- Email: [email protected]
GaAs tandem solar cells using carbon doped GaAs in tunnel junction
Microelectronic Engineering 87 (2010) 677–681
Contents lists available at ScienceDirect
Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee
Enhanced efficiency in GaInP/GaAs tandem solar cells using carbon doped GaAs in tunnel junction Chang Zoo Kim a, Hogyoung Kim b,*, Keun Man Song a, Dong Hwan Jun a, Ho Kwan Kang a, Wonkyu Park a, Chul Gi Ko a a b
Korea Advanced Nano Fab Center, 906-10, Iui-Dong, Yeongtong-Gu, Suwon, Gyeonggi-Do 443-270, Republic of Korea College of Humanities and Sciences, Hanbat National University, San 6-1, Dukmyung-Dong, Yuseong-Gu, Daejeon 305-719, Republic of Korea
a r t i c l e
i n f o
Article history: Received 21 May 2009 Received in revised form 28 August 2009 Accepted 25 September 2009 Available online 4 October 2009 Keywords: Carbon doping GaAs MOCVD Tandem solar cells
a b s t r a c t Carbon doping of GaAs using CBr4 (carbon tetrabromide) in metal-organic chemical vapor deposition (MOCVD) was investigated to obtain very high and sharp doping profiles required for tunnel junction in tandem solar cells. It was found that the hole concentration increased with decreasing growth temperature and V/III ratio. Hole doping profiles versus distance from the sample surface showed that the hole concentration near the surface was very low in comparison with that far below the surface. As a postgrowth treatment, CBr4 was supplied during the cool down process and produced almost constant hole concentration of 1 1020 cm3 regardless of the depth, when CBr4 flow rate was 9.53 lmol/min. Based on these results, solar cells were fabricated using both carbon (C) and zinc (Zn) as a p-type dopant. It was shown that C doping exhibits higher efficiency and lower series resistance than those of Zn doping in GaInP/GaAs tandem solar cells. Ó 2009 Elsevier B.V. All rights reserved.
1. Introduction Carbon (C) has been researched extensively as a p-type dopant in GaAs and related materials for applications in electronic and optoelectronic devices such as heterojunction bipolar transistors (HBTs), vertical cavity surface emitting lasers (VCSELs) and solar cells [1–3]. Doping levels up to 1 1020 cm3 have also been realized with less morphological degradation using metal-organic chemical vapor deposition (MOCVD) [4,5]. Carbon has lower diffusion coefficient, higher solubility, and smaller memory effect than zinc (Zn) [6–8]. Thus, in contrast to Zn doping, highly C-doped layers can be obtained during overgrowth with negligible tailing effect [9]. The low diffusion coefficient and the lack of tailing make C doping more attractive than Zn doping for devices that require highly doped and well-defined p-type doping profiles, such as tunnel junctions [3] and HBTs [10]. Carbon doping can be performed using two different methods such as intrinsic and extrinsic doping. In intrinsic carbon doping, carbon atoms from the group III precursor, trimethylgallium (TMGa) can be incorporated into the deposited material by reducing the growth temperature and the V/III precursor concentration ratio [11–13], which requires a tight control of growth parameters when using arsine (AsH3) as the group V precursor. On the other hand, carbon atoms can be introduced extrinsically into the depos* Corresponding author. E-mail address: [email protected] (H. Kim). 0167-9317/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.mee.2009.09.014
ited material using halomethanes such as carbon tetrachloride (CCl4), carbon tetrabromide (CBr4) and carbon bromidetrichloride (CBrCl3) because they have high doping efficiencies due to weak C–Cl or C–Br bonds [14–16]. This method makes it easier to control the carbon doping density widely without changing growth conditions such as growth temperature and V/III ratio. High carbon doping levels (>1020 cm3) have been achieved using CCl4 as a carbon source [17]. However, a strong reduction of growth rate was observed due to the in situ etching of GaAs by Cl [18,19]. The reduction in the growth rate of C-doped GaAs was also reported when using CBr4 due to the HBr etching [14]. The use of CBrCl3 was reported by Uchida et al. [20] to reduce the etching of GaAs. When fabricating tandem solar cells, very thin and wide-bandgap tunnel junction is necessary to minimize optical absorption. However, since the tunnel peak current density decreases exponentially with the increase in bandgap energy, it is not easy to obtain low-resistance wide-bandgap tunnel junction. An GaInP tunnel junction has been tried in GaInP/GaAs tandem cell using double heterostructure (DH) to prevent the impurity diffusion (Zn) [3]. The AlGaAs/GaAs tandem solar cells were reported to reduce the voltage losses effectively across the tunnel junction by incorporation of semimetallic ErAs nanoparticles into the GaAs tunnel junction, where Be and Si were used for p-dopant and ndopant, respectively [21]. Although lots of research has been done to develop the GaAs-based solar cells, there are relatively few reports on the GaInP/GaAs tandem solar cells using carbon doped GaAs in the tunnel junction. For the present work, we describe
678
C.Z. Kim et al. / Microelectronic Engineering 87 (2010) 677–681
the carbon doping profiles of GaAs grown by MOCVD under variation of the growth parameters such as CBr4 flow rate, growth temperature, V/III ratio and post-growth treatment. In addition, the conversion efficiency for the GaInP/GaAs tandem solar cells with GaAs tunnel junction using both C and Zn as a p-type dopant is discussed.
3.0
(a)
2.5
2.0
0.002
0.003
0.004
0.005
CBr4 flow rate ( µmol/min)
2. Experimental
19
-3
Hole concentration ( x 10 cm )
8
Growth was carried out in a multiwafer MOCVD reactor (AIX2600G3 IC) on semi-insulating GaAs (100) substrates with a misorientation of 2° toward <1 1 1> plane. TMGa and AsH3 were used as the precursors of Ga and As, respectively. The carrier gas was hydrogen and the reactor pressure was kept at 50 mbar. Carbon doping was achieved using the extrinsic source of CBr4. The CBr4 bath temperature was maintained at 25 °C. All samples used in this study were grown with a constant TMGa flow rate of 265.05 lmol/min. The total flow rate in the reactor was 58 slm. The influence of three growth parameters such as growth temperature, CBr4 flow rate and V/III ratio on the hole concentration was first investigated to obtain very high doping levels up to 1 1020 cm3. After growth process, AsH3 was supplied to all samples until the temperature cooled down to 350 °C in order to prevent As outdiffusion. Thereafter, the effect of CBr4 flow rate under the cool down process on the hole concentration was studied in order to obtain well-defined carbon doping profiles. The CBr4 flow rate was changed from 0 to 9.53 lmol/min. Electrochemical capacitance–voltage (ECV) measurements were used to determine the hole concentration of carbon doped GaAs layers. Atomic force microscopy (AFM) was used to investigate the surface morphology.
6
(b)
4 2
540
560
580
600
o
Growth temperature ( C) 10 8
(c)
6 4 2
20
40
60
V/III ratio
10
(d) 5
reference
3. Results and discussion Fig. 1a shows the hole concentration versus CBr4 flow rate. The growth temperature and the V/III ratio were 600 °C and 57.25, respectively. When CBr4 flow rate increases, the hole concentration remains almost constant, although there is a slight increase. Kurpas et al. [22] demonstrated that there is a sublinear relationship between hole concentration and CBr4 partial pressure which is expressed as p / [CBr4]0.7. This relation shows that the hole concentration increases linearly at relatively low doping levels and tends to saturate at high doping levels. Watanabe et al. [23] investigated the behavior of the hole concentration in carbon doped GaAs grown by MOCVD using CBr4 as a carbon source and concluded that the hole concentration tends to saturate at higher CBr4 flow rates in the very high doping region due to the saturation of carbon incorporation efficiency, not the deactivation of carbon acceptors. The same mechanism can be applied to explain the saturation of the hole doping levels. Fig. 1b shows the hole concentration as a function of growth temperature. The CBr4 flow rate and the V/III ratio were kept at 21.43 lmol/min and 57.25, respectively. The growth temperature was varied from 550 to 600 °C. It is clearly observed that the hole concentration was increased by decreasing the growth temperature. Fig. 1c shows the hole concentration versus V/III ratio. The growth temperature and the CBr4 flow rate were 550 °C and 21.43 lmol/min, respectively. When the V/III ratio (or AsH3 flow rate) decreases, the hole concentration increases accordingly and reaches to 1 1020 cm3. These results imply that the hole concentration increases with decreasing growth temperature and V/III ratio, which are generally observed in the carbon doped p-type GaAs epilayer. In intrinsic carbon doping, carbon incorporation is possible from the hydrogen radicals generated by the decomposition of TMGa [16]. When AsH3 is present, these hydrogen radicals can be present
0
50
100
150
Depth (nm) Fig. 1. Hole concentration as a function of (a) CBr4 flow rate, (b) growth temperature and (c) V/III ratio, and (d) hole doping profile versus depth for the reference sample that CBr4 flow was turned off after growth process.
in the form of methane and be desorbed at the GaAs surface, reducing carbon incorporation in GaAs. This mechanism can explain the rapid increase in the hole concentration when V/III ratio decreases. As discussed by Watanabe et al. [23], lower growth temperature makes the adsorption coefficient larger and a lower V/III ratio provides more sites for CBrk to adsorb. Therefore, carbon incorporation can be enhanced further with the presence of CBr4 at low growth temperature and low V/III ratio. In addition, many carbon atoms from hydrogen radicals by TMGa will incorporate with decreasing V/III ratio, increasing hole concentration further. As discussed above, very high carbon doping (1020 cm3) could be achieved at the growth temperature of 550 °C, the CBr4 flow rate of 21.43 lmol/min, and the V/III ratio of 16.84, respectively, and these growth conditions were used to investigate the effect of in situ CBr4 treatment after growth process on the carbon doping profiles. Fig. 1d shows the hole concentration as a function of depth from the surface without supplying CBr4 after growth process (denoted as reference). As the depth increases, the hole concentration increases and reaches to the value of 1020 cm3 when the depth is around 150 nm. When making tunnel junction by growing Sidoped n-type GaAs layer on the carbon doped p-type layer, such a doping profile which has quite low doping levels near the p-type GaAs surface cannot be used for the fabrication of tunnel junctions with high performance and low resistance. In this respect, we examined the influence of CBr4 flow rate under the cool down process on the hole doping profiles.
C.Z. Kim et al. / Microelectronic Engineering 87 (2010) 677–681
679
Fig. 2. Schematic growth process as a function of time. After growth process, CBr4 was supplied during the cool down process.
Fig. 2 shows the growth process graphically as a function of time. Instead of turning off the CBr4 flow after growth process, CBr4 was supplied during the cool down process until the temperature dropped down to 350 °C. Three different CBr4 flow rates such as 2.38 lmol/min (denoted as C1), 4.76 lmol/min (denoted as C2) and 9.53 lmol/min (denoted as C3) were employed. Fig. 3 shows the AFM images presenting the change of surface morphology of the CBr4 treated samples after growth process. All the samples were scanned over 2 lm 2 lm areas. The surface morphology of the reference sample was mirror-like regardless of the etching effect due to bromide radicals. After CBr4 treatment, the surface morphology was changed only marginally and was mirror-like, too. The root mean square (rms) roughness of all the samples was in the range of 0.2–0.4 nm. Fig. 4a shows the hole doping profiles versus depth from the sample surface. It is clearly observed that the hole concentration near the sample surface increased with increasing CBr4 flow rate. When the CBr4 flow rate was 9.53 lmol/
Fig. 4. Hole doping profiles versus depth after annealing at 500 °C for 10 min in an N2 ambient (a) for the samples with different CBr4 flow rate under the cool down process and (b) for the reference sample.
min, the hole concentration near the surface reached to a value of 1 1020 cm3 and was almost constant, regardless of the depth. When only AsH3 is supplied during the cool down process, two possible reasons for the reduction in hole concentration near the
Fig. 3. AFM images of the surface morphology obtained form the samples (a) reference, (b) C1 sample, (c) C2 sample and (d) C3 sample. All the samples were scanned over 2 lm 2 lm area.
680
C.Z. Kim et al. / Microelectronic Engineering 87 (2010) 677–681
surface can be considered. One is the passivation of carbon acceptors by hydrogen atoms from AsH3. Fig. 4b shows the hole doping profiles versus depth for the reference sample after annealing at 500 °C for 10 min in an N2 ambient. Compared to the unannealed sample, annealing did not give any significant difference. Therefore, hydrogenation is not the main reason for the reduction in hole concentration. Another possible mechanism is the desorption of carbon atoms from the GaAs epilayer and incorporation of As atoms into the As sites. When carbon atoms at As sites outdiffuse to the surface, they can be present in the form of methane by bonding with hydrogen atoms from AsH3 and are desorbed easily from the GaAs surface, and then, As atoms from AsH3 will occupy the empty As sites. Thus the carbon concentration (or hole concentration) will be reduced especially near the surface. On the other hand, when both AsH3 and CBr4 are supplied, carbon atoms at As sites will leave and make those sites empty like only AsH3 is supplied. However, at the same time, As atoms from AsH3 and carbon atoms from CBr4 will compete to occupy the empty As sites. When the CBr4 flow rate increases, therefore, there are more chances for carbon atoms to occupy As sites, resulting in higher hole concentrations. After optimizing carbon doped GaAs layers, GaInP/GaAs tandem solar cells were fabricated using conventional photolithographic techniques such as ohmic metallization, mesa isolation, antireflecting coating and scribing (denoted as C-cell). To compare the solar cell performance, p-type doping with dimethylzinc (DMZn) was also employed when fabricating solar cells (denoted as Zn-cell). Fig. 5 shows a schematic diagram of the solar cell structure. In order to minimize contact resistance, a Ti/Pt/Au metallization with a resistivity of rc < 1 105 X cm2 was applied to the front p-side layer and the n-type backside was covered entirely with an AuGe/Ni/Au metallization with a resistivity of rc < 1 106 X cm2. A double layer anti-reflection coating of SiO2 (600 nm)/SiNx(500 nm) was applied to minimize the optical loss due to the reflection. The current–voltage (I–V) measurements
were utilized using a solar simulator (Wacom: WXS-220S-L2) under one-sun air-mass (AM) 1.5 global illumination with the cell area of 1.0 cm2. Fig. 6 presents the photovoltaic I–V characteristics of the GaInP/ GaAs tandem solar cells and the extracted device parameters are depicted in Table 1. Under the illumination of 100 mW/cm2, the VOC and the JSC were 2.37 V and 12.83 mA/cm2, respectively, for the Zn-cell. The fill factor (FF), which is defined as (VMJM/VOCJSC), was 79%, where VM and JM are the voltage and the current density at the maximum power output, respectively. The solar cell conversion efficiency, which is defined as g = VOCJSCFF/JPh, was determined to be 24.10%, where JPh is the incident photon flux. The VOC and the JSC were 2.42 V and 12.39 mA/cm2, respectively, for the C-cell. The FF and the efficiency were 84% and 25.58%, respectively. Except the JSC value, all other parameters were improved for the C-cell. The series resistance (RS) in solar cell arises from the resistance of the cell material to current flow, specifically, the movement of current through the emitter, base and window of the solar cell, the contact resistance between the metal contact and the GaAs, the resistance of the top and rear metal contacts, and the resistance in the tunnel junction. The obtained series resistance was 17.20 X and 14.65 X for the Zn-cell and the C-cell, respectively. Since we fabricated solar cells using the same cell parameters for both the Zn-cell and C-cell such as doping level, ohmic metal formation, anti-reflecting coating, geometry of grid line and so forth, the reduced series resistance of C-cell probably comes from the reduced series resistance in the tunnel junction. By assuming that the current value is about 13 mA (from JSC value) and using the difference in the series resistance between C-cell and Zn-cell (2.55 X), the voltage drop across the tunnel junction is calculated to be 0.033 V. This is close to the VOC difference between two cells (0.05 V). These results thus imply that C doping provides a more stable tunnel junction with lower resistance compared to that of Zn doping. Although, to improve the solar cell conversion efficiency further, n-GaAs layer in tunnel junction should also be investigated more, this work suggests that GaAs tunnel junction with C doping can provide an improved performance in GaAs based multijunction solar cells.
20 2
Area = 1.0 cm
2
Current density (mA/cm )
C-cell Zn-cell 15
10
5
0 0.0
0.5
1.0
1.5
2.0
2.5
Voltage (V) Fig. 6. Measured photovoltaic I–V curves for the GaInP/GaAs tandem solar cells under one-sun illumination (AM1.5, 100 mW/cm2) using a solar simulator.
Table 1 Measured solar cell performance from Zn-cell and C-cell.
Fig. 5. A schematic diagram of the fabricated GaInP/GaAs tandem solar cells with GaAs tunnel junction.
Cells
JSC (mA/cm2)
VOC (V)
FF (%)
RS (X)
Efficiency, g (%)
Zn-cell C-cell
12.83 12.39
2.37 2.42
79 84
17.20 14.65
24.10 25.58
C.Z. Kim et al. / Microelectronic Engineering 87 (2010) 677–681
4. Conclusion We investigated the behavior of hole concentration in carbon doped GaAs using CBr4 in order to obtain very high and well-defined doping profiles in tunnel junction. The decrease of the growth temperature and V/III ratio was found to produce the increase of the hole concentration. When both AsH3 and CBr4 were supplied during the cool down process, the decrease of the hole concentration near the surface was not observed and the hole doping levels were almost constant regardless of the depth. The GaInP/GaAs tandem solar cells fabricated with both Zn and C doping showed that C doping produces an improved conversion efficiency compared to that of Zn doping, which can be attributed to the reduced series resistance in the GaAs tunnel junction with C doping. Acknowledgements
[4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]
This work was supported in part by the Gyeonggi Provincial Office, Korea. References [1] F. Ren, C.R. Abernathy, S.J. Pearton, T.R. Fullowan, J.R. Lothian, P.W. Wisk, K. Chen, W.S. Smith, Electron. Lett. 27 (1991) 2391. [2] M.G. Feters, B.J. Thibeault, D.B. Young, A.C. Gossard, L.A. Coldren, J. Vac. Sci. Technol. B 12 (1994) 3075. [3] T. Takamoto, E. Ikeda, H. Hurita, M. Ohmori, Appl. Phys. Lett. 70 (1997) 38.
[19] [20] [21]
[22] [23]
681
L.W. Yang, P.D. Wright, V. Eu, Z. Lu, A. Majerfeld, J. Appl. Phys. 72 (1992) 2063. P.M. Enquist, Appl. Phys. Lett. 57 (1990) 2348. N. Kobayashi, T. Makimoto, Y. Horikoshi, Appl. Phys. Lett. 50 (1987) 1435. T.F. Kuech, M.A. Thschler, P.-J. Wang, G. Schilla, R. Potemski, F. Cardone, Appl. Phys. Lett. 53 (1988) 1317. A.J. Moll, K.M. Yu, W. Walukiewicsz, W.L. Hansen, E.E. Haller, Appl. Phys. Lett. 60 (1992) 2383. K.A. Bertness, S.K. Kurtz, D.J. Friedman, A.E. Kibbler, C. Kramer, J.M. Olson, Appl. Phys. Lett. 65 (1994) 989. R. Zhang, S.F. Yoon, K.H. Tana, Z.Z. Suna, Q.F. Huanga, J. Jiangb, L.H. Leeb, SolidState Electron. 47 (2003) 1339. M. Konagai, T. Yamada, T. Akatsuka, K. Saito, E. Tokumitsu, K. Takahashi, J. Cryst. Growth 98 (1989) 167. N. Watanabe, H. Ito, T. Ishibashi, J. Cryst. Growth 147 (1995) 256. M. Pristovsek, B. Han, J.-T. Zettler, W. Richter, J. Cryst. Growth 221 (2000) 149. J.-S. Lee, I. Kim, B.-D. Choe, W.G. Jeong, J. Appl. Phys. 76 (1994) 5079. K. Taneno, Y. Kohama, C. Amano, J. Cryst. Growth 172 (1995) 5. F. Brunner, T. Bergunde, E. Richter, P. Kurpas, M. Achouche, A. Maabdorf, J. Wurfl, M. Moriski, J. Cryst. Growth 221 (2000) 53. S.-I. Kim, Y. Kim, M.-S. Kim, S.-K. Min, C. Lee, J. Cryst. Growth 141 (1994) 324. B.T. Cunningham, J.E. Baker, G.E. Stillman, Appl. Phys. Lett. 56 (1990) 836. H.Q. Hou, B.E. Hammons, H.C. Chiu, Appl. Phys. Lett. 70 (1997) 3600. K. Uchida, S. Bhunia, N. Sugiyama, M. Furiya, M. Katoh, S. Katoh, S. Nozaki, H. Moriski, J. Cryst. Growth 248 (2003) 124. J.M.O. Zide, A. Kleiman-Shwarsctein, N.C. Strandwitz, J.D. Zimmerman, T. Steenblock-Smith, A.C. Gossard, A. Forman, A. Ivanovskaya, G.D. Stucky, Appl. Phys. Lett. 88 (2006) 162103. P. Kurpas, E. Richter, D. Grutsche, F. Brunner, M. Weyers, Inst. Phys. Conf. Ser. 145 (1996) 177. N. Watanabe, H. Ito, J. Cryst. Growth 182 (1997) 30.
Contents lists available at ScienceDirect
Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee
Enhanced efficiency in GaInP/GaAs tandem solar cells using carbon doped GaAs in tunnel junction Chang Zoo Kim a, Hogyoung Kim b,*, Keun Man Song a, Dong Hwan Jun a, Ho Kwan Kang a, Wonkyu Park a, Chul Gi Ko a a b
Korea Advanced Nano Fab Center, 906-10, Iui-Dong, Yeongtong-Gu, Suwon, Gyeonggi-Do 443-270, Republic of Korea College of Humanities and Sciences, Hanbat National University, San 6-1, Dukmyung-Dong, Yuseong-Gu, Daejeon 305-719, Republic of Korea
a r t i c l e
i n f o
Article history: Received 21 May 2009 Received in revised form 28 August 2009 Accepted 25 September 2009 Available online 4 October 2009 Keywords: Carbon doping GaAs MOCVD Tandem solar cells
a b s t r a c t Carbon doping of GaAs using CBr4 (carbon tetrabromide) in metal-organic chemical vapor deposition (MOCVD) was investigated to obtain very high and sharp doping profiles required for tunnel junction in tandem solar cells. It was found that the hole concentration increased with decreasing growth temperature and V/III ratio. Hole doping profiles versus distance from the sample surface showed that the hole concentration near the surface was very low in comparison with that far below the surface. As a postgrowth treatment, CBr4 was supplied during the cool down process and produced almost constant hole concentration of 1 1020 cm3 regardless of the depth, when CBr4 flow rate was 9.53 lmol/min. Based on these results, solar cells were fabricated using both carbon (C) and zinc (Zn) as a p-type dopant. It was shown that C doping exhibits higher efficiency and lower series resistance than those of Zn doping in GaInP/GaAs tandem solar cells. Ó 2009 Elsevier B.V. All rights reserved.
1. Introduction Carbon (C) has been researched extensively as a p-type dopant in GaAs and related materials for applications in electronic and optoelectronic devices such as heterojunction bipolar transistors (HBTs), vertical cavity surface emitting lasers (VCSELs) and solar cells [1–3]. Doping levels up to 1 1020 cm3 have also been realized with less morphological degradation using metal-organic chemical vapor deposition (MOCVD) [4,5]. Carbon has lower diffusion coefficient, higher solubility, and smaller memory effect than zinc (Zn) [6–8]. Thus, in contrast to Zn doping, highly C-doped layers can be obtained during overgrowth with negligible tailing effect [9]. The low diffusion coefficient and the lack of tailing make C doping more attractive than Zn doping for devices that require highly doped and well-defined p-type doping profiles, such as tunnel junctions [3] and HBTs [10]. Carbon doping can be performed using two different methods such as intrinsic and extrinsic doping. In intrinsic carbon doping, carbon atoms from the group III precursor, trimethylgallium (TMGa) can be incorporated into the deposited material by reducing the growth temperature and the V/III precursor concentration ratio [11–13], which requires a tight control of growth parameters when using arsine (AsH3) as the group V precursor. On the other hand, carbon atoms can be introduced extrinsically into the depos* Corresponding author. E-mail address: [email protected] (H. Kim). 0167-9317/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.mee.2009.09.014
ited material using halomethanes such as carbon tetrachloride (CCl4), carbon tetrabromide (CBr4) and carbon bromidetrichloride (CBrCl3) because they have high doping efficiencies due to weak C–Cl or C–Br bonds [14–16]. This method makes it easier to control the carbon doping density widely without changing growth conditions such as growth temperature and V/III ratio. High carbon doping levels (>1020 cm3) have been achieved using CCl4 as a carbon source [17]. However, a strong reduction of growth rate was observed due to the in situ etching of GaAs by Cl [18,19]. The reduction in the growth rate of C-doped GaAs was also reported when using CBr4 due to the HBr etching [14]. The use of CBrCl3 was reported by Uchida et al. [20] to reduce the etching of GaAs. When fabricating tandem solar cells, very thin and wide-bandgap tunnel junction is necessary to minimize optical absorption. However, since the tunnel peak current density decreases exponentially with the increase in bandgap energy, it is not easy to obtain low-resistance wide-bandgap tunnel junction. An GaInP tunnel junction has been tried in GaInP/GaAs tandem cell using double heterostructure (DH) to prevent the impurity diffusion (Zn) [3]. The AlGaAs/GaAs tandem solar cells were reported to reduce the voltage losses effectively across the tunnel junction by incorporation of semimetallic ErAs nanoparticles into the GaAs tunnel junction, where Be and Si were used for p-dopant and ndopant, respectively [21]. Although lots of research has been done to develop the GaAs-based solar cells, there are relatively few reports on the GaInP/GaAs tandem solar cells using carbon doped GaAs in the tunnel junction. For the present work, we describe
678
C.Z. Kim et al. / Microelectronic Engineering 87 (2010) 677–681
the carbon doping profiles of GaAs grown by MOCVD under variation of the growth parameters such as CBr4 flow rate, growth temperature, V/III ratio and post-growth treatment. In addition, the conversion efficiency for the GaInP/GaAs tandem solar cells with GaAs tunnel junction using both C and Zn as a p-type dopant is discussed.
3.0
(a)
2.5
2.0
0.002
0.003
0.004
0.005
CBr4 flow rate ( µmol/min)
2. Experimental
19
-3
Hole concentration ( x 10 cm )
8
Growth was carried out in a multiwafer MOCVD reactor (AIX2600G3 IC) on semi-insulating GaAs (100) substrates with a misorientation of 2° toward <1 1 1> plane. TMGa and AsH3 were used as the precursors of Ga and As, respectively. The carrier gas was hydrogen and the reactor pressure was kept at 50 mbar. Carbon doping was achieved using the extrinsic source of CBr4. The CBr4 bath temperature was maintained at 25 °C. All samples used in this study were grown with a constant TMGa flow rate of 265.05 lmol/min. The total flow rate in the reactor was 58 slm. The influence of three growth parameters such as growth temperature, CBr4 flow rate and V/III ratio on the hole concentration was first investigated to obtain very high doping levels up to 1 1020 cm3. After growth process, AsH3 was supplied to all samples until the temperature cooled down to 350 °C in order to prevent As outdiffusion. Thereafter, the effect of CBr4 flow rate under the cool down process on the hole concentration was studied in order to obtain well-defined carbon doping profiles. The CBr4 flow rate was changed from 0 to 9.53 lmol/min. Electrochemical capacitance–voltage (ECV) measurements were used to determine the hole concentration of carbon doped GaAs layers. Atomic force microscopy (AFM) was used to investigate the surface morphology.
6
(b)
4 2
540
560
580
600
o
Growth temperature ( C) 10 8
(c)
6 4 2
20
40
60
V/III ratio
10
(d) 5
reference
3. Results and discussion Fig. 1a shows the hole concentration versus CBr4 flow rate. The growth temperature and the V/III ratio were 600 °C and 57.25, respectively. When CBr4 flow rate increases, the hole concentration remains almost constant, although there is a slight increase. Kurpas et al. [22] demonstrated that there is a sublinear relationship between hole concentration and CBr4 partial pressure which is expressed as p / [CBr4]0.7. This relation shows that the hole concentration increases linearly at relatively low doping levels and tends to saturate at high doping levels. Watanabe et al. [23] investigated the behavior of the hole concentration in carbon doped GaAs grown by MOCVD using CBr4 as a carbon source and concluded that the hole concentration tends to saturate at higher CBr4 flow rates in the very high doping region due to the saturation of carbon incorporation efficiency, not the deactivation of carbon acceptors. The same mechanism can be applied to explain the saturation of the hole doping levels. Fig. 1b shows the hole concentration as a function of growth temperature. The CBr4 flow rate and the V/III ratio were kept at 21.43 lmol/min and 57.25, respectively. The growth temperature was varied from 550 to 600 °C. It is clearly observed that the hole concentration was increased by decreasing the growth temperature. Fig. 1c shows the hole concentration versus V/III ratio. The growth temperature and the CBr4 flow rate were 550 °C and 21.43 lmol/min, respectively. When the V/III ratio (or AsH3 flow rate) decreases, the hole concentration increases accordingly and reaches to 1 1020 cm3. These results imply that the hole concentration increases with decreasing growth temperature and V/III ratio, which are generally observed in the carbon doped p-type GaAs epilayer. In intrinsic carbon doping, carbon incorporation is possible from the hydrogen radicals generated by the decomposition of TMGa [16]. When AsH3 is present, these hydrogen radicals can be present
0
50
100
150
Depth (nm) Fig. 1. Hole concentration as a function of (a) CBr4 flow rate, (b) growth temperature and (c) V/III ratio, and (d) hole doping profile versus depth for the reference sample that CBr4 flow was turned off after growth process.
in the form of methane and be desorbed at the GaAs surface, reducing carbon incorporation in GaAs. This mechanism can explain the rapid increase in the hole concentration when V/III ratio decreases. As discussed by Watanabe et al. [23], lower growth temperature makes the adsorption coefficient larger and a lower V/III ratio provides more sites for CBrk to adsorb. Therefore, carbon incorporation can be enhanced further with the presence of CBr4 at low growth temperature and low V/III ratio. In addition, many carbon atoms from hydrogen radicals by TMGa will incorporate with decreasing V/III ratio, increasing hole concentration further. As discussed above, very high carbon doping (1020 cm3) could be achieved at the growth temperature of 550 °C, the CBr4 flow rate of 21.43 lmol/min, and the V/III ratio of 16.84, respectively, and these growth conditions were used to investigate the effect of in situ CBr4 treatment after growth process on the carbon doping profiles. Fig. 1d shows the hole concentration as a function of depth from the surface without supplying CBr4 after growth process (denoted as reference). As the depth increases, the hole concentration increases and reaches to the value of 1020 cm3 when the depth is around 150 nm. When making tunnel junction by growing Sidoped n-type GaAs layer on the carbon doped p-type layer, such a doping profile which has quite low doping levels near the p-type GaAs surface cannot be used for the fabrication of tunnel junctions with high performance and low resistance. In this respect, we examined the influence of CBr4 flow rate under the cool down process on the hole doping profiles.
C.Z. Kim et al. / Microelectronic Engineering 87 (2010) 677–681
679
Fig. 2. Schematic growth process as a function of time. After growth process, CBr4 was supplied during the cool down process.
Fig. 2 shows the growth process graphically as a function of time. Instead of turning off the CBr4 flow after growth process, CBr4 was supplied during the cool down process until the temperature dropped down to 350 °C. Three different CBr4 flow rates such as 2.38 lmol/min (denoted as C1), 4.76 lmol/min (denoted as C2) and 9.53 lmol/min (denoted as C3) were employed. Fig. 3 shows the AFM images presenting the change of surface morphology of the CBr4 treated samples after growth process. All the samples were scanned over 2 lm 2 lm areas. The surface morphology of the reference sample was mirror-like regardless of the etching effect due to bromide radicals. After CBr4 treatment, the surface morphology was changed only marginally and was mirror-like, too. The root mean square (rms) roughness of all the samples was in the range of 0.2–0.4 nm. Fig. 4a shows the hole doping profiles versus depth from the sample surface. It is clearly observed that the hole concentration near the sample surface increased with increasing CBr4 flow rate. When the CBr4 flow rate was 9.53 lmol/
Fig. 4. Hole doping profiles versus depth after annealing at 500 °C for 10 min in an N2 ambient (a) for the samples with different CBr4 flow rate under the cool down process and (b) for the reference sample.
min, the hole concentration near the surface reached to a value of 1 1020 cm3 and was almost constant, regardless of the depth. When only AsH3 is supplied during the cool down process, two possible reasons for the reduction in hole concentration near the
Fig. 3. AFM images of the surface morphology obtained form the samples (a) reference, (b) C1 sample, (c) C2 sample and (d) C3 sample. All the samples were scanned over 2 lm 2 lm area.
680
C.Z. Kim et al. / Microelectronic Engineering 87 (2010) 677–681
surface can be considered. One is the passivation of carbon acceptors by hydrogen atoms from AsH3. Fig. 4b shows the hole doping profiles versus depth for the reference sample after annealing at 500 °C for 10 min in an N2 ambient. Compared to the unannealed sample, annealing did not give any significant difference. Therefore, hydrogenation is not the main reason for the reduction in hole concentration. Another possible mechanism is the desorption of carbon atoms from the GaAs epilayer and incorporation of As atoms into the As sites. When carbon atoms at As sites outdiffuse to the surface, they can be present in the form of methane by bonding with hydrogen atoms from AsH3 and are desorbed easily from the GaAs surface, and then, As atoms from AsH3 will occupy the empty As sites. Thus the carbon concentration (or hole concentration) will be reduced especially near the surface. On the other hand, when both AsH3 and CBr4 are supplied, carbon atoms at As sites will leave and make those sites empty like only AsH3 is supplied. However, at the same time, As atoms from AsH3 and carbon atoms from CBr4 will compete to occupy the empty As sites. When the CBr4 flow rate increases, therefore, there are more chances for carbon atoms to occupy As sites, resulting in higher hole concentrations. After optimizing carbon doped GaAs layers, GaInP/GaAs tandem solar cells were fabricated using conventional photolithographic techniques such as ohmic metallization, mesa isolation, antireflecting coating and scribing (denoted as C-cell). To compare the solar cell performance, p-type doping with dimethylzinc (DMZn) was also employed when fabricating solar cells (denoted as Zn-cell). Fig. 5 shows a schematic diagram of the solar cell structure. In order to minimize contact resistance, a Ti/Pt/Au metallization with a resistivity of rc < 1 105 X cm2 was applied to the front p-side layer and the n-type backside was covered entirely with an AuGe/Ni/Au metallization with a resistivity of rc < 1 106 X cm2. A double layer anti-reflection coating of SiO2 (600 nm)/SiNx(500 nm) was applied to minimize the optical loss due to the reflection. The current–voltage (I–V) measurements
were utilized using a solar simulator (Wacom: WXS-220S-L2) under one-sun air-mass (AM) 1.5 global illumination with the cell area of 1.0 cm2. Fig. 6 presents the photovoltaic I–V characteristics of the GaInP/ GaAs tandem solar cells and the extracted device parameters are depicted in Table 1. Under the illumination of 100 mW/cm2, the VOC and the JSC were 2.37 V and 12.83 mA/cm2, respectively, for the Zn-cell. The fill factor (FF), which is defined as (VMJM/VOCJSC), was 79%, where VM and JM are the voltage and the current density at the maximum power output, respectively. The solar cell conversion efficiency, which is defined as g = VOCJSCFF/JPh, was determined to be 24.10%, where JPh is the incident photon flux. The VOC and the JSC were 2.42 V and 12.39 mA/cm2, respectively, for the C-cell. The FF and the efficiency were 84% and 25.58%, respectively. Except the JSC value, all other parameters were improved for the C-cell. The series resistance (RS) in solar cell arises from the resistance of the cell material to current flow, specifically, the movement of current through the emitter, base and window of the solar cell, the contact resistance between the metal contact and the GaAs, the resistance of the top and rear metal contacts, and the resistance in the tunnel junction. The obtained series resistance was 17.20 X and 14.65 X for the Zn-cell and the C-cell, respectively. Since we fabricated solar cells using the same cell parameters for both the Zn-cell and C-cell such as doping level, ohmic metal formation, anti-reflecting coating, geometry of grid line and so forth, the reduced series resistance of C-cell probably comes from the reduced series resistance in the tunnel junction. By assuming that the current value is about 13 mA (from JSC value) and using the difference in the series resistance between C-cell and Zn-cell (2.55 X), the voltage drop across the tunnel junction is calculated to be 0.033 V. This is close to the VOC difference between two cells (0.05 V). These results thus imply that C doping provides a more stable tunnel junction with lower resistance compared to that of Zn doping. Although, to improve the solar cell conversion efficiency further, n-GaAs layer in tunnel junction should also be investigated more, this work suggests that GaAs tunnel junction with C doping can provide an improved performance in GaAs based multijunction solar cells.
20 2
Area = 1.0 cm
2
Current density (mA/cm )
C-cell Zn-cell 15
10
5
0 0.0
0.5
1.0
1.5
2.0
2.5
Voltage (V) Fig. 6. Measured photovoltaic I–V curves for the GaInP/GaAs tandem solar cells under one-sun illumination (AM1.5, 100 mW/cm2) using a solar simulator.
Table 1 Measured solar cell performance from Zn-cell and C-cell.
Fig. 5. A schematic diagram of the fabricated GaInP/GaAs tandem solar cells with GaAs tunnel junction.
Cells
JSC (mA/cm2)
VOC (V)
FF (%)
RS (X)
Efficiency, g (%)
Zn-cell C-cell
12.83 12.39
2.37 2.42
79 84
17.20 14.65
24.10 25.58
C.Z. Kim et al. / Microelectronic Engineering 87 (2010) 677–681
4. Conclusion We investigated the behavior of hole concentration in carbon doped GaAs using CBr4 in order to obtain very high and well-defined doping profiles in tunnel junction. The decrease of the growth temperature and V/III ratio was found to produce the increase of the hole concentration. When both AsH3 and CBr4 were supplied during the cool down process, the decrease of the hole concentration near the surface was not observed and the hole doping levels were almost constant regardless of the depth. The GaInP/GaAs tandem solar cells fabricated with both Zn and C doping showed that C doping produces an improved conversion efficiency compared to that of Zn doping, which can be attributed to the reduced series resistance in the GaAs tunnel junction with C doping. Acknowledgements
[4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]
This work was supported in part by the Gyeonggi Provincial Office, Korea. References [1] F. Ren, C.R. Abernathy, S.J. Pearton, T.R. Fullowan, J.R. Lothian, P.W. Wisk, K. Chen, W.S. Smith, Electron. Lett. 27 (1991) 2391. [2] M.G. Feters, B.J. Thibeault, D.B. Young, A.C. Gossard, L.A. Coldren, J. Vac. Sci. Technol. B 12 (1994) 3075. [3] T. Takamoto, E. Ikeda, H. Hurita, M. Ohmori, Appl. Phys. Lett. 70 (1997) 38.
[19] [20] [21]
[22] [23]
681
L.W. Yang, P.D. Wright, V. Eu, Z. Lu, A. Majerfeld, J. Appl. Phys. 72 (1992) 2063. P.M. Enquist, Appl. Phys. Lett. 57 (1990) 2348. N. Kobayashi, T. Makimoto, Y. Horikoshi, Appl. Phys. Lett. 50 (1987) 1435. T.F. Kuech, M.A. Thschler, P.-J. Wang, G. Schilla, R. Potemski, F. Cardone, Appl. Phys. Lett. 53 (1988) 1317. A.J. Moll, K.M. Yu, W. Walukiewicsz, W.L. Hansen, E.E. Haller, Appl. Phys. Lett. 60 (1992) 2383. K.A. Bertness, S.K. Kurtz, D.J. Friedman, A.E. Kibbler, C. Kramer, J.M. Olson, Appl. Phys. Lett. 65 (1994) 989. R. Zhang, S.F. Yoon, K.H. Tana, Z.Z. Suna, Q.F. Huanga, J. Jiangb, L.H. Leeb, SolidState Electron. 47 (2003) 1339. M. Konagai, T. Yamada, T. Akatsuka, K. Saito, E. Tokumitsu, K. Takahashi, J. Cryst. Growth 98 (1989) 167. N. Watanabe, H. Ito, T. Ishibashi, J. Cryst. Growth 147 (1995) 256. M. Pristovsek, B. Han, J.-T. Zettler, W. Richter, J. Cryst. Growth 221 (2000) 149. J.-S. Lee, I. Kim, B.-D. Choe, W.G. Jeong, J. Appl. Phys. 76 (1994) 5079. K. Taneno, Y. Kohama, C. Amano, J. Cryst. Growth 172 (1995) 5. F. Brunner, T. Bergunde, E. Richter, P. Kurpas, M. Achouche, A. Maabdorf, J. Wurfl, M. Moriski, J. Cryst. Growth 221 (2000) 53. S.-I. Kim, Y. Kim, M.-S. Kim, S.-K. Min, C. Lee, J. Cryst. Growth 141 (1994) 324. B.T. Cunningham, J.E. Baker, G.E. Stillman, Appl. Phys. Lett. 56 (1990) 836. H.Q. Hou, B.E. Hammons, H.C. Chiu, Appl. Phys. Lett. 70 (1997) 3600. K. Uchida, S. Bhunia, N. Sugiyama, M. Furiya, M. Katoh, S. Katoh, S. Nozaki, H. Moriski, J. Cryst. Growth 248 (2003) 124. J.M.O. Zide, A. Kleiman-Shwarsctein, N.C. Strandwitz, J.D. Zimmerman, T. Steenblock-Smith, A.C. Gossard, A. Forman, A. Ivanovskaya, G.D. Stucky, Appl. Phys. Lett. 88 (2006) 162103. P. Kurpas, E. Richter, D. Grutsche, F. Brunner, M. Weyers, Inst. Phys. Conf. Ser. 145 (1996) 177. N. Watanabe, H. Ito, J. Cryst. Growth 182 (1997) 30.