Luminescent characteristics of InGaAsP/InP multiple quantum well structures by impurity-free vacancy disordering

(1)

Luminescent characteristics of InGaAsP/InP multiple quantum well

structures by impurity-free vacancy disordering

J. Zhao

a

, Z.C. Feng

b,

*, Y.C. Wang

a

, J.C. Deng

a

, G. Xu

c a

College of Physics and Electronic Information, Tianjin Normal University, Tianjin 300074, PR China

b

Graduate Institute of Electro-Optical Engineering and Department of Electrical Engineering, National Taiwan University, Taipei, Taiwan 106-17, ROC

c

Department of Materials Science and Engineering, McMaster University, Hamilton, Canada L8S 4L7 Available online 11 August 2005

Abstract

InGaAsP/InP multiple quantum wells with quantum well intermixing (QWI) have been prepared by Impurity-Free Vacancy Disordering (IFVD). The luminescent characteristics were investigated using photoluminescence (PL) and photoreflectance (PR), from which the band gap blueshift was observed. Si3N4, SiO2and SOG (spin on glass) were used for the dielectric layer to enhance intermixing from the out-diffusion of Group III atoms. All samples were annealed by rapid thermal annealing (RTA). The results indicate that the band gap blueshift varies with the dielectric layers and the annealing temperature. The SiO2capping was successfully used with an InGaAs cladding layer to cause larger band tuning effect in the InGaAs/InP MQWs than the Si3N4capping with an InGaAs cladding layer. On the other hand, samples with the Si3N4– InP cap layer combination also show larger energy shifts than that with SiO2– InP cap layer combination.

PACS: 78.40.Fy; 78.55.Cr; 81.05.Ea; 81.15.Hi

Keywords: InP; InGaAsP; Molecular beam epitaxy; Quantum well; Impurity-free vacancy disordering (IFVD); Optical properties

1. Introduction

Quantum well intermixing (QWI) technology has attracted much interest in recent years for the fabrication of various optoelectronic devices, such as high power semiconductor lasers and photonic integrated devices and circuits[1 – 6]. In order to realize the monolithic integration of active and passive optoelectronic devices and compo-nents, different techniques are currently under investigation, such as selective area growth and growth – etch – regrowth. Regrowth requires many steps of etching and it is often associated with low efficiency and poor yield[2]. Also, the contamination and defects at etched and regrowth interface are difficult to avoid. As an alternative approach, post growth QWI becomes a more attractive technique, which can change the QW shape and composition by intermixing wells and barriers in QWs giving rise to a blueshifted band

gap. This postgrowth controlling of QW profiles, or the post-tuning of optical band gap energy can be realized by impurity-induced intermixing or impurity-free vacancy-enhanced intermixing [7]. Several technical approaches have been explored to achieve this purpose, including Impurity Induced Disordering (IID) [8,9], Implant Induced Composition Disordering (IICD) [2,10 – 12], and Impurity-Free Vacancy Disordering (IFVD)[2,13 – 15]. Among them, IFVD technique is more promising because it can maintain the high crystal quality and low optical propagation loses without introducing free-carrier concentration.

IFVD usually involves a deposition of a dielectric capping layer such as SiO2 and Si3N4on the top of QWs

and post rapid thermal annealing (RTA) at elevated temper-ature. Vacancies can be created at the dielectric-semi-conductor interface due to the out-diffusion, for example Ga atoms, from the semiconductor layers into the dielectric region. The diffusion of these vacancies into the structure can enhance the QW intermixing, leading to a large blueshift of the QW band gap energy with minimum effect

* Corresponding author. Tel.: +886 2 33663543; fax: +886 2 23677467. E-mail address: [email protected] (Z.C. Feng).

(2)

on their electrical and optical properties[16,17]. By using the IFVD technique, QW intermixing has been realized on multiple quantum well (MQW) structures of In0.53Ga0.47As/

InP [13], Al0.3Ga0.7As/GaAs and In0.2Ga0.8As/GaAs [14],

In1xGaxAs/InP [2,3], In1xGaxAs/In1xGaxAs1yPy [2],

In0.76Ga0.24As0.85P0.15/In0.76Ga0.24As0.52P0.48 [4], and

In0.2Ga0.8As/GaAs[15].

It is found that by selecting a proper combination, the properties of capping layers have important influence on the degree of QWI and band gap blueshifts. The SiO2

dielectric cap is porous to out-diffusion of Ga atoms, thus generating group III vacancies which results in diffusion of group III atoms from the barrier to the well [18]. As a result, effective band gap of the QW is widened. If the cap does not react with significant Ga atoms, QWI is suppressed. For the GaAs – AlGaAs QW system, SiO2

cap has been generally used to promote QWI while SiNxis

used as a mask to prevent or suppress QWI because the SiO2 cap layer induces a relatively larger blueshift than

SiNx cap layer [5,6,19]. A very thin SiO2 cap is also

expected to suppress QWI because of saturation of the diffused Ga atoms [20].

For the InGaAs/InP system, the SiO2 cap is also

employed to enhance QWI while Si3N4, Ga2O3 and SrF2

are used to suppress the intermixing[15]. By changing the deposition pressure of SiOx capping layers, variable

blueshifts can be obtained [15]. It is also reported for this system that the SiO2– In0.53Ga0.47As combination produces

a band gap blueshift while SiO2– InP or SiNx– InGaAs cap

layer combinations do not show significant energy shift

[13].

The effect of semiconductor-capping layer combination on QWI for the In1xGaxAs/In1xGaxAs1yPy MQW

structures showed similar trends[2]. The samples with the InGaAs/SiO2capping layer combination exhibited a higher

degree of intermixing than those with the InP/SiO2capping

layer combination after RTA treatment. This is attributed to the fact that InP has no Ga atom and possesses a lower thermal expansion coefficient than InGaAs[2].

However, more penetrating and systematic studies on the QWI in various quantum well systems and the factors influencing the band gap blueshifts as well as their mechanisms are still needed to be explored. In this paper, a systematic investigation of the luminescent character-istics of InGaAsP/InP MQW system using SiO2, Si3N4

and SOG (spin on glass) as dielectric layers with different cladding layer in IFVD is reported. Photoluminescence (PL) was measured by a Fourier Transform Infrared (FT-IR) PL system. Photoreflectance (PR) measurements on these samples were used to investigate further the behavior of band gap blueshift. To our knowledge, only a few work [5] was reported in the literature with measuring band gap blueshift by PR for this material system. We also attempt to analyze the band gap luminescent characteristics and the effects from different combinations of cladding layer and dielectric layer, such as InP – SiO2 and InP – Si3N4.

2. Experiment

Two typical samples were chosen in this paper. Both samples A and B are with InGaAsP/InP MQW layer structures, consisting of three quantum wells, designed to emit at wavelength of 1.57 Am and 1.55 Am, respectively. They were grown by Gas Source Molecular Beam Epitaxy (GSMBE). The details of the samples are shown in Table 1. Sample A was cut and divided into three groups with capped dielectric layers of SiO2 and Si3N4 by way of

plasma enhance chemical vapor deposition (PECVD) and SOG (spin on glass) by spin coating at 3000 rpm for 45 s. The thickness of all the dielectric layers is about 200 nm.

Table 1

Schematic layer structure of the InGaAsP/InP MQW samples studied

Sample A Sample B

Material Band gap kg

(Am) Carrier concentration (cm 3) Thickness d (nm) Material Band gap (Am) Carrier concentration (cm 3) Thickness (nm)

InP (Heimplanted) p = 1e18 100 InP (undoped) 100

InP p = 6e17 25 InGaAs (undoped) 5

InGaAsP 1.15 p = 5e17 80 InP p = 1e18 200

InGaAsP 1.24 p = 5e17 70 InP p = 5e17 200

InGaAsP 1.58 p = 5e17 5 InP (Heimplanted) p = 5e17 40

InGaAsP 1.24 p = 5e17 10 InP p = 5e17 5

InGaAsP 1.58 p = 5e17 5 InGaAsP 1.15 p = 5e17 100

InGaAsP 1.24 p = 5e17 10 InGaAsP 1.24 (undoped) 60

InGaAsP 1.58 p = 5e17 5 QWs: In0.758Ga0.242As0.83P0.17 3 * 5

Barriers: In0.758Ga0.242As0.525P0.475 2 * 10

InGaAsP 1.24 p = 5e17 70 InGaAsP 1.24 (undoped) 30

InGaAsP 1.15 n = 5e17 80 InGaAsP 1.15 n = 5e17 30

InP n = 1e18 500 InP n = 1e18 500

n+-InP substrate n+-InP substrate

(3)

Samples with the SOG (a mixture of organic and inorganic compounds commercially purchased) cap were then baked at 200 -C for 2 h under pure nitrogen ambit protection. Sample B was divided into two groups remarked B1 and B2. B1 was same as B listed in Table 1, while B2 was with the InP cladding layer etched away using a corrosive acid solution (HCl/H3PO4=1:1). Subsequently, samples B1

and B2 were deposited with SiO2and Si3N4, respectively,

with a thickness of about 200 nm, both by PECVD. After the GSMBE growth, samples were then annealed in a rapid thermal annealing (RTA) furnace in the temperature range of 650-C–850 -C in 50 -C steps. The annealing time for all samples was kept for 30 s. During RTA the samples were covered with a piece of semi-insulating-GaAs face to face to minimize the decomposition of InP and possible contamination. All annealing processing are under pure nitrogen.

Photoluminescence measurements were performed at the temperature 300 K. The excitation source was an Argon ion laser with the wavelength of 514.5 nm. The Photoreflec-tance (PR) spectra were measured by a modulation system. The modulation source was an He – Ne laser with the wavelength of 632.8 nm.

3. Results and discussion

Room temperature PL spectra are shown inFig. 1for the as-grown sample A and disordered InGaAsP multi-quantum well structures after RTA at 800-C for 30 s with SiO2and

Si3N4encapsulating layer, respectively. The peak position of

the PL spectrum for the as grown sample A is at 1.571 Am (0.789 eV), corresponding to the electron transition from the first level of electronic subband to the first level of heavy hole (E1-HH1) and light hole subband (E1-LH1). FromFig. 1, we can find that the band gap blueshift depends on the dielectric layer. Sample A with an Si3N4 capped layer

obtained larger blueshift. In order to find the dependence of

the band gap blueshift on the annealing temperature, the samples covered with Si3N4, SiO2and SOG were annealed at

the temperature of 650, 700, 750, 800 and 850 -C, respectively, under pure nitrogen protection at atmospheric pressure.

Fig. 2 shows the annealing temperature dependence of band gap shift. It can be observed that the band gap of PL peak varies with the RTA temperature. For low annealing temperature range of 650 – 750 -C, the PL peak has little change, however, when the annealing temperature was beyond 750 -C, the PL peak moves to short wavelength evidently. This may be due to the impurity-free vacancy enhanced interdiffusion created by the dielectric layer deposition and annealing.

On the other hand, we also performed photoreflectance (PR) measurements in accordance with the results of PL spectra, as shown in Fig. 3. It shows that PR results are consistent with the PL results. This indicates that PR can be used as a supplementary way to study the luminescent

0.76 0.78 0.80 0.82 0.84 0.00 0.05 0.10 0.15 0.20 0.25 Si3N4

control sample _SiO

2

PL Intensity (a.u.)

Energy (eV)

Fig. 1. The PL spectra of the as-grown control sample and SiO2, Si3N4

covered samples (Group A).

650 700 750 800 850 0 4 8 12 16 20

Blue Shift of Pl Peak (meV)

Annealing Temperature (°C) Si₃N₄

SiO SOG

Fig. 2. The temperature dependence of blueshift for different dielectric covered samples (Group A).

0.70 0.75 0.80 0.85 0.90 850°C,30s 800°C,30s 750°C,30s 700°C,30s 650°C,30s control sample PR Intensity (a.u) Energy (eV)

Fig. 3. Band gap blueshift measured by PR with different annealing temperatures (for samples in Group A).

(4)

characteristics of the band gap shift. Furthermore, PR spectra can provide some other information, for example, some detailed data about different layer luminescent characteristics through analyzing PR spectra[5].

In order to analyze the effects of the combination of the cladding layer and dielectric covered layer, the annealing temperature dependence of the band gap blueshift for samples B1 and B2 with different dielectric layers were further studied, based upon the PL data. Sample B1 has an InP cladding structure, and sample B2 has an InGaAs cladding layer. Both samples B1 and B2 have the same MQWs except for the cladding layer. These two types of samples were measured under the same experimental condition in order to examine their band gap blueshift.

Fig. 4shows the dependences of the blueshift PL peak on the annealing temperature, caused by different covered layer of Si3N4and SiO2, respectively. It can be observed that the

induced blueshift from the sample with the InP – Si3N4

combination is larger than that with the InP – SiO2

combi-nation. For example, the blueshift of the sample with the

InP – Si3N4 cap layer combination reaches 50 meV at 850

-C, but the blueshift of the sample with the InP–SiO2cap

layer combination is only 40 meV at the same annealing temperature. On the other hand, the combination of InGaAs – Si3N4 covered layer caused a blueshift of 20

meV, however the combination of InGaAs – SiO2reached a

value of 43 meV.

From these results, we can conclude that the combination layer of InP – Si3N4or InGaAs – SiO2can create larger band

gap blueshift in comparison to the combination of InP – SiO2

and InGaAs – Si3N4. The reason for these experimental

results is not very clear yet, but in our opinion, it can be explained as follows: the vacancies are produced in both groups III and V. It is known that in GaAs material the following reactions are prompted to produce large number of vacancies at the interface:

4GaAsþ 3SiO2S4Ga þ 2As2O3þ 3Si ð1Þ

As2O3þ 2GaAsSGa2O3þ 4As: ð2Þ

Our results indicate that the vacancies generated by the InP – SiO2combination are less than that by the InGaAs –

SiO2combination layers because of the absence of Ga. As

for the group V vacancies, we cannot provide the detailed reactions as to how the vacancies are produced. This may be due to the effects on the combination of the cladding layer and dielectric layers. Further penetrating investigation is needed.

4. Conclusion

In conclusion, we have studied the dependence of the band gap blueshift on the annealing temperature and dielectric layers from the InGaAsP/InP multiple quantum wells prepared by Impurity-Free Vacancy Disordering (IFVD). Both PL and PR results, which are consistent each other, showed that this band gap blueshift increases with the annealing temperature for all samples especially when the temperature is over 750-C. Our results indicate that PR can also be used as a supplementary technique, in addition to PL, to study the band gap blueshift. On the other hand, to obtain larger energy shift the optimal selected cap layer combination with the dielectric layer is necessary. From our experiments the combination of InP – Si3N4 or InGaAs –

SiO2 is better than the combination of InGaAs – Si3N4 or

InP – SiO2.

Acknowledgments

The authors would like to thank Dr. B. Robinson at McMaster University Canada for providing the GSMBE samples, Dr. P. Jin at Institute of Semiconductors, Beijing, for PR measurements, Prof. S. J. Chua at Institute of Materials Research and Engineering, Singapore for help and

650 700 750 800 850 0 5 10 15 20 25 30 35 40 45 50 55

Blue Shift of PL Peak (meV)

Temperature (°C) Temperature (°C) InP+SiO₂ InP+Si3N4 (a) 650 700 750 800 850 0 10 20 30 40 50 InGaAs+SiO2 InGaAs+Si3N4 (b)

Fig. 4. The temperature dependences of blueshift for the samples with (a) InP – SiO2and InP – Si3N4combination layers, and (b) InGaAs – SiO2and

(5)

support of PL measurements, and X.D. Zhang and H.H. Liu for technical assistants. This project at Tianjin Normal University is supported by the Nature Science Foundation of China (NSFC) with a contract number of 60276013. The work at National Taiwan University was supported by funds from National Science Council of Republic of China, NSC 93-2218-E-002-011 and 93-2215-E-002-035.

References

[1] D. Hofstetter, B. Maisenho¨lder, H.P. Zappe, IEEE J. Sel. Top. Quantum Electron. 4 (1998) 794.

[2] H.-S. Kim, J.W. Park, D.K. Oh, K.R. Oh, S.J. Kim, I.-H. Choi, Semicond. Sci. Technol. 15 (2000) 1005.

[3] J. Zhao, Y. Wang, Surf. Coat. Technol. 131 (2000) 340.

[4] J.H. Teng, J.R. Dong, S.J. Chua, M.Y. Lai, B.C. Foo, D.A. Thompson, B.J. Robinson, A.S.W. Lee, John Hazell, Irwin Sproule, J. Appl. Phys. 92 (2002) 4330.

[5] R. Kudrawiec, G. Sek, W. Rudno-Rudzinski, J. Misiewicz, J. Wojcik, B.J. Robinson, D.A. Thompson, P. Mascher, Mater. Sci. Eng., B, Solid-State Mater. Adv. Technol. 101 (2003) 137.

[6] B.E. Gordon, A.S.W. Lee, D.A. Thompson, B.J. Robinson, Semicond. Sci. Technol. 18 (2003) 782.

[7] J.H. Marsh, Semicond. Sci. Technol. 8 (1993) 1136.

[8] D. Gdeepe, N. Holonyak Jr., J. Appl. Phys. 64 (1988) R93. [9] P.G. Piva, R.D. Goldberg, I.V. Mitchell, H. Chen, R.M. Feenstra, G.C.

Weatherly, D.W. McComb, G.C. Aers, P.J. Poole, S. Charbonneau, Appl. Phys. Lett. 72 (1998) 1599.

[10] S.R. Andrew, J.H. Marsh, M.C. Holland, A.H. Kean, IEEE Photonics Technol. Lett. 4 (1992) 426.

[11] B.B. Elenkrig, D.A. Thompson, J.G. Simmons, D.M. Bruce, Yu Si, Jie Zhao, J.D. Evans, I.M. Templeton, Appl. Phys. Lett. 65 (1994) 1239. [12] M. Paquette, J. Beauvais, J. Beerens, P.J. Poole, S. Charbonneau, C.J.

Miner, C. Blaauw, Appl. Phys. Lett. 71 (1997) 3749.

[13] J.H. Lee, S.K. Si, Y.B. Moon, E.J. Yoon, S.J. Kim, Electron. Lett. 33 (1997) 1179.

[14] G. Li, S.J. Chua, S.J. Xu, X.C. Wang, A. Saher Helmy, Mao-Long Ke, J.H. Marsh, Appl. Phys. Lett. 73 (1998) 3393.

[15] J.S. Yu, J.D. Song, Y.T. Lee, J. Appl. Phys. 92 (2002) 1386. [16] S. Burkner, M. Maier, E.C. Larkins, W. Rothemund, E.P. O’Reilly,

J.D. Ralston, J. Electron. Mater. 24 (1995) 805.

[17] G. Li, S.J. Chua, J.H. Teng, W. Wang, Z.C. Feng, H. Huang, T. Osipowicz, J. Vac. Sci. Technol., B 17 (1999) 1507.

[18] F. Camacho, E.A. Avrutin, P. Cusumano, A.S. Helmy, A.C. Bryce, J.H. Marsh, IEEE Photonics Technol. Lett. 9 (1997) 1208.

[19] W.J. Choi, S.M. Han, S.I. Shah, S.G. Choi, D.H. Woo, S. Lee, S.H. Kim, J.I. Lee, K.N. Kang, J. Cho, IEEE J. Sel. Top. Quantum Electron. 4 (1998) 624.

[20] N. Shimada, Y. Fukumoto, M. Uemukai, T. Suhara, H. Nishihara, A. Larsson, Jpn. J. Appl. Phys. Part 1 39 (2000) 5914.