes [1-9]. CdS is one of the most studied materials due to its exten-sive applications in photoelectric conversion in solar cells and light-emitting diodes in flat panel displays [10]. CdS nanowires have been synthesized by several tech-niques. For instance, the growth of thin CdS nanowires (-20 nm thick) has been achieved by a laser ablation technique or chemical vapor deposition (CVD) process based on a gold nanocluster catalyzed vapor-liquid-solid (VLS) growth mechanism [11]. CdS nanowires with di-ameters of 30-70 nm have also been synthesized simply by thermal evaporation of CdS powders [12]. Neverthe-less, the above mentioned methods need some special instruments, harsh conditions, and/or relatively high per-formance temperature (over 800). Uniform nanowires of CdS could also be obtained in the channels of various templates, such as anodic aluminum oxide (AAO) mem-brane [13], polymer gels [14], micelles [15], and so on [16]. Although the template-directed methods are effec-tive in preparing nanowires with uniform and controlla-ble dimensions, they usually lead to a complicated proc-ess and also impurities due to the incomplete removal of

the templates, and the yields are relatively low. Moreover, high aspect ratio (length to diameter) CdS nanowires have also been prepared by solvothermal process [9, 17-19], which may provide a more promising technique for preparing CdS nanowires than conventional methods in terms of cost and potential for large-scale production. The modification in the properties of the semiconducting nanomaterials can be done by tailoring their energy band structure [20,21] with ion implantation, ion doping, chemical vapor ion doping [22]. Nanomaterials doped with optically active luminescence centers create new opportunities for luminescence research and also for the application of nanometer-scale structured materials. Many research groups have studied the optical, magnetic and fluorescent properties of Mn-doped CdS nanocrys-tals [23-26]. But most studies focus on doped CdS nanoparticles. However, there is a need of studying the changes in the properties of the 1D nanoforms (undoped as well as doped) for their potential application in nanoscale optoelectronic devices.

This paper describes the successful synthesis of high aspect ratio of CdS nanowires by the solvothermal tech-nique, using ethylenediamine (En). The effect of Mn doping on the growth of the nanowires and the optical properties have also been discussed. Optical properties of the doped nanoforms indicate that the dopant incorpora-tion in the host plays an active part in controlling the



211

luminescence properties.

2. Experimental

The synthesis of CdS nanowires has been carried out in a closed cylindrical teflon-lined stainless steel chamber. All of the chemical reagents used in this experiment were of analytical grade and used without further purification. Cadmium foils were used as a substrate and also as an additional source of cadmium. 0.005 M cadmium nitrate [Cd(NO3)2.4H2O] along with 0.015 M thiourea [CSN2H4] were taken with 70ml of ethylenediamine (En) which acted as the solvent in the teflon chamber (capacity ~ 100 ml). Mn doped CdS nanostructures were also prepared by adding 5 mmol and 10 mmol manganese acetate (Mn (CH3COO)2). The properly sealed teflon-lined stainless chamber was maintained at temperature of 120 for 24 hours in an electric oven and afterwards, it was allowed to normally cool down to room temperature. The foil and the yellow colored precipitates were collected from the reaction vessel and were washed with de-ionized water and ethanol several times and subsequently dried in air at 50 for 6-12 hours.

The products were characterized by Panalytical`s X`Perto Pro X-ray diffraction machine using the copper characteristic wavelength of 1.5418 Å. Microstructures of the nanoforms were studied through scanning electron microscopy (SEM, FEI, Nova 200 NanoLab ) and trans-mission electron microscopy (TEM, Hitachi, H-7500). Optical absorption spectra of the products, dispersed in spectroscopic grade ethanol, were recorded by a Hitachi 330 UV-Vis spectrophotometer. Photoluminescence (PL) measurements were carried out at room temperature with a luminescence spectrometer (Varian Cary Eclipse fluo-rescence spectrophotometer) using 336 nm as the excita-tion wavelength.


Figure 1 shows the XRD pattern of the synthesized CdS nanowires with all the diffraction peaks corresponding to the hexagonal wurzite phase of CdS. These match well with those in the JCPDS Card (Joint Committee on Powder Diffraction Standards, Card no. 41-1049), as shown in Figure 1. No impurity peaks were detected, indicating high purity product. In addition, the intense and sharp diffraction peaks suggest that the obtained product is well crystallized. The d-spacing of the CdS nanowires have been calculated using the XRD analysis and com-pared with the standard JCPDS data (Table 1). The cor-responding (hkl) values are illustrated in the table.

In case of solvothermal synthesis, temperature and concentration plays an important role in the formation of crystal structure, shape and size of the nanoforms. The

20 30 40 50 60 70

0

200

400

600

800

1000

1200

1400

(100

)

(002

)(1

01)

(102

)

(110

)

(103

)

(112

)

Inte

nsi

ty (

arb

. un

its)

2 (Degree)

20 30 40 50 60 700

20

40

60

80

100

JCPDS card 41-1049

Figure 1. XRD pattern of the CdS nanowires revealing their hexagonal wurzite phase, and the standard JCPDS card No. 41-1049. Table 1. The comparison of d-values, obtained from XRD and JCPDS, and illustration of the corresponding (hkl) val-ues.

Peak 2θo dXRD(Å) dJCPDS(Å) (hkl)

1. 25.04 3.5561 3.5861 100

2. 26.78 3.3289 3.3599 002

3. 28.52 3.1296 3.1630 101

4. 37.07 2.4251 2.4519 102

5. 43.88 2.0632 2.0705 110

6. 48.08 1.8924 1.8998 103

7. 52.09 1.7557 1.7627 112

mechanism behind the formation of nanorods, in pres-ence of ethylenediamine (En) as chelating agent, has al-ready been discussed by many researchers [18]. En reacts with the Cd2+ ions to form Cd-En complex lamellar products, which react with the S2- ions to produce CdS-En lamellar materials. The high temperature leads to the breakage of volatile amine groups giving rise to lamel-lar-to-rod transitions. This is known to proceed via the rolling mechanism [27].

The morphology and the dimensions of the nanowires were studied through scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Figures 2(a, b), 2(c) and 2(d) show the SEM images of undoped, 5 mmol and 10 mmol Mn doped CdS nanos-tructures, respectively. Figures 2(a, b) show the growth of highly dense CdS nanowires having diameters varying between 9-40 nm and lengths varying from 0.5 to 2 μm. The ends of the nanowires are still attached to adjacent nanoforms, possibly due to the incomplete transforma-tion of the lamellae to the nanowires. Whereas, on dop-



212

(a) (b)

(c) (d)

Figure 2. SEM images of (a, b) undoped and (c) 5 mmol (d) 10 mmol Mn doped CdS nanostructures. ing with Mn, and on increasing the concentration of Mn from 5 to 10 mmol (as per the experiment performed), it has been observed that this lamellar-to-rod transition has decreased. Figure 2(c) shows the fragmentation of these Mn doped CdS-En (5 mmol Mn) lamellae to higher level as compared with those of figure 2(d) (10 mmol Mn), where this fragmentation is still in the initial stage. The reason behind this inhibition of CdS nanowire growth on addition of Mn dopant is not clear at this moment. The Mn dopant is considered to bind to the most stable sur-face sites formed during the nanowire nucleation, which inhibits the advancement of the growth in the particular direction [28].

Figures 3(a, b, c, d) show the formation of high aspect ratio (up to 100) CdS nanowires. The diameter of the synthesized nanowires ranges from 9 to 18 nm, whereas the length varies from 0.5 to 1.5 μm. Figure 3(a, b) shows the bundles of the CdS nanowires where the indi-vidual nanowires can be well distinguished, whereas Fig-ure 3(c, d) shows the individual nanowires. Moreover, the flexibility of the CdS nanowires can be observed from Figure 3(c, d), by their wavy nature.

Figure 4 shows the optical absorption spectra of the as synthesized undoped and Mn-doped (5 mmol Mn) CdS nanowires. The maximum absorption peak positions of CdS and CdS:Mn nanoforms are at 460 and 469 nm re-



213

(a) (b)

(c) (d)

Figure 3. (a-d) TEM images of undoped CdS nanowires. spectively, as compared with that of CdS bulk materials (515 nm). The band gap energies were calculated from the differential minima, which varied from 2.48 eV to 2.44 eV on doping with Mn. The observed diameters of the CdS nanowires are well above its Bohr`s exciton ra-dius (-2.8 nm), therefore, this shift in the band gap ener-gies may not be related with quantum confinement effect. The small change in the band gap values may be attrib-uted to the direct energy transfer between the semicon-ductor excited states and the 3d levels of the Mn2+ ions,

that are coupled by energy transfer processes [29]. The room temperature photoluminescence (PL) meas-

urement results of the CdS nanowires (undoped and doped with Mn) are shown in Figure 5. The excited wavelength was 336 nm, and no filter was used. In the past several decades, the luminescence mechanisms of CdS have been studied. Usually, two emissions are ob-served from the semiconductor nanoparticles – excitonic and trapped luminescence [30]. The excitonic emission is sharp and located near the absorption edge of the parti-



214

400 500 600 700400 450 500 550 600 650 700

CdS pure

469 nm

460 nm

Ab

sorb

ance

(a

rb. u

nit

s)

Wavelength (nm)

CdS:Mn

Figure 4. Optical absorbance of the as synthesized CdS, undoped and doped with Mn (5 mmol), nanowires.

460 480 500 520 540 560 580 600 620 640500 600500 600

(i)

(iii)

(ii)

481 nm

599 nm

595 nm

555 nm

517 nm

Inte

nsi

ty (

arb

. un

its)

Wavelength (nm)

Figure 5. Room temperature PL spectra of (i) undoped CdS and (ii) 5 mmol (iii) 10 mmol Mn doped CdS nanowires (λex ~ 336 nm). cles, while the trapped emission is broad and stokes- shifted. CdS is a wide-band-gap (Eg = 2.42 eV) semi-conductor and has typically two emission bands: green band (excitonic emission) around 518 nm and the red band (ascribed to trap of surface states) at about 732 nm [31]. But due to their 1D geometrical characteristic at the nanometer scale, CdS nanowires are expected to have different physical properties from their bulk counterparts [32]. Moreover, it is also believed that nanowires with high aspect ratio have more surface and subsurface de-fects such as grain boundaries and sulfur/cadmium re-lated defects. These would definitely exert an influence on the PL properties of the CdS nanowires. Room tem-perature PL spectra of undoped CdS nanowires exhibit a weak and sharp emission at -481 nm and a broad green emission band centered at -517 nm. The weak emission

band at shorter wavelengths is attributed to the direct transition from the conduction to the valence band [33]. This indicates that the particle crystallinity is rather high. The main luminescence band is broad and is attributed to CdS trap emission. The electrons and holes, after excita-tion across the band edge, trickle down non-radiatively to the surface states lying in the bandgap region. Radiative de-excitation across the surface states in CdS nanowires gives rise to green fluorescence observed at around 517 nm. On addition of Mn (5 mmol) dopant, the intensity of the direct transition has been found to decrease and the broad band, red shifted to 595 nm, which is similar to the Mn emission in bulk CdS:Mn due to an internal Mn2+ transition (4T1 – 6A1). On increase in the concentration of Mn (5 to 10 mmol), this broad band has red shifted to ~ 599 nm, indicating that the Mn2+ concentration is suffi-cient to influence the crystal-field splitting between 4T1

and 6A1 states [34]. Moreover, a broad band, centered around 555 nm, has evolved on the increase of Mn con-centration, which may be attributed to the deep surface trap recombination, unlike from defect related states [35].

4. Conclusions

In summary, we have studied the effect of Mn doping on the solvothermal synthesis of CdS nanowires. The mor-phological study showed that the lamellar-to-rod transi-tion of CdS has decreased on increase in concentration of Mn dopant, while keeping all the other reaction condi-tions same, like: temperature, time duration, solvent, concentration of Cd and S precursors. This might be due to the binding of Mn to the most stable sites during the nanowires nucleation, leading to the inhibition of the growth in a particular direction. CdS nanowires exhibit broad green emission and on addition of Mn, 5 mmol, this band has red shifted to characteristic 595 nm. On further increase of Mn concentration to 10 mmol, there is additional red shift of 4 nm of this broad band. And an-other emission centered around 555 nm has been ob-served, possibly due to deep surface trap recombination. In future, due to the potential applications of CdS:Mn nanowires in the field of optoelectronics, there is need of more detailed study regarding the mechanism.

5. Acknowledgement

We acknowledge Defence Research & Development Organisation (DRDO), Government of India, for their generous funding for the research work vide their letter No. ERIP/ER/0504321/M/01/855 dated 16th December 2005.

REFERENCES [1] S. Iijima, Nature, Vol. 354, 1991, pp. 56.



215

[2] Z. L. Wang and J. H. Song, Science, Vol. 312, 2006, pp. 242.

[3] Y. N. Xia, P. D. Yang, Y. G. Sun, Y. Y. Wu, B. Mayers, B. Gates, Y. D. Yin, F. Kim and Y. Q. Yan, Adv. Mater., Vol. 15, 2003, pp. 353.

[4] W. Q. Han, S. S Fan, Q. Q. Li and Y. D. Hu, Science, Vol. 277, 1997, pp. 1287.

[5] A. M. Morales and C. M. Lieber, Science, Vol. 279, 1998, pp. 208.

[6] J. R. Heath, P. J. Kuekes, G. S. Snider and R. S. Williams, Science, Vol. 280, 1998, pp. 1716.

[7] J. T. Hu, O. Y. Min, P. D. Yang and C. M Lieber, Nature, Vol. 399, 1999, pp. 48.

[8] C. N. R. Rao, F. L. Deepak, G. Gundiah and A. Govinda-raj, Prog. Solid State Chem. Vol. 31, 2003, pp. 5.

[9] S. Kar, S. Santra and H. Heinrich, J. Phys. Chem. C, Vol. 112, 2008, pp. 4036.

[10] M. Morkel, L. Weinhardt, B. Lohmu¨ller, C. Heske, E. Umbach, W. Riedl, S. Zweigart and F. Karg, Appl. Phys. Lett., Vol. 79, 2001, pp. 4482.

[11] X. F. Duan and C. M. Lieber, Adv. Mater., Vol. 12, 2000, pp. 298.

[12] C. Ye, G. Meng, Y. Wang, Z. Jiang and L. Zhang, J. Phys. Chem. B, Vol. 106, 2002, pp. 10338.

[13] D. Routkevitch, T. Bigioni, M. Moskovits and J. M. Xu, J. Phys. Chem., Vol. 100, 1996, pp. 14037.

[14] J. Zhan, X. Yang, D. Wang, S. Li, Y. Xie, Y. N. Xia and Y. T. Qian, Adv. Mater., Vol. 12, 2000, pp. 1348.

[15] Y. J. Xiong, Y. Xie, J. Yang, R. Zhang, C. Z. Wu and G. A. Du, J. Mater. Chem., Vol. 12, 2002, pp. 3712.

[16] F. Gao, Q. Y. Lu and D. Y. Zhao, Adv. Mater., Vol. 15, 2003, pp. 739.

[17] Y. D. Li, H. W. Liao, Y. Ding, Y. Fan, Y. Zhang and Y. T. Qian, Inorg. Chem., Vol. 38, 1999, pp. 1382.

[18] A. Datta, S. Kar, J. Ghatak, and S. Chaudhari, J. Nanosci.

Nanotechnol., Vol. 7, 2007, pp. 677.

[19] F. W. G. Li and Z. Zhang, J. Nanoparticle Research, Vol. 7, 2005, pp. 685.

[20] R.N. Bhargava, D. Gallagher and T. Welker, J. Lumin., Vol. 275, 1994, pp. 60.

[21] L. Brus, J. Phys. Chem., Vol. 90, 1998, pp. 2555.

[22] A. Javey, R. Tu, D. B. Farmer, J. Guo, R. G. Gordon and H. J. Dai, Nano. Lett., Vol. 5, 2005, pp. 345.

[23] M. Tanaka, J. F. Qi and Y. Masumoto, J. Crystal Growth, Vol. 214/215, 2000, pp. 410.

[24] S. M. Liu, F. Q. Liu, H. Q. Guo, Z. H. Zhamg and Z. G. Wang, Solid State Comuun., Vol. 115, 2000, pp. 615.

[25] M. Tanaka and Y. Masumoto, Solid State Commun., Vol. 120, 2001, pp. 7.

[26] D. M. Hofmann, A. Hofstaetter, U. Leib, B. K. Meyer and G. Counio, J. Crystal Growth, Vol. 184/185, 1998, pp. 383.

[27] J. S. Jang, S. H. Choi, H. Park, W. Choi and J. S. Lee, J. Nanosci. Nanotechnol., Vol. 6, 2006, pp. 3642.

[28] P. V. Radovanovic, K. G. Stamplecoskie and B. G. Pautler, J. Am. Chem. Soc., Vol. 129, 2007, pp. 10980.

[29] Balram Tripathia, F. Singhb, D.K. Avasthib, D. Dasc, Y.K. Vijaya, Physica B, Vol. 400, 2007, pp. 70-76.

[30] Y. Wang, G. Meng, L. Zhang, C. Liang and J. Zhang, Chem. Mater., Vol. 14, 2002, pp. 1773.

[31] R. L. Morales, O. Z. Angel and G. T. Delgado, Appl. Sur. Sci., Vol. 562, 2001, pp. 175.

[32] T. Gao, G. W. Meng and T. H. Wang, Chin. Phys. Lett., Vol. 21, 2004, pp. 959.

[33] L. Levy, N. Feltin, D. Ingert, and M. P. Pileni, J. Phys. Chem. B, Vol. 101, 1997, pp. 9153.

[34] D. S. Kim, Y. J. Cho, J. P. J. Yoon, Y. Jo and M.-H. Jung, J. Phys. Chem. C, Vol. 111, 2007, pp. 10861.

[35] S. Sapra , A. Prakash , A. Ghanghrekar, N. Perasamy and D. D. Sarma, J. Phys. Chem. B, Vol. 109, 2005, pp. 1663.




Y. Mazaheri, F. Karimzadeh*, M. H. Enayati

Department of Materials Engineering, Nanotechnology and Advanced Materials Institute (NAMI), Isfahan University of Technology, Isfahan, Iran. Email: [email protected] Received May 4th, 2010; May 31st, 2010; June 6th, 2010.

ABSTRACT

In this study ball milling of Al356 and Al2O3 powder mixture was carried out in order to produce Al356-Al2O3 nano-composite containing 20 vol.% Al2O3. The structural evolution and morphological changes of powder particles during ball milling were studied by X-ray diffractometery and scanning electron microscopy analysis. As a result of ball mill-ing Al2O3 particles were uniformly dispersed in Al356 matrix. Furthermore the crystallite size of the Al356 decreased to 25 nm after ball milling for 10 h. Morphological studies of powder particles indicated that the powder particle size con-tinuously decreases with increasing milling time. Hardness and elastic modulus values of powder particles were meas-ured by nanoindentation method. It was found that the hardness and elastic modulus of Al356-20 vol.% Al2O3 nano-composite were about 216 Hv and 86 GPa, respectively which is higher than 75 Hv and 74 GPa for as-received Al356. Keywords: Al-Al2O3 Nanocomposite, Nanocrystalline Structure, Ball Milling, Nanoindentation

1. Introduction

Metal matrix composites (MMCs) are under attention for many applications in aerospace, defense, and automobile industries. These materials have been considered for us-ing in automobile brake rotors and various components in internal combustion engines because of its high strength/weight ratio and wear resistance [1]. Al is the most popular matrix for MMCs because of its low density, good corrosion resistance and high thermal and electrical conductivity [2,3]. Conventional Al matrix composites (AMCs) reinforced with ceramic particulates, especially Al2O3 exhibit high strength, hardness and elastic modulus [4].

AMCs have been widely studied since the 1920s [2]. A survey of the previous studies indicates that a ho-mogenous dispersion of fine particles in a fine grained matrix is beneficial to the mechanical properties of MMCs [5-10]. The use of Al-Al2O3 has been limited due to high processing cost [11]. Solid state processes such as ball milling (BM) can be readily used to fabricate Al-Al2O3 composite with improved properties [12]. For instance; Tavoosi et al. [13] used high energy BM to prepare Al-Al2O3 nanocomposite and showed that the hardness and wear resistance increased with increasing Al2O3 content of the nanocomposites. BM is well recog-

nized as a potential method for achieving better disper-sion of reinforcing particles in the matrixes of micro- and nanocomposites. The BM process involves repeated plastic deformation, welding and fracture of powder particles [4]. Addition of ceramic reinforcements into a ductile matrix has a great effect on the structural evolution dur-ing BM. Although there have been several research stud-ies about the effect of milling parameters, such as ball sizes, number of balls and milling time on the micro-structure of Al-Al2O3 composites , for example [1,14-18], the effect of nanocrystalline structure reinforced with ceramic particulates on properties of Al-Al2O3 nano-composites is not well investigated yet. The objective of the present work is to investigate the properties of mi-crometric Al2O3 reinforced Al356 matrix composite pre-pared by BM technique. The addition of Al2O3 particles to residual machining chips of Al356 display an effective cost saving in this work.

2. Materials and Methods

2.1. Samples Preparation

Residual machining chips of A356 aluminum alloy (Al356) and α-Al2O3 powder with purity of 99% were used as starting materials. Table 1 lists chemical analysis of the Al356 chips. Figure 1 shows scanning electron micros-



217

(a) (b)

Figure 1. SEM images of as-received materials. (a) Al356 chips; (b) Al2O3 powder particles.

Table 1. Chemical composition of Al356 chips.

Element Al Si Mg Fe Mn Cu Ti

Composition (wt. %)

Rem 7.44 0.44 0.26 0.07 0.05 0.02

copy micrographs of as-received materials. Al356 chips were irregular in shape with a size distribution of 200- 300 μm and Al2O3 powder particles had an angular shape with a size distribution of 100-200 μm.

The Al356 chips and Al2O3 powder particles were mixed to achieve Al356-20 vol. % Al2O3 composition. BM was carried out in a high energy planetary ball mill (PM 100), nominally at room temperature and under Ar atmosphere. The milling media consisted of twenty 20 mm diameter balls confined in a 500 ml volume vial. The ball and vial materials were hardened chromium steel. Ball to powder weight ratio and rotation speed of vial was 6:1 and 300 rpm, respectively. The total powder mass was 100 gr and 0.3 wt. % stearic acid was added as a process control agent (PCA).

2.2. Analysis Techniques

Samples were taken at selected time intervals and char-acterized by X-ray diffraction (XRD) in a Philips XPERT MPD diffractometer using filtered Cu Kα radiation (λ = 0.1542 nm). Morphology and microstructure of powder particles were characterized by scanning electron mi-croscopy (SEM) in a Philips XL30.

The crystallite size and lattice strain of powders were estimated using the Williamson-Hall method by follow-ing Equation [19]:

2cos 2 sin

KA

D

(1)

where θ is the Bragg diffraction angle, D the crystallite

size, ε the average internal strain, λ the wave length of the radiation used, β the diffraction peak width at half maximum intensity, K the Scherrer constant (0.9) and A is the coefficient which depends on the distribution of strain; it is near to unity for dislocations.

2.3. Nanoindentation Method

Depth sensing indentation (DSI) is commonly referred to as nanoindentation since the technique usually operates in the submicron depth range with nanometer resolution [20-24]. DSI differs from classical hardness measure-ments (Vickers, Brinell and Knoop), where the impres-sions are first generated, and then imaged using a mi-croscopy technique. The nanoindentation test involves indenting a specimen with a very low load using a high precision instrument, which records the load and penetra-tion depth continuously. The mechanical properties can be derived from the measured load-penetration depth curves under loading/unloading through appropriate data analysis. Figure 2 shows a typical load-penetration depth curve obtained in a nanoindentation test. The peak in-dentation depth is denoted by hm and includes elastic and plastic deformation. The depth at which the applied loads become zero on unloading is the final indentation depth hf and represents the plastic deformation. S represents the contact stiffness measured during the first moments of the unload operation. S = dF/dh is the slope of the tan-gent of the load-penetration depth curve during the unloading cycle. The depth hc is the contact depth at which the cross-section area Ap is taken to calculate hardness [25].

The contact depth hc and the hardness are calculated by a standard procedure according to the method of Oliver and Pharr [26]; hc can be written as:

cm

mF

hhs

(2)



218

Figure 2. Load versus penetration depth curve obtained from a nanoindentation test.

Knowing hc, Ap is calculated. The instrumented hard-ness HIT is determined from peak load Fm and projected area Ap of contact as:

ITm

p

FH

A (3)

Whereas the Vickers hardness HV is calculated from the developed area Ad:

m

d

FHV

A (4)

The difference between an instrumented hardness and Vickers hardness resides in definition of the contact area between the indenter and the tested material.

A reduced modulus, EIT*, is used to account for the fact

that the elastic displacements occur in both the indenter and the sample. This reduced elastic modulus can be linked to the measured stiffness S by the relation:

*

2ITp

SE

A

(5)

Knowing S and Ap, EIT* is calculated.

The instrumented elastic modulus in the test material, EIT, is determined by the relation:

2

2

*

1

11IT

i

iIT

E

EE

(6)

where ν is the Poisson’s ratio for the sample, Ei and νi are the elastic modulus and Poisson’s ratio, respectively, of the indenter.

The hardness and elastic modulus of Al356 and Al356- Al2O3 composite was evaluated from the load-penetration depth curves obtained in nanoindentation tests using a nanoindentation tester (NHTX S/N: 01-03119, CSM In-struments) with a Berkovich diamond indenter (B-J87). The elastic constants Ei = 1141 GPa and νi = 0.07 are often used for a diamond indenter [27]. The indentation was made to a maximum load of about 70 mN and under loading and unloading rate of 140 mN/min. In order to take the repeatability into account, the test results were acquired from the average of four indentations.


3.1. Structural Evolution

Figure 3 shows XRD patterns of Al356 and Al2O3 powder mixture at different milling times. As can be seen with increasing milling times the intensity of Al356 and Al2O3 diffraction peaks decreases and their width increases progressively as a result of refinement of crystallite size and enhancement of lattice strain. With increasing milling time the brittle particles (Al2O3) are uniformly dispersed in the ductile matrix (Al356) [28].

The variation of Al356 crystallite size and lattice strain as a function of milling time is shown in Figure 4. As can be seen in Figure 4(a), with increasing milling time Al crystallite size gradually redused reaching a value of 25 nm after 10h of milling time. Moreover, the lattice strain induced by milling increased up to 0.43% (Figure 4(b)). The crystallite size of the Al2O3 particles was calculated to be about 60 nm after 10h of milling time.

SEM images of powder particles at different milling times are shown in Figure 5. As seen after 2 h of milling time the powder particles had a flake morphology. with increasing milling time the powder particles size decreased to 10-20 μm due to the predominance of the fracturing of powder particles over the cold welding process. Also flake morphology changed to equeiaxed morphology with

Figure 3. XRD patterns of Al356 and Al2O3 powder mixture at different milling times.



219

(a)

(b)

Figure 4. The variation of (a) Al356 crystallite size; (b) lat-tice strain as a function of milling time. increasing milling time. At longer milling times the powder particles were more uniform in size compared to the early stages of milling. The larger particles at longer milling times appeared to be an agglomaration of many smaller particles.

3.2. Nanoindentation Profile

Figure 6 shows the load-penetration depth curves ob-tained from nanoindentation test of as-received Al356 and Al356-20 vol.% Al2O3 nanocomposite after 10 h of milling times. The difference in hardness of the materials is apparent from the large difference in the peak depth. The data obtained from the analysis of load/unload curve, are given in Table 2. Hardness and elastic modulus val-ues of Al356-Al2O3 nanocomposite showed considerable increase compared with Al356.

The possible strengthening mechanisms which may operate in particle-reinforced MMCs [29]:

1) Orowan strengthening. 2) Grain and substructure strengthening. 3) Quench hardening resulting from the dislocations

generated to accommodate the differential thermal con-traction between the reinforcing particles and the matrix.

4) Work hardening, due to the strain misfit between

Table 2. The results obtained from nanoindentation tests.

Value Parameter

Al356 Nanocomposite Dimension

Fm 70.28 70.23 mN

hm 2066 1279 nm

hf 1625 702 nm

S 0.2068 0.2353 mN/nm

hc 1811 1062 nm

Ap 8.7×107 3×107 nm²

HIT 807 2334 MPa

HV 75 216 Vickers

EIT 74 86 GPa

Epsilon 0.75 0.73

the elastic reinforcing particles and the plastic matrix.

According to the characteristics of the microstructure, the better mechanical properties of Al356-Al2O3 nano-composite can be attributed to 1) the nano grain size of the Al matrix following the classical Hall-Petch rela-tionship, and 2) the Orowan strengthening due to the fine dispersion of Al2O3 particles. Rule of mixtures can be applied to calculate the hardness and elastic modulus of Al356-Al2O3 nanocomposite [30]:

c m m r rH H F H F (7)

c m m r rE E F E F (8)

Hc, Hm, and Hr, show the hardness of the composite, matrix and reinforcement, respectively. Ec, Em, and Er, show the elastic modulus of the composite, matrix and reinforcement, respectively. Fm and Fr are fractional volumes of matrix and reinforcement. Nanoindentation results show that the addition of 20vol. % Al2O3 in Al356 matrix increased the hardness and elastic modulus from 75 Hv and 74 GPa to 216 Hv and 86 GPa, respectively. Nanoindentation tests showed that hardness and elastic modulus of Al2O3 were about 880 Hv and 150 GPa, re-spectively [31]. Taking the data in Table 2 for HAl356 (75 Hv), EAl356 (74 GPa) and FAl356 (0.8), FAl2O3 (0.2), equa-tion 7 and 8 give Hc = 236 Hv and Ec = 89.2 GPa, which are in good agreement with the experimental values of 216 Hv and 86 GPa, respectively.

4. Conclusions

Ex-situ Al356-Al2O3 nanocomposite was produced by ball milling process. Structural evolution indicated that as a result of ball milling the Al2O3 particles are uni-formly dispersed in ductile Al356 matrix. Crystallite size



220

(a) (b)

(c) (d)

Figure 5. SEM images of powder particles after (a) 2 h, (b) 5 h, (c) 7 h and (d) 10 h of milling times.

Figure 6. Load versus penetration depth curves of Al356 and Al356-20 vol.% Al2O3 nanocomposite as-milled for 10 h. of Al matrix was 25 nm after 10 h of milling time. This microstructure led to a remarkable improvement of me-chanical characteristics so that, for instance, the hardness

and elastic modulus of Al356-20vol.% Al2O3 powder increased to 216 Hv and 86GPa, respectively.

REFERENCES [1] S. M. Zebarjad and S. A. Sajjadi, “Dependency of Physi-

cal and Mechanical Properties of Mechanical Alloyed Al-Al2O3 Composite on Milling Time,” Materials & De-sign, Vol. 28, No. 7, 2007, pp. 2113-2120.

[2] V. Sethi, “Effect of Aging on Abrasive Wear Resistance of Silicon Carbide Particulate Reinforced Aluminum Ma-trix Composite,” MSc. Thesis, Univ. of Cincinnati, 2007.

[3] A. Pyzalla, B. Camin, B. Lehrer, M. Wichert, A. Koch, K. Zimnik, E. Boller and W. Reimers, “In-Situ Observation of Creep Damage in Al-Al2O3 MMCs by Synchrotron X-Ray Tomography,” International Centre for Diffraction Data, 2006, pp. 1097-2102.

[4] S. C. Tjong, “Novel Nanoparticle-Reinforced Metal Ma-trix Composites with Enhanced Mechanical Properties,” Adv. Eng. Mater., Vol. 8, 2007, pp. 639-652.

[5] I. A. Ibrahim, F. A. Mohamed and E. J. Lavernia, “Par-ticulate Reinforced Metal Matrix Composite,” J. Mater.



221

Sci. A, Vol. 26, 1991, pp. 1137-1156.

[6] A. Bhaduri, V. Gopinathan and P. Ramakrishnan, “Mi-crostructural Changes in a Mechanically Alloyed Al-6.2 Zn-2.5 Mg-1.7 Cu Alloy (7010) with and without Par-ticulate SiC Reinforcement,” Metal Trans. A, Vol. 27, 1995, pp. 3718-3726.

[7] A. Olszowka-Myalska, S. Janusz and J. Cwajna, “Char-acterization of Reinforcement Distribution in Al/Al2O3 System Composite Obtained from Composite Powder,” Mater. Charact., Vol. 46, 2001, pp. 189-195.

[8] K. M. Shorowordi, T. Laoui, A. S. M. A. Haseeb and J.P. Celis, “Microstructure and Interface Characteristics of B4C, SiC and Al2O3 Reinforced Al Matrix Composite: A Comparative Study,” J. Mater. Process. Tech., Vol. 142, 2003, pp. 738-743.

[9] C. Shuangje and W. Renjie, “The Structure and Bending Properties of Squeeze-Cast Composites of A356 Alumi-num Alloy Reinforced with Alumina Particles,” Comp. Sci. Tech., Vol. 59, 1999, pp. 157-162.

[10] B. Prabhu, C. Suryanarayana, L. An and R. Vaidyanathan, “Synthesis and Characterization of High Volume Fraction Al-Al2O3 Nanocomposite Powders by High-Energy Mill-ing,” Mater. Sci. Eng. A, Vol. 425, 2006, pp. 192-200.

[11] G. J. Howell and A. Ball, “Dry Sliding Wear of Particu-late-Reinforced Aluminum Alloys against Automobile Friction Materials,” Wear, Vol. 181, 1995, pp. 379-390.

[12] P. R. Soni, “Mechanical alloying: fundamentals and ap-plications,” Cambridge International Science Publishing, 1999.

[13] M. Tavoosi, F. Karimzadeh and M. H. Enayati, “Wear Behaviour of Al-Al2O3 Nanocomposites Prepared by Mechanical Alloying and Hot Pressing,” Mater. Sci. Technol., In press.

[14] M. M. Moshksar and S. M. Zebarjad, “Effect of Milling Parameters on Milling of Metallic Powders,” Proceeding of Nonferrous Metals, Iran, Kerman, 1996, pp. 913-917.

[15] M. M. Moshksar and S. M. Zebarjad, “Morphology and size Distribution of Aluminum Powder during Milling Processing,” Iran. J. Sci. Technol., Vol. 23, 1996, pp. 150-154.

[16] S. M. Zebarjad and S. A. Sajjadi, “Microstructure Evalua-tion of Al–Al2O3 Composite Produced by Mechanical Alloying Method,” Mater. Des., Vol. 27, 2006, pp. 684-688.

[17] Z. Razavi Hesabi, A. Simchi and S. M. Seyed Reihani, “Structural Evolution during Mechanical Milling of Na- nometric and Micrometric Al2O3 Reinforced Al Matrix Composite,” Mater. Sci. Eng. A, Vol. 428, 2006, pp. 159-168.

[18] E. M. Ruiz-Navas, J. B. Fogagnolo, F. Velasco and L. Froyn, “One Step Production of Aluminum Matrix Com-posite Powder by Mechanical Alloying,” Comp. Part A., Vol. 37, 2006, pp. 2114-2120.

[19] G. K. Williamson and W. H. Hall, “X-ray Line Broaden-ing from Filed Aluminium and Wolfram,” Acta Metall., Vol. 1, 1953, pp. 22-31.

[20] W. C. Oliver and G. M. Pharr, “Measurement of Hardness and Elastic Modulus by Instrumented Indentation: Ad-vances in Understanding and Refinements to Methodol-ogy,” J. Mater. Res., Vol. 19, 2004, pp. 3-20.

[21] A. C. Fischer-Cripps, “Review of Analysis and Interpre-tation of Nanoindentation Test Data,” Surf. Coat. Tech-nol., Vol. 200, 2006, pp. 4153-4165.

[22] C. A. Schuh, “Nanoindentation Studies of Materials,” Mater. Today., Vol. 9, 2006, pp. 32-49.

[23] A. Gouldstone, N. Chollacoop, M. Dao, J. Li, A. M. Mi-nor and Y.L. Shen, “Indentation across Size Scales and Disciplines: Recent Developments in Experimentation and Modeling,” Acta Mater., Vol. 55, 2007, pp. 4015-4039.

[24] Y. Golovin, “Nanoindentation and Mechanical Properties of Solids in Submicrovolumes, Thin Near-Surface Layers, and Films: A Review,” Phys. Solid. State., Vol. 50, 2008, pp. 2205–2236.

[25] C. L. Chen, A. Richter and R. C. Thomson, “Mechanical Properties of Intermetallic Phases in Multi-Component Al-Si Alloys Using Nanoindentation,” Intermetallics, Vol. 17, 2009, pp. 634–641.

[26] W. C. Oliver and G. M. Pharr, “An Improved Technique for Determining Hardness and Elastic-Modulus Using Load and Displacement Sensing Indentation Experi-ments,” J. Mater. Res., Vol. 7, 1992, pp. 1564-1583.

[27] G. Simmons and H. Wang, “Single Crystal Elastic Con-stants and Calculated Aggregate Properties,” A Handbook, Cambridge, MIT Press, 1971.

[28] C. Suryanarayana, “Mechanical Alloying and Milling,” Prog. Mater. Sci., Vol. 46, 2001, pp. 1-184.

[29] D. J. Lloyd, “Particle Reinforced Aluminium and Magne-sium Matrix Composites,” Int. Mater. Rev., Vol. 39, 1994, pp. 1-24.

[30] G. E. Dieter, “Mechanical metallurgy,” Third Ed, McGraw-Hill, Berlin, 1976.

[31] T. C. Chou, T. G. Nieh, S. D. McAdams, G. M. Pharr and W. C. Oliver, “Mechanical Properties and Microstruc-tures of Metal/Ceramic Microlaminates: Part II. A Mo/Al2O3 System,” J. Mater. Res., Vol. 10, 1992, pp. 2774-2784.




Levan Chkhartishvili1, Ivane Murusidze2

1Department of Physics, Georgian Technical University, Tbilisi, Georgia; 2Institute of Applied Physics, Ilia State University, Tbilisi, Georgia. Email: [email protected] Received August 4th, 2010; August 27th, 2010; September 6th, 2010.

ABSTRACT

Molar binding energy of the boron nitride single-walled zigzag and armchair nanotubes is calculated within the quasi-classical approach. We find that, in the range of ultra small radii, the binding energy of nanotubes exhibit an oscillatory dependence on tube radius. Nanotubes (1,1), (3,0), and (4,0) are predicted to be more stable species among single-walled boron nitride nanotubes. The obtained binding energies of BN single-walled nanotubes corrected with zero-point vibration energies lies within the interval (12.01-29.39) eV. In particular, molar binding energy of the ul-tra-large-radius tube is determined as 22.95 eV. The spread of the molar zero-point vibration energy of BN nanotubes itself is (0.25-0.33) eV and its limit for ultra-large-radius tubes is estimated as 0.31 eV. The binding energy peak lo-cated at 2.691 Å corresponds to the equilibrium structural parameter of all realized stable BN nanotubular structures. Keywords: Binding Energy, Zigzag and Armchair Nanotubes, BN

1. Introduction

Boron nitride with the chemical formula BN can be found in the form of one-dimensional diatomic molecules, two-dimensional nanotubes and fullerenes, three-dimensional crystals like the layered hexagonal h-BN and rhombohedral r-BN as well as turbostratic t-BN, cubic zinc-blende c-BN and wurtzite w-BN modifications as well as their nanostructures etc. Boron and nitrogen at-oms are surrounded tetrahedrally in both denser c-BN and w-BN crystals.

Any constituent atom of an h-BN crystal, which is be-lieved to correspond to the boron nitride ground state, may be considered as a 3-coordinated atom because the strong chemical binding (covalent with some deal of ionic) occurs only within the layers, while weak van der Waals forces seem to be responsible mostly for interlayer binding. The h-BN crystal has a “graphitic” structure with a two-layer stacking sequence (r-BN is character-ized by a three-layer stacking). These layers consist of regular hexagons (i.e., 6-membered atomic rings) with vertexes alternatively occupied by B and N atoms. In the h-BN crystal, B atoms are placed directly above N atoms and vice versa. Therefore one might suppose that the ionicity contributes to interlayer bonding as well. How-

ever, actually, the electrostatic component is insignificant due to the large interlayer distances. It is argued also by the existence of a layered r-BN crystal, in which each subsequent layer is turned around an angle of / 3 , and also by the isolated plane defects and their bundles in-cluded in real h-BN crystals, in which any given atom can be placed above the same atom. In addition, it is pos-sible to obtain turbostratic t-BN and amorphous struc-tures in the form of mixes of various boron nitride crys-talline phases, and multi-walled nanotubular and multi- shelled fullerene-like BN structures. Strong chemical bonding between atoms in a given layer and weak inter-layer interaction in layered boron nitrides specify an op-portunity of physical and chemical intercalations by various atoms and molecules.

Based upon the similarity of structures of the boron nitride layered phases with graphite, it was assumed [1] that along with carbon C nanotubes, stable BN nanotubes – fragments of hexagonal or mixed BN layers wrapped into cylinders – could also exist.

Indeed, by means of arc discharge BN nanotubes had been obtained both from carbon nanotubes [2] and in carbon-free plasma [3]. The arc discharge methods were used to produce BxCyNz nanotubular structures identified by the high-resolution transmission-electron-microscopy



223

(TEM) together with K-edge electron-energy-loss-spectrometry (EELS) determining the local atomic composi-tion, while, in a carbon-free plasma discharge area be-tween a BN-packed W-rod and a cooled Cu-electrode, multi-walled pure BN tubes were produced with inner diameters on the order of 1 to 3 nm and with lengths up to 200 nm; EELS of individual tubes yielded B/N ratio of approximately 1.

At present various methods of synthesis of the BN nanotubular structures are developed.

The arc discharge in a molecular nitrogen atmosphere between electrodes made of graphite and refractory bo-ron compounds, e.g., hafnium diboride HfB2, forms BN-C nanotubes with strong phase separation between BN and C layers along the radial direction [4,5]. If both elec-trodes are made of HfB2 rods, single- or double-walled chemically pure BN nanotubes are formed with a struc-ture close to stoichiometric B/N ~ 1 [6]. Most obtained tube ends are closed by flat layers perpendicular to the tube axis. A closure by a triangular facet resulting from three 120-disclinations was proposed to account for this specific shape. For the most part, the multi-walled BN nanotubes are formed with electrodes made of zirconium diboride ZrB2 [7,8]. In this case most of the nanotubes have diameters from 3 to 40 nm and lengths on the order of 100 nm. Single-layer tubes with diameters of 2 to 5 nm are also formed rarely. The morphology of the tube tips suggests the presence of pentagons and heptagons which are energetically less favorable compared with squares.

A laser melting of the solid-state BN (of any crystal structure, not only layered but also amorphous) at high nitrogen pressure, (5-15) GPa, forms nanotubes free from inclusions, containing from 3 up to 8 walls, and having a characteristic outer diameter of (3-15) nm [9].

BN nanotubes have been also obtained by pyrolysis of the molecular precursor with the use of Co catalysts [10].

Bundles of single-walled (or containing few layers) BN nanotubes with almost stoichiometric structure can be formed in substitution reactions – by thermal treat-ment of a mixture of boron trioxide or trichloride, and bundles of single-walled C-nanotubes at high tempera-tures, (1250-1350), in a nitrogen flow [11,12].

BN nanotubes also grow in solid-state process that in-volves neither deposition from the vapor phase nor chemical reactions [13]. The nanotubes were produced by first ball-milling of the layered h-BN powder to generate highly disordered or amorphous nanostructures and followed by the product annealing at temperatures up to 1300. The annealing leads to the nucleation and growth of hexagonal BN nanotubes both of cylindrical and bamboo-like morphologies.

Multi-walled BN nanotubes have been obtained by

carbothermal reduction of the ultra-dispersive amorphous boron oxide B2O3 at simultaneous nitriding at high tem-peratures, (1100-1450) [14,15]. For large tubes, it is found that the ratio of length to radius is preserved.

Besides of arc-melting, pyrolysis, and chemical reac-tions, boron nitride nanotubular structures were created by means of ballistic nuclear displacements caused in a h-BN layered crystal structure by electron irradiation in TEM [16,17].

High growth temperatures (above 1100), a low pro-duction yield, and impurities have prevented progress in applications of BN nanotubes in the past decade. Rather recently, it has been shown that these tubes can be grown on substrates at lower temperatures (of about 600) [18]. High-order tubular structures were constructed, which can be used without further purification.

For synthesizing BN nanotubular material, some meth-ods were inspired from carbon. In particular, these in-clude techniques such as laser ablation and non-ablative laser heating [19]. Transformation of the compressed powders of the fine-grained h-BN into nanotubular form can be induced [20] by the concentrated light energy in nitrogen flow. Fiber-like clusters synthesized by evapo-ration of the layered BN in a nitrogen atmosphere and obtained in powders formed on substrate or chamber surfaces contained nanotubes with diameters and lengths equal to (0.05-200) and (100-3000) m, respectively. Applying TEM, there were obtained their associations in tree- and coral-like aggregates.

BN nanotubes can be grown from a nanococoon seed as well [21].

Recently the development of a new method for pro-ducing long, small-diameter, single- and few-walled, BN nanotubes in macroscopic quantities has been reported [22]. The pressurized vapor/condenser (PVC) method produces highly crystalline, very long, small-diameter nanotubes without catalysts. Their palm-sized, cotton- like masses of raw material were grown by this technique and spun directly into centimeters-long yarn. Nanotube lengths were observed to be ~ 100 times that of those grown by the most closely related method.

Soon after synthesizing the first BN nanotubes, it was proposed a number of their possible applications in tech-nique [3]. For example, a system of the collinear BN nanotubes forms a boron nitride fiber. At the same time, the theory [23] developed for structural and electronic properties of nanotubular heterojunctions, in which one of the layers is nanotubular boron nitride (namely, for C/BN and BC2N/BN systems), leads to a conclusion that on basis of it a different electronic devices can be de-signed. In particular, nanostructures with C-layers both in the center and at the periphery separated by a few BN- layers may allow the creation of sandwich nanotubular



224

devices [4]. Within the special semiempirical approach [24] C/BN superlattices and isolated junctions have been investigated as specific examples by the wide variety of electronic devices that can be realized using such nano-tubes. The bottom of the conduction bands in pure BN nanotubes is controlled by a nearly-free-electron-state localized inside the tube suggesting interesting electronic properties under doping.

Other opportunities of application of the BN nanotubes are connected with the features of their phonon spectrum [25]. Such dielectric tubes without inversion center can be used as a phonon laser in GHz − THz range or hyper-sound quantum generator. Because of the presence of special nanotubular oscillatory modes, there is a strong enhancement of electron-phonon interaction in compari-son with a bulk material. It is not excluded that close- packed one-dimensional BN nanotubes will serve as high-temperature superconductors. The GHz oscillatory behavior of double-walled BN nanotubes was also pre-dicted [26,27]. This system can also be employed for making good shock absorbers because application of low pressure leads to its significant compression.

The band gap progression with BN nanotube diameter (which is of crucial importance for device applications) was presented and analyzed in detail in [28]. In zigzag BN nanotubes, radial deformations that give rise to transverse pressures decrease the gap from 5 to 2 eV, allowing for optical applications in the visible range [29]. Importantly, both the zigzag and armchair tubes are found [30] (see also [31]) to exhibit large second-order nonlinear optical behavior with the second-harmonic generation and linear electro-optical coefficients being up to 15-times larger than that of bulk BN in both denser zinc-blende and wurtzite structures. This indicates that BN nanotubes are promising materials for nonlinear op-tical and optoelectronic applications.

The electronic structure of BN nanotubes can be tuned within a wide range through covalent functionalization [32] (see also [33]). The ultraviolet (UV) and visible ab-sorption spectra indicate that their electronic structure undergoes drastic changes under functionalization. First principle calculations revealed that the covalently func-tionalized BN nanotubes can be either n- or p-doped de-pending on the electronegativity of molecules attached. Their energy gap can be adjusted from UV to visible optical range by varying concentration of functionalizing species.

One-dimensional crystals of potassium halides, including KI, KCl, and KBr, were inserted into BN nanotubes [34]. High-resolution TEM and energy-dispersive X-ray spectrometry were used to characterize their microstruc-tures and compositions. The fillings are usually single crystals with lengths up to several m. The wetting

properties (static contact angles of the liquids and surface tension) of individual BN nanotubes were studied [35] experimentally using a nanotube-based force to measure the interactions between nanotubes and liquids in situ.

First principles simulations on the interaction of mo-lecular hydrogen H2 with the native and substitutional defects in small-diameter (8,0) BN nanotubes were per-formed in [36]. The adsorption of H2 in structures found to be endothermic with respect to dissociation, with the small-diameter nanotube possessing the smaller barrier. Although chemisorption along the tube axis is energeti-cally preferred, the barrier for dissociation is lower for chemisorption across the tube axis. This implies that chemisorbed hydrogen can be kinetically trapped in a higher energy state. Dopants that maximize the localiza-tion of the higher-occupied-molecular-orbital (HOMO) and lower-unoccupied-molecular-orbital (LUMO) states maximize hydrogen binding energies. C-dopants do not enhance H2 binding, whereas Si-dopants substituting for N provide H2 binding energies of 0.8 eV, at the upper end of the range required for hydrogen storage. The for-mation energy of most defects is reduced with increasing curvature except for the C-substitutionals. Vacancies do not reduce the barriers for H2 dissociation for strongly curved nanotubes. The surface stress induced by the nanotube curvature boosts the hydrogen storage capabili-ties of vacancies with the nitrogen vacancy chemisorbing 4H and allowing a H2 molecule to enter the interior of the tube. The hydrogen binding properties of BN systems strongly depend on existing defects and dopants. Pre-treating of these systems so as to partially remove nitro-gen should enhance H2 adsorption properties. The hy-drogen absorption capacity of Ti-covered single-walled BN nanotube was investigated using first principles plane-wave (PW) method [37]. The weak interaction of H2 molecules with the outer surface of bare nanotube can be significantly enhanced upon functionalization by Ti atoms: each Ti atom adsorbed on tube can bind up to four H2 molecules with average binding energy suitable for room temperature storage.

The morphology of BN nanotubes with a collapsed structure has been discovered by a metal-catalyzed treatment [38]. The collapse causes the dramatic enlarge- ment of a specific surface area of BN nanotubes and re-markably enhances the hydrogen storage capacity of BN nanotubes.

It was reported [39] that proteins are immobilized on boron nitride nanotubes. There is a natural affinity of a protein to BN nanotube: it can be immobilized on tube directly, without using of an additional coupling reagent. Besides, boron nitride nanotubes may be dissolved in organic solvents by wrapping them with a polymer [40].



225

It was proposed [41] that BN polymers, having the structures similar to organic polymers, can serve as a cheap alternative to inorganic semiconductors in design-ing modern electronic devices. Some related potential innovations, including band gap tuning, were also dem-onstrated.

The observed giant Stark effect significantly reduces the band gap of BN nanotubes and thus greatly enhances their utility for nanoscale electronic, electromechanical, and optoelectronic applications [42]. In particular, this effect may be important for tuning the band gap of BN nanotubes for applications as a nanoscale field-effect- transistor (FET).

Boron nitride nanotubes have also manifested stable currents in field emission geometry and may be more stable than carbon nanotubes at high temperatures [43,44].

As it was mentioned, boron nitride nanotubes exhibit many similarities with the carbon ones (such as high Young modulus etc.) and might have superior unique mechanical, thermal, and electronic properties [45]. In addition, BN nanotubes are characterized by chemical inertness and poor wetting [46].

The factor that distinguishes BN from C is partial het-eropolarity of the chemical bonding. For this reason, one more sphere of possible applications of BN nanotubes can be new pyroelectric and piezoelectric materials promising for applications in nanometer-scale sensors and actuators. The 3-fold symmetry of a BN sheet, the III-V analog to graphite, prohibits an electric polarization in its ground state. However, this symmetry is broken when the sheet is wrapped to form a BN tube. It was shown [47] that this leads to an electric polarization along the nanotube axis which is controlled by the quan-tum-mechanical boundary conditions of its electronic states around the tube circumference. Thus, the macro-scopic dipole moment has an intrinsically nonlocal quantum-mechanical origin from the wrapped dimension. Combining first principles, tight-binding methods and analytical theory, the piezoelectricity of heteropolar (in particular, BN) nanotubes was found [48] to depend on their chirality and radius. This effect can be understood starting from the piezoelectric response of an isolated sheet along with a structure specific mapping from the sheet onto the tube surface. It was demonstrated that a linear coupling between the uniaxial and shear deforma-tions occurs for chiral nanotubes, and the piezoelectricity of nanotubes is fundamentally different from its coun-terpart in a bulk material. First principles calculations of the spontaneous polarization and piezoelectric properties of BN nanotubes have shown [49] that they are excellent piezoelectric systems with response values larger than those of piezoelectric polymers. The intrinsic chiral symmetry of the nanotubes induces an exact cancellation of

the total spontaneous polarization in ideal, isolated nanotubes of arbitrary indexes. But the breaking of this symmetry by the intertube interaction or elastic deforma-tions induces spontaneous polarization comparable to that of wurtzite bulk semiconductors [50].

Multielement nanotubes comprising multiple SiC-core, an amorphous SiO2-intermediate layer, and outer shells made of BN and C layers separated in the radial direction with diameters of a few tens of nm and lengths up to 50 m were synthesized by means of reactive laser ablation [51]. They resemble a coaxial nanocable with a semi-conductor-insulator-semiconductor (SIS) geometry and suggest applications in nanoscale electronic devices that take advantage of this self-organization mechanism for multielement nanotube formation.

A theoretical description of electron irradiation of sin-gle-walled BN nanotubes was presented in [52]. As a first step, the anisotropy of the atomic emission energy threshold was obtained within extended molecular-dyna- mical (MD) simulations based on the density-functional- theory (DFT) tight-binding method. As a second step, total cross section for different emission sites as a func-tion of the incident electron energy was numerically de-rived. Two regimes were then described: at low irradia-tion energies (below 300 keV), atoms are ejected mostly from the upper and lower parts of the tube while at high energies (above 300 keV) atoms are ejected mostly from the side walls. Typical values of the total cross section of knock-on processes are obtained to vary from a fraction of barn (at side wall for 150 keV electrons) up to around 20 barns (for 1 MeV electrons). In BN nanotubes, the emission energy threshold maps were reported to show B sputtering to be more favorable for low irradiation ener-gies, while N sputtering is more favorable at high ener-gies. These calculations of the total knock-on cross sec-tion for nanotubes can be used as a guideline for TEM experimentalists using high energy focused beams to shape nanotubes, and also, more generally, if electron irradiation is used to change nanotube properties such as their optical behavior or conductivity.

Such wide field of possible technical and technological applications of boron nitride nanotubes makes useful theoretical research determining their main physical characteristics. In particular, for purposeful design of some materials and devices based on nanotubular BN, like the fibrous composites, tubular heterojunctions, other nanoelectronic devices, nanoreservoirs for hydro-gen storage etc, it is very important to be able to predict reliably values of the ground-state parameters, especially, the molar binding energies and sizes of the nanotubes with given indexes and their relative stability. In present work, this task is solved for the most stable – achiral (zigzag and armchair) – single-walled forms.



226

Paper is organized as follows. In Section 1, we have introduced methods of synthesis of the boron nitride nanotubes and their technological applications. Section 2 is a brief summary of the structural and binding data available on nanotubular boron nitride. In Section 3, the theoretical approach based on the quasi-classical ap-proximation to the binding energy calculation and ge-ometries is presented. In Section 4, results of the per-formed calculations are presented in the form of curves “molar binding energy – structural parameter”. Section 5 is devoted to estimation of the nanotube lattice zero-point vibration energy. And finally, Section 6 discusses rela-tive stability of the boron nitride nanotubes of various radii and makes an attempt to generalize the obtained results.

2. Structural and Binding Data

Let us start with a brief overview of the structural and binding data available on a boron nitride diatomic mole-cule, isolated sheet, and nanotubular form.

2.1. Molecular Boron Nitride

The diatomic molecule BN can be considered as a sim-plest (degenerated) form for boron nitride nanotubes. In general, electronic theory of substance considers a dia-tomic molecule as a special problem for its intermediate structural and, consequently, electronic properties be-tween mono- and polyatomic systems. Peculiarities are related mainly with the system axial symmetry and uniqueness of the structural parameter – interatomic dis-tance d . Unlike the solid state or nanoscale boron ni-trides, which are materials with a diversity of technical and industrial applications, BN molecule, which exists under the extreme conditions, is only of academic inter-est as a “building block” for two- and three-dimensional boron nitride structures. From the standard thermochemical data, the energy of B–N bond at the equilibrium length is known to be considerably higher compared with those of B–B and N–N bonds. In addition, any stable regular BN structure is a network of atomic rings with alternating atoms such that the nearest-neighbor envi-ronment of both B and N atoms consists of only B–N bonds. Therefore, the B–N bond length is a key intera-tomic distance in the analysis of boron-nitrogen binding.

There are known some old first principles and semiempirical investigations for boron–nitrogen interac-tion (see [53,54]). Applying a self-consistent-field (SCF) procedure to the BN molecule in [55], it was calculated molecular orbitals (MOs) in order to minimize total en-ergy of the diatomic system. Then using the spectro-scopic data available for the corresponding ground state, the BN molecule dissociation energy value E was

found to be 4.6 eV. According to the original theoretical approach of [56], the equilibrium interatomic distance in this molecule equals to 1.307 Å. At the same time, spec-troscopic parameters characterizing the calculated bo-ron–nitrogen interaction potential curve lead to the dis-sociation energy estimation of 5.05 eV. Nearly the same theoretical value for the bond length of 1.320 Å was suggested in [57]. In [58], a short-ranged classical-force- field (CFF) modeling of BN modifications was per-formed on the basis of experimental and first principles solid-state and diatomic-molecular data. In particular, assuming that CFF can be correctly determined by a sum of only two-body interaction terms, the B−N potential energy had been expressed analytically via Morse poten-tial, which gave 1.32521d Å and 5.0007E eV. However, it was noted [25] that standard forms of the pair interatomic potentials, such as the Morse, Mee–Grüneisen, Buckingham, and other potentials, con-verge slowly and, therefore, a cutoff procedure should be used. But, in such a case a non-physical jump on the po-tential radial function can arise. In order to eliminate this problem, based on the embedded atom method, a new B−N interatomic potential was designed which fulfills the conditions for smooth end: the potential function and its derivative (i.e., the interatomic force) vanish at the cutoff radius. The equilibrium bond length of 1.4457 Å and binding energy of 4.00 eV were found to reproduce correctly relative stabilities of the boron nitride layered structures.

We also suggested [59] a theoretical, namely, quasi- classical method of calculation of the dependence the B–N interatomic binding energy E upon the bond length d . The constructed ( )E E d curve was shown to be useful for estimations of BN crystalline structures cohesion parameters as well. This function reveals standard behavior characteristic for the central pair potentials. (0)E , and ( ) 0E d if d is equal or greater than the sum of B and N quasi-classical atomic radii ( ) 2.30r and ( ) 1.70r Å, i.e.,

( ) ( ) 4.00d r r Å (note that quasi-classical B−N interatomic potential automatically fulfills the conditions for the smooth end at ( ) ( )d r r ), while within the intermediate region ( ) ( )0 d r r it is an oscillating function with several maxima. Among these maxima only one is available kinetically and, therefore, it corre-sponds to the equilibrium. Analysis of the piece of the quasi-classical ( )E E d curve for BN diatomic mole-cule, in the vicinity of this maximum, yields the values of bond length of 1.55 Å and binding energy of 4.51 eV. Same dependence determined earlier [60,61] within the frames of another quasi-classical parameterization scheme (for this purpose the screening factor of the po-tential affecting the given electron in interacting atom



227

was approximated by the radial polynomial, not by the constant) is relatively flat and leads to the estimations of 1.58 Å and 4.79 eV.

Thus, the spread in theoretical and semiempirical val-ues for BN molecule binding energy is (4.0-5.5) eV, which overlaps with the recommended [62] experimental dissociation energy value of (4.0 ± 0.5) eV. The available first principles and semiempirical calculations and ther-mochemical experimental data lead to the binding energy values of about (4-7) eV per B−N bond for various, dif-ferently coordinated, BN modifications (for sheet and nanotubular structures see below). Such kind of estima-tions may be considered to be in qualitative agreement with the quasi-classically calculated B–N bond energy as the ground state energetic parameters are quite sensitive to the atomic coordination. For this reason, we focus our attention on the differences in the bond length values between BN molecular and crystalline phases.

The quasi-classical values for the isolated B−N bond length and other relevant theoretical and semiempirical data, which lie over the range (1.307-1.580) Å, are overes-timated in comparison with 1.281 Å measured in 11B14N molecule [63]. An explanation may be that BN molecular spectra [64] verify triplet ground state, but at the same time reveal a low-lying singlet state with longer bond.

Quasi-classically calculated interatomic vibration en-ergy in a B−N diatomic system of 0.178 eV/mole (the corresponding vibration quantum equals to 1435 cm1) was found by fitting the quasi-classical B–N potential curve with parabola [60,65]. This value is in good agreement (accurate within 5 %) with the values experimen-tally obtained for a neutral BN molecule of 0.187 (1514.6) [63] and 0.188 eV/mole ((1519.0 ± 0.2) cm1, from the absorption spectra Fourier analysis for laser-induced molecular fluorescence) [64].

According to the SCF theoretical method of [56], the ground state vibration energy in molecular BN estimated as 0.179 eV/mole (1446 cm1), which is almost the quasi- classical result.

In [57], it was suggested the higher theoretical value of 0.217 (1750), what is close with 0.216 eV/mole (1740 cm1) measured in ionized molecule BN+ [66].

Studies of more complex molecular clusters of B and N are also interesting to get deeper insight into the defect formation processes in boron nitride nanotubes. High- temperature Knudsen cell mass spectrometry was used to study the equilibria involving the B2N molecule [67]. The thermal functions necessary to evaluate the mass spectrometric equilibrium data had been calculated from available experimental and theoretical molecular pa-rameters. In particular, in some B2N formation reactions changes in enthalpy have been measured. Room tem-perature atomization and formation enthalpies were de-

termined to be 10.84 and 5.71 eV, respectively. At the same time, first-principles calculations were performed to estimate the electronic parameters of B2N, such as ionization energy and electron affinity.

Mixed clusters of B and N atoms – B2N, BN2, B3N, B4N, B2N2, and B3N2 – can be produced by sputtering of a solid state BN [68]. Atom ordering in assumed linear species had been derived from measurements of the mass distribution of both the positive and the negative prod-ucts from the fragmentation of the anionic clusters in a gas target. As for neutral configurations, they were cal-culated. A tendency was found that a structure with the highest number of B−N bonds is most stable both in neu-tral and anionic species (an exception is the BN2 mole-cule). In contrast to this, the species with the highest number of adjacent same atoms (except for triatomic chains) had the largest electron affinity.

2.2. Boron Nitride Sheet

The facts that boron nitride layered crystals and nano-tubes may be prepared suggest the necessity of analyzing the hypothetic isolated infinite hexagonal layer, i.e., the BN sheet. Corresponding two-dimensional BN crystal is represented as a planar layer composed of regular hexa-gons with vertexes alternately occupied by B and N at-oms. Classification and discussion of the BN haeckelite sheet structures, consisting of not only hexagonal atomic rings but also other even-membered rings, one can find in [69].

For the first time, the truncated crystal approach in the form of two semiempirical (standard and extended itera-tive Hückel) methods was applied to a two-dimensional hexagonal boron nitride structure [70]. The bond length was found to be 1.48 or 1.50 Å. However, when semiempirical calculations were performed on a two-dimensional periodic small cluster of the h-BN layer the equilibrium B−N distance was computed as 1.441 Å [71]. In [72], the 3-coordinated B12N12 network of 6-membered atomic rings was examined theoretically. Namely, the total en-ergy calculated using Hartree–Fock (HF) approach and DFT in local and gradient-corrected forms was mini-mized with respect to the B−N bond length. But, “graph-itic” isomer B12N12 is only a fragment of the BN sheet and its geometry appears to be somewhat distorted be-cause of finite sizes. As is to be expected, slight devia-tions of the bonds’ angles from the ideal value of 120 were observed for the bonds of atoms forming the central hexagon: −(2.52-2.65) for B atoms and +(2.52-2.65) for N atoms. There were also obtained number of unequal bond lengths: (1.266-1.283), (1.371-1.378), (1.427- 1.442), (1.434-1.444), (1.520-1.536), and (1.553-1.576) Å. The finiteness of the quasi-classical atomic radii al-lowed us to obtain the B−N bond length for an infinite



228

boron nitride sheet within the initial quasi-classical ap-proximation [73]. The calculated dependence of the mo-lar binding energy on the lattice constant exhibits a maximum of 23.0 eV at 2.64 Å, which should correspond to the equilibrium state for an isolated hexagonal layer (analytical optimization [74] of the lattice parameter us-ing the binding energy calculated in quasi-classical ap-proximation, which is possible only by neglecting the vibration energy, yields slightly different values: 23.2 eV and 2.66 Å, respectively). The lattice constant of 2.64 Å implies B−N bond length of 1.52 Å. Correction in energy introduced by zero-point vibrations was estimated as 0.242 eV/mole [65,73].

The quasi-classical bond length of 1.52 Å in an iso-lated BN sheet is in reasonable agreement (accurate within 4.6 %) with the bond length of 1.45 Å observed in layers of real h-BN crystals. At first glance, the surpris-ing thing is that the theoretical result for the isolated layer is in better agreement (with the accuracy of (2.6-3.8) %) with the bond lengths in tetrahedrally coordinated modifications c-BN (1.57 Å) and w-BN (1.56 and 1.58 Å). However, it is worth noting that to a certain extent two-dimensional boron nitride looks like three-dimensional crystals c-BN and w-BN: these three structures do not contain weak interlayer bonds, which occur in the h-BN layered modification. The lengths of (1.52-1.54) and (1.55-1.58) Å obtained in [72] for the bonds of atoms forming the central (almost undistorted) hexagon in B12N12 plane-fragment are also in good agreement with the quasi-classical result found for an idealized infinite BN sheet. Another quasi-classical approach using a different scheme of parameterization, employed to calculate h-BN binding and zero-point vibration energies, slightly un-derestimates the intralayer bond length [75]. The plausi-ble reason may be that the crystalline equilibrium con-figuration was selected to maximize its static binding energy with respect only to the layer lattice parameter, while the interlayer distance was fixed.

Summarizing other theoretical and semiempirical re-sults concerning intralayer bond lengths in h-BN (and r-BN), one can state that all of them are in agreement with the experimental value of 1.446 Å [76]. For instance, in [77] the total energy of h-BN crystal as a function of unit cell volume V had been calculated using orthogo-nalized linear-combinations-of-atomic-orbitals (LCAOs) method within the local-density-approximation (LDA). The equilibrium was found at exp/ 0.998V V , where

expV is the experimental value of V . Such result corre-sponds to the intralayer B−N distance of 1.438 Å. The calculations of [78] were also based on DFT within LDA, but PW expansion was used both for the pseudo- potential (PP) and the wave-function. The computed total energies and, consequently, the intralayer bond lengths in

h-BN and r-BN were nearly the same: 1.441 and 1.439 Å, respectively. The short-ranged CFF modeling of boron nitrides leads to exactly the same intralayer B–N bond lengths in both layered structures: 1.454 Å [58]. The re-sults presented show satisfactory accuracy for the quasi- classically determined boron-nitrogen binding character-istics: accuracies of quasi-classical approach to deter-mine isolated B−N bond length and length of bonds in solid state structure amount a few percents, 7.2 and 5.1 %, respectively. Thus, the quasi-classically obtained B–N binding curve and its parameters mentioned above (namely, equilibrium bond length, binding energy, and vibration frequency) would be useful for investigations of compounds containing B−N bonds and, especially, BN nanosystems.

As for the BN sheet binding and vibration energies, it is also reasonable to analyze correctness of the given predictions by comparing them with data available on the cohesion characteristics of h-BN layered crystals. As follows from standard thermochemical data, the binding energy of h-BN equals to 13.0 eV/mole [79]. The binding energies of 14.5, 16.0, and 14.4 eV/mole were deter-mined from semiempirical calculations performed using two variants of the semiempirical LCAOs method and an approach based on a periodic small-sized cluster [70,71]. Within the CFF potential model, the lower semiempirical estimate of 11.5 eV/mole was obtained [58]. In the framework of DFT, optimization of the structural pa-rameters led to the theoretical binding energy of 12.5 eV/mole [78]. Therefore, it can be expected that the mo-lar binding energy for h-BN layered crystal lies in the range from 11.5 to 16.0 eV. The binding energy of 23.0 eV/mole found by the quasi-classical method for the iso-lated layer is considerably higher. However, when com-paring these energies, one should take into account that interlayer bonds are substantially weaker than intralayer ones and that each atom in layered BN structures is in-volved in the formation of 5 bonds, of which only 3 are intralayer bonds. Consequently, if the interlayer energy is ignored as compared to the intralayer energy, we can assume that the binding energy per B−N bond of similar modifications is equal to 3/5 of the molar binding energy of the isolated layer. Making use of the result 23.0 eV/mole for layer, we find the molar binding energy of 3/5 23.0 eV = 13.8 eV. Indeed, this energy is close to the midpoint (13.75 eV) of the aforementioned energy range. On the other hand, the vibration energies of the isolated layer and layered crystals can be directly com-pared because the atoms of the low-dimensional system can execute vibrations in three independent directions in physical space. The quasi-classical result of 0.242 for two-dimensional BN agrees well with analogous calcula-tions of 0.266 [65], with the semiempirical estimate of



229

0.225 for zero-point vibrations energy in h-BN [58], and coincides in order of magnitude with the estimate of 0.350 eV/mole from the theoretical phonon spectrum [78].

A few words about the defects in BN layer. Using PP and expanded unit cell methods, it was found that N- vacancies in BN sheet, as well as di- and trivacancional clusters including neighbor defects in BN layer, are characterized by small binding energies [80]. Calculated spectra and oscillator strengths allow to interpret local bands of the optical absorption in pyrolitic h-BN crystals before and after irradiation by fast-neutrons, protons, and C-ions. By first-principles calculations the 20 structures BxCyNz, derived from a hexagonal layer by placing B, N, or C atoms on each site, were considered [81] to investi-gate their relative stabilities. First-principles simulations of the interaction of molecular hydrogen H2 with the na-tive and substitutional defects of a single hexagonal BN sheet were performed in [36]. The adsorption of H2 on structures found to be endothermic with respect to disso-ciation. Vacancies reduce the barriers for H2 dissociation.

The geometries of haeckelite BN sheets were con-structed by DFT [82]. Their molar energy of cohesion is found to be higher (by ~ 0.6 eV/mole) than that of regu-lar one.

2.3. Nanotubular Boron Nitride

The elementary form of a BN nanotube is a wrapped closed hexagonal surface inscribed in the cylinder. Such BN nanotubes can be found in regular – achiral, i.e., zig-zag (n,0) or armchair (n,n), and also in chiral (n,m) forms, 0 m n . Here n and m are the tube indexes. Their symmetry operators have been identified in [83]: each type belongs to different family of the non-symmorphic rod groups; armchair tubes with even n are found to be centro-symmetric. The types and structures of the non- carbon, in particular, BN nanotubes were reviewed in [84]. In addition, the deformed regular or haeckelite nanotubes can exist. Concerning the haeckelite structures of BN tubes, a variety of chiral angles, including zigzag and armchair types, were observed. Depending on the structure formation kinetics characteristic for a given technology, BN nanotubes quite often take the bamboo- like morphology, form of a nanoarch (i.e., half-tube at the ends closed by planes) etc. Real nanotubular struc-tures are not infinite in length: they are definitely trun-cated.

The three main different possible morphologies of the cylindrical tube closing with flat [8], conical, and amor-phous ends, as observed in experiments, were shown [85] to be directly related to the tube chirality. There are also possible rectangular BN nanotubes with linear defects on edges and with tips in the form of triangular flags. Such kinds of morphologies suggest the presence of energeti-

cally unfavorable odd-membered atomic rings (i.e., pen-tagons and heptagons) in addition to favorable even-membered rings (e.g. squares).

As the growth of BN nanotubes cannot be directly ob-served and, consequently, the underlying microscopic mechanism is a controversial subject, in [86] first-principles MD simulation of the single-walled nanotube edges was performed. The behavior of growing BN nanotubes was found to strongly depend on the nanotube network chirality. In particular, open-ended zigzag tubes close rapidly into an amorphous tip, preventing further growth. In the case of armchair tubes, formation of squares traps the tip into a flat cap presenting a large central even- membered ring. This structure is meta-stable and is able to revert to a growing hexagonal framework by incorpo-rating incoming atoms. These findings are directly re-lated to frustration effects, namely that B−N bonds are energetically favored over B–B and N–N bonds.

The expressions of radii, ( ,0)nR and ( , )n nR , of the zig-zag and armchair, (n,0) and (n,n), BN nanotubes in terms of the index and the structure parameter a were ob-tained [87]. The parameter a corresponds to the lattice constant of the boron nitride layered crystals, i.e., intra-layer B–B or N–N bonds lengths. Therefore, the B–N bond length d equals to / 3a . The nanotube index

1, 2,3,...n determines the number of atoms as nano-tube unit cell consists of 2n formula units BN. The estimations of radii of the single-walled BN nanotubes, for their part, can be used for predicting their most prob-able combinations in multi-walled structures.

Analyzing this problem, it is necessary to take into account that actually the question involves the average radii. A detailed study using the generalized tight-binding MD method has revealed [85] that, as a result of the dy-namical relaxation, the structure acquires a wave-like or “rippled” surface in which B atoms are displaced inward, while N atoms are displaced outward. This relaxation is similar to the reconstruction occurring at clean surfaces of III-V type crystalline semiconductors. However a general feature of BN nanotubular systems is that stronger surface potentials are associated with regions of higher curvature [88]. Thus, the interlayer interaction in BN nanotubes differs from bonding in three-dimensional layered crystals. However, most probably, these distinc-tions for nanotubular BN are weak enough to change essentially the equilibrium interlayer distances which are observed in h-BN and r-BN crystals. This conclusion is also confirmed by the results of an experimental study of the multi-walled nanotubes by high-resolution electronic microscopy [89]. In these structures, like in three-dimensional layered BN crystals, hexagonal and rhombohedral stacking sequences can freely coexist in nanotube wall- assembly. There are also possible some different cross-



230

section flattening, as well as ordering of layers in non- spiral zigzag. According to first-principles total-energy calculations [90], the most favorable double-walled BN nanotubes are structures in which the inter-wall distances are about 3 Å, i.e., as interlayer distances in layered BN crystals. Therefore, due to the weakness of the interlayer van der Waals forces, various types of multi-walled BN nanotubes can exist. Consequently, it is more probable the formation of such multi-walled BN nanotubes in which the difference between the radii of adjacent regu-lar nanotubes is close to the interlayer distance in a lay-ered h-BN crystal, i.e., to half of the height of the hex-agonal unit cell 6.6612c Å [91]. Thus, the similarity between types of the adjacent nanotubes has no crucial importance.

In view of these factors, from the calculated single- walled nanotubes [87] pairs most suitable for the forma-tion of the stable double-walled BN nanotubes have been chosen. Remaining small divergences in sizes of the neighboring regular nanotubes can be compensated by defects and small chiral distortions. Such transformations of the zigzag and armchair nanotubes into chiral one will be accompanied, respectively, by the increase and de-crease in their radii. If the difference in radius between regular nanotubes is more (less) than / 2c , the realiza-tion of structure in which the internal wall will be zigzag (armchair) and external – armchair (zigzag) is more probable. Hence, based on estimations of sizes of the single-walled BN nanotubes, it is possible to predict successfully the most stable double-walled forms. But, how can be solved the same problem for multi-walled nanotubes? A few words on the task. In this case, all over, it will be necessary to calculate radii of nanotubes with high indexes to choose sequences of single-walled nano-tubes, whose radii are close to terms of arithmetic pro-gression with common difference of / 2c . However, now only geometrical considerations will be insufficient. The point is that unlike double-walled nanotubes in multi-walled ones there are also medial layers. For this reason, the choice of the most stable multi-walled struc-ture should be based on the comparison between the gains in energy, which are caused by the deviation from the equilibrium interlayer distance, on the one hand, and by chiral distortions, on the other hand.

We can mention some theoretical results available on binding properties and stabilities of BN nanotubes. Sta-bilities of the boron nitride nanotubular structures were studied by means of non-orthogonal tight-binding for-malism [92]. The radii and energies of the BN nanotubes also were estimated by MD simulation [25] within the embedded atom model in which parameter d took the experimental value 1.4457 Å of the intralayer B–N bond length in real h-BN crystals. In [69], the binding energy

of the regular BN nanotubes has been calculated within the DFT in generalized gradient approximation (GGA). Seeking equilibrium values of B–N bond length and radii, the geometry of the tubular 32-atom supercell was opti-mized. For (8,0), (10,0) and (4,4) tubes, it has been found

1.46d Å, and for (5,5) tube, 1.45d Å. Within the frame of semiempirical calculations of the nanotubular piezoelectric characteristics performed by the method of modified neglecting of diatomic overlapping (MNDO) [50], their radii also were determined. In this case the dependence of energies on the bond length was calcu-lated for the molecular fragments containing 3 or 4 ele-mentary layers (presumably, in this work for d the empirical value known for h-BN crystals was fixed as an equilibrium value).

The possible contribution of ionicity of bonds in boron nitride structures is important to explain the binding dif-ferences between BN tubes and similar C tubes [1]. In order to facilitate understanding and prediction of nano-tube interactions in a multi-walled structure, the electro-static potentials on both outer and inner surfaces of some single-walled BN nanotubes have been calculated at a HF Slater-type-orbital level [88]. Structures were opti-mized computationally. Fictitious hydrogen atoms were introduced at the ends of the open tubes to satisfy the unfulfilled valences. It was found that BN tubes have stronger and more variable surface potentials than graph-itic ones. There are characteristic patterns of positive and negative sites on the outer lateral surfaces, while the in-ner ones are markedly positive.

The binding and vibrations in small-radius single- walled BN nanotubes in [93] were studied by DFT using LDA. The results show that the chirality preference ob-served in experiments may be explained from the relative stability of the corresponding BN strips: the zigzag strips have larger binding energies and thus may be more easily formed. The smallest stable BN nanotube is found to be the (5,0) zigzag nanotube. The dependence of the tube deformation energy on its radius R was approximated by the formula E [eV/mole] 2.095.82 / R [Å]. The phonon dispersions of BN nanotubes were calculated and the frequency of the radial breathing mode was found to be inversely proportional to the nanotube radius. The geometries of the BN nanotubes were also constructed in DFT [82]. Based on DFT calculations [69], it was found that the energies of haeckelite BN nanotubes exceed by ~ 0.6 eV/mole those of corresponding hexagonal nanotubes. They are less stable in comparison with corresponding haeckelite sheets as well. However, still they are stable and can be synthesized. Energy of deformation (i.e., en-ergy needed to wrap nanotube from its sheet prototype) for large haeckelite BN tubes extrapolated by the formula ~ /C R , where R is the tube radius, with different



231

parameters C and 2 for different structures. Using the symmetry properties in [83], it was deter-

mined the numbers of Raman- and infrared (IR)-active vibrations in single-walled BN nanotubes. In contrast to the regular carbon nanotubes, zigzag boron nitride tubes possess almost twice as many vibrations as armchair ones. An extensive first principles study of the phonons in BN nanotubes using perturbation DFT in the LDA was performed in [94], where, based on the non-symmorphic rod group symmetry of tubes, the Raman- and IR-active modes at the point of the one-dimensional Brillouin zone were evaluated. For zigzag and chiral nanotubes, the set of IR-active modes is a subset of the Raman-active modes. In particular, the radial breathing mode is not only Raman-, but also IR-active. However, for armchair tubes, the sets of IR- and Raman-active modes are dis-joint. The frequencies of the active modes of zigzag, chiral, and armchair tubes were presented as a function of the tube diameter. They were compared with the fre-quencies obtained by the zone-folding method (i.e., by rolling of a BN sheet into a tube). Except for the high- frequency tangential modes, the zone-folding results are in very good agreement with the first principles calcula-tions. The radial breathing mode frequency can be de-rived by folding a sheet of finite width. Finally, the ef-fects of bundling on the phonon frequencies are shown to be small. First principles calculations of the nonresonant Raman spectra of zigzag and armchair BN nanotubes were presented in [95]. In comparison, a generalized bond-polarizability model, where the parameters are ex-tracted from first principles calculations of the polariza-bility tensor of a BN sheet, was implemented. For light polarized parallel to the tube axis, the agreement between model and first principles spectra is almost perfect, but for perpendicular polarization, depolarization effects have to be included in the model in order to reproduce Raman intensities. The possible dislocation dipoles as defect nuclei under tension in BN nanotubes were identified by dislocation theory and MD simulations [96]. Formation energies of the dipoles evaluated by gradient-corrected DFT are high and remain positive at large strains, thus suggesting great yield resistance of BN nanotubes. The dipole appears to be more favorable in spite of its homoelemental B−B and N−N bonds. The resonant photoabsorption and vibration spectroscopy combined with scanning tunneling micros-copy unambiguously identify the presence of Stone–Wales defects in BN nanotubes [97]. Based on extensive time- dependent DFT calculations, it was proposed to reso-nantly photoexcite such defects in the IR and UV re-gimes as a means of their identification. Intrinsic defects in zigzag BN nanotubes, including single vacancy, diva-cancy, and Stone–Wales defects, were systematically

investigated using DFT calculation in [98]. It was found that the structural configurations and formation energies of the topological defects are dependent on tube diameter. The results demonstrate that such properties are origi-nated from the strong curvature effect in BN nanotubes. The scanning tunneling microscope images of intrinsic defects in the BN nanotubes also were predicted. The defected BN tubes with C-substitutions were considered in [50].

The theoretical studies of the elastic properties of sin-gle-walled BN nanotubes, carried out using the total -energy non-orthogonal tight-binding parameterization, were reported in [99]. Tubes of different diameters, ranging from 0.5 to 2 nm, were examined. The study found that in the limit of large diameters the mechanical properties of nanotubes approach those of the graph-ite-like sheet. The stiffness and plasticity of BN nano-tubes was investigated [100] using generalized tight- binding MD and first principles total-energy methods. Due to the B−N bond rotation effect, the compressed zigzag nanotubes were found to undergo anisotropic strain release followed by anisotropic plastic buckling. The strain is preferentially released toward N atoms in the rotated B−N bonds. The tubes buckle anisotropically toward only one end when uniaxially compressed from both ends. Based on these results, a skin-effect-model of smart nanocomposite materials is proposed, which local-izes the structural damage toward the surface side of the material. B−N bond-rotation mode of plastic yield in BN nanotubes in [101] was investigated combining first principles computations with a probabilistic rate approach to predict the kinetic and thermodynamic strength. BN nanotubes yield defects have low activation, but high formation energies. In [50], elastic characteristics of BN nanotubes also were calculated applying MNDO method.

3. Theoretical Basis

Our calculations are based on the quasi-classical expres-sion for binding molar energy of a substance, on the one hand, and on the geometric characterization of nano-tubular boron nitride, on the other hand.

3.1. Quasi-Classical Binding Energy of Substance

Under the term ‘substance’ we imply polyatomic struc-tures at the ground state, i.e., molecules, various clusters, and crystals. Consequently, any substance is considered as a non-relativistic electron system affected by the static external field of nuclei, which are fixed at the sites in structure, and the averaged SCF of electrons. Because of singularities at the points, where the nuclei are located, and electron shell effects as well the inner potential of



232

substance does not satisfy the standard Wentzel–Kramers– Brillouin (WKB) quasi-classical condition of spatial smoothness. Nevertheless, beginning from Bohr’s fun-damental work ‘On the constitution of atoms and mole-cules’ up to the present the semi-classical analysis of the electronic spectrum has been widely used for light atoms and their small complexes. Besides, heavy atoms, large molecules, and crystals can be treated within the LDA using the total energy functional in the form of quasi- classical expansion. Success of quasi-classical approaches can be attributed to the diffuseness of atomic poten-tials. The expression for bounded electron states energies obtained by Maslov yields that precise and quasi-classi-cal spectra are close to each other if the characteristic values of potential 0 and the radius of its action 0R meet requirement 2

0 02 1R (here and below all rela-tions are given in atomic units). For atomic potential

0 ~ /Z R and 0 ~R R where 1Z is the atomic number and R is the radius of electron cloud. Therefore, in case of atoms it is required that 2 1ZR . Even for light atoms their radii are several times larger than Bohr radius, 1R . Thus atoms and all polyatomic struc-tures indeed are quasi-classical electron systems and their structural and electronic characteristics can be calculated based on the quasi-classically parameterized electric charge density and electrical field potential distributions in atoms.

The values of i -th electron classical turning point ra-dii ir and ir , i ir r , are obtained by solving the equations

2

( 1)( ) 1, 2,3,...,

2i i

i i

l lE r i Z

r

,

where r denotes the distance from the center of atom, ( )i r is the potential affecting the given electron,

0iE and il are its energy and orbital quantum num-ber, respectively.

In the ground state the inner classical turning point for relative motion of atomic nucleus and electron cloud coincides with the center of system. As for the corre-sponding outer classical turning point radius r , it is obtained by solving the equation

( )E Z r ,

where 0E denotes the energy associated with rela-tive motion and

2

1

1( ) ( )

1

i Z

ii

Zr r

Z r

is the electron cloud potential affecting the nucleus. In particular, using the quasi-classical parameteriza-

tion based on the Coulomb-like atomic potentials ( )i r

iZ r we are able to get exact formulas

2 ( 1)i i i i i

ii

n n n l lr

Z

,

2 ( 1)i i i i i

ii

n n n l lr

Z

,

22 2

1

2( 1)i Z

ii

Zr

Z Z Z

,

23 2

1

22( 1)

i Z

ii

Z Z Z

EZ

.

Here 2i i iZ n E is the effective charge of the screened nucleus and in is the principal quantum num-ber of i -th electron. The numerical values of iZ , E ,

ir , ir and r can be found by fitting the quasi-classical energy levels iE to ab initio (for instance HF) ones.

Quasi-classical limit implies the truncation of electron states charge densities outside the classical turning points and space-averaging within the range between them. In this case i -th electron partial charge density is approxi-mated by the piecewise-constant radial function

3 3

( ) 0

3

4 ( )

0

i i

i ii i

i

r r r

r r rr r

r r

.

As for the nucleus charge density, it should be aver-aged inside the r -sphere:

3

3( ) 0

4

0

Zr r r

r

r r

.

Consequently, the full atomic charge density is ex-pressed by the step-like radial function

1

( ) ( ) ( )i Z

i ki

r r r

1 1, 2,3,...,k kr r r k q ,

where kr and k denote known constants which de-

pend on parameters ir , ir and r ,

0 1 20 qr r r r , 2q Z is the number of different homogenous-charge-density spherical layers in atom.

Using the Poisson equation, the radial dependence of the full atomic potential also can be approximated by the



233

step-like function, 2 2 5 5

1 13 3 3 3

1 1

3 ( ) 3 ( )( )

2( ) 5( )k k k k k k

k kk k k k

a r r b r rr c

r r r r

1 1, 2,3,...,k kr r r k q ,

3 3 311 1

1

4 ( ) 4

3 3

i ki i i k k

ki

r r ra

,

2

3k

kb

,

2 2 21

1

2 ( ) 2i q

k i i i k ki k

c r r r

,

if it is substituted by the space-averaged values inside each of the 1k kr r r intervals.

In the region qr r , both the charge density and po-tential vanish identically, ( ) 0r and ( ) 0r . Thus finite parameter qr acquires a meaning of the quasi- classical atomic radius.

Based on the presented step-like parameterization of

the charge density and potential distributions in an atom, its quasi-classical total energy can be expressed in the following form:

3 3( ) 1

1

( ) 03

i q

Atom k k k kk

E r r

.

Note that it includes the non-physical energy of self-action, ( ) 0Atom Self ActionE which arises from sub-stituting the charge density for the probability density. Its value can be easily calculated in the quasi-classical approximation and then excluded from the total energy.

When the molecular or crystalline charge densities and potentials are expressed by the superposition of the step- like atomic charge densities and potentials, respectively, the molar (i.e., per chemical formula unit of the sub-stance) ground state static energy and its zero-point vi-bration correction are calculated as

( ) ( )( ) ( )

( ) ( ) ( ) ( )( ) 1 ( ) 1 1 1

1( ) 0

4

i kj q l qi N k N

Static i j k l ik jl ik tt i k j l

E V r

,

( ) ( )( ) ( )( ) ( ) ( ) ( ) ( ) ( )/

( ) 1 ( ) 1 1 1 ( ) ( ) ( )

( )30.

2 2

i kj q l qi N k Ni j k l k l i j ik jl ik t

Vibrationi t k j l i ik t ik t

V rE

M r r

Here the primed summation symbol denotes the elimination of the terms with 0t

and i k ; in-

dexes in parentheses ( )i and ( )k denote the atoms in the molecule or crystal unit cell, N is the full

number of atoms, ( )iM is the mass of i -th atom, t

is the crystal translational vector – in case of a mole-cule 0t

, ( )ik tr is a distance between atomic sites

and

( ) ( ) ( ) ( ) 1 ( ) 1 ( ) ( ) 1 ( ) 1 ( )( ) ( ) ( ) ( ) ( )( ) ( , , ) ( , , ) ( , , ) ( , , )ik jl i j k l i j k l i j k l i j k lik t ik t ik t ik t ik tV r V r r r V r r r V r r r V r r r .

We have introduced an universal geometric function

1 2 12( , , )V R R D which expresses the volume of the inter-section of two spheres as a function of their radii 1R

and 2R , and the inter-center distance 12D . 1 2 12( , , )V R R D and its partial derivative 1 2 12 12( , , ) /V R R D D both are continuous piecewise algebraic functions as follows:

31

1 2 12

4( , , )

3

RV R R D

12 2 1D R R ,

324

3

R 12 1 2D R R ,

2 2 2 21 2 12 1 2 12 1 1 2 2

12

( ) (( ) 4( ))

12

R R D R R D R R R R

D

1 2 12 1 2| |R R D R R ,

0 1 2 12R R D ;

1 2 12

12

( , , )0

V R R D

D

12 2 1D R R ,

0 12 1 2D R R , 2 2 2 2

1 2 12 12 1 2212

(( ) )( ( ) )

4

R R D D R R

D

1 2 12 1 2| |R R D R R ,

0 1 2 12R R D .

In the lowest quasi-classical approximation, the equi-librium structure of substance is obtained by maximizing the molar binding energy of expected structures.

( )

( ) ( )( ) 1

( )

( ) 0

i N

Binding i i Self Actioni

Static Self Action Vibration

E E E

E E E



234

with respect to their structural parameters. However, neglecting the insignificant redistribution of

valence electrons arisen from association of atoms into a molecular or crystalline structure, the quasi-classical self- action energy of the substance is approximated by the sum of self-action energies of constituent atoms:

( )

( )( ) 1

i N

Self Action i Self Aactioni

E E

.

And consequently, the binding energy approximately can be calculated without excluding of the self-action terms in advance:

( )

( )( ) 1

0i N

Binding i Static Vibrationi

E E E E

.

The expected errors of the quasi-classical approach can be estimated for the model inner potential in the form of the analytical solution of the Thomas–Fermi (TF) equation for the semi-classical atomic potential: structural and energy parameters of the electronic system deter-mined within the initial quasi-classical approximation are shown to differ from their exact values by factors

1/3(10 / 3 ) 1.02 ~ 1 and 2/3(3 /10) 0.96 ~ 1 , respectively. Thus the expected errors of the quasi-classical approach amount to a few percents. Even more, within the initial quasi-classical approximation there are no un-controllable calculation errors due the finiteness of quasi- classical atomic radii − the pair interactions without se-ries termination are truncated at the distances exceeded the sums of atomic radii.

A complete quasi-classical theory of substance in-cluding calculation schemes for structural and binding, as well as for electronic spectrum characteristics, one can find in [102,103]. These schemes have been applied suc-cessfully for Na molecular and crystalline structures [104], various diatomic molecules [60,61], boron nano-tubes [105,106], and mainly for one-, two- and three- dimensional structural modifications of boron nitride – diatomic molecule, isolated plane sheet, hexagonal h-BN, cubic c-BN, and wurtzite-like w-BN crystals [59,60,64, 73-75,107,108].

3.2. Geometries of the Boron Nitride Regular Single-Walled Nanotubes

Summarizing previous subsection, one can conclude that equilibrium structural and binding parameters of the bo-ron nitride nanotubes can be calculated quasi-classically based on analytical expressions describing their geome-tries. This task has been solved in [87,108,109] for regu-lar (achiral), i.e., zigzag (n,0) and armchair (n,n) BN nanotubes. A model of regular nanotubes used here as-sumes that all atomic sites are located on cylindrical sur-face at the vertexes of regular hexagons broken along

B–N or B–B and N–N diagonals, i.e., the expected small differences in bond length distinguished by their orienta-tion toward the tube axis are neglected.

Namely, radii ( ,0)nR and ( , )n nR of the zigzag and armchair nanotubes have been obtained [87,108] in terms of the nanotube index 1,2,3,...n and the structure parameter a :

( ,0) 4sin / 2n

aR

n ,

( , )

5 4cos / 2

4 3 sin / 2n n

n aR

n

.

As it was mentioned, the parameter a corresponds to a lattice constant of the boron nitride layered crystals, i.e., to an intralayer B–B or N–N bond lengths. Therefore, the B–N bond length d equals to / 3a . The nanotube index n determines the number of atoms, as a nanotube unit cell consists of 2n formula units, B2nN2n.

Detailed regular geometries of zigzag and armchair BN nanotubes have been described in [108,109] using cylindrical coordinates ( , , )z , which are useful for calculating binding energy.

A unit cell of zigzag nanotubes consists of 4 atomic rings in parallel planes perpendicular to the axis. There are 2 pairs of rings, each consisting of 2 planes with n boron or n nitrogen atoms. Obviously, the cylindrical coordinate for all atomic sites equals to the tube ra-dius:

( ,0)nR .

As for the coordinates and z in the first and sec-ond pairs of atomic rings, they equal to

2 /l n ,

(6 1) / 2 3z m a ,

(6 1) / 2 3z m a ,

and

(2 1) /l n ,

(3 1) / 3z m a ,

(3 1) / 3z m a ,

respectively. Here 0,1,2,..., 1l n and 0, 1, 2,...m number atomic pairs in a given pair of the atomic rings and these rings themselves, respectively.

The unit cell of armchair nanotubes consists of 2 atomic rings in parallel planes perpendicular to the tube axis. For its part, each ring consists of n boron and n ni-trogen atoms. The coordinate for all atomic sites again equals to the tube radius:

( , )n nR ,



235

while the rest cylindrical coordinates in the first and second atomic rings equal to

1 2 /l n ,

1 2 /l n ,

z z ma ,

and

1 22 2 /l n ,

1 22 2 /l n ,

(2 1) / 2z z m a ,

respectively. Here

1

2sin / 2sin

5 4cos / 2

n

n

,

2

sin / 2sin

5 4cos / 2

n

n

,

and 0,1,2,..., 1l n and 0, 1, 2,...m number B or N atoms in atomic rings and these rings themselves.

Based on the above discussion, the distances between a given atomic site and the sites in the so-called central atomic pairs (with 0l m ) in zigzag and armchair nanotubes have been found.

For zigzag ( 0l m : 0 , B / 2 3z a , and N / 2 3z a ) tubes

00 2 2( ,0) ( ,0) 2

2 2

( 1 1) sin /3

4sin / 2

lmn n l n

ma n

,

00 2 2 2( ,0) ( ,0)

2 2

( 2 1) sin (2 1) / 2 3(2 1)

44sin / 2

lmn n l n m

a n

,

00 2 2 2( ,0) ( ,0)

2 2

( 1 1) sin / (3 1)

34sin / 2

lmn n l n m

a n

,

00 2 2 2( ,0) ( ,0)

2 2

( 2 1) sin (2 1) / 2 (6 1)

124sin / 2

lmn n l n m

a n

,

00 2 2 2( ,0) ( ,0)

2 2

( 1 1) sin / (3 1)

34sin / 2

lmn n l n m

a n

,

00 2 2 2( ,0) ( ,0)

2 2

( 2 1) sin (2 1) / 2 (6 1)

124sin / 2

lmn n l n m

a n

,

00 2 2( ,0) ( ,0) 2

2 2

( 1 1) sin /3

4sin / 2

lmn n l n

ma n

,

00 2 2 2( ,0) ( ,0)

2 2

( 2 1) sin (2 1) / 2 3(2 1)

44sin / 2

lmn n l n m

a n

.

For armchair ( 0l m : 1 , 1 , and 0z z ) tubes,

00 2 2( , ) ( , ) 2

2 2

( 1 1) (5 4cos / 2 )sin /

12sin / 2

lmn n n n n l n

ma n

,

00 2( , ) ( , )

2

2 2

2

( 2 1)

(5 4cos / 2 )sin (2 1) / 2 (2 1)

412sin / 2

lmn n n n

a

n l n m

n

.

00 2( , ) ( , )

2

22

2

( 1 1)

(2sin(2 1) / 2 sin / )

12sin / 2

lmn n n n

a

l n l nm

n

,

00 2( , ) ( , )

2

2 2

2

( 2 1)

(sin(2 1) / 2 2sin / ) (2 1)

412sin / 2

lmn n n n

a

l n l n m

n

,

00 2( , ) ( , )

2

22

2

( 1 1)

(2sin(2 1) / 2 sin / )

12sin / 2

lmn n n n

a

l n l nm

n

,

00 2( , ) ( , )

2

2 2

2

( 2 1)

(sin(2 1) / 2 2sin / ) (2 1)

412sin / 2

lmn n n n

a

l n l n m

n

,

00 2 2( , ) ( , ) 2

2 2

( 1 1) (5 4cos / 2 )sin /

12sin / 2

lmn n n n n l n

ma n

,

00 2( , ) ( , )

2

2 2

2

( 2 1)

(5 4cos / 2 )sin (2 1) / 2 (2 1)

412sin / 2

lmn n n n

a

n l n m

n

.

4. Binding Energies in Dependence on Structural Parameter

At first, based on the above stated relations and HF val-ues of the atomic electron energy-levels tabulated in [110], the required quasi-classical parameters kr , k , and k for constituent atoms B and N have been calcu-lated. They are given in Tables 1 and 2, respectively.

Here the values are shown in atomic units with 7 sig-nificant digits in accordance with the accuracy of input data (HF energies). Such high accuracy is useful in in-terim calculations. As for the final results, they should be expressed in rounded figures with 3 or 4 significant digits (in Å or eV for structure or energy parameters, respec-tively) because it corresponds to the usual experimental errors when determining structure and energy parameters of a substance, and the relative errors of the semi-classical calculations aimed at finding theoretically these parameters for polyatomic systems amount to a few percents.



236

Table 1. Quasi-classical parameters of step-like radial distributions of electron-charge-density and electric-field-potential in boron atom (in a.u.).

k 1 2 3 4 5

kr 2.758476E − 02 5.098016E − 01 7.441219E − 01 4.021346E + 00 4.337060E + 00

k 5.686514E + 04 −3.610951E + 00 −7.342212E − 03 −1.028341E − 02 −2.941197E − 03

k 2.105468E + 02 8.882329E + 00 3.652920E + 00 2.060720E − 01 6.135348E − 04

Table 2. Quasi-classical parameters of step-like radial distributions of electron-charge-density and electric-field-potential in nitrogen atom (in a.u.).

k 1 2 3 4 5

kr 9.446222E − 03 3.577244E − 01 5.498034E − 01 2.909074E + 00 3.204489E + 00

k 1.982589E + 06 −1.044967E + 01 −1.939444E − 02 −4.126981E − 02 −2.187537E − 02

k 8.784581E + 02 2.022523E + 01 8.464698E + 00 5.096684E − 01 3.993358E − 03

Using these parameters and expressions for the com-

ponents of the quasi-classical molar binding energy and squared interatomic distances in zigzag and armchair BN nanotubes (Figures 1 and 2), their binding energies were calculated versus the structural parameter a with spac-ing of 0.001 a.u., i.e., within the accuracy of 4 significant digits. The vibration energy is assumed to be zero when radicand in its formula becomes negative.

In order to carry out these massive calculations, a spe-cial computer code has been designed. Calculations were performed within the range of the structure parameter a which varied from 1 to 13 a.u. with a step 0.001a a.u. (that is quite enough to cover values having any physical sense). The nanotube indexes varied from 1n , covering zigzag and armchair nanotubes up to (18,0) and (10,10), respectively. The radii of the largest calculated species are approximately equal. They are sufficiently large for the tube molar binding energy to almost reach the “saturation” value, which is given by the binding energy of the planar hexagonal BN sheet. In order to make sure that such “saturation” indeed takes place, test species with very large indexes (45,0) and (26,26) (again with approximately equal radii) have also been calcu-lated.

Figures 3-6 show ( )BindingE a curves (and their trends near the peaks) for some zigzag and armchair BN nano-tubes, respectively. One can see that at sufficiently small interatomic distances binding energy might take a large negative value that implies that the structure is unstable, while at sufficiently large interatomic distances the binding energy always equals to zero which reveals atomization of a structure. As for the intermediate distances, the mo-lar binding energy is positive that is a signature of struc-tural stability. In this case, general trend in binding en-ergy value is decreasing. However, ( )BindingE a curves

are not monotonous, but with several extremes. Such kind of oscillatory behavior of the molar biding energy of

Figure 1. Structure of a zigzag BN nanotube.

Figure 2. Structure of an armchair BN nanotube.



237

(a)

(b)

Figure 3. Molar binding energy of zigzag BN nanotubes vs. structural parameter a for different nanotube indexes n. any atomic structure against the inter-atomic distances reflects electron-shell-structure of the constituent atoms (note that the interaction between particles of matter with forces non-monotonously decreasing with distance was foreseen as early as in 18th century by Boscovich [111], whose atomic theory was based only on abstract philoso- phical speculations).

Figures 7 and 8 show the binding energies for the two types of achiral nanotubes with the same index n (namely (3,0) and (3,3) nanotubes).

Let us discuss which of the peaks in these figures cor-respond to the equilibrium structure. The first peak from the right is lower than the successive one. This second peak for all tubes located at 5.085a a.u. (2.691 Å) seems to correspond to the realized stable BN nanotubu-lar structures (the detailed behavior of ( )BindingE a curves in its vicinity is shown in Figures 4, 6 and 8). The next peak, even being higher than this, can not been reached kinetically in standard laboratory conditions because they correspond to lower interatomic distances and these two peaks are separated by very deep and sufficiently wide

(a)

(b)

Figure 4. Molar binding energies of zigzag BN nanotubes with different indexes n near the peak a = 2.691 Å.

Figure 5. Molar binding energy of armchair BN nanotubes vs structural parameter a for different nanotube indexes n. minima, i.e., by high and wide potential barriers which can be overcome only at ultrahigh temperatures or tun-neled only at ultrahigh pressures.

The obtained equilibrium binding energies of BN nanotubes of both achiral types are summarized in Table 3.



238

Figure 6. Molar binding energies of armchair BN nanotubes with different indexes n near the peak a = 2.691 Å.

Figure 7. Dependence of molar binding energies of zigzag and armchair BN nanotubes with the nanotube index n = 3 on the structural parameter a. together with their radii calculated from formulas for equilibrium value of the structural parameter 2.691a Å.

Thus, the molar binding energies of small-sized nano-tubes, both zigzag and armchair have peaks at (3,0) and (2,2), respectively. Then for large indexes the binding energies decrease toward the same constant value (see Figures 9 and 10).

Figure 11 presents the dependence of the molar bind-ing energy of achiral BN nanotubes on their radii R . It reveals pairs of minima at (1,0) and (2,0), and maxima at (1,1) and (3,0), i.e., all the extremes are located in low-radii-region. At higher radii, the molar binding en-ergy slowly decreases to the value of 23.26 eV, which, apparently, corresponds to that of the plane hexagonal BN sheet.

The obtained dependence ( )BindingE a in its domain of monotonicity seems to be quite smooth. It allows us to extrapolate this curve also to chiral BN nanotubes (Figure 12) because the radius of a chiral tube ( , )n mR

Figure 8. Molar binding energies of zigzag and armchair BN nanotubes with the index n = 3 near the peak a = 2.691 Å.

Figure 9. Molar binding energy of zigzag BN nanotubes for different nanotube indexes n at the peak a = 2.691 Å.

Figure 10. Molar binding energy of armchair BN nanotubes for different nanotube indexes n at the peak a = 2.691 Å. ( 0 m n ) and radii of the corresponding achiral tubes always meet the condition ( ,0) ( , ) ( , )n n m n nR R R .

The results of the carried out calculations are pre-



239

Table 3. Quasi-classically calculated radii and binding en-ergies of BN nanotubes.

Nanotube Radius, Å Binding Energy, eV/mole

(1,0) 0.673 12.26

(1,1) 0.868 26.08

(2,0) 0.951 23.17

(3,0) 1.345. 29.72

(2,2) 1.537 27.59

(4,0) 1.758 26.40

(5,0) 2.177 24.74

(3,3) 2.260 24.62

(6,0) 2.599 24.19

(4,4) 2.993 23.94

(7,0) 3.023 23.92

(8,0) 3.448 23.76

(5,5) 3.729 23.68

(9,0) 3.874 23.64

(10,0) 4.300 23.57

(6,6) 4.668 23.54

(11,0) 4.727 23.51

(12,0) 5.154 23.47

(7,7) 5.207 23.46

(13,0) 5.581 23.43

(8,8) 5.947 23.41

(14,0) 6.008 23.41

(15,0) 6.436 23.39

(9,9) 6.687 23.37

(16,0) 6.863 23.37

(17,0) 7.291 23.35

(10,10) 7.428 23.35

… … …

(45,0) 19.276 23.26

(26,26) 19.290 23.26

sented in Figures 13 and 14 in the form of surface plots where the molar binding energy ( , )BindingE a n of zigzag and armchair BN nanotubes is shown as a function of the structural parameter a and the nanotube index n .

5. Zero-Point Vibration Energies

First, let us emphasize some features characteristic for the quasi-classical procedure of estimation of the zero-point vibration energies.

Figure 11. Molar binding energy of achiral BN nanotubes vs the nanotube radius R.

Figure 12. Structure of a chiral BN nanotube.

On the one hand, within the above formulated quasi- classical approach, all binding energy maxima are related with the onset of overlapping between certain regions of homogeneity of electric charge density and electric field potential in interacting atoms, constituents of the struc-ture under the consideration. Namely, equilibrium point at 5.085a a.u. corresponds to the B–N bond length of 2.937d a.u. which is a sum of radii B 1 0.028r a.u. and N 4 2.909r a.u. (Tables 1 and 2).

On the other hand, quasi-classical expression of the vibration energy is based on the parabolic approximation of the ( )BindingE a curve and formula for the volume of the intersection of two spheres, 1 2 12( , , )V R R D , which is a continuously differentiable function of the inter-central distance 12D . However, one can see readily from ex-pression of its first (continuous) derivative that second derivative is not a continuous function.

That is the reason why the parabolic approximation of the ( )BindingE a curve in the immediate vicinity of the



240

1 2

3 4

5 6

7 8

9 10

11 12

13 14

15 16

17 18

n

2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 a (Å)

-30-20-10

0102030

EBinding (eV)

-30 -20 -10 0 10 20 30

(eV)

Figure 13. Surface plot of the molar binding energy ( , )BindingE a n of a zigzag BN nanotube as a function of the

structural parameter a and nanotube index n.

1 2

3 4

5 6

7 8

9 10

n 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 a (Å)

-30

-20

-10

0

10

20

30

EBinding (eV)

-30 -20 -10 0 10 20 30

(eV)

Figure 14. Surface plot of the molar binding energy ( , )BindingE a n of an armchair BN nanotube as a function of

the structural parameter a and nanotube index n. binding energy peak is impossible. According to formula,

VibrationE is identically zero at the equilibrium point and is assumed to be identically zero also on the left in the vicinity of that point, where radicand in its formula be-comes negative. As for the right side in the vicinity of the equilibrium point, the binding energy can be estimated for the nearest domain allowing parabolic approximation. Its half, i.e., arithmetic mean of the vibration energy left- and right-sided values can be considered as estima-tion for the vibration energy correction in the equilibrium. These values together with the correspondingly corrected binding energy are presented in Table 4.

The dependence of the molar vibration energy on the BN nanotube radius qualitatively reproduces that for the binding energy. However, this dependence is very weak and, thus, the molar vibration energy can be considered as almost independent from the tube radius, ~ 0.3 eV/mole.

Table 4. Quasi-classically calculated vibration energies of BN nanotubes.

Nanotube Vibration Energy,

eV/mole Corrected Binding Energy,

eV/mole

(1,0) 0.25 12.01

(1,1) 0.33 25.75

(2,0) 0.32 22.85

(3,0) 0.33 29.39

(2,2) 0.32 27.27

(4,0) 0.32 26.08

(5,0) 0.32 24.42

(3,3) 0.32 24.30

(6,0) 0.32 23.87

(4,4) 0.31 23.67

(7,0) 0.31 23.61

(8,0) 0.31 23.45

(5,5) 0.31 23.37

(9,0) 0.31 23.33

(10,0) 0.31 23.26

(6,6) 0.31 23.23

(11,0) 0.31 23.20

(12,0) 0.31 23.16

(7,7) 0.31 23.15

(13,0) 0.31 23.12

(8,8) 0.31 23.10

(14,0) 0.31 23.10

(15,0) 0.31 23.08

(9,9) 0.31 23.06

(16,0) 0.31 23.06

(17,0) 0.31 23.04

(10,10) 0.31 23.04

… … …

(45,0) 0.31 22.95

(26,26) 0.31 22.95

Of course, the vibration corrections to the binding energy are too weak to change character of the ( )BindingE R de-pendence.

6. Concluding Remarks

The quasi-classically calculated structure parameter a 2.691 Å of single-walled boron nitride nanotubes is in satisfactory agreement with experimental value for the h-BN layered crystals exp 2.504a Å [91], i.e., the diff-



241

erence is about 7%. As is mentioned above the overesti-mations in the structural parameter is characteristic for the quasi-classical approach. However, at least partially, this overestimation seems to be related with expansion of lattice of the single hexagonal layer (plane or cylindrical) if compared with that of the three-dimensional layered crystal.

It is also instructive to analyze the obtained spread of the molar zero-point vibration energy of BN nanotubes (0.25-0.33) eV and, in particular, its limit for ultra-large- radius tubes 0.31 eV using the data available on the vi-bration characteristics of h-BN layered crystal. The vi-bration energies of an isolated tubular layer and layered crystals can be directly compared as the atoms of the low-dimensional system can execute vibrations in three independent directions in physical space. Our quasi-classical estimations made for BN nanotubes agree well with analogous calculations (but in tight-binding approximation) for BN plane sheet of 0.27, the semiempirical estimate of 0.23 for zero-point vibrations energy in h-BN, and the estimate of 0.35 eV/mole from the theoretical phonon spectrum (see Subsection 2.2).

We have found that the binding energies of BN sin-gle-walled nanotubes corrected with zero-point vibration energies lies within the interval (12.01-29.39) eV. In par-ticular, the calculated corrected binding energy of the ultra-large-radius tube is predicted as 22.95 eV. Previous quasi-classical calculations (but in tight-binding approximation) performed for BN isolated plane sheet have given the binding energy 23.00 eV/mole, which coin-cides in order of magnitude with this interval and agrees very well with present result obtained for large tubes. As it was demonstrated in Subsection 2.2, for its part the binding energy ~ 23 eV/mole for single-layer boron ni-tride structures should be in good agreement with bind-ing energy data available for BN multi-layered structures. Summarizing the obtained results, it should be empha-sized that a complex dependence of the BN nanotube molar binding energy on its radius is found out, though all the binding energy values are found to be positive, i.e., all tubes should be stable. However, they have rather different degrees of stability.

On the one hand, ultra-small-radius BN nanotubes (1,0) and (2,0) seem to be meta-stable, though their molar binding energies are positive, they are less than that for isolated hexagonal boron nitride layer. Especially the smallest (1,0) tube structure degenerated in zigzag atomic strip should be meta-stable because its binding en-ergy is only about half of this value. Such a structure can be realized only as an inner wall in a multi-walled tube. On the other hand, the formation probabilities for BN tubes with indexes (1,1), (3,0), and (4,0) should exceed that for isolated sheet. Among them the (3,0) tube is well

pronounced, formation of which is predicted to be ener-getically most preferable than the layer growth. Molar binding energies for other BN nanotubes slightly exceed that of sheet and their formation probabilities should be almost same as for layered crystal growth.

Finally, it should be noted that, in addition to the en-ergy considerations concerning the relative stability of tubular structure, it is also necessary to take into account features characteristic to BN nanosystems, in view of the general equations derived for energy fluctuations of small completely open (incompressible) systems [112]. They show that the fluctuations should be unusually large because there are no constraints on the size of a system and, in addition, the fluctuations of the total or partial number of atoms in binary systems indirectly con-tribute to the fluctuations in their energy.

7. Acknowledgements

L. Chkhartishvili acknowledges the financial support from the Georgia National Science Foundation (GNSF) under the Project # GNSF/ST 08/4-411-Geometry of the boron nitride nanostructures.

REFERENCES [1] A. Rubio, J. L. Corkill and M. L. Cohen, “Theory of

Graphitic Boron Nitride Nanotubes,” Physical Review B, Vol. 49, No. 7, 1994, pp. 5081-5084.

[2] Z. Weng-Sieh, K. Cherrey, N. G. Chopra, X. Blasé, Y. Miyamoto, A. Rubio, M. L. Cohen, S. G. Louie, A. Zettl and R. Gronsky, “Synthesis of BxCyNz Nanotubules,” Physical Review B, Vol. 51, No. 16, 1995, pp. 11229- 11232.

[3] N. G. Chopra, R. J. Luyken, K. Cherrey, V. H. Crespi, M. L. Cohen, S. G. Louie and A. Zettl, “Boron-Nitride Nano-tubes,” Science, Vol. 269, No. 5226, 1995, pp. 966-967.

[4] K. Suenaga, C. Colliex, N. Demoncy, A. Loieseau, H. Pascard and F. Willaime, “Synthesis of Nanoparticles and Nanotubes with Well-Separated Layers of Boron Nitride and Carbon,” Science, Vol. 278, No. 5338, 1997, pp. 653- 655.

[5] K. Suenaga, F. Willaime, A. Loieseau and C. Colliex. “Organization of Carbon and Boron Nitride Layers in Mixed Nanoparticles and Nanotubes Synthesized by Arc Discharge,” Applied Physics A, Vol. 68, No. 3, 1999, pp. 301-308.

[6] A. Loiseau, F. Willaime, N. Demoncy, G. Hug and H. Pascard, “Boron Nitride Nanotubes with Reduced Num-bers of Layers Synthesized by Arc Discharge,” Physical Review Letters, Vol. 76, No. 25, 1996, pp. 4737-4740.

[7] Y. Saito, M. Maida and T. Matsumoto, “Structures of Boron Nitride Nanotubes with Single-Layer and Multi- Layers Produced by Arc Discharge,” Japanese Journal of Applied Physics, Vol. 38, No. 1A, 1999, pp. 159-163.

[8] Y. Saito and M. Maida, “Square, Pentagon, and Heptagon



242

rings at BN Nanotube Tips,” Journal of Physical Chem-istry A, Vol. 103, No. 10, 1999, pp. 1291-1293.

[9] D. Golberg, Y. Bando, M. Eremets, K. Takemura, K. Kurashima and H. Yusa “Nanotubes in Boron Nitride Laser Heated at High Pressure,” Applied Physics Letters, Vol. 69, No. 14, 1996, pp. 2045-2047.

[10] R. Sen, B. C. Satishkumar, A. Govindaraj, K. R. Hariku-mar, G. Raina, J.-P. Zhang, A. K. Cheetham and C. N. R. Rao, “B-C-N, C-N and B–N Nanotubes Produced by the Pyrolysis of Precursor Molecules over Co Catalysts,” Chemical Physics Letters, Vol. 287, No. 5-6, 1998, pp. 671-676.

[11] D. Golberg, Y. Bando, W. Han, K. Kurashima and T. Sato, “Single-Walled B-Doped Carbon, B/N-Doped Car-bon, and BN Nanotubes Synthesized from Single-Walled Carbon Nanotubes through a Substitution Reaction,” Chemical Physics Letters, Vol. 308, No. 3-4, 1999, pp. 337-342.

[12] D. Golberg, W. Han, Y. Bando, L. Bourgeois, K. Kura-shima and T. Sato, “Fine Structure of Boron Nitride Nanotubes Produced from Carbon Nanotubes by a Sub-stitution Reaction,” Journal of Applied Physics, Vol. 86, No. 4, 1999, pp. 2364-2366.

[13] Y. Chen, L. T. Chadderton, J. F. Gerald and J. S. Wil-liams, “A Solid-State Process for Formation of Boron Ni-tride Nanotubes,” Applied Physics Letters, Vol. 74, No. 20, 1999, pp. 2960-2962.

[14] T. S. Bartnitskaya, G. S. Oleinik, V. V. Pokropivnyi, N. V. Danilenko, V. M. Vershchaka and A. V. Kotko, “Na- notubes from Graphite-Like BN,” Superhard Materials, No. 6, 1998, pp. 71-74.

[15] T. S. Bartnitskaya, G. S. Oleinik, A. V. Pokropivnyi and V. V. Pokropivnyi, “Synthesis, Structure, and Formation Mechanism of Boron Nitride Nanotubes,” JEPT Letters, Vol. 69, No. 2, 1999, pp. 163-168.

[16] C. Colazo-Davila, E. Bengu, L. D. Marks and M. Kirk, “Nucleation of Cubic Boron Nitride Thin Films,” Dia-mond Related Materials, Vol. 8, No. 6, 1999, pp. 1091- 1100.

[17] T. Oku, T. Hirano, M. Kuno, T. Kusunose, K. Niihara and K. Suganuma, “Synthesis, Atomic Structures, and Proper-ties of Carbon and Boron Nitride Fullerene Materials,” Materials Science and Engineering B, Vol. 74, No. 1-3, 2000, pp. 206-217.

[18] J. Wang, V. K. Kayastha, Y. K. Yap, Z. Fan, J. G. Lu, Z. Pan, I. N. Ivanov, A. A. Puretzky and D. B. Geohegan, “Low Temperature Growth of Boron Nitride Nanotubes on Substrates,” Nano Letters, Vol. 5, No. 12, 2005, pp. 2528-2532.

[19] L. L. Sartinska, S. Barchikovski, N. Wagenda, B. M. Rud’ and I. I. Timofeeva, “Laser Induced Modification of Surface Structures,” Applied Surface Science, Vol. 253, No. 9, 2007, pp. 4295-4299.

[20] L. L. Sartinska, A. A. Frolov, A. Yu. Koval’, N. A. Danilenko, I. I. Timofeeva and B. M. Rud’, “Transforma-tion of Fine-Grained Graphite-Like Boron Nitride In-duced by Concentrated Light Energy,” Materials Chem-

istry and Physics, Vol. 109, No. 1, 2008, pp. 20-25.

[21] M. V. P. Altoe, J. P. Sprunck, J.-C. P. Gabriel and K. Bradley, “Nanococoon Seeds for BN Nanotube Growth,” Journal of Materials Science, Vol. 38, No. 24, 2003, pp. 4805-4810.

[22] M. W. Smith, K. C. Jordan, C. Park, J.-W. Kim, P. T. Lillehei, R. Crooks and J. S. Harrison, “Very Long Sin-gle- and Few-Walled Boron Nitride Nanotubes via the Pressurized Vapor/Condenser Method,” Nanotechnology, Vol. 20, No. 50, 2009, pp. 505604-505610.

[23] X. Blasé, J.-C. Charlier, A. de Vita and R. Car, “Theory of Composite BxCyNz Nanotube Heterojunctions,” Ap-plied Physics Letters, Vol. 70, No. 2, 1997, pp. 197-199.

[24] X. Blasé, J.-C. Charlier, A. de Vita and R. Car, “Struc-tural and Electrical Properties of Composite BxCyNz Nanotubes and Heterojunctions,” Applied Physics A, Vol. 68, No. 3, 1999, pp. 293-300.

[25] V. V. Pokropivnyj, V. V. Skorokhod, G. S. Oleinik, A. V. Kurdyumov, T. S. Bartnitskaya, A. V. Pokropivnyj, A. G. Sisonyuk and D. M. Sheichenko, “Boron Nitride Analogs of Fullerenes (the Fulborenes), Nanotubes, and Fullerites (the Fulborenites),” Journal of Solid State Chemistry, Vol. 154, No. 1, 2000, pp. 214-215.

[26] J. H. Lee, “A Study on a Boron-Nitride Nanotube as a Gigahertz Oscillator,” The Journal of the Korean Physi-cal Society, Vol. 49, No. 1, 2006, pp. 172-176.

[27] V. Verma and K. Dharamvir, “BNNT in Contact with h-BN Sheet and Other BNNT and DW-BNNT as GHz Oscillator,” International Journal of Nanosystems, Vol. 1, No. 1, 2008, pp. 27-34.

[28] B. Baumeier, P. Krüger and J. Pollmann, “Structural, Elastic, and Electronic Properties of SiC, BN, and BeO Nanotubes,” Physical Review B, Vol. 76, 2007, p. 085407.

[29] Y.-H. Kim, K. J. Chang and S. G. Louie, “Electronic Structure of Radially Deformed BN and BC3 Nanotubes,” Physical Review B, Vol. 63, 2001, p. 205408.

[30] G. Y. Guo and J. C. Lin, “Second-Harmonic Generation and Linear Electro-Optical Coefficients of BN Nano-tubes,” Physical Review B, Vol. 72, 2005, p. 075416.

[31] G. Y. Guo and J. C. Lin, “Erratum: Second-Harmonic Generation and Linear Electro-Optical Coefficients of BN Nanotubes,” Physical Review B, Vol. 77, 2008, p. 049901 (E).

[32] C. Zhi, Y. Bando, C. Tang and D. Golberg, “Engineering of Electronic Structure of Boron-Nitride Nanotubes by Covalent Functionalization,” Physical Review B, Vol. 74, 2006, p. 153413.

[33] C. Zhi, Y. Bando, C. Tang and D. Golberg, “Publisher’s Note: Engineering of Electronic Structure of Boron-Ni- tride Nanotubes by Covalent Functionalization,” Physical Review B, Vol. 74, 2006, p. 199902 (E).

[34] W.-Q. Han, C. W. Chang and A. Zettl, “Encapsulation of One-Dimensional Potassium Halide Crystals within BN Nanotubes,” Nano Letters, Vol. 4, No. 7, 2004, pp. 1355- 1357.

[35] K. Yum and M.-F. Yu, “Measurement of Wetting Proper-



243

ties of Individual Boron Nitride Nanotubes with the Wil-helmy Method Using a Nanotube-Based Force Sensor,” Nano Letters, Vol. 6, No. 2, 2006, pp. 329-333.

[36] S. A. Shevlin and Z. X. Guo, “Hydrogen Sorption in De-fective Hexagonal BN Sheets and BN Nanotubes,” Physi-cal Review B, Vol. 76, 2007, p. 024104.

[37] E. Durgun, Y.-R. Jang and S. Ciraci, “Hydrogen Storage Capacity of Ti-Doped Boron-Nitride and B/Be-Substitut- ed Carbon Nanotubes,” Physical Review B, Vol. 76, 2007, p. 073413.

[38] Ch. Tang, Y. Bando, X. Ding, Sh. Qi and D. Golberg, “Catalyzed Collapse and Enhanced Hydrogen Storage of BN Nanotubes,” Journal of the American Chemical Soci-ety, Vol. 124, No. 49, 2002, pp. 14550-14551.

[39] C. Zhi, Y. Bando, C. Tang and D. Golberg. “Immobiliza-tion of Proteins on Boron Nitride Nanotubes,” Journal of the American Chemical Society, Vol. 127, No. 49, 2005, pp. 17144-17145.

[40] C. Zhi, Y. Bando, C. Tang, R. Xie, T. Sekiguchi and D. Golberg, “Perfectly Dissolved Boron Nitride Nanotubes Due to Polymer Wrapping,” Journal of the American Chemical Society, Vol. 127, No. 46, 2005, pp. 15996- 15997.

[41] M. Côté, P. D. Haynes and C. Molteni, “Boron Nitride Polymers: Building Blocks for Organic Electronic De-vices,” Physical Review B, Vol. 63, 2001, p. 125207.

[42] M. Ishigami, J. D. Sau, Sh. Aloni, M. L. Cohen and A. Zettl, “Observation of the Giant Stark Effect in Bo-ron-Nitride Nanotubes,” Physical Review Letters, Vol. 94, 2005, p. 056804.

[43] J. Cumings and A. Zettl, “Mass-Production of Boron Nitride Double-Wall Nanotubes and Nanococones,” Chemical Physics Letters, Vol. 316, No. 3-4, 2000, pp. 211-216.

[44] Y. Chen, J. Zou, S. J. Campbell and G. le Caer, “Boron Nitride Nanotubes: Pronounced Resistance to Oxidation,” Applied Physics Letters, Vol. 84, No. 13, 2004, pp. 2430- 2432.

[45] E. Bengu and L. D. Marks, “Single-Walled BN Nanos-tructures,” Physical Review Letters, Vol. 86, No. 11, 2001, pp. 2385-2387.

[46] C. Tang, Y. Bando, Y. Huang, S. Yue, C. Gu, F. F. Xu and D. Golberg, “Fluorination and Electrical Conductivity of BN Nanotubes,” Journal of the American Chemical Society, Vol. 127, No. 18, 2005, pp. 6552-6553.

[47] E. J. Mele and P. Král, “Electric Polarization of Heter-opolar Nanotubes as a Geometric Phase,” Physical Re-view Letters, Vol. 88, 2002, p. 056803.

[48] N. Sai and E. J. Mele, “Microscopic Theory for Nanotube Piezoelectricity,” Physical Review B, Vol. 68, 2003, p. 241405.

[49] S. M. Nakhmanson, A. Calzolari, V. Meunier, J. Bernholc and M. Buongiorno Nardelli, “Spontaneous Polarization and Piezoelectricity in Boron Nitride Nanotubes,” Physi-cal Review B, Vol. 67, 2003, p. 235406.

[50] N. G. Lebedev and L. A. Chernozatonskiĭ, “Quantum-

Chemical Calculations of the Piezoelectric Characteristics of Boron Nitride and Carbon Nanotubes,” Physics of the Solid State, Vol. 48, No. 10, 2006, pp. 2028-2034.

[51] Y. Zhang, K. Suenaga, C. Colliex and S. Iijima, “Coaxial Nanocable: Silicon Carbide and Silicon Oxide Sheathed with Boron Nitride and Carbon,” Science, Vol. 281, No. 5379, 1998, pp. 973-975.

[52] A. Zobelli, A. Gloter, C. P. Ewels, G. Seifert and C. Col-liex, “Electron Knock-on Cross Section of Carbon and Boron Nitride Nanotubes,” Physical Review B, Vol. 75, 2007, p. 245402.

[53] I. V. Weiz, “Supplement II,” In: K.-P. Huber and G. Herzberg, Constants of Diatomic Molecules, Part 2: Molecules N2–ZrO, Mir, Moscow, 1984, pp. 295-366.

[54] O. I. Bukhtyarov, S. P. Kurlov and B. M. Lipenskikh, “Calculation of the Structure and Physicochemical Prop-erties of B-Based Systems by the Computer Modeling Method,” In: Abstracts of the 8th International Sympo-sium on Boron, Borides, Carbides, Nitrides, and Related Compounds, Tbilisi, October 1984, pp. 135-136.

[55] J. L. Masse and M. Bärlocher, “Etude par la Méthode du Autocohérent (Self-Consistent) de la Molécule BN,” Helvetica Chimia Acta, Vol. 47, No. 1, 1964, pp. 314- 318.

[56] Yu. G. Khajt and V. I. Baranovskij, “Ab Initio Calcula-tions of the BN Molecule,” Journal of Structural Chem-istry, Vol. 21, No. 1, 1980, pp. 153-154.

[57] H. Bredohl, J. Dubois, Y. Houbrechts and P. Nzo-habonayo, “The Singlet Bands of BN,” Journal of Phys-ics B, Vol. 17, No. 1, 1984, pp. 95-98.

[58] C. M. Marian, M. Gastreich and J. D. Gale, “Empirical Two-Body Potential for Solid Silicon Nitride, Boron Ni-tride, and Borosilazane Modifications,” Physical Review B, Vol. 62, No. 5, 2000, pp. 3117-3124.

[59] L. Chkhartishvili, “Quasi-Classical Analysis of Boron- Nitrogen Binding,” In: Proceedings of the 2nd Interna-tional Boron Symposium, Eskişehir, Turkey, September 2004, pp. 165-171.

[60] L. Chkhartishvili, D. Lezhava, O. Tsagareishvili and D. Gulua, “Ground State Parameters of B2, BC, BN, and BO Diatomic Molecules,” Transactions of the AMIAG, Vol. 1, 1999, pp. 295-300.

[61] L. Chkhartishvili, D. Lezhava and O. Tsagareishvili, “Quasi-Classical Determination of Electronic Energies and Vibration Frequencies in Boron Compounds,” Jour-nal of Solid State Chemistry, Vol. 154, No. 1, 2000, pp. 148-152.

[62] A. Gaydon, “Dissociation Energies and Spectra of Dia-tomic Molecules,” Chapman & Hall, London, 1947.

[63] K. P. Huber and G. Herzberg, “Molecular Spectra and Molecular Structure: IV. Constants of Diatomic Mole-cules,” van Nostrand Reinhold Co, New York, 1979.

[64] M. Lorenz, J. Agreiter, A. M. Smith and V. E. Bondybey, “Electronic Structure of Diatomic Boron Nitride,” Jour-nal of Chemical Physics, Vol. 104, No. 8, 1996, pp. 3143- 3146.



244

[65] L. Chkhartishvili, “Zero-Point Vibration Energy within Quasi-Classical Approximation: Boron Nitrides,” Pro-ceedings of the TSU (Physics), Vol. 40, 2006, pp. 130-138.

[66] V. I. Vedeneev, L. V. Gurvich, V. N. Kondrat’ev, V. A. Medvedev and E. A. Frankevich, “Chemical Bonds Breaking Energies, Ionization Potentials and Electron Af-finities,” Academy of Sciences of USSR, Moscow, 1962.

[67] G. Meloni, M. Sai Baba and K. A. Gingerich, “Knudsen Cell Mass Spectrometric Investigation of the B2N Mole-cule,” Journal of Chemical Physics, Vol. 113, No. 20, 2000, pp. 8995-8999.

[68] R. Vanderbosch, “Gas-Phase Anions Containing B and N,” Physical Review A, Vol. 67, 2003, p. 013203.

[69] S. V. Lisenkov, G. A. Vinogradov, T. Y. Astakhova and N. G. Lebedev, “Non-Spiral ‘Haeckelite-Type’ BN-Nano-tubes,” JETP Letters, Vol. 81, No. 7, 2005, pp. 431-436.

[70] A. Zunger, “A Molecular Calculation of Electronic Prop-erties of Layered Crystals: I. Truncated Crystal Approach for Hexagonal Boron Nitride,” Journal of Physics C, Vol. 7, No. 1, 1974, pp. 76-95.

[71] A. Zunger, “A Molecular Calculation of Electronic Prop-erties of Layered Crystals: II. Periodic Small Cluster Calculation for Graphite and Boron Nitride,” Journal of Physics C, Vol. 7, No. 1, 1974, pp. 96-106.

[72] D. L. Strout, “Structure and Stability of Boron Nitrides: Isomers of B12N12,” Journal of Physical Chemistry A, Vol. 104, No. 15, 2000, pp. 3364-3366.

[73] L. S. Chkhartishvili, “Quasi-Classical Estimates of the Lattice Constant and Band Gap of a Crystal: Two-Di-mensional Boron Nitride,” Physics of the Solid State, Vol. 46, No. 11, 2004, pp. 2126-2133.

[74] L. S. Chkhartishvili, “Analytical Optimization of the Lat-tice Parameter Using the Binding Energy Calculated in the Quasi-Classical Approximation,” Physics of the Solid State, Vol. 48, No. 5, 2006, pp. 846-853.

[75] L. Chkhartishvili and D. Lezhava, “Zero-Point Vibration Effect on Crystal Binding Energy: Quasi-Classical Cal-culation for Layered Boron Nitride,” Transactions of the GTU, Vol. 439, 2001, pp. 87-90.

[76] K.-H. Hellwege and O. von Madelung, Eds., “Boron Ni-tride (BN),” In: Landolt-Börnstein, Numerical Data and Functional Relationships in Science and Technology, New Series, Group III: Crystal and Solid State Physics, Volume 17: Semiconductors, Subvolume g: Physics of Non-Tetrahedrally Bonded Binary Compounds III, Springer-Verlag, Berlin, 1982.

[77] Y.-N. Xu and W. Y. Ching, “Calculation of Ground-State and Optical Properties of Boron Nitrides in the Hexagonal, Cubic, and Wurtzite Structures,” Physical Review B, Vol. 44-I, No. 15, 1991, pp. 7784-7798.

[78] K. Albe, “Theoretical Study of Boron Nitride Modifica-tions at Hydrostatic Pressures,” Physical Review B, Vol. 55-II, No. 10, 1997, pp. 6203-6210.

[79] D. D. Wagmann, W. H. Evans, V. B. Parker, J. Halow, S. M. Bairly and R. H. Shumn, Eds., Selected Values of Chemical Thermodynamic Properties, National Bureau of

Standards, Washington, 1968.

[80] S. N. Grinyaev, F. V. Konusov and V. V. Lopatin, “Deep Levels of Nitrogen Vacancy Complexes in Graphite-Like Boron Nitride,” Physics of the Solid State, Vol. 44, No. 2, 2002, pp. 286-293.

[81] M. S. C. Mazzoni, R. W. Nunes, S. Azevedo and H. Chacham, “Electronic Structure and Energetics of BxCyNz Layered Structures,” Physical Review B, Vol. 73, 2006, p. 073108.

[82] S. V. Lisenkov, G. A. Vonogradov, T. Y. Astakhova and N. G. Lebedev, “Geometric Structure and Electronic Properties of Planar and Nanotubular BN Structures of the Haeckelite Type,” Physics of the Solid State, Vol. 48, No. 1, 2006, pp. 192-198.

[83] O. E. Alon, “Symmetry Properties of Single-Walled Bo-ron Nitride Nanotubes,” Physical Review B, Vol. 64, 2001, p. 153408.

[84] V. V. Pokropivnyi, “Non-Carbon Nanotubes (Review). 2. Types and Structure,” Powder Metallurgy and Metal Ce-ramics, Vol. 40, No. 11-12, 2002, pp. 582-594.

[85] M. Menon and D. Srivastava, “Structure of Boron Nitride Nanotubes: Tube Closing Versus Chirality,” Chemical Physics Letters, Vol. 307, No. 5-6, 1999, pp. 407-412.

[86] J.-C. Charlier, X. Blase, A. de Vita and R. Car, “Micro-scopic Growth Mechanisms for Carbon and Boron-Ni- tride Nanotubes,” Applied Physics A, Vol. 68, No. 3, 1999, pp. 267-273.

[87] L. S. Chkhartishvili, “On Sizes of Boron Nitride Nano-tubes,” In: Proceedings of the 18th International Sympo-sium “Thin Films in Optics in Nanoelectronics,” Kharkiv, October 2006, pp. 367-373.

[88] Z. Peralta–Inga, P. Lane, J. S. Murray, S. Boyd, M. E. Grice, C. J. O’Connor and P. Politzer, “Characterization of Surface Electrostatic Potentials of Some (5,5) and (n,1) Carbon and Boron/Nitrogen Model Nanotubes,” Nano Letters, Vol. 3, No. 1, 2003, pp. 21-28.

[89] D. Golberg, Y. Bando, L. Bourgeois, K. Kurashima and T. Sato, “Insights into the Structure of BN Nanotubes,” Ap-plied Physics Letters, Vol. 77, No. 13, 2000, pp. 1979- 1981.

[90] S. Okada, S. Saito and A. Oshiyama, “Inter-Wall Interac-tion and Electronic Structure of Double-Walled BN Nanotubes,” Physical Review B, Vol. 65, No. 16, 2002, p. 165410.

[91] Y. B. Kuz’ma and N. F. Chaban, “Boron Containing Bi-nary and Ternary Systems: Handbook,” Metallurgiya, Moscow, 1990.

[92] E. Hernández, C. Goze, P. Bernier and A. Rubio, “Elastic properties of C and BxCyNz Composite Nanotubes,” Physical Review Letters, Vol. 80, No. 20, 1998, pp. 4502- 4505.

[93] H. J. Xiang, J. Yang, J. G. Hou and Q. Zhu, “First-Prin- ciples Study of Small-Radius Single-Walled BN Nano-tubes,” Physical Review B, Vol. 68, 2003, p. 035427.

[94] L. Wirtz, A. Rubio, R. A. de la Concha and A. Loiseau, “Ab initio Calculations of the Lattice Dynamics of Boron



245

Nitride Nanotubes,” Physical Review B, Vol. 68, 2003, p. 045425.

[95] L. Wirtz, M. Lazzeri, F. Mauri and A. Rubio, “Raman Spectra of BN Nanotubes: Ab Initio and Bond-Polari- zability Model Calculations,” Physical Review B, Vol. 71, 2005, p. 241402 (R).

[96] H. F. Bettinger, T. Dumitrică, G. E. Scuseria and B. I. Yakobson, “Mechanically Induced Defects and Strength of BN Nanotubes,” Physical Review B, Vol. 65, 2002, p. 041406.

[97] Y. Miyamoto, A. Rubio, S. Berber, M. Yoon and D. To- mánek, “Spectroscopic Characterization of Stone–Wales Defects in Nanotubes,” Physical Review B, Vol. 69, 2004, p. 121413 (R).

[98] G. Y. Gou, B. C. Pan and L. Shi, “Theoretical Study of Size-Dependent Properties of BN Nanotubes with Intrin-sic Defects,” Physical Review B, Vol. 76, 2007, p. 155414.

[99] E. Hernandez, C. Goze, P. Bernier and A. Rubio, “Elastic Properties of Single-Wall Nanotubes,” Applied Physics A, Vol. 68, No. 3, 1999, pp. 287-292.

[100] D. Srivastava, M. Menon and K.-J. Cho, “Anisotropic Nanomechanics of Boron Nitride Nanotubes: Nanostruc-tured “Skin” Effect,” Physical Review B, Vol. 63, 2001, p. 195413.

[101] T. Dumitrică and B. I. Yakobson, “Rate Theory of Yield in Boron Nitride Nanotubes,” Physical Review B, Vol. 72, 2005, p. 035418.

[102] L. Chkhartishvili, “Quasi-Classical Theory of Substance Ground State,” Technical University Press, Tbilisi, 2004.

[103] L. Chkhartishvili, “Quasi-Classical Method of Calcula-tion of Substance Structural and Electronic Energy Spec-

trum Parameters,” Tbilisi University Press, Tbilisi, 2006.

[104] L. Chkhartishvili, “Selection of Equilibrium Configura-tions for Crystalline and Molecular Structures Based on Quasi-Classical Inter-Atomic Potential,” Transactions of the GTU, No. 427, 1999, pp. 13-19.

[105] L. Chkhartishvili, “On Quasi-Classical Estimations of Boron Nanotubes Ground-State Parameters,” Journal of Physics: Conference Series, Vol. 176, 2009, p. 012013.

[106] L. Chkhartishvili, “Molar Binding Energy of the Boron Nanosystems,” In: Proceedings of the 4th International Boron Symposium, Osmangazi University Press, Eskişe-hir, Turkey, 2009, pp. 153-160.

[107] L. Chkhartishvili, “Ground State Parameters of Wurtzite Boron Nitride: Quasi-Classical Estimations,” In: Pro-ceedings of the 1st International Boron Symposium, Dumlupinar University Press, Kütahya, 2002, pp. 139- 143.

[108] L. Chkhartishvili, “Boron Nitride Nanosystems of Regu-lar Geometry,” Journal of Physics: Conference Series, Vol. 176, 2009, p. 012014.

[109] L. S. Chkhartishvili, “Equilibrium Geometry of Ultra- Small-Radius Boron Nitride Nanotubes,” Material Sci-ence of Nanostructures, No. 1, 2009, pp. 33-44.

[110] Ch. Froese–Fischer, “The Hartree–Fock Method for At-oms: A Numerical Approach,” Wiley, New York, 1977.

[111] R. J. Boscovich, “Philosophiae Naturalis Theoria Redacta ad Inicam Legem Vitium in Natura Existentiam,” Colle-gio Romano, Milano, 1758.

[112] T. L Hill and R. V. Chamberlin, “Fluctuations in Energy in Completely Open Small Systems,” Nano Letters, Vol. 2, No. 6, 2002, pp. 609-613.




Lígia Maria Manzine Costa, Rosário Elida Suman Bretas, Rinaldo Gregorio Jr

Department of Materials Engineering, Federal University of São Carlos, Rod. Washington Luís, São Carlos-SP, Brazil. Email: [email protected] Received March 19th, 2010; June 17th, 2010; June 21st, 2010.

ABSTRACT

A study was conducted regarding the effect of concentration of poly (vinylidene fluoride) (PVDF)/N,N-dimethylformamide (DMF) and PVDF/DMF/acetone solutions on the transition between electrospray and electrospinning and on the for-mation of the and crystalline phases of PVDF. The crystalline phases present in the samples, crystallinity and morphology were determined by Fourier transform infrared spectroscopy (FTIR), differential scanning calorimetry (DSC) and scanning electron microscopy (SEM), respectively. Low concentration solutions resulted in films consisting of small droplets (electrospray) containing predominantly the phase. High concentration solutions resulted in a non-woven mesh of nano-to-micron diameter fibers (electrospinning) containing exclusively the phase. These results showed that, the formation of this phase in the electrospinning is related mainly to the solvent evaporation rate, and not to drawing experienced by the polymer during the process. Solvent type affected the amount of crystalline phase present, the boundary concentration between the two processes and the average diameter of fibers. Meshes processed by elec-trospinning display a degree of crystallinity higher than the films obtained by electrospray. Keywords: Poly(Vinylidene Fluoride) (PVDF), Electrospinning, Electrospray, Crystalline Phase, Fibers, Morphology

1. Introduction

PVDF, poly(vinylidene fluoride), is a widely investigated polymer due to its excellent mechanical properties, chemical stability and ferroelectricity. This polymer has a simple chemical formula, -CH2-CF2-, and may crystal- lize in at least four crystalline phases known as α, β, γ and δ. The structure and property of these phases are well documented in the literature [1]. Each crystalline phase confers characteristic properties to the polymer and therefore distinct applications. The β phase is that which displays the strongest pyro and piezoelectric activities and is therefore technologically more interesting. Many techniques have been used to obtain this phase, being the most common uni or biaxial drawing of originally α phase films [2-6]. Crystallization of PVDF from solution (casting) may also result in the phase, depending on the evaporation rate of the solvent [7-9]. Low rates result predominantly in the phase, whereas high rates favor the phase. Electrospinning [10,11] also produces pre-

dominantly the phase [12-18]. In this technique the polymer solution is added to a capillary (which can be a syringe with needle). The solution forms a droplet at the tip of the needle due to surface tension. If a sufficiently high electric voltage (5-30kV) is applied to the solution electric charges accumulate in the droplet. When the electrostatic repulsion between the charges overcomes the surface tension and viscoelasticity of the droplet this assumes the shape of a cone. When the force exerted by the electric field, formed between the droplet and a grounded collector, overcomes the surface tension, a thin jet is formed. The jet of the charged solution is accelerated towards the collector under the action of the electric field. If the concentration of the solution is sufficiently high (viscous) to stabilize the jet, the polymer solution is severely stretched and the solvent evaporates to form ultrathin fibers which solidify and are deposited on the collector forming a non-woven mesh of fibers. Some authors [13-17] attribute the formation of β phase in electrospun PVDF meshes to this severe drawing ex-



247

perienced by the polymer jet solution. This elongation process would be similar to the well known transition caused by mechanical drawing at T 90 [2-6].

A variant of electrospinning is electrospray [19]. The basic difference between these two processes lies in the concentration of the solution. In electrospray the concentration is sufficiently low to destabilize the charged jet which then breaks down into small spherical droplets that solidify during the course and are deposited on the collector. In this case the polymer solution does not experi- ence severe drawing and the formed film consists of small droplets instead of fibers.

The objectives of the current investigation were: 1) determine the solution concentration at which the transition between electrospray and electrospinning of PVDF occurs and 2) verify through the electrospray technique, where no stretching occurs, that the formation of the phase depends fundamentally on the solvent evaporation rate. The effect of the type of solvent in the process was also verified.

2. Experimental Methods

2.1. Materials

The PVDF used was Foraflon® 4000HD, from Elf Ato- chem. The solvents used were N,N-dimethylformamide (DMF, Merck 99.5%) and acetone (Merck, 99.7%). So- lutions were prepared at concentrations of 5, 7, 10 and 15 wt% PVDF using as solvent pure DMF and a 3:1 v/v mixture of DMF with acetone. The ratio of mixture DMF and acetone was selected because they produce thinner and more homogeneous nanofibers in the electrospinning process [13]. Solubilization was carried out at 70 under stirring for one hour. Table 1 shows some properties of the solvents used.

2.2. Electrospray/Electrospinning

A scheme of the system used in the electrospray/electrospinning process is shown in Figure 1. A 20-mL glass syringe was used with a steel needle with internal diameter of 0.7 mm. The distance between needle and collector was 3 cm and the electric voltage applied was 10 kV, using a Bertan 210 30R (0-30 kV) high voltage source. The collector used was an aluminum disk with diameter of 15 cm and width of 5 cm, at an angular velocity of 60 rpm. The process was carried out at 25 and RH 55%.

2.3. Characterizations

The crystalline phases present in the samples were characterized by transmission infrared spectroscopy (FTIR, Perkin Elmer Spectrum 1000) in the range between 400 and 1000 cm-1 and with resolution of 2 cm-1. Calorimetric analyses were conducted in a Perkin Elmer DSC-7, cali-

Table 1. Properties of the solvents used (20).

Solvent Formula Density (g/mL)

Boiling point ()

Viscosity (mPa.s)

Acetone C3H6O 0.786 56.5 0.326

DMF C3H7NO 0.944 153 0.820

Figure 1. Scheme of the setup used in the Electrospinning/ Electrospray process. brated with In, using samples of 8 mg and heating rate of 10/min. Degree of crystallinity was determined by comparing the enthalpy of fusion of the sample obtained by DSC with that of 100% crystalline PVDF- (104.5 J/g) [20]. Sample surface morphology was observed by using a scanning electron microscope (Philips 30 FEG Model XL, operating at 15 kV) after gold coating. Size distribution of the fibers diameters was determined by the soft- ware “image J” developed by the National Institute of Health.

3. Results and Discussions

The micrographs in Figures 2(a) to (d) and 3(a) to (d) present the morphology of the samples processed with PVDF/DMF and PVDF/DMF/acetone solutions at different concentrations, respectively.

One can observe in Figures 2(a)-(b) (5 and 7 wt% in DMF) and 3(a) (5 wt% in DMF/acetone) the formation of a film consisting of small droplets, characteristic of the electrospray process. The morphology of this film depends among other things on droplet volume at impact and on evaporation rate of the solvent [21]. In Figures 2(c)-(d) (10, 15 wt% in DMF) and 3(b)-(d) (7, 10 and 15 wt% in DMF/acetone) fibers predominate, characteristic of electrospinning. In micrographs 2(c) (10 wt% in DMF) and 3(b) (7 wt% in DMF/acetone) the electrospun sheets present some small droplets, demonstrating that concentration was not yet ideal for the formation of homogeneous fibers. These results show that the boundary concentration between electrospray and electrospinning ranges from 7 to10 wt% PVDF in DMF and from 5 to 7 wt% in



248

(a) (b)

(c) (d)

Figure 2. Micrographs of the samples processed from PVDF/ DMF solution at the following concentrations (wt%): (a) 5; (b) 7; (c) 10 and (d) 15.

(a) (b)

(c) (d)

Figure 3. Micrographs of samples processed from solution of PVDF with DMF/acetone (3:1 v/v) at the following con-centrations (wt%): (a) 5; (b) 7; (c) 10 and (d) 15. DMF/acetone mixture, at the conditions used of voltage and distance between needle and collector. The solution containing acetone is less viscous than that of pure DMF and therefore the electrospinning process should be initi- ated at a higher concentration. However, the high volatil- ity of acetone increased the jet concentration rapidly during its course towards the collector, impeding formation of droplets and resulting in electrospinning at a lower initial concentration. Micrographs 2(d) (15 wt% in DMF) and 3(c)-(d) (10 and 15 wt% in DMF/acetone) show fibers with improved homogeneity and no droplets, indicating that these concentrations are the optimum for obtaining electrospun meshes.

Figures 4 and 5 contain the size distribution of fibers diameters prepared with 15 wt% PVDF in DMF and DMF/acetone, respectively.

Figure 4. Size distribution of the fibers diameters processed from 15 wt% PVDF/DMF solution.

Figure 5. Size distribution of the fibers diameters processed from 15 wt% PVDF/DMF/acetone.

Average diameter of the nanofibers prepared from DMF solution (514 nm) were superior to that obtained from DMF/acetone solution (330 nm), at the same concentration. This has been caused by the lower viscosity of the acetone solution, since average fiber diameter increases with solution viscosity [12]. That effect can be observed in Figures 2(c)-(d) and 3(b)-(d).

The phases present in the samples were analyzed by FTIR and results are presented in Figures 6 and 7 for DMF and DMF/acetone solutions at different concentrations, respectively.

All samples presented bands at 445, 510 and 840 cm-1, characteristic of the phase of PVDF [22,23]. Therefore, the β phase predominated both in the electrosprayed films and the electrospun meshes. Since no stretching occurred during electrospray, we may conclude that for



249

Figure 6. FTIR spectra of the samples with different wt% PVDF in DMF solution. The characteristic bands of the α and β phase are pointed out in the figure.

Figure 7. FTIR spectra of the samples with different wt% PVDF in 3:1 DMF/acetone solution. The characteristic bands of the α and β phase are pointed out in the figure. mation of this phase can not be attributed to drawing experienced by the polymer during the process. The oc- currence of weak bands at 614, 764 and 976 cm-1 indicate the presence of a small amount of the phase [22,23], observed at concentrations of 5 to 10 wt % in DMF/acetone solution. In the solution containing pure DMF the presence of the phase was only observed for 5 wt%. Since solvent evaporation rate increases with decreasing concentration of the solution, appearance of this phase may be related to the higher solvent evaporation rate in low-concentration solutions. The evaporation rate of the solutions containing acetone increased in relation to those containing pure DMF, since acetone has a lower evapo-

ration temperature, and this resulted in the formation of a small amount of in almost all concentrations. Thus, the formation of the β phase in the electrospinning/electrospray process is related mainly to the solvent evaporation rate, as verified in cast films [8,9], and not to drawing experienced by the polymer during the electro-spinning process, as suggested by some authors [13-17]. Low evaporation rates result predominantly in the phase, thermodynamically more favorable, intermediate rates in a mixture of and and high rates in the phase, kinetically more favorable.

Figure 8 shows DSC curves for samples prepared with 5 and 10 wt% PVDF in DMF and DMF/acetone. The DSC measurements give the melting point (Tm) (endo- therm peak), enthalpy of fusion (H), and degree of crystallinity (%C) of samples. These values are summarized in Table 2.

It is easy to observe that meshes produced by electrospinning display a degree of crystallinity higher when compared to films obtained by electrospray. This result probably happens due to the preferential orientation of the chains in the direction of the nanofibers, caused by the drawing experienced by the polymer during the electrospinning and that facilitates the crystallization. In this process the addition of acetone promoted a small increase in the degree of the mesh crystallinity. The melting points, in all cases correspond to the temperature range related to the PVDF α and β phases [7,23], which confirm FTIR results.

4. Conclusions

The current investigation showed that formation of the β crystalline phase of PVDF in the electrospun meshes is

Figure 8. DSC curves for samples prepared with 5 wt% PVDF in DMF (a) and DMF/acetone (b); with 10 wt% PVDF in DMF (c) and DMF/acetone (d).



250

Table 2. Melting point (Tm), enthalpy of fusion (H) and degree of cristallinity (%C) of the samples processed with PVDF/DMF and PVDF/DMF/acetone solutions at different concentrations.

Sample Process Tm () H (J/g) %C

5 wt% PVDF/DMF 166 46.2 44.6

5 wt% PVDF/DMF/acetone Electrospray

167 46.8 45.2

10 wt% PVDF/DMF 168 53.4 51.1

10 wt% PVDF/DMF/acetone Electrospinning

167 55.0 53.1

not related to drawing experienced by the polymer during the process, as suggested by some authors. Formation of the and phase is related primarily to the evaporation rate of the solvent. High rates favor formation of the phase, whereas low rates favor the phase. Under the same conditions of voltage and needle-collector distance, the boundary concentration between electrospray and electrospinning depends on the solvent used. For PVDF this concentration lies between 7 and 10 wt% with DMF as solvent and between 5 and 7 wt% for the 3:1 DMF/ acetone mixture. Moreover, addition of acetone to the solution reduced average fiber diameter. Meshes produced by electrospinning present degree of crystallinity higher than the films obtained by electrospray, for both used solvents.

5. Acknowledgements

The following agencies are acknowledged for financial support: CAPES, CNPq and FAPESP.

REFERENCES [1] K. Tashiro, “Crystal Structure and Phase Transition of

PVDF and Related Copolymers,” In: H. S. Nalwa, Ed., Ferroelectric Polymers: Chemistry, Physics and Applica- ons, Marcel Dekker, New York, 1995, pp. 63-180.

[2] R. Gregorio Jr., and E. M. Ueno, “Effect of Crystalline Phase, Orientation and Temperature on the Dielectric Properties of Poly(Vinylidene Fluoride) (PVDF),” Jour-nal of Materials Science, Vol. 34, 1999, pp. 4489-4500.

[3] V. Sencadas, V. M. Moreira, S. Lanceros-Mendez, A. S. Pouzada and R. Gregorio Jr., “Alpha-to-Beta Transforma-tion on PVDF Films Obtained by Uniaxial Stretch,” Ad-vanced Materials Forum III, Vol. 514-516, 2006, 872- 876.

[4] P. Sajkiewicz, A. Wasiak and Z. Goclowski, “Phase Tran-sitions during Stretching of Poly(Vinylidene Fluoride),” European Polymer Journal, Vol. 35, No. 3, pp. 423-429.

[5] D. T. Grubb and F. R. Kearney, “Phase Transitions dur-ing Stretching of Poly(Vinylidene Fluoride),” Journal of Polymer Science Part B: Polymer Physics, Vol. 28, No. 11, 1990, pp. 2071-2078.

[6] M. C. Branciforti, V. Sencadas, S. Lanceros-Mendez and R. Gregorio Jr., “New Technique of Processing Highly

Oriented Poly(Vinylidene Fluoride) Films Exclusively in the Phase,” Journal of Polymer Science Part B: Poly-mer Physics, Vol. 45, 2007, pp. 2793-2801.

[7] R. Gregorio Jr., and M. Cestari, “Effect of Crystallization Temperature on the Crystalline Phase Content and Mor-phology of Poly(Vinylidene Fluoride),” Journal of Poly-mer Science Part B: Polymer Physics, Vol. 32, No. 5, pp. 859-870.

[8] R. Gregorio Jr. and D. S. Borges, “Effect of Crystalliza-tion Rate on the Formation of the Polymorphs of Solution Cast Poly(Vinylidene Fluoride),” Polymer, Vol. 49, No. 18, 2008, pp. 4009-4016.

[9] D. L. Chinaglia, R. Gregorio Jr., J. C. Stefanello, R. A. P. Altafin, W. Wirges, F. Wang and R. Gerhard, “Influence of the Solvent Evaporation Rate on the Crystalline Phases of Solution-Cast Poly(Vinylidene Fluoride) films,” Jour-nal of Applied Polymer Science, Vol. 116, No. 2, 2010, pp. 785-791.

[10] A. Formhals, “Process and Apparatus for Preparing Artificial Threads,” U.S. Patent 1,975,504, 1934.

[11] J. Doshi and D. H. Reneker, “Electrospinning Process and Application of Electrospun Fibers,” Journal of Electro- statics, Vol. 35, No. 2-3, 1995, pp. 151-160.

[12] M. Nasir, H. Matsumoto, T. Danno, M. Minagawa, T. Irisawa, M. Shioya and A. Tanioka, “Control of Diameter, Morphology, and Structure of PVDF Nanofiber Fabri- cated by Electrospray Deposition,” Journal of Polymer Science Part B: Polymer Physics, Vol. 44, No. 5, pp. 779-786.

[13] W.A. Yee, M. Kotaki, Y. Liu and X. Lu, “Morphology, Polymorphism Behavior and Molecular Orientation of Electrospun Poly(Vinylidene Fluoride) Fibers,” Polymer, Vol. 48, No. 2, 2007, pp. 512-521.

[14] E. Ogut, O. S. Yordem, Y. Z. Menceloglu and M. Papila, “Poly(Vinylidene Fluoride)/Zinc Oxid Smart Composite Material,” Proceeding of SPIE, Vol. 6526, 2007, p. 65260Q-1-10

[15] J. Zheng, A. He, J. Li and C. C. Han, “Polymorphism Control of Poly(Vinylidene Fluoride) through Electro-spinning,” Macromolecular Rapid Communications, Vol. 28, No. 22, 2007, pp. 2159-2162.

[16] W. A. Yee, A. C. Nguyen, P. S. Lee, M. Kotaki, Y. Liu, B. T. Tan, S. Mhaisalkar and X. Lu, “Stress-Induced Struc-tural Changes in Electrospun Polyvinylidene Difluoride Nanofibers Colleted Using a Modified Rotating Disk,” Polymer, Vol. 49, No. 19, 2008, pp. 4196-4203.



251

[17] J. S. Andrew and D. R. Clarke, “Effect of Electrospin- ning on the Ferroelectric Phase Content of Polyvinylidene Difluoride Fibers,” Langmuir, Vol. 24, No. 3, 2008, pp. 670-672.

[18] H. Na, Y. Zhao, C. Zhao, C. Zhao and X. Yuan, (2008) Effect of Hot-Press on Electrospun Poly(Vinylidene Fluo- ride) Membranes,” Polymer Engineering and Science, Vol. 48, No. 5, pp. 934-940.

[19] G. Taylor, “Electrospray,” Proceedings of Royal Society of London, Vol. A280, 1964, pp. 383-387.

[20] K. Nakagawa and Y. Ishida, “Dielectric Relaxations and Molecular Motions in Poly(Vinylidene Fluoride) with Crystal Form II,” Journal of Polymer Science Part B:

Polymer Physics, Vol. 11, No. 8, pp. 1503-1533.

[21] I. B. Rietveld, K. Kobayashi, H. Yamada and K. Matsu- shige, “Model Supported Morphology Control of Elec-trospray Deposited Poly(Vinylidene Fluoride) Film,” Macromolecular Symposium, 2007, pp. 249-250, 322- 329.

[22] M. Kobayashi, K. Tashiro and H. Tadokoro, “Molecular Vibrations of Three Crystal Forms of Poly(Vilidene Fluo-ride),” Macromolecules, Vol. 8, No. 2, 1975, pp. 158-170.

[23] R. Gregorio Jr., “Determination of the , and Crystal-line Phases of Poly(Vinylidene Fluoride) Films Prepared at Different Conditions,” Journal of Applied Polymer Sci-ence, Vol. 100, No. 4, 2006, pp. 3272-3279.




Amelia Carolina Sparavigna1, Luca Florio2, Jamshid Avloni3, Arthur Henn4

1Physics Department, Politecnico di Torino, Torino, Italy; 2Laboratorio Tessili Innovativi, Biella, Italy; 3Eeonyx Corporation, Pinole, USA; 4Marktek Inc., Chesterfield, USA. Email: [email protected] Received July 7th, 2010; July 26th, 2010; July 27th, 2010.

ABSTRACT

Polypyrrole can be chemically synthesized on PET fabrics, giving rise to textiles with high electric conductivity. These textiles are suitable for several applications from antistatic films to electromagnetic interference shielding devices. Here we discuss the thermal-electric performance and the heat generation of polypyrrole coated PET fabric samples, previously studied because of their electric conductivity and electromagnetic interference shielding effectiveness. The measured Seebeck effect is comparable with that of metallic thermocouples. Since polypyrrole shows extremely low thermal diffusivities regardless of the electrical conductivity, the low thermal conductivity gives significant advantage to the thermoelectric figure-of-merit ZT, comparable with that of some traditional inorganic thermoelectric materials. The heat generation is also investigated for possible heating textile devices. The results confirm polypyrrole as a promising material for thermal electric applications due to its easy preparation in low cost processing. Keywords: Conductive Polymers, Polypyrrole, Thermoelectric Materials

1. Introduction

In the late 1970s, MacDiarmid, Heeger and Shirakawa discovered how to get polymers conducting electricity.1 The first material becoming an intrinsically conducting polymer (ICP) was polyacetylene, after a doping with iodine. The announcement of this discovery quickly re- verberated around scientific community, and intensity of the research for other conducing polymers magnified dramatically [1-3]. A new generation of polymers was then developed, exhibiting the electrical and optical properties of metals or semiconductors, at the same time retaining the attractive mechanical properties and processing advantages of polymers.

Intrinsically conducting polymers were immediately seen as a new route to mimic metallic conductivity, besides the well-know approach to insert conductive fillers into an inherently insulating resin, or to coat a plastic substrate with a conductive metal solution [4]. In this manner, conductive fibers can be prepared to obtain conductive fabrics, or, fabrics already produced can be metalized with a conductive coating. By the way, let us

note that metal coated textiles remain fundamental materials, because they generally show a high electromagnetic interference shielding effectiveness (EMI-SE) [5,6].

In fact, intrinsically conducting polymeric materials can be used to obtain rather innovative textiles [6]. These textiles are able to absorb as well as reflect electromagnetic waves, and then can exhibit certain advantages over metallic materials. Actually, the most prominent ICPs in EMI-SE are polypyrrole and polyaniline, where electrical conductivity can have values comparable to those observed for poorly conducting metals and alloy [5-8].

Among the first commercial products incorporating polypyrrole there was Contex®, a conductive textile product originally manufactured by Milliken [9], starting around 1990, and now produced by Eeonyx Corp., as EeonTex™. An early application, involving the coating of polyester fibers with polypyrrole (PPy) was the creation of an antistatic fabric. Here, we will show the results of measurements of thermal electric properties of the PPy-coated EeonTex. In particular, we deduce its Seebeck coefficient, which turns out to be comparable with that of metal thermocouple materials. Because polypyrrole shows extremely low thermal diffusivity, the value of the figure-of-merit turns out to be of the same

1The Nobel Prize in Chemistry 2000 was awarded jointly to Alan J. Heeger, Alan G. MacDiarmid and Hideki Shirakawa “for the discovery and development of conductive polymers”.



253

order of magnitude of some traditional inorganic thermoelectric materials [10].

Thermoelectric effects in ICPs are deserving a special attention, due to the considerable effort actually paid to have thermoelectric effects from low-weight but high- reliable materials. The reason is that the thermoelectric cooling is a good strategy in semiconductor electronics operating at high frequencies, where the thermal man- agement becomes crucial. Let us remember that, according to Kelvin relations between Seebeck and Peltier coefficients, an evaluation of the Seebeck effect allows estimating the cooling power of materials.

The target of this paper is then in investigating the thermoelectric effects in PPy coated samples, which have been previously studied from the point of view of their electric conductivity and electromagnetic interference shielding effectiveness [5,6], We will also discuss the heat generation obtained from PPy coated fabrics, for suitable applications in textile heating systems. Polypyr- role is in fact one of ICPs very promising for wide thermal electric applications because of its easy preparation with a low cost processing.

2. Samples Preparation

A PPy coated textile could represent a possible solution for heating and cooling and for temperature monitoring. As previously told, one of the first commercial textile products incorporating conductive polypyrrole was the Contex conductive textile. This textile evolved in a new material, with a modified PPy coating, more conductive and thermally stable. While imparting electrical conductivity and a dark color to the substrates, the PPy coating process barely affects the strength, drape, flexibility, and porosity of the starting substrates.

For the measurements discussed in this paper, we used an EeonTex PPy-coated PET fabric which was prepared similarly as described previously in [11,12], with raw chemicals purchased from Sigma-Aldrich and used without further purification. Stochiometric molar ratio of organic acid dopant, anthraquinone-2-sulfonic acid to pyrrole-monomer (i.e., 0.33:1) was used to ensure complete doping level. The molar ratio of polymerisation catalyst, iron (III) nitrate, to monomer (pyrrole) equal to 2.3 mol/ mol was used for all reactions. The macroscopic texture of this fabric is shown in Figure 1: it is a net with a structure quite useful for application in heating systems, as we shall demonstrate in Section 4 of this paper.

Simultaneous in-situ polymerization and deposition of conductive polypyrrole leads to production of conductive, smooth and uniform coating with thickness under 1 m, according to transmission electronic microscope measurements (see Figure 2). As observed in Ref. 13, it is possible a formation of insoluble polymers in the bulk

Figure 1. Polypyrrole coated PET net. The image sizes are 4.3 cm 5 cm.

solution and on the surface of the substrate simultaneously. The bulk polymerization produces dendritic polymer particles in the solution and the surface polymerization forms a polymer film on the substrate surface. Some of the bulk polymer precipitates on the surface of the substrate and then the SEM analysis shows these particles on the fibers.

Figures 3 and 4 are reproducing scanning electron microscope SEM images of PPy coated fibers in a non-woven textile sample and in a twill textile sample, re-spectively. These materials possess a good conductivity and then are quite useful for electromagnetic shielding applications, as reported in Ref. 5 and 6. In that paper, the performances of PPy coated fabrics were compared with that of a leno nylon fabric, metalized with a Ni/Ag coating.

In the following section we will discuss the Seebeck effect of the PPy-coated net shown in Figure 1 and of the leno Ni/Ag metalized textile, for comparison. For what concerns the heat generation, this will be discussed in Section 4 for the PPy coated net. The metalized leno sample is not useful due to its low resistance. In fact, the PPy/PET net has a DC surface resistivity of 306.0 Ω/sq, whereas the metalized leno has a resistance of 0.22 Ω/sq. The electrical DC surface resistances were measured by using a four-in-line point probe in combination with computerized Loresta-AP meter from Mitsubishi Petro-chemical Co., LTD.

3. Thermal-Electric Effects

Researches on organic materials for thermoelectric applications have not been so attractive, probably because of their poor electronic transporting characters, till the



254

Figure 2. Scanning electron microscope SEM images of the PPy-coated PET net shown in Figure 1. The dust is due to polymer particles in solution, deposited on the fibers of the net. In the lower panel of the image a detail of fibers. discovery of ICPs. Some of these electrically conducting polymers have gained attention because of their considerable thermal stability. These materials, which are polyaniline and polypyrrole, are considered suitable for ap-plications to electronic devices and sensors [14-17].

For thermal sensors, a systematic investigation on thermoelectric performances of polyaniline [18] and polypyrrole [19] is then interesting. In fact, supposing for polymeric compounds a low thermal conductivity, we can obtain significant advantage of the thermoelectric figure-of-merit. Let us remember that the figure-of-merit is defined as 2 /ZT S T ( ) , where S, , and T are Seebeck coefficient, electric conductivity, thermal conductivity, and absolute temperature, respectively. The value of ZT for polypyrrole is comparable with that of

Figure 3. SEM images of PPy coated fibers in a nonwoven textile sample.

Figure 4. SEM images of PPy coated fibers in a twill textile sample. some traditionally used inorganic thermoelectric materials [8,16]. Besides polyaniline and polypyrrole, other materials, such as polythiophene has been recently invest- tigated, in the form of nanofilms too [20-21].

We investigated the behavior of the electrical resistive- ity as a function of temperature. Starting from room temperature, the resistance of a PPy/PET sample placed in a



255

thermostage, was checked till a temperature of 70. The resistance behavior with temperature is typical of a semiconductor with the resistance decreasing linearly as the temperature increases. At room temperature, the resistance was of 172 in a sample with a length of 4 cm, composed by 10 yarns, each yarn with a diameter of 0.05 cm. At 70°C, the resistance of the sample was of 145 . Assuming as in Ref.23 a thickness in polypyrrole coating of around 1 m, the electric conductivity turns out to be 4 1 110 m . This estimation of the electrical conductivity of PPy coating is in good agreement with the value of 4 1 11.7 10 m , given in Ref. 17.

Thermoelectric Seebeck coefficient ( S ) and its temperature dependence were determined by connecting a stripe of PPy/PET net 0.5 cm wide with a copper wire. The two materials are electrically connected by the pressure of a very small silver clip, insulated from the junction. The hot junction was placed in the thermostage with a reference Chromel/Alumel thermocouple and the cold junction between the PPy/PET stripe and copper was thermally anchored at room temperature (26). The same anchoring was used for the cold junction of the Chro- mel-Alumel thermocouple (a diagram of the experiment- tal set-up in Figure 5). Variations in monitored room temperature during measurements were negligible (around 5%).

In Figure 6, the behavior of the electro-motive force measured for two such PPy/Copper thermocouples is given as a function of temperature difference T between the actual hot junction temperature and the room temperature. Assuming a value of Copper e.m.f. vs. Pla- tinum of 0.0076 mV/K [24], we can estimate a value of 0.0133 mV/K for the PPy vs. Platinum e.m.f. and posi-tive.

To obtain the figure-of-merit ( ZT ), we estimate the PPy thermal conductivity in the following manner. Ther- mal diffusivity of PPy films was deduced from measurements by the laser flash method [25]. In Ref. 25, PPy films exhibit a thermal diffusivity ranging from 4 to

7 2 18. 10 m s at room temperature (even lower than the value of 6 2 11.3 10 m s measured for polyaniline [26]). The very low thermal diffusivity of polypyrrole films is originated in the lattice structure, in particular from a dominant amorphous character of the chain structure. An amorphous structure strongly reduces the thermal phonon transport, because strong phonon scattering mechanisms appear [27,28].

We can then assume a value of thermal diffusivity of about 7 2 15. 10 m s and a specific heat capacity

1 10.4 J g Kc , in the range of the values for organic polymers [26]. The thermal conductivity is then given by

c . With a density of 3 31.3 10 kg / m [29], the thermal conductivity of polypyrrole film turns out to

Figure 5. Diagram of the experimental set–up for measur-ing the Seebeck effect. The electromotive force of two ther-mocouples are compared: a thermocouple is used as a ref-erence to determine the temperature in the thermostage, the other to determine the unknown electromotive force of the material.

Figure 6. Electromotive force measured for a PPy-PET/ Copper thermocouple as a function of the temperature dif-ference T between the actual hot junction temperature and the room temperature. The figure reports two diffeent scanning in temperature. be 1 10.3 W m K . This low thermal conductivity is at least in one order of magnitude lower than that of the best inorganic thermoelectric materials. Using 4 1 110 m ,

1 10.3 W m K and our estimate of Seebeck coeffi-cient, we obtain 31.5 10ZT at 300 KT in agreement with data of polypyrrole films [17].

The power factor 2S is approximately 6 1 22. 10 W m K . This power factor is better than

those reported for polypyrrole in Ref. 30. Note from Figure 6 the linear behaviour of the ther-

moelectric power with respect to temperature: this behaviour has been observed in highly doped polyacetylene and in some cases for PANi [30].

Investigations of the electromotive forces with non- woven and twill samples of Figures 3 and 4 were also performed, but in this case, the results of measurements are questionable due to a strong dependence on the contact between textile and wires. The best result we obtained was



256

of 0.008 mV/K for the non-woven sample, with a linear behavior with temperature. The nonwoven sample had a DC surface resistivity of 30.0 Ω/sq: we could estimate a quite optimistic value of the electric conductivity of

5 1 110 m . Assuming the value 0.005 mV/K as a possible estimate of Seebeck coefficient, we could reach a power factor 2S of 5 1 20.5 10 W m K , and a figure-of-merit 34.8 10ZT at a temperature of

300KT . The Seebeck electromotive force of the nylon leno

sample, coated Ni/Ag, connected with Cu was also mea-- sured and the behavior is shown in Figure 7 (curve b). The same figure shows that a thermocouple built with Ni/Ag/Nylon and PPy/PET can give the higher electromotive force (curve a). We have also prepared a thermocouple with PPy/PET and a yarn composed of comer- cial carbon fibers (curve c).

As shown by the measurements here reported, PPy coating can be successfully used with other conductive yarns, for instance Copper, Ni/Ag coated yarns or carbon fibers, to obtain stable textile thermopiles, which are able to provide thermo-powers as typical semiconductors. The proposal of use coated fabrics in thermal-electric devices is not new [31]. Let us remark that we used a commercial PPy-coated textile, with a stable coating, not an ad-hoc prepared film, and this, in our opinion, is relevant for industrial applications.

4. Heat Generation from Fabrics

A possible application of the PPy-coated net, due to its fabric structure, is in heating devices. The following set- up for detecting heat generation is used [32,33]. This set-up was previously developed to study polypyrrole samples prepared in Biella laboratory for innovative textiles. Those samples were too dusty to allow them to be used in applications. An adjustable Variac power supply was used to generate an AC current/voltage over the fabric. Voltage and current were monitored by Keithley voltmeter and amperometer. A square shape fabric (6 cm 6 cm) was positioned between two pressed electric

contacts (a diagram of the arrangement in Figure 8). The temperature rise was measured using an Omega infrared thermometer, placed to control the center of the sample.

In Figure 9, the behavior of the temperature as a function of the current is given, with the rise of the voltage. According to the power law, the maximum theoretical power achieved from the fabrics is: P = VI, where P is the power developed and V,I the voltage and current. The AC current frequency is 50 Hz. In Figure 10, the power and the impedance as a function of current are shown.

In Ref. 12, the power density per unit area is assumed to be: 2 2/ SP V R l , where SR is the surface resis-tance and l the size of the sample. Our highest value is

Figure 7. Electromotive force measured for Ni/Ag/Nylon/ Contex. (a) for Ni/Ag/Nylon/Copper, (b) and for PPy/PET/ Carbon Fibers; (c) as a function of the temperature difference T between the actual hot junction temperature and cold junction at room temperature.

Figure 8. Experimental set-up diagram for measuring the heating effect of a textile. Current and voltage across the sample must be monitored.

Figure 9. Behavior of the voltage and temperature of the PPy coated sample as a function of the current, measured with voltmeter and thermometer.



257

Figure 10. Behavior of the impedance and power developed by the PPy coated sample as a function of the current. The values of impedance and power are estimated from data of Figure 8. 370 2W / m in agreement with the value obtained in Ref.12.

5. Conclusions

The conversion electricity-heat has always attracted a great attention because of applications in heaters, coolers and thermoelectric power generators. The parameter measuring the suitability of a material for these applications is the figure-of-merit. In order to have high values of the figure-of-merit, a material must have high charge transport conductivity, high Seebeck coefficient and low thermal conductivity. The intrinsically conducting polymers can be considered as a new generation of thermoelectric materials, due to their characteristic that often are achieving a figure-of-merit comparable with that of typi-cal semiconductors. Other attractive features are the low cost of material resources, an easy synthesis and proc-essing into desired forms.

Many conducting polymers have been investigated as thermal materials, among them polyaniline, polythio- phenes and polypyrroles. Here we have studied the properties of a commercially available polypyrrole coated fabric. As shown by measurements, a Seebeck effect can be achieved by using a PPy conducting coating of a PET fabric. According to the Kelvin relation between Seebeck S and Peltier coefficients, ST , we can also ima- gine a possible application in cooling devices of polypyrrole coated fabrics.

We have also seen that with PPy/PET fabrics, it is possible to easily make heating fabrics. Since the coating with polypyrrole is possible on many different fibers [34], the potential applications of polypyrrole in the building

of heating pads is relevant. We suggest then that PPy- coated fabrics may be practically useful for many applications, including flexible, portable surface-heating elements for medical or other applications.

6. Acknowledgements

Authors thank Angelica Chiodoni for SEM analysis.

REFERENCES [1] Y. Cao, P. Smith, A. J. Heeger, “Counter-Ion Induced

Processibility of Conducting Polyaniline,” Synthetic Met-als, Vol. 57, No. 1, 1993, pp. 3514-3519.

[2] C. K. Chiang, C. R. Fincher, Y. W. Park, A. J. Heeger, H. Shirakawa, E. J. Louis, S. C. Gau and A. G. MacDiarmid, “Electrical Conductivity in Doped Polyacetylene,” Phys-ics Review Letters, Vol. 39, No. 17, 1997, pp. 1098- 1101..

[3] A. G. MacDiarmid, “Polyaniline and Polypyrrole: Where are we Headed?” Synthetic Metals, Vol. 84, No. 1-3, 1997, pp. 27-34.

[4] J. Edenbaum, Ed., “Plastics Additives and Modifiers Handbook,” Van Nostrand Reinhold, New York, 1992.

[5] J. Avloni, R. Lau, M. Ouyang, L. Florio, A. R. Henn and A. Sparavigna, “Shielding Effectiveness Evaluation of Metallized and Polypyrrole-Coated Fabrics,” Journal of Thermoplastic Composite Materials, Vol. 20, No. 3, 2007, pp. 241-254.

[6] J. Avloni, M. Ouyang, L. Florio, A. R. Henn and A. Sparavigna, “Polypyrrole-Coated Nonwovens for Elec- tromagnetic Shielding,” Journal of Industrial Textiles, Vol. 38, No. 3, 2008, pp. 55-68.

[7] A. R. Henn and R. M. Cribb, “Modelling the Shielding Ef- fectiveness of Metallized Fabrics,” Interference Technology Engineering Master (ITEM) Update, 1993, pp. 49-57.

[8] Y. Y. Wang and X. L. Jing, “Intrinsically Conducting Po- lymers for Electromagnetic Interference Shielding,” Polymers for Advanced Technologies, Vol. 16, No. 4, 2005, pp. 344-351.

[9] Milliken and Co., Milliken. http://www.milliken.com

[10] A. F. Ioffe, “Semiconductor Thermoelements and Ther- moelectric Cooling,” Infosearch, London, 1957.

[11] H. H. Kuhn and A. D. Child, “Handbook of Conducting Polymers,” Marcel Dekker, New York, 1998.

[12] H. H. Kuhn, A. D. Child and W. C. Kimbrell, “Toward Real Applications of conductive Polymers, Synthetic Metals,” Vol. 71, No. 1-3, 1995, pp. 2139-2142.

[13] E. Hakansson, A. Kaynak, T. Lin, S. Nahavandi, T. Jones and E. Hu, “Characterization of Conducting Polymer Coated Synthetic Fabrics for Heat Generation,” Synthetic Metals, Vol. 144, No. 1, 2004, pp. 21-28.

[14] D. Kumar and R. C. Sharma, “Advances in Conductive Polymers,” European Polymer Journal, Vol. 34, No. 8, 1998, pp. 1053-1060.

[15] D. Braun, “Semiconducting polymer LEDs,” Materials Today, Vol. 5, No. 6, 2002, pp. 32-39.



258

[16] J. V. Hatfield, P. Neaves, K. Persaud and P. Travers, “Towards an Integrated Electronic Nose Using Conduct-ing Polymer Sensors,” Sensors and Actuators B, Vol. 18, No. 1-3, 1994, pp. 221-228.

[17] H. Yan, T. Ishida and N. Toshima, “Thermoelectric Prop-erties of Electrically Conductive Polypyrrole Film, Ther-moelectrics, 2001,” Proceedings ICT 2001, XX Interna-tional Conference on Thermoelectrics, Beijing, 2001, pp. 310-313.

[18] F. Yakuphanoglu and B. F. Şenkal, “Electronic and Thermoelectric Properties of Polyaniline Organic Semi-conductor and Electrical Characterization of Al/PANI MIS Diode,” Journal of Physical Chemistry C, Vol. 111, No. 4, 2007, pp. 1840-1846.

[19] A. B. Kaiser, O. Mercier, H. J. Trodahl, N. T. Kemp, C. J. Liu, B. Chapman, A. M. Carr, R. G. Buckley, A. C. Par- tridge, J. Y. Lee, C. Y. Kim, A. Bartl, L. Dunsch, W. T. Smith and J. S. Shapiro, “Thermoelectric Power and Con- ductivity of Different Types of Polypyrrole,” Journal of Polymer Science: Part B: Polymer Physics, Vol. 37, No. 9, 1999, pp. 953-960.

[20] M. Scholdt, H. Do, J. Lang, A. Gall, A. Colsmann, U. Lemmer, J. D. Koenig, M. Winkler and H. Boettner, “Or- ganic Semiconductors for Thermoelectric Applications,” Journal of Electronic Materials, Springer, Boston, 2010.

[21] Y. Shinohara, K. Hiraishi, H. Nakanishi, Y. Isoda and Y. Imai, “Study on Thermoelectric Properties of Conductive Polymers,” Transactions on Materials Research Society of Japanese, Vol. 30, No. 4, 2005, pp. 963-969.

[22] B. Y. Lu, C. C. Liu, S. Lu, J. K. Xu, F. X. Jiang, Y. Z. Li and Z. Zhang, “Thermoelectric Performances of Free- standing Polythiophene and Poly(3-Methylthiophene) Nanofilms,” Chinese Physics Letters, Vol. 27, No. 5, 2010, pp. 057201.1-057201.4.

[23] A. R. Henn, “Calculating the Surface Resistivity of Con-ductive Fabrics,” Interference Technology Engineering Master (ITEM) Update, 1996, pp. 66-72.

[24] D. E. Gray, “American Institute of Physics Handbook,” McGraw-Hill, New York, 1963.

[25] M. Y. Lim, W. M. M. Yunus, A. Kassim and H. N. M. E.

Mahmud, “Photoacoustic Measurement of Thermal Diffu-sivity of Polypyrrole Conducting Polymer Composite Films,” American Journal of Applied Sciences, Vol. 6, No. 2, 2009, pp. 313-316.

[26] H. Yan, N. Sada and N. Toshima, “Thermal Transporting Properties of Electrically Conductive Polyaniline Films as Organic Thermoelectric Materials,” Journal of Thermal Analysis and Calorimetry, Vol. 69, No. 3, 1992, pp. 881- 887.

[27] M. Omini and A. Sparavigna, “Role of Grain Boundaries as Phonon Diffraction Gratings in the Theory of Thermal Conductivity,” Physics Review B, Vol. 61, No. 10, 2000, pp. 6677-6688.

[28] A. Sparavigna, M. Omini, A. Pasquarelli and A. Strigazzi, “Thermal Diffusivity and Conductivity in Low-Con- ducting Materials: A New Technique,” International Journal of Thermophysics, Vol. 13, No. 2, 1992, pp. 351- 360.

[29] J. Tietje-Girault, C. Ponce de León, F. C. Walsh, “Elec-trochemically Deposited Polypyrrole Films and their Characterization,” Surface & Coatings Technology, Vol. 201, No. 12, 2007, pp. 6025-6034.

[30] A. Shakouri and S. Li, “Thermoelectric Power Factor for Electrically Conductive Polymers,” Proceedings of In-ternational Conference on Thermoelectrics, Baltimore, September 1999.

[31] E. Hu, A. Kaynak and Y. Li, “Development of a Cooling Fabric from Conducting Polymer Coated Fibres: Proof of Concept,” Synthetic Metals, Vol. 150, No. 2, 2005, pp. 139-143.

[32] L. Florio, “Dissertation for Master Degree, Studio dei Polimeri Intrinsecamente Conduttori: Applicazione al tessile Elettronico (in Italian),” Politecnico di Torino, 2004.

[33] L. Florio and A. Sparavigna, “Thermoelectric Properties of Conducting Polymers,” INF Meeting, National Con-ference on the Physics of Matter, Genova, 2004.

[34] D. T. Seshadri and N. V. Bhat, “Synthesis and Properties of Cotton Fabrics Modified with Polypyrrole,” Sen’I Gakkaishi, Vol. 61, No. 4, 2005, pp. 103-108.

Documents

Volume 1 Number 4 October 2010 - Scientific Research Publishing · 2011. 9. 7. · used to study the temperature dependent conductivity by the ITC 502S Oxford temperature controller