5
Physica B 327 (2003) 15–19 Electron traps in metalorganic chemical vapor deposition grown Al 0.2 Ga 0.8 As R. Ajjel a, *, L. Bouzrara a , M.A. Za . ıdi a , H. Maaref a , G. Br ! emond b a Laboratoire de Physique des Semiconducteurs et des Composants ! electronique, Facult ! e des Sciences, 5019 Monastir, Tunisia b Laboratoire de Physique de la Mati ! ere (UMR 5511), INSA de Lyon, Bat. 502, 20 Av. A. Einstein, 69621Villeurbanne, France Received 11 July 2002; received in revised form 9 October 2002 Abstract The effect of the growth temperature on deep electron traps present in n-type Al 0.2 Ga 0.8 As layers grown by metalorganic chemical vapor deposition was investigated using deep level transient spectroscopy. Four electron traps have been found and were labeled A–D. Trap A is present only in metalorganic chemical vapor deposition grown epilayers, while electron traps B–D are present in all of AlGaAs layers independent of the growth technique. A correlation of the growth temperature to both the electron trap concentration, and the minority carrier lifetime leaded us to suggest that the electron trap D behaves as a recombination center, which controls the minority carrier lifetime. r 2002 Elsevier Science B.V. All rights reserved. Keywords: Electron traps; Metalorganic chemical vapor deposition; Al 0.2 Ga 0.8 As 1. Introduction Advanced material growth technology such as metalorganic chemical vapor deposition (MOCVD) technique enables the growth of high quality Al x Ga 1x As layers technologically impor- tant for device applications (heterostructure field effect transistors, laser diodes, light emitting diodes and high efficiency solar cells). The characterization of deep levels is an important indicator of the material quality, since they may control the minority carrier lifetime and hence the radiative recombination efficiency, which is im- portant for optoelectronic device performance. Several experimental techniques can be used for defect characterization. Among these techniques, deep level transient spectroscopy (DLTS) is very prominent [1]. For Al x Ga 1x As, most of the studies have been focused on the properties of the DX center [2–4], which is the dominant electron trap commonly present in n-type doped layers with an Al content x > 0:22; irrespective of the growth technique. However, only a few papers have been published on the other traps in MOCVD and MBE grown Al x Ga 1x As. Moreover, there are serious discre- pancies among the activation energies of the traps reported in the literature. Johnson et al. [5] have detected the presence of four electron traps in addition to the EL2 center. They have reported a monotonical increase in the activation energy for the latter center with increased alloy composition. *Corresponding author. Fax: +216-73-500-278. E-mail address: [email protected] (R. Ajjel). 0921-4526/03/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved. PII:S0921-4526(02)01694-0

Electron traps in metalorganic chemical vapor deposition grown Al0.2Ga0.8As

  • Upload
    r-ajjel

  • View
    216

  • Download
    2

Embed Size (px)

Citation preview

Physica B 327 (2003) 15–19

Electron traps in metalorganic chemical vapor depositiongrown Al0.2Ga0.8As

R. Ajjela,*, L. Bouzraraa, M.A. Za.ıdia, H. Maarefa, G. Br!emondb

aLaboratoire de Physique des Semiconducteurs et des Composants !electronique, Facult!e des Sciences, 5019 Monastir, TunisiabLaboratoire de Physique de la Mati!ere (UMR 5511), INSA de Lyon, Bat. 502, 20 Av. A. Einstein, 69621Villeurbanne, France

Received 11 July 2002; received in revised form 9 October 2002

Abstract

The effect of the growth temperature on deep electron traps present in n-type Al0.2Ga0.8As layers grown by

metalorganic chemical vapor deposition was investigated using deep level transient spectroscopy. Four electron traps

have been found and were labeled A–D. Trap A is present only in metalorganic chemical vapor deposition grown

epilayers, while electron traps B–D are present in all of AlGaAs layers independent of the growth technique. A

correlation of the growth temperature to both the electron trap concentration, and the minority carrier lifetime leaded

us to suggest that the electron trap D behaves as a recombination center, which controls the minority carrier lifetime.

r 2002 Elsevier Science B.V. All rights reserved.

Keywords: Electron traps; Metalorganic chemical vapor deposition; Al0.2Ga0.8As

1. Introduction

Advanced material growth technology such asmetalorganic chemical vapor deposition(MOCVD) technique enables the growth of highquality AlxGa1�xAs layers technologically impor-tant for device applications (heterostructure fieldeffect transistors, laser diodes, light emittingdiodes and high efficiency solar cells). Thecharacterization of deep levels is an importantindicator of the material quality, since they maycontrol the minority carrier lifetime and hence theradiative recombination efficiency, which is im-portant for optoelectronic device performance.

Several experimental techniques can be used fordefect characterization. Among these techniques,deep level transient spectroscopy (DLTS) is veryprominent [1].For AlxGa1�xAs, most of the studies have been

focused on the properties of the DX center [2–4],which is the dominant electron trap commonlypresent in n-type doped layers with an Al contentx > 0:22; irrespective of the growth technique.However, only a few papers have been publishedon the other traps in MOCVD and MBE grownAlxGa1�xAs. Moreover, there are serious discre-pancies among the activation energies of the trapsreported in the literature. Johnson et al. [5] havedetected the presence of four electron traps inaddition to the EL2 center. They have reported amonotonical increase in the activation energy forthe latter center with increased alloy composition.

*Corresponding author. Fax: +216-73-500-278.

E-mail address: [email protected] (R. Ajjel).

0921-4526/03/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved.

PII: S 0 9 2 1 - 4 5 2 6 ( 0 2 ) 0 1 6 9 4 - 0

On the other hand, Wagner et al. [6] havemeasured an activation energy ofB0.84 eV, whichremains nearly independent on x in the range 0–0.35. Andr!e et al. [7] have reported the presence ofthree electron traps at 0.3, 0.5 and 0.8 eV and theyfound that their concentrations increase withdecreasing growth temperature.In this paper, we report DLTS measurements

performed on n-type Al0.2Ga0.8As grown byMOCVD. We have detected four electron trapslabeled A–D. We have determined their character-istics, i.e., their corresponding ionization energies,their concentrations, their capture cross sectionsand their barriers associated with the electroncapture. Attention has been also paid to the defectD and its minority carrier lifetime. This shows thatthis electron trap acts as a recombination center,which controls the minority carriers lifetime.

2. Results and discussion

The samples under investigation were grown onGaAs by MOCVD technique at three differentgrowth temperatures: 7101C, 7501C and 7801C.The III/V ratio was around 85 with a growth rateof B3 mm/h. The sample numbers as well as theirgrowth temperatures are summarized in Table 1.The choice of an Al content x ¼ 0:2 has been madein order to avoid the manifestation of the DXcenter which might otherwise obscure otherobservable traps. The doping concentration, ob-tained from capacitance–voltage CðV Þ measure-ments, is in the order of 1017 cm�3 in all samples.

DLTS measurements allow us to determine theapparent ionization energy Ei of the trap. This isobtained by performing the measurement of therate window en versus temperature. The slope ofthe ln(T2=en) versus T�1 gives Ei: Our measure-ments were performed in the temperature range ofabout 77–400K.A typical DLTS spectrum is shown in Fig. 1.

The spectrum exhibits four peaks, labeled A–Dwith increasing energy. DLTS spectra of the threesamples have been recorded under the sameconditions, i.e., for a bias voltage V0 ¼ �1V, afilling pulse DV ¼ 1V, a filling durationtp ¼ 0:5ms. The arrhenius plots of the electrontraps observed in the three samples studiedare shown in Fig. 2. The physical parameters,obtained from the analysis of these data forlevels A–D, are summarized in Table 1. Theslope of an Arrhenius plot (the electrical signatureof the defect) gives an apparent ionization energy,which must be eventually corrected for thetemperature variation of the cross section s toobtain the true energy level of the trap. Indeed, inthe case of thermally activated capture crosssection, i.e., s ¼ s0 exp �EB=kT

� �; the apparent

ionization energy Ei is the sum of the actualionization energy ET and a capture barrier EB:Ei ¼ ET þ EB:Since the electron capture cross section is

thermally activated, the obtained largevariation of s0 is due to the strong temperaturedependence of the filling factor f for a givenvalue of tp [8]. We attribute this effect to thepossible presence of an electric field in agreementwith Ref. [8].

Table 1

Physical quantities obtained from the analysis of the DLTS spectra

Samples 671 672 673

Growth temperature7101C 7501C 7801C

Defects DEi(eV)

NT

(cm3) � 1014s0(cm2)

DEi(eV)

NT

(cm3) � 1014s0(cm2)

DEi(eV)

NT

(cm3) � 1014s0(cm2)

A 0.13 0.9 7.2� 10�18 0.19 1 5.7� 10�15 0.15 1 2.3� 10�17

B 0.23 1.1 1.5� 10�18 0.23 1 5.8� 10�17 0.28 0.2 3.78� 10�14

C 0.26 0.5 1.68� 10�15 0.3 0.50 3.31� 10�18 0.3 0.4 9.7� 10�18

D 0.5 1 3.1� 10�16 0.56 0.47 7.85� 10�19 0.57 0.1 6.18� 10�16

R. Ajjel et al. / Physica B 327 (2003) 15–1916

We have measured the capture barrierusing the method proposed by Criado et al. [9].In this method, the barrier EB can be de-duced from the temperature dependence of theDLTS spectrum amplitude DC0; i.e., from theslope of the logarithm of the quantityT�1=2 ln 1� ðDC0ðtp;TÞ=DCiÞ

� �where DCi is the

peak amplitude corresponding to completely filledcenters (Fig. 3). The measured barriers, as well asto the apparent and actual ionization energies, arereported in Table 2. The large values of the capturebarriers indicate a strong electron–phonon cou-pling, their variations are related to the nature ofthe impurity [10] and to the temperature, showing

that in a capture process which is mediated bymultiphonon emission, EB is temperature depen-dent [11].A comparison of the concentration of the

defects A–D for the three studied samples (seeTable 1) shows that the concentration of levels Aand C remains nearly constant for the threegrowth temperatures. Based on the energeticposition and taking into account the capturebarrier height, we ascribe the electron trap A tothe Ec�0.13 eV one reported by Johnson et al. [5]and which seems to be present only in MOCVDgrown epilayers. Indeed, reported DLTS results onMBE grown AlxGa1�xAs did not reveal thepresence of a similar electron trap [12]. Deeplevels B–D are present in all n-type AlGaAs layersirrespective of the crystal growth technique [13].However, their concentrations depend on growthconditions and they are particularly sensitive tothe mole flux ratio g ¼ As=Ga: These electrontraps can be identified with traps ME2, ME3 andME6 reported by Yamaka et al. [12] andNaritsuka et al. [13]. Moreover, these electrontraps are similar to the electron irradiation induceddefects reported by Pons [14]. Consequently, wecan attribute them to intrinsic defects, i.e., isolatedinterstitials, vacancies, their binary complexes orintrinsic defects complexed with doping or residualimpurities. We expect that A and C traps levelsoriginate from the same center which is related toan arsenic vacancy. B and D traps are also believedto arise from the same center and they areprobably formed by impurities involving oxygen[15,16]. Indeed, both O2 and water vapor can existin vapor phase reactors and may also be intro-duced by impurities in source materials. Oxygenrelated deep levels act as a recombination centersand then reduce the minority carrier lifetimes [17–19].It has been demonstrated that oxygen incor-

poration is proportional to Al content [20]. Theoxygen concentration in each sample was mea-sured by secondary ion mass spectroscopy (SIMS)using a Cs+ primary ion beam. Absolute concen-trations were obtained by comparison with animplanted GaAs standard. The oxygen concentra-tion decreases with increased growth temperature.Minority carrier lifetimes were deduced from PL

50 100 150 200 250 300 350

0.00

0.02

0.04

0.06

0.08

AB

C

D

DL

TS

SIG

NA

L (

pF

)

TEMPERATURE (K)

Fig. 1. Typical DLTS spectrum for MOCVD grown n-type

Ga0.8Al0.2As.

2 4 6 8 10

2

4

6

8

10

ln (

T2 /e

n) (

K2 s)

1000/T (K-1)

Fig. 2. Arrehnius plots of the electron traps observed in

MOCVD grown n-type Ga0.8Al0.2As, (’) 671, (�) 672, (,) 673.

R. Ajjel et al. / Physica B 327 (2003) 15–19 17

decay measurements. Table 3 gives the PL decaytimes measured at room temperature. The increasein the effective lifetime with increasing growthtemperature agrees with the trends reported in theliterature [21–27]. These results imply that theoxygen introduction decreases as the growth

temperature increases, results in agreementwith those reported by Goorsky et al. [28]and by Ahrenkiel et al. [29]. It is worth noticingthat Bhattacharya et al. [25] have shown thatthe PL intensity of MOCVD grown AlGaAsis inversely proportional to the concentrationof the two traps ET ¼ 0:2570:04 eV andET ¼ 0:3570:04 eV. Thus, it has been concludedthat these two traps, thought to be related tooxygen, are mainly responsible of the reduction inPL intensity. An inspection of data listed in Table3 shows that the concentration of the electron trapD is correlated to the oxygen concentration andalso the minority carrier lifetime reduces as theconcentration of this trap increases. Thus, wesuggest that the electron trap D controls theminority carrier lifetime.

3 4 5 6 7 8 9 10 11

2

3

4

5

AB

C

-ln (

-T-1

/2ln

(1-

∆C0/ ∆

Ci)

1000/T (K-1)

Fig. 3. Variation of the logarithm of �T�1=2 ln 1� ðDC0ðtp;TÞ=DCiÞ� �

as a function of 1000/T :

Table 3

D level defect concentration, oxygen concentration and

minority carrier lifetime measured by PL decay as a function

of the growth temperature

Samples 671 672 673

Growth temperature (K) 710 750 780

D level concentration � 1014 (cm�3) 1 0.47 0.1

Oxygen concentration � 1017 (cm�3) 1.85 1.50 1.2

Lifetime (ns) 2 3 13

Table 2

Barrier energies and actual ionization energies of electron traps detected in MOCVD grown n-type Ga0.8Al0.2As

Defect A B C

671 672 673 671 672 673 671 672 673

Ei (eV) 0.13 0.19 0.15 0.23 0.23 0.28 0.26 0.30 0.30

EB (eV) 00 0.006 0.03 0.06 0.05 0.02 0.08 0.05

ET (eV) 0.13 0.13 0.20 0.17 0.23 0.24 0.22 0.25

R. Ajjel et al. / Physica B 327 (2003) 15–1918

3. Conclusion

In this study, we have examined the effects ofthe growth temperature on the electron traps inn-type MOCVD grown Al0.2Ga0.8As with Alfraction of x ¼ 0:2: Four electron traps have beenfound and were labeled A–D. While B–D electrontraps are present in all AlGaAs layers independentof the growth technique, with growth conditionsdependent concentrations, the electron trap Aseems to be present only in MOCVD grownepilayers. It has been shown that B and D trapsare associated with oxygen. Also the minoritylifetime is suggested to be controlled by theelectron trap D.

References

[1] D.V. Lang, J. Appl. Phys. 45 (1974) 3023.

[2] D.J. Chadi, Phys. Rev. B 46 (1992) 6777.

[3] A. Triki, H. Mejri, F. Rziga, A. Selmi, Phys. Status Solidi

(b) 227 (2001) 541.

[4] F. Rziga-Ouaja, H. Mejri, A. Triki, A. Selmi, A. Rebey, J.

Appl. Phys. 88 (2000) 2583.

[5] N.M. Johnson, R.D. Burnham, D. Feteke ans Yinling,

Defects in semiconductors, Proceedings of the Materials

Research Meeting Society, November, 1980, North Hol-

land Publishing, New York, Oxford, 1981, pp. 481–486.

[6] D. Wagner, D.E. Mars, G. Hom, C.B. Stringfellow, J.

Appl. Phys. 20 (1981) L.

[7] J.P. Andr!e, M. Boulou, A. Mircea-Roussel, J. Cryst.

Growth 55 (1981) 192.

[8] M. Zazoui, S.L. Feng, J.C. Bourgoin, Semicond. Sci. Tech.

6 (1991) 973.

[9] J. Criado, A.I. Gomez, E. Calleja, E. Munoz, Appl. Phys.

Lett. 52 (1988) 660.

[10] M. Zazoui, S.L. Feng, J.C. Bourgoin, Phys. Rev. B 44

(1991) 10898.

[11] M. Zazoui, V. Dontchev, J.C. Bourgoin, Phys. Rev. B 47

(1993) 4296.

[12] K. Yamaka, S. Narisuka, K. Kanamoto, M. Mihara, M.

Ishii, J. Appl. Phys. 61 (1987) 5062.

[13] S. Naritsuka, K. Yamaka, M. Mihara, M. Ishii, Jpn.

J. Appl. Phys. 23 (1984) LI.

[14] D. Pons, Physica B 116 (1983) 388.

[15] M. Sakamoto, T. Okada, Y. Mori, J. Appl. Phys. 58 (1985)

337.

[16] J.H. Neave, P. Blood, B.A. Joyce, Appl. Phys. Lett. 36

(1980) 311.

[17] J. Zhang, B.M. Keyes, S.E. Asher, R.K. Ahrenkiel,

M.L. Timmons, Appl. Phys. Lett. 63 (1993) 1369.

[18] J.W. Huang, T.F. Kuech, Appl. Phys. Lett. 65 (1994) 604.

[19] R.A.J. Thomeer, P.R. Hageman, L.J. Giling, Appl. Phys.

Lett. 64 (1994) 1561.

[20] E.H. Bakraji, Thesis, Paris, 1990.

[21] K. Akimoto, M. Kamada, K. Taira, M. Aira, N.

Watanabe, J. Appl. Phys. 59 (1986) 2833.

[22] M. Yamaguchi, C. Amano, H. Sigiura,A. Yamamoto,

Proceedings of the 19th IEEE Photovoltaics Specialists

Conference, IEEE, 1987, p. 1484.

[23] H. Tereo, H. Sunakawa, J. Cryst. Growth 68 (1984) 157.

[24] J.M. Ryan, J.W. Huang, T.K. Kuech, K.L. Bray, J. Appl.

Phys. 76 (1994) 1175.

[25] P.K. Bhattacharya, S. Subramanian, M.J. Ludowise,

J. Appl. Phys. 55 (1984) 3664.

[26] M. Hata, H. Takata, T. Yako, N. Fukuhara, T. Maeda,

Y. Uemura, J. Cryst. Growth 124 (1992) 427.

[27] C.T. Faxon, J.B. Clegg, K. Woodbridge, D. Hilton, P.

Dauson, P. Blood, J. Vac. Sci. Technol. B 3 (1985) 703.

[28] M.S. Goorsky, T.F. Kuech, F. Cordane, P. Mooney, G.J.

Scilla, R.M. Potemski, Appl. Phys. Lett. 58 (1991) 1979.

[29] R.K. Ahrenkiel, B.M. Keys, T.S. Shen, J.I. Chyi, H.

Morkoc, J. Appl. Phys. 69 (1991) 3094.

R. Ajjel et al. / Physica B 327 (2003) 15–19 19