Temporal evolution of the laser-induced breakdown spectroscopy spectrum of aluminum metal in different bath gases

Temporal evolution of the laser-inducedbreakdown spectroscopy spectrum of aluminummetal in different bath gases

Thuvan N. Piehler, Frank C. DeLucia, Jr., Chase A. Munson, Barrie E. Homan,Andrzej W. Miziolek, and Kevin L. McNesby

The spectral emission of gas-phase aluminum and aluminum oxide was measured during and immedi-ately after exposure of a bulk-aluminum sample to a laser-induced spark produced by a focused, pulsedlaser beam (Nd:YAG, 10�ns pulse duration, 35 mJ�pulse, � � 1064 nm). The spectral emission wasmeasured as a function of time after the onset of the laser pulse, and it was also measured in differentbath gases (air, nitrogen, oxygen, and helium). © 2005 Optical Society of America

OCIS codes: 140.3440, 300.6360.

1. Introduction

Aluminum is a common ingredient of explosives andpropellants. In explosives, aluminum is used to aug-ment air blast, raise reaction temperature, and createincendiary effects.1 In rocket propellants, aluminumis used to increase thermal energy and elevate theflame temperature.2 A proposed mechanism for thecombustion of aluminum in oxygen is shown below3:

Al(l) → Al(g) �H � 317.7 kJ, (1)

AlO � Al(l) → Al2O �H � �224.0 kJ, (2)

Al(g) � O2 → AlO � O �H � 6.53 kJ, (3)

Al(g) � O � M → AlO � M �H � �492.1 kJ, (4)

AlO � O2 → AlO2 � O �H � �78.6 kJ, (5)

AlO � AlO2 → Al2O3(l) �H � �1557.2 kJ,(6)

Al2O � O2 → Al2O3(l) �H � �1562 kJ. (7)

For some explosive materials, aluminum may beused to tailor performance to specific needs. Measure-ments of relative amounts of aluminum metal �Al�and aluminum oxide �AlO� during explosions of en-ergetic materials may provide insight into increasingthe performance of aluminum-containing explosives.The experiments described here are a preliminarystudy of the application of laser-induced breakdownspectroscopy (LIBS) to this problem.

Although best known for high selectivity for metalsanalysis,4 laser-induced breakdown spectroscopy(LIBS) also has been used to detect energetic mate-rials,5 trace elements in liquids,6 organic compoundsin ambient air,7 and some biological materials.8 InLIBS, a pulsed laser focused onto a target materialconverts some of the material into a plasma of ionsand electrons, with temperatures that may approach20 000 K.9 As the plasma cools, some of the energyis radiated as light. When measured using a spectro-graph, the wavelengths of the emitted light are char-acteristic to the elemental components of the target,whereas the intensity of light over a given wave-length range may yield the proportion of that elementwithin the target material.10 Additionally, the timeevolution of the emission following the laser pulsemay be used to identify certain chemical reactionsoccurring in the plasma as it cools.

In this paper, we measure the emission from alaser-induced spark produced by focusing a pulsedNd:YAG laser onto the surface of an aluminum rod.The emission is spectrally and temporally resolved,the effect of different bath gases (air, O2, N2, and He)

The authors are with the U.S. Army Research Laboratory, Ab-erdeen Proving Ground, Maryland 21005-5066. T. Piehler can bereached at the e-mail address [email protected].

Received 29 July 2004; revised manuscript received 5 January2005; accepted 17 January 2005.

0003-6935/05/183654-07$15.00/0© 2005 Optical Society of America

3654 APPLIED OPTICS � Vol. 44, No. 18 � 20 June 2005

on the emission is measured, and temperature andelectron density are calculated.

2. Experiment

A schematic of the simple LIBS system used in thiswork is shown in Fig. 1. Briefly, a light pulse (�10 ns,35 mJ per pulse) from an actively Q-switchedNd:YAG laser (Big Sky Laser Technologies, Inc.,Bozeman, Montana) emitting at a wavelength of1064 nm was focused by a 50�mm convex lens ontothe surface of an aluminum rod (0.6 cm in diameter,3.5 cm in length). The laser beam was perpendicularwith respect to the Al rod cylinder axis, with the focuson the center of the rod at its apogee. Each laser pulsewas triggered manually. A Si–Si optical fiber(600��m core diameter) collected the emission fromthe plasma spark. A lens was placed in front of thefiber so that the light emitted by the plasma wasfocused onto the fiber surface within the acceptancesolid angles of the fiber optic. An echelle spectrometer(Catalina Scientific Corporation, Tucson, Arizona) fit-ted with a gated, intensified CCD camera (AndorTechnology, Model DH 734-18-03) was used to mea-sure the emitted light. The entire experiment, includ-ing background measurement, laser control, dataacquisition, and data processing, was controlled by alaptop computer (Dell).

Before the measurement of each LIBS spectrum, abackground spectrum was measured and subse-quently subtracted from the sample spectral data. Inan attempt to minimize errors from shot-to-shot vari-ations in the laser output power ��5%�, each spec-trum used in the data analysis is the average of 50“single shot” spectra. For each LIBS spectrum mea-sured, the aluminum rod was repositioned so onlynew sample was exposed to the laser-induced spark.To enable comparison with previous LIBS studies ofaluminum,11 a detector gate width of 2 �m was usedfor these experiments. Detector gate delays (relative

to the Q switch of the Nd:YAG laser) ranged from zeroto 30 �m. The composition of aluminum rods used inthis study was Al 91.4%, Cu 5.67%, Fe 1.28%,Li 1.11%, and minor constituents (Mg, Mn, Ti, andZn percentage less than 0.5% by weight). Bath gases(N2, O2, and He) were obtained from Matheson(industrial grade, ultrahigh purity, 99.9995%, re-spectively) and were used without any further puri-fication. Typical flow rates were approximately2 liters/min. The gas flow was delivered via 4 mm i.d.Tygon tubing. The exit port of the tubing was approx-imately 5 mm from the location of the plasma vol-ume.

3. Results and Discussion

A. Emission Spectra of an Aluminum Rod in Air

The most intense regions of the aluminum rod LIBSspectrum (bath gas, air [ambient]; gate width, 2 �s;and gate delay, 20 �s) are shown in Figs. 2, 3, and 4.The first spectral window (Fig. 2) from 300 to 420 nmincludes emission from gas-phase aluminum �Al I�at wavelengths of 308.34, 309.44, 394.56, and396.26 nm. The second spectral window (Fig. 3; 420

Fig. 1. Schematic of the experimental setup used to measureLIBS spectra.

Fig. 2. Portion of the LIBS spectrum of an aluminum rod in airwith a 20��s gate delay and a 2��s gate, in the 300 to 420 nmregion.


20 June 2005 � Vol. 44, No. 18 � APPLIED OPTICS 3655

to 580 nm) includes emission from the gas-phase mo-lecular species AlO, with the most intense emissionnear 484.58 nm. The third spectral window (Fig. 4;740 to 760 nm) includes emission from gas-phase alu-minum �Al II� at a wavelength of 747.14 nm. For alu-minum combustion in air, previous investigatorshave suggested that, above the melting point of Al2O3(2315 K12), the aluminum species with highest par-tial pressures are Al and AlO.

B. Temporal Evolution of Al and AlO Emission

Figure 5 shows the LIBS spectrum of an aluminumrod in air at various gate delays (gate width, 2 �s).The Al I line �396.2 nm� reaches its maximum inten-sity in air immediately following the laser pulse (ap-proximately one microsecond after the laser pulse,and is not represented in the figure). The band fromAlO emission �484.4 nm� reaches its maximum inten-sity approximately 20 �s after the laser shot. This isqualitatively consistent with the combustion mecha-nism for aluminum in oxygen, outlined in reactions1–7 above.

LIBS spectra of the aluminum rod (measured from350 nm to 580 nm), at various gate delays (gatewidth, 2 �s) for the bath gases oxygen, helium, andnitrogen are shown in Figs. 6, 7, and 8, respectively.

For comparison, the peak intensities of the Al I linesin each figure have been normalized with respect tothe maximum of the Al I 396.2 nm line intensity foreach gate pulse delay. It is worth noting that theemission near 484 nm (from AlO) in Fig. 7 (He bathgas) and Fig. 8 (N2 bath gas) is vanishingly smallcompared with the emission near 484 nm in Fig. 5(air bath gas) and Fig. 6 (O2 bath gas). AlO emissionobserved in helium and nitrogen bath gases was fromthe oxide coating on the rod surface.

Figure 9 shows a pseudo first-order plot of loga-rithm of intensity at 396 nm versus time. From thisplot, the deactivation of Al (fastest to slowest) as afunction of bath gas is O2 � He � air � N2. Thisunusual intensity decay behavior during the earlytime period �5–10 �s� of aluminum observed in nitro-gen bath gas may come from the formation of AlN inthe gas phase because of the reaction of nitrogen withAl vapor at high temperatures, as the plasma tem-perature was high in this early time period.3 Figures10 and 11 (expanded by a factor of 1000) show the


Fig. 5. LIBS spectrum of an aluminum rod in air, at various gatedelays. The gate pulse width is 2 �s.

Fig. 6. Temporal emission evolution of Al LIBS in oxygen. Thegate pulse width is 2 �s.

Fig. 7. Temporal emission evolution of Al LIBS in helium. Thegate pulse width is 2 �s.


maximum emission intensity of the AlO band near484 nm as a function of time for reactive (air, O2) andnonreactive �He, N2� bath gases, respectively. Figure10 shows that the maximum emission from AlO oc-curs 10 �s after the laser pulse in the pure oxygenatmosphere, whereas in air the intensity reached a

maximum 20 �s after the laser pulse. Therefore webelieve the main source of AlO emission in bath gasesof O2 and air is AlO formed by the reaction of Al �g�with ambient oxygen, analogous to reactions (3) and(4) above for the combustion of aluminum in oxygen.This is also supported by the increase in AlO emis-sion as the bath gas is changed from air (Fig. 5) tooxygen (Fig. 6).

Figure 11 shows that, in the absence of ambientoxygen, the temporal behavior of the AlO emission issimilar to that of the Al emission. That is, the tem-poral behavior of the AlO emission in the unreactivebath gases is similar to emission from material �Al�native to the aluminum rod. Therefore we believe thesource of the AlO emission in the absence of ambientoxygen is the Al2O3 layer on the aluminum metal.

C. Temperature Calculations

For the temperature calculations reported here, weassume that, for the gate width used �2 �s�, the timerate of change of the plasma temperature is small,and that light emission collected and analyzed isemitted from a gas region that is approximately ho-mogeneous in temperature and composition. This as-sumption of “local thermodynamic equilibrium” isnecessary when calculating temperatures using aBoltzmann distribution. The intensities of Al I spec-tral lines at wavelengths of 308.34, 309.44, 394.56,and 396.26 nm were used to calculate temperaturesat different gate pulse delays according to the follow-ing equation:

Ln(Iki�(gkAki)) � �(Ek�kT) � Ln(CF�U(T)) (8)

where I is the peak line intensity of atomic species �with concentration C, Ek �eV� is the upper energylevel, T �K� is the plasma temperature, U�T� is thepartition function of the species �, k is the Boltzmannconstant, F is a constant depending on experimentalconditions, Aki is the transition probability, and gk isthe statistical weight for the upper level. Spectro-scopic data (Table 1) were obtained from the NationalInstitute of Standards and Technology (NIST) data-base.13 Our measured wavelength values differedfrom the NIST database � 0.2 nm. A plot of

Fig. 8. Temporal emission evolution of Al LIBS in nitrogen. Thegate pulse width is 2 �s.

Fig. 9. Plot of logarithm of intensity at 396 nm versus time for thedifferent bath gases used in these experiments.

Fig. 10. Maximum emission intensity of the aluminum oxideband near 484 nm as a function of time for the bath gases air andO2.

Fig. 11. Maximum emission intensity of the aluminum oxideband near 484 nm as a function of time for the bath gases He andN2.


(Iki��gk Aki�) as a function of Ek will have a slope

equal to �1�kT. A typical Boltzmann plot using Eq.(8) is shown in Fig. 12.

Temperatures calculated using Eq. (8) and spectralline intensities from LIBS spectra measured in dif-ferent bath gases are shown in Fig. 13. These calcu-lated temperatures are in good agreement withpreviously reported calculated temperatures for sim-ilar systems.14–16 In general, the calculated temper-ature exhibits an approximately exponential decayover the emission lifetime.

D. Electron Density

The electron density �Ne, cm�3� was determined usingthe Stark-broadening effect17 and assuming the

plasma to be optically thin (negligible self-absorption) for the Al II emission line at 747.14 nm.Stark-broadening parameters are available for thelines at 747.14 and 466.3 nm.18 The Al II line at466.3 nm was not used because this line is partiallyobscured by the AlO band near 484 nm. The relationbetween the line width (full width at half-maximum,FWHM) of the Stark-broadened line and the electrondensity is given by Eq. (9):

��1�2 � 2�(Ne�1016) � 3.5A(Ne�1016)1�4

(1 � BND�1�3)�(Ne�1016), (9)

where ��1�2 �nm� is the line width (FWHM), � �nm� isthe Stark-broadening parameter, A is the ion-broadening parameter, ND is the number of particlesin the Debye sphere, and B is a coefficient equal to 1.2for ions and 0.75 for neutral lines. The values of �were taken from literature.18–22

The measured line width was corrected to first or-der by subtracting the contribution of the instrumen-tal line broadening. The instrument line broadeningwas found to be 0.1 nm, as determined by measuringthe emission lines from a calibrated mercury lamp.The first term on the right side of Eq. (9) is the con-tribution of electron broadening. The second term onthe right side of Eq. (9) is the quasi-static ion-broadening contribution, which can to be neglected inthis analysis.23 Equation (9) then reduces to

��1�2 � 2�(Ne�1016). (10)

In order to make a determination as to whether thelocal thermodynamic equilibrium conditions weresatisfied for the selected spectral lines, the criticalvalue of electron-density distribution �Ne� was evalu-ated by following the procedure described in previouspublications.24,25 The critical limit of electron-densitydistribution was determined from Eq. (11), which de-scribes a necessary condition for local thermody-namic equilibrium.

Ne � 1.6 1012T1�2(Ek � Ei)3. (11)

For the experiments reported here, the criticalelectron densities varied from 3 1015 to 9.5 1015 cm�3 for the temperature range 4000–14500 K.The lowest calculated electron-density value ex-ceeded these critical values by a factor of 20 for therange of temperatures calculated using the Boltz-mann equation [Eq. (8)]. As seen in Fig. 14, there is ageneral trend toward lower electron densities at laterdecay times as the plasma cools.

As seen in Fig. 15, the highest temperature valuewas observed in the helium atmosphere, whereas thehighest value for electron density was seen in oxygenatmosphere. The higher value of thermal conductiv-ity for helium and the lower value of the ionizationpotential for oxygen may be a key factor contributingto these observations. Further, it is noted that there

Table 1. Spectroscopic Parameters for Al I and Al II Investigated Lines

Line

MeasuredWavelength

(nm) Aki �108 s�1� Ek �eV� gk � �nm�

Al I 308.34 0.63 4.021485 4 –309.44 0.74 4.021650 6 –394.56 0.49 3.142721 2 –396.26 0.98 3.142721 2 –

Al II 466.30 0.53 13.25646 3 6.85 10�3

747.14 0.94 15.30840 7 1.26 10�2

Fig. 12. Boltzman plot for 308.34, 309.44, 394.56, and396.26 nm Al I lines in oxygen. The gate pulse width is 2 �s. Thegate pulse delay is 15 �s.

Fig. 13. Excitation temperature versus gate pulse delay with a2��s gate width.


are formation and decomposition processes of manydifferent species in a gaseous medium in the hotplasma. This may result in a lack of correlation be-tween the plasma temperature and electron densityin a particular bath gas.

4. Conclusions

Measurements of the emission of AlO following ex-posure of an aluminum metal surface to a laser-induced spark have been carried out for bath gases ofair, oxygen, nitrogen, and helium. Results of theseexperiments indicate that virtually all of the AlOemission is from AlO formed by the reaction of Alvapor with oxygen from the bath gas (if present).Emission from AlO initially present as an Al2O3 oxidelayer on the metal sample was vanishingly small foremission spectra measured in bath gases of nitrogenand helium, when compared with the AlO emissionmeasured in air and in oxygen bath gases. However,it is possible to distinguish the AlO emission from theAl2O3 oxide layer from AlO formed by reaction withambient oxygen by examining the temporal behaviorof the emission. The temporal behavior of Al and AlOemission following Al metal exposure to a laser-induced spark (in air and oxygen) is consistent withknown chemical mechanisms for aluminum combus-tion in oxygen. Finally, calculations of temperatureassuming a Boltzmann distribution of Al emissionlines gives results in good agreement with calcula-tions by previous investigators.

References1. P. Brousseau, H. Dorsett, and M. Cliff, “Detonation properties

of explosives containing nanometric aluminum powder,” pre-

sented at the Twelfth International Detonation Symposium,DOE and DoD laboratories, San Diego, Calif., 11–16 August2002.

2. D. C. Mueller and S. R. Turns, “A theoretical evaluation ofsecondary atomization effects on engine performance for alu-minum gel propellants,” J. Propul. Power 12, 591–597 (1996).

3. S. Yuasa, Y. Zhu, and S. Sogo, “Ignition and combustion ofaluminum in oxygen�nitrogen mixture streams,” Combust.Flame 108, 387–396 (1997).

4. F. Capitelli, F. Colao, M. R. Provenzano, R. Fantoni, G. Bru-netti, and N. Senesi, “Determination of heavy metals in soils bylaser induced breakdown spectroscopy,” Geoderma 106, 45–62(2002).

5. F. C. DeLucia, Jr., R. S. Harmon, K. L. McNesby, R. J. Winkel,Jr., and A. W. Miziolek, “Laser induced breakdown spectros-copy analysis of energetic materials,” J. Appl. Opt. 42, 6148–6152 (2003).

6. F. Y. Yueh, R. C. Sharma, J. P. Singh, H. Zhang, and W. A.Spencer, “Evaluation of the potential of laser-induced break-down spectroscopy for detection of trace element in liquid,” J.Air Waste Manag. Assoc. 52, 1307–1315 (2002).

7. A. Portnov, S. Rosenwaks, and I. Bar, “Emission followinglaser-induced breakdown spectroscopy of organic compoundsin ambient air,” J. Appl. Opt. 42, 2835–2842 (2003).

8. A. C. Samuel, F. C. DeLucia, Jr., K. L. McNesby, and A. W.Miziolek, “Laser induced breakdown spectroscopy of bacterialspores, molds, pollens, and protein: initial studies of discrim-ination potential,” J. Appl. Opt. 42, 6205–6209 (2003).

9. J. B. Simeonsson and A. W. Miziolek, “Time-resolved emissionstudies of ArF laser produced microplasmas,” J. Appl. Opt. 32,939–947 (1993).

10. R. T. Wainner, R. S. Harmon, A. W. Miziolek, K. L. McNesby,and P. D. French, “Analysis of environmental lead contamina-tion: comparison of LIBS field and laboratory instruments,”Spectrochem. Acta Part B 56, 777–793 (2001).

11. A. K. Rai, H. Zhang, F. Y. Yueh, J. P. Singh, and A. Weisburg,“Parametric study of a fiber-optic laser-induced breakdownspectroscopy probe for analysis of aluminum alloys,” Spectro-chim. Acta Part B 56, 2371–2383 (2001).

12. A. Dokhan, E. W. Price, J. M. Seitzman, and R. K. Sigman,“Combustion mechanisms of bimodal and ultra fine aluminumin ammonium perchlorate solid propellant,” Rep. AIAA-2002-4173 (AIAA, Reston, Va., 2002).

13. http://physics.nist.gov/cgi-bin/AtData/lines_form.14. B. Le Drogoff, J. Margot, M. Chaker, M. Sabsabi, O. Bar-

thelemy, T. W. Johnston, S. Laville, F. Vidal, and Y. vonKaenel, “Temporal characterization of femtosecond laserpulses induced plasma for spectrochemical analysis of alumi-num alloys,” Spectrochim. Acta Part B 56, 987–1002 (2001).

15. H. C. Liu, X. L. Mao, J. H. Yoo, and R. E. Russo, “Early phaselaser induced plasma diagnostic and mass removal duringsingle pulse laser ablation of silicon,” Spectrochim. Acta Part B54, 1607–1624 (1999).

16. S. Amoruso, “Modeling of UV-pulsed laser ablation of silion,”Appl. Phys. A 69, 323–332 (1999).

17. E. V. Saramdaev, and M. K. H. Salakhov, “Regularities in theStark widths and shifts of spectral lines of singly-ionized alu-minum,” J. Quant. Spectrosc. Radiat. Transfer 56, 399–407(1996).

18. F. Coloa, V. Lazic, F. Fantoni, and S. Pershin, “A comparisonof single and double pulse laser-induced breakdown spectros-copy of aluminum samples,” Spectrochim. Acta Part B 57,1167–1179 (2002).

19. J. T. Knudtson, W. B. Green, and D. G. Sutton, “The UV-visiblespectroscopy of laser produced aluminium plasma,” J. Appl.Phys. 61, 4771–4780 (1987).

20. C. Colon, G. Hatem, E. Verdugo, P. Ruiz, and J. Campos,“Measurement of the Stark broadening and shift parameters

Fig. 15. Excitation temperature versus electron-density profile indifferent atmospheres.

Fig. 14. Electron density of Al versus gate pulse delay with a 2��sgate width.


for several ultraviolet lines of singly ionized aluminum,”J. Appl. Phys. 73, 4752–4758 (1993).

21. E. V. Sarandaev and M. K. H. Salakhov, “Regularities in the Starkwidths and shifts of spectral lines of singly ionized aluminum,” J.Quant. Spectrosc. Radiat. Transfer 56, 399–407 (1996).

22. N. Konjeviv and W. Wiese, “Experimental Stark widths andshifts for spectral lines of neutral and ionized atoms,” J. Phys.Chem. Ref. Data 19, 1307–1385 (1990).

23. G. Abdellatif and H. Imam, “A study of the laser plasma pa-rameters at different laser wavelengths,” Spectrochim. ActaPart B 57, 1155–1165 (2002).

24. H. R. Griem, Plasma Spectroscopy (McGraw-Hill, New York,1964), pp. 267–313.

25. C. Aragon, J. Bengoechea, and J. A. Aguilera, “Influence of theoptical depth on spectral line emission from laser-inducedplasmas,” Spectrochim. Acta Part B 56, 619–628 (2001).


Documents

Temporal evolution of the laser-induced breakdown spectroscopy spectrum of aluminum metal in different bath gases