4
Normal-incidence far-infrared detectivity of InAs/GaAs QDIPs doped in dots and barriers S.J. Lee a , S.K. Noh a, * , S.C. Hong b , J.I. Lee c a Quantum-Dot Technology Laboratory, Korea Research Institute of Standards and Science, Daedeok Science Town, Daejeon 305-600, South Korea b Department of Electrical Engineering, Korea Advanced Institute of Science and Technology, Daejeon 305-701, South Korea c Nano-Device Research Center, Korea Research Institute of Science and Technology, Seoul 136-791, South Korea Received 30 March 2004; received in revised form 26 August 2004; accepted 7 December 2004 Available online 25 January 2005 Abstract We report some comparative results on the normal-incidence device characteristics accomplished with a couple of self-assembled InAs/GaAs quantum-dot infrared photodetectors (QDIPs) doped in InAs QDs and GaAs barriers. The peak values of responsivity and detectivity for the barrier-doped device are 650 mA/W and 3.2 · 10 8 cm Hz 1/2 /W (18 K) at k p 5 lm, respectively, which are approximately two and ten times higher than those for the QD-doped one. In addition, while there is no spectral response over 6 lm in the QD-doped structure, a strong photoresponse is extended up to around 10 lm in the barrier-doped one. Although the direct doping in InAs QDs is effective for blocking the dark current, the doping in GaAs barriers has better device performance of QDIP. Ó 2005 Elsevier B.V. All rights reserved. PACS: 78.30.Fs; 85.35.Be; 85.60.Gz Keywords: Infrared photodetector; InAs/GaAs; Quantum dot 1. Introduction Remarkable progress has been in developments of self-assembled quantum-dot (QD)-based laser diodes (QDLDs) and infrared photodetectors (QDIPs) taking aims at high-frequency direct-modulation and normal- incidence room-temperature operation to overcome the limitation of quantum well (QW)-based devices [1–5]. While the 1.0–1.3 lm operation of QDLDs was already achieved at room temperature a few years ago [6,7], the reliable operation temperature of QDIPs is at a stand- still around 100 K [8,9]. Through the theoretical analysis on a direct comparison of limitations in dark current and detectivity among HgCdTe, QWIPs, and QDIPs, Phillips [10] has made a conclusion that QDIPs are ex- pected to possess the fundamental abilities to achieve the highest infrared (IR) detector performance if QD ar- rays with high size uniformity and optimal bandstruc- ture can be achieved. Lee et al. [11] newly proposed a modulation-doped QDIP structure with an AlGaAs spacer based on lateral transport, and Stiff-Roberts et al. [8] demonstrated raster-scanned images using a (13 · 13) QDIP array at 80 K. Recently, Jiang et al. [9] reported an InGaAs/GaAs QDIP device with excellent device performance as high as 10 11 cm Hz 1/2 /W in the detectiviy, and Krishna et al. [12] presented a two-color detector (k p 4.2/7.6 lm) based on the InAs/InGaAs dot-in-well (DWELL) structure. They have normally adapted the device structures doped directly in QDs instead of the barrier doping. Though a variety of efforts have continued to realize the room-temperature operation QDIP with higher 1567-1739/$ - see front matter Ó 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.cap.2004.12.001 * Corresponding author. Tel.: +82 42 868 5127; fax: +82 42 868 5047. E-mail address: [email protected] (S.K. Noh). www.elsevier.com/locate/cap www.kps.or.kr Current Applied Physics 6 (2006) 37–40

Normal-incidence far-infrared detectivity of InAs/GaAs QDIPs doped in dots and barriers

  • Upload
    sj-lee

  • View
    212

  • Download
    0

Embed Size (px)

Citation preview

www.elsevier.com/locate/cap

www.kps.or.kr

Current Applied Physics 6 (2006) 37–40

Normal-incidence far-infrared detectivity of InAs/GaAsQDIPs doped in dots and barriers

S.J. Lee a, S.K. Noh a,*, S.C. Hong b, J.I. Lee c

a Quantum-Dot Technology Laboratory, Korea Research Institute of Standards and Science, Daedeok Science Town, Daejeon 305-600, South Koreab Department of Electrical Engineering, Korea Advanced Institute of Science and Technology, Daejeon 305-701, South Korea

c Nano-Device Research Center, Korea Research Institute of Science and Technology, Seoul 136-791, South Korea

Received 30 March 2004; received in revised form 26 August 2004; accepted 7 December 2004

Available online 25 January 2005

Abstract

We report some comparative results on the normal-incidence device characteristics accomplished with a couple of self-assembled

InAs/GaAs quantum-dot infrared photodetectors (QDIPs) doped in InAs QDs and GaAs barriers. The peak values of responsivity

and detectivity for the barrier-doped device are 650 mA/W and 3.2 · 108 cm Hz1/2/W (18 K) at kp ffi 5 lm, respectively, which are

approximately two and ten times higher than those for the QD-doped one. In addition, while there is no spectral response over

6 lm in the QD-doped structure, a strong photoresponse is extended up to around 10 lm in the barrier-doped one. Although

the direct doping in InAs QDs is effective for blocking the dark current, the doping in GaAs barriers has better device performance

of QDIP.

� 2005 Elsevier B.V. All rights reserved.

PACS: 78.30.Fs; 85.35.Be; 85.60.GzKeywords: Infrared photodetector; InAs/GaAs; Quantum dot

1. Introduction

Remarkable progress has been in developments of

self-assembled quantum-dot (QD)-based laser diodes

(QDLDs) and infrared photodetectors (QDIPs) taking

aims at high-frequency direct-modulation and normal-

incidence room-temperature operation to overcome the

limitation of quantum well (QW)-based devices [1–5].While the 1.0–1.3 lm operation of QDLDs was already

achieved at room temperature a few years ago [6,7], the

reliable operation temperature of QDIPs is at a stand-

still around 100 K [8,9]. Through the theoretical analysis

on a direct comparison of limitations in dark current

and detectivity among HgCdTe, QWIPs, and QDIPs,

1567-1739/$ - see front matter � 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.cap.2004.12.001

* Corresponding author. Tel.: +82 42 868 5127; fax: +82 42 868

5047.

E-mail address: [email protected] (S.K. Noh).

Phillips [10] has made a conclusion that QDIPs are ex-

pected to possess the fundamental abilities to achieve

the highest infrared (IR) detector performance if QD ar-

rays with high size uniformity and optimal bandstruc-

ture can be achieved. Lee et al. [11] newly proposed a

modulation-doped QDIP structure with an AlGaAs

spacer based on lateral transport, and Stiff-Roberts et

al. [8] demonstrated raster-scanned images using a(13 · 13) QDIP array at 80 K. Recently, Jiang et al. [9]

reported an InGaAs/GaAs QDIP device with excellent

device performance as high as �1011 cm Hz1/2/W in the

detectiviy, and Krishna et al. [12] presented a two-color

detector (kp ffi 4.2/7.6 lm) based on the InAs/InGaAs

dot-in-well (DWELL) structure. They have normally

adapted the device structures doped directly in QDs

instead of the barrier doping.Though a variety of efforts have continued to realize

the room-temperature operation QDIP with higher

38 S.J. Lee et al. / Current Applied Physics 6 (2006) 37–40

device performance, especially at the far-IR atmospheric

window of a 8–12 lm range [13–19], it is true that the

device performance is still a little inferior to QWIPs

[2]. Of current necessity, therefore, is QDIP device with

enhanced detectivity working at higher temperature in a

longer spectral range. In this letter, as a part of efforts toenhance the device performance, we report some com-

parative results on the dark current and the spectral

photoresponse characteristics obtained from a couple

of InAs/GaAs QDIP devices doped in InAs QDs and

GaAs barriers. In this study, we demonstrate that the

barrier-doped device has much higher performance in

spite of larger dark current in comparison with the

QD-doped one.

2. Experimental procedures

In the present experiment, we prepared two kinds of

QDIP devices with an active layer of five-period InAs-

QD/GaAs-barrier whose doping position is different.

Both structures were grown under the same conditionsby a molecular beam epitaxy (MBE) technology via

the Stranski–Krastanow (S–K) growth mode, and basi-

cally have the same layer profile except for the position

of Si dopants (1 · 1017 cm�3) for n-type conduction, as

shown in Fig. 1(a). The first sample was directly doped

Fig. 1. (a) Schematic layer structure for a single period of the InAs-

QD/GaAs-barrier active layer doped in InAs QDs (n-InAs:Si-QD) or

GaAs barriers (n-GaAs:Si) and (b) the cross-sectional TEM image

showing the full active layer stacked by five periods.

in InAs QDs during the growth, and the second one

doped in GaAs barriers after the QD formation. The

growth was conducted at temperatures of 580 �C from

the GaAs buffer to bottom n+-GaAs layer and of

460 �C from the active region to top n+-GaAs layer,

and each layer of InAs QDs was formed with an equiv-alent thickness of 2.5 monolayers (MLs) on the 40 nm

GaAs spacer. In the barrier-doped structure, a 3 nm-

thick Si-doped layer was positioned at 6 nm above the

QD layer in the GaAs spacer for each stack. In order

to be occupied by approximately one electron in each

dot, we have designed the layer thickness (3 nm) and

the doping level (1 · 1017 cm�3) which give the areal

electron density of �3 · 1010 cm�2 comparable to theQD density (mid-1010 cm�2). The epitaxial growth and

the related basic properties have been reported in detail

elsewhere [13–15].

The dot formation was monitored in situ during the

growth by observing the 2-dimensional-to-3-dimen-

sional (2D–3D) transition in the reflection high-energy

diffraction (RHEED) patterns [13]. The QD profile

was confirmed by the cross-sectional transmission elec-tron microscope (TEM) image presented in Fig. 1(b),

and the dot density of �5 · 1010 cm�2 was identified

from the atomic force microscopy (AFM) surface profile

of a single-layered uncapped structure. Circular-ring-

and square-shaped electrodes were formed on the top

and the bottom n+-GaAs layers, respectively, by using

the standard photolithography and the wet-chemical

etching procedures. The diameter of aperture for anindividual device is 450 lm (A = 1.6 · 105 lm2), and

the dimension of electrodes is 100 · 100 lm2. All the

spectral photoresponse measurements were performed

at normal-incidence configuration by using the far-IR

spectrometer system with a pair of gratings of 60 and

120 gr/mm and a broadband SiC globar IR source cali-

brated by an HgCdTe detector, which has been used in

examinations of QWIPs.

3. Results and discussion

The current–voltage (I–V) characteristic curves for

dark current taken at 10 K are compared between two

QDIP devices, as shown in Fig. 2. The device directly

doped in InAs QDs (open circles) shows very small darkcurrent of approximately 10�11 A at a bias voltage of

0.1 V in contrast with a level of 10�6 A for the barrier-

doped one (solid circles). It can be simply explained by

the doping position of Si dopants. In the barrier-doped

structure with a Si-donor level (DE = 6 meV) located

just below the GaAs conduction band edge, bound elec-

trons can be easily activated into the conduction channel

at any appropriate temperatures, so quite large level ofdark current can be generated. On the other hand, in

the QD-doped device, electrons bound to InAs QDs

-1.0 -0.5 0.0 0.5 1.0

10-3

10-5

10-7

10-9

10-11

10-1

10-13

InAs/GaAs QDIPT = 10 K

Dar

k C

urre

nt (A

)

Applied Bias Voltage (V)

Doped in GaAs BarriersDoped in InAs QDs

Fig. 2. The I–V characteristic curves of dark current for a couple of

QDIP devices. The device doped in GaAs barriers has relatively large

dark current compared with the barrier-doped one.

S.J. Lee et al. / Current Applied Physics 6 (2006) 37–40 39

have to be thermally excited to the conduction state with

a quite large energy in order to participate in vertical

transport. Thus, we can expect a small amount of dark

current in this structure. In the viewpoint of I–V charac-

teristics, we can say that the QDIP device doped in QDshas a great advantage in the dark current blocking com-

pared with the doped-barrier one.

The photocurrent response curves normalized to

the spectral photon flux taken at 18 K are depicted

in Fig. 3 for the same QDIP samples used in the I–V

measurement. Both spectra show similar IR detection

behaviour, but the QDIP device doped in GaAs barriers

presents higher photoresponse feature in the spectral

2 6 10 12

Doped in GaAs BarriersDoped in InAs QDs

InAs/GaAs QDIPT=18 K

Nor

mal

ized

Pho

tore

spon

se (

a.u.

)

Absorption Wavelength (µm)4 8

Fig. 3. The spectral photoresponse curves normalized to the spectral

photon flux taken at 18 K for a couple of QDIP devices. While there is

no spectral response above 6 lm in the QD-doped sample, a strong

photoresponse is extended up to around 10 m in the barrier-doped one.

range of 3–10 lm with a peak at kp ffi 5 lm compared

with the QD-doped one. While there is no spectral

response above 6 lm in the QD-doped sample, a strong

photoresponse is extended up to around 10 lm in the

barrier-doped one, as shown in Fig. 3. The reason for

no long-wavelength response in QD-doped sample isnot clear at the present stage.

Fig. 4 presents the temperature dependences of

responsivities (open symbols) measured at a peak wave-

length of kp ffi 5 lm, together with corresponding detec-

tivities (solid symbols) in which the device area and the

noise figure are taken into account. The peak values of

responsivity (R) and detectivity (D*) of the barrier-

doped device (squares) are fairly high as 650 mA/Wand 3.2 · 108 cm Hz1/2/W at 18 K, respectively, and sur-

vive up to 190 K, in comparison with the values of

R = 350 mA/W and D* = 3.7 · 107 cm Hz1/2/W for the

QD-doped one (circles). (For reference, R = 2 mA/W/

D* = 2.9 · 109 cm Hz1/2/W (T = 100 K, kp = 3.7 lm)

for Stiff-Roberts et al. [8]; R = 220 mA/W/D* = 1.1 ·1010 cm Hz1/2/W (T = 77 K, kp = 7.6 lm) for Jiang

et al. [9].) Comparing with the previously reported data,the reason for relatively low D*-values contrary to much

high R-values can be interpreted as a little large dark

current due to no use of additional current blocking

barrier. Table 1 is a comparative summary on the

normal-incidence device characteristics of the two QDIP

structures doped in InAs QDs and GaAs barriers.

The values of R and D* for the barrier-doped device

are approximately two and ten times higher than thoseof the QD-doped one, respectively. The difference be-

tween a pair of R�s or D*�s for two samples can be

understood by the position of dopants, as previously

mentioned in the I–V characteristics. Since the conduc-

tion electrons activated from Si donors in GaAs barrier

can be easily swept into and occupied by lower QD sub-

level in the barrier-doped structure under a bias, a fairly

0

2

4

6

8

10

0 50 100 150 200 250104

105

106

107

108

109

Res

pons

ivity

(x 1

00 m

A/W

)

Det

ectiv

ity (

cm.H

z1/2 /W

)

Doped inGaAs Barriers Doped inInAs QDs

Temperature (K)

InAs/GaAs QDIPλ = 5 µm

Fig. 4. The peak responsivity (open symbols) and the corresponding

detectivity (solid symbols) taken at kp ffi 5 lm plotted as a function of

temperature. Both devices survive up to around 200 K.

Table 1

A comparative summary of the normal-incidence device characteristics

for a couple of QDIP structures doped in InAs QDs and GaAs barriers

QDIP device doped in InAs QDs GaAs barriers

Dark current (10 K, 0.1 V) (A) �10�11 �10�6

Photoresponse spectral range (lm) 3 � 6 3 � 10

Peak responsivity (18 K) (mA/W) 350 650

Peak detectivity (18 K) (cm Hz1/2/W) 3.7 · 107 3.2 · 108

Maximum operation temperature (K) 180 190

40 S.J. Lee et al. / Current Applied Physics 6 (2006) 37–40

strong absorption is expected by a transition from the

occupied sublevel to the above conduction-state. In the

QD-doped device, on the other hand, we can expect

small amount of electrons in the QD sublevel that con-

tribute to the bound-to-continuum absorption, becauseelectrons bound to Si donors in InAs QDs need quite

a large energy in order to be activated onto the sublevel,

as discussed in Fig. 3. Thus, the intersublevel absorption

in the QD-doped structure may be weaker than that of

the barrier-doped one. On the basis of results on the

photoresponse characteristics, we can conclude that

the QDIP device doped in GaAs barriers has better de-

vice performance in comparison with the QD-doped onebecause of difference in the absorption efficiency associ-

ated with the intersublevel transition.

4. Summary and conclusions

We accomplished a comparative study on the normal-

incidence device characteristics in a couple of self-assem-bled InAs/GaAs QDIP structures doped in InAs QDs

and GaAs barriers through dark current and photore-

sponse investigations. The device directly doped in InAs

QDs showed very low dark current of approximately

10�11 A at a bias voltage of 0.1 V in contrast with a level

of 10�6 A for the barrier-doped one. The peak values of

R and D* of the barrier-doped device were fairly high as

650 mA/W and 3.2 · 108 cm Hz1/2/W at 18 K, respec-tively, which were approximately two and ten times lar-

ger than the values of R = 350 mA/W and D* = 3.7 ·107 cm Hz1/2/W for the QD-doped one. In addition,

while there was no spectral response over 6 lm in the

QD-doped structure, a strong photoresponse was ex-

tended up to around 10 lm in the barrier-doped one.

Although the direct doping in InAs QDs was effective

for blocking the dark current, we made a conclusion thatthe QDIP device doped in GaAs barriers had better de-

vice performance in comparison with the QD-doped

one on the basis of results on the photoresponse charac-

teristics. In order to enhance the characteristic of D* by

reducing the dark current, modified device structures

inserted by current blocking barriers of high-bandgap

AlGaAs are now in measurements.

Acknowledgment

This work was supported in part by MOIC

(IMT2000-B4-1), and mainly carried out at the National

Research Laboratory on Quantum Dot Technology at

KRISS designated by MOST (M1-0104-00-0127). One

of the authors (SKN) acknowledges the partial support

provided by KOSEF through the Quantum-functionalScience Research Center at Dongguk University.

References

[1] N.N. Ledentsov, Semiconductors 33 (1999) 946.

[2] B.F. Levine, J. Appl. Phys. 74 (1993) R1–R81.

[3] G. Walter, T. Chung, N. Holonyak Jr., Appl. Phys. Lett. 80

(2002) 1126.

[4] H. Hwang, K. Park, J.-H. Kang, S. Yoon, E. Yoon, Curr. Appl.

Phys. 3 (2003) 465.

[5] Y.J. Park, K.M. Kim, Y.M. Park, E.K. Kim, Curr. Appl. Phys. 1

(2001) 187.

[6] K. Kamath, P. Bhattacharya, T. Sosnowski, T. Norris, J. Phillips,

Electron. Lett. 32 (1996) 1374.

[7] G. Park, O.B. Shchenkin, S. Csutak, D.L. Huffaker, D.G. Deppe,

Appl. Phys. Lett. 75 (1999) 3267.

[8] A.D. Stiff-Roberts, S. Chakrabarti, S. Pradhan, B. Kochman,

P. Bhattacharya, Appl. Phys. Lett. 80 (2002) 3265.

[9] L. Jiang, S.S. Li, N.T. Yeh, J.I. Chyi, C.E. Ross, K.S. Jones,

Appl. Phys. Lett. 82 (2003) 1986.

[10] J. Phillips, J. Appl. Phys. 91 (2002) 4590.

[11] S.W. Lee, K. Hirakawa, Y. Shimada, Appl. Phys. Lett. 75 (1999)

1428.

[12] S. Krishna, S. Raghaven, G. von Winckel, P. Rotella, A. Stintz,

C.P. Morath, D. Le, S.W. Kennerly, Appl. Phys. Lett. 82 (2003)

2574.

[13] M.D. Kim, S.K. Noh, S.C. Hong, T.W. Kim, Appl. Phys. Lett.

82 (2003) 553.

[14] S.K. Kang, S.J. Lee, M.D. Kim, S.K. Noh, Y.H. Kang, U.H.

Lee, S.C. Hong, H.S. Kim, C.G. Park, J. Kor. Phys. Soc. 42

(2003) 418.

[15] S.J. Lee, S.K. Noh, J.W. Choe, U.H. Lee, S.C. Hong, J.I. Lee,

Jpn. J. Appl. Phys. 43 (2004) 1218.

[16] S.Y. Wang, S.D. Lin, H.W. Wu, C.P. Lee, Infrared Phys.

Technol. 42 (2001) 473.

[17] O.B. Shchekin, D.G. Deppe, D. Lu, Appl. Phys. Lett. 78 (2001)

3115.

[18] S.F. Tang, S.Y. Lin, S.C. Lee, IEEE Trans. Electron Dev. 49

(2002) 1341.

[19] Z. Chen, E.T. Kim, A. Madhukar, Appl. Phys. Lett. 80 (2002)

2490.