p-type InGaAs/Si heterojunctions were fabricated through a wafer fusion bonding process. The relative band alignment between the two materials at the heterointerface was determined using current-voltage (I-V) measurements and applying thermionic emission-diffusion theory. The valence and conduction band discontinuities for the InGaAs/Si interface were determined to be 0.48 and ?0.1 eV, respectively, indicating a type-II band alignment. Band discontinuity measurements of the wafer bonded InGaAs/Si heterojunction Kyle S. McKay,a Felix P. Lu, Jungsang Kim, Changhyun Yi, and April S. Brown Fitzpatrick Institute for Photonics, Electrical and Computer Engineering Department, Duke University, Durham, North Carolina 27708 Aaron R. Hawkins Electrical and Computer Engineering Department, Brigham Young University, 459 Clyde Building, Provo, Utah 84602 Received 24 January 2007; accepted 9 May 2007; published online 31 May 2007 p-type InGaAs/Si heterojunctions were fabricated through a wafer fusion bonding process. The relative band alignment between the two materials at the heterointerface was determined using current-voltage I-V measurements and applying thermionic emission-diffusion theory. The valence and conduction band discontinuities for the InGaAs/Si interface were determined to be 0.48 and ?0.1 eV, respectively, indicating a type-II band alignment. © 2007 American Institute of Physics. DOI: 10.1063/1.2745254 In0.53Ga0.47As InGaAs lattice matched to InP has a direct band gap of 0.74 eV at room temperature, corresponding to a photon wavelength of 1.68 m.1 It is an ideal semiconductor material for realizing photodetectors in the near infrared wavelength range relevant for optical fiber communications at 1.31 and 1.55 m. Si, which has excellent carrier multiplication properties, has been integrated with In- GaAs to create high performance photodiodes at these wavelengths.2–6 Due to the large mismatch in the lattice constants 7.7% and coefficients of thermal expansion CTE of InGaAs and Si, low defect epitaxial growth of InGaAs on Si is very difficult. The wafer fusion process allows materials with different lattice constants and CTE to be permanently joined with a relatively low defect density and low oxide content at the heterointerface. This process was employed to fabricate the photodiodes cited above, as well as other solid state devices.7 The band discontinuities are critical for understanding the carrier transport properties of the heterointerface and are essential to the optimal design of these and future devices. The measured valence band discontinuity of the wafer fused InGaAs/Si interface is reported in this letter. p-type heterojunctions were fabricated to investigate the hole transport properties across the InGaAs/Si interface. We used boron doped, mirror polished, p-type 100 Si wafers with a resistivity of 4.5 cm 3 1015 cm?3 . The InGaAs structures were grown using solid source molecular beam epitaxy on a semi-insulating 100 InP substrate with a highly doped 1000 Å InGaAs contact layer 1019 cm?3 , a 5000 Å moderately doped InGaAs active layer 5 1016 cm?3 , and a 1000 Å InAlAs cap layer. The InAlAs cap layer is a sacrificial protective layer that is removed before bonding. Si wafers were cleaved into 8 8 mm2 pieces and cleaned using the standard RCA process NH4OH: H2O2:H2O 1:1:5, RCA1; and HCl:H2O2:H2O 1:1:5, RCA2 to remove any organic and metallic contaminations from the wafer surface.8 The InGaAs wafers were cleaved into 7 7 mm2 pieces, and then a hydrochloric acid solution HCl:H2O 3:1 was used to remove the InAlAs cap layer. The InGaAs pieces were patterned with a 300 300 m2 square grid with 10 m channels etched using an InGaAs selective etch H3PO4:H2O2:H2O 1:1:11 . The channels were etched to allow any trapped gases that may form during the high temperature annealing to escape from the interface.7 The Si and InGaAs wafer pieces were then cleaned in acetone and isopropanol until a microscope inspection revealed a clean surface free from particulate contaminants. The wafers were then transferred to a nitrogen glovebox for bonding. In the nitrogen ambient, both wafer pieces were dipped into a 5% hydrofluoric acid HF solution to remove any surface oxides. The InGaAs piece was then pressed onto the Si, visually aligned to match crystallographic orientation, and then pressure was applied while the sample was dried with a nitrogen gun. The nitrogen glovebox, where the Si and InGaAs were bonded together, had an oxygen concentration of 1 ppm and a moisture content of 10 ppm. The HF dip and the low oxygen concentration ensured that minimal oxides were present at the interface. After contact, the bonded pair was loaded into a graphite fixture and transported in nitrogen to the annealing furnace, where it encountered room air for about 1 min before the furnace was pumped down and backfilled with nitrogen or hydrogen. The bonded sample was then subjected to a two-stage annealing.6 The first stage started with a slow ramp 2 °C/min up to 300 °C, and the sample was then annealed for 1 h to allow any gas at the interface to escape through the channels etched into the In- GaAs. The second stage consisted of a faster ramp 6 °C/min up to 650 °C and held for 1 h during which covalent bonds were formed between Si and InGaAs. After the annealing, the InP substrate was removed from the bonded sample in a HCl solution HCl:H2O 3:1 . Through this process, the InGaAs layers were bonded and transferred to the Si surface for device processing. The valence band discontinuity of the heterojunction was characterized by measuring I-V characteristics and applying thermionic emission-diffusion theory9,10 to calculate the barrier height.11,12 Devices of different shapes and sizes were fabricated to measure the impact of surface and edge a Electronic mail: ksm18@duke.edu effects on the I-V measurements. Ohmic contacts 70 Å APPLIED PHYSICS LETTERS 90, 222111 2007 0003-6951/2007/90 22 /222111/3/$23.00 90, 222111-1 © 2007 American Institute of Physics Downloaded 10 Feb 2009 to 128.187.0.164. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp Cr/3000 Å Au in square and circle shapes with characteristic sizes ranging from 100 to 250 m were evaporated onto the transferred InGaAs. The Cr/Au contacts were used as self-aligned masks for subsequent wet etching of the InGaAs layers that defined the size of the devices. Aluminum was evaporated on the back side of the Si to provide a substrate ohmic contact. I-V measurements were taken over a range of temperatures using a semiconductor parameter analyzer Keithley 4200 . Current transport across a barrier can be described by thermionic emission-diffusion theory for the moderate temperatures and bias values of our heterojunction. In semiconductor materials with low doping levels or low mobility, the carrier transport properties in the space charge regions dominate, and the current density J reduces to J qNC E exp ? q b kBT exp qV1 nkBT ? exp ? qV2 nkBT , 1 where q is the electronic charge, NC is the majority carrier concentration, is the hole mobility, E is the electric field near the interface, b is the barrier height, n is the ideality factor, kB is Boltzmann constant, T is the temperature, and V1 and V2 are the fractions of voltage supported in each material in the junction where the applied voltage V=V1+V2 Fig. 1 a .9 The current density as a function of applied voltage is shown in Fig. 2 where the InGaAs is biased with respect to grounded Si. The curve has three regions of interest. The first region extends from ?2 to ?0.3 V and is a region where series resistance effects dominate. The second region is the region from ?0.3 to 0 V where the current density is exponentially dependent on the applied voltage second term in Eq. 1 . The third is the positively biased region where the dependence on the applied voltage is weak first term in Eq. 1 . The strong exponential voltage dependence for hole injection from Si to InGaAs as compared to the weak voltage dependence for hole injection from InGaAs to Si is a result of the difference in the V1 and V2 terms in Eq. 1 . The relationship between V1 and V2 is determined from the continuity of electric displacement at the interface,9 and in this heterojunction V2 is 20V1. In order to find the barrier height, the current density in region 2 was fitted to the second term in Eq. 1 , ignoring the first term in the limit of positive bias. The barrier height was then extracted from the temperature dependence.11 The barrier height was determined to be 0.59 with a standard deviation of 0.04 eV, as shown in Fig. 3. The reduced scatter in the data points at larger device area suggests that edge and surface effects are responsible for some of the variations in the barrier height. The valence band discontinuity EV can be deduced from the measurement of the barrier height using the relationship EV = qVD ? 1 + 2, 2 where VD is the diffusion potential VD=VD1+VD2 and 1 and 2 are the distances from the valence band edge to the Fermi level for InGaAs and Si, respectively Fig. 1 a .12 The diffusion potentials are determined from the calculated barrier heights and from boundary conditions imposed by the continuity of the electric displacement field.9 The valence FIG. 1. a Definitions of selected terms for the valence band referred to in this letter. EV is the valence band, EF0 is the equilibrium Fermi level, EFP1 and EFP2 are the quasi-Fermi levels, 1 and 2 are the energy differences between the Fermi level and valence band, V1 and V2 are the fractions of the applied voltage supported in each semiconductor, VD1 and VD2 are the diffusion potentials, EV is the valence band discontinuity, and b is the barrier height. b Band alignment of the p-type InGaAs/Si heterojunction. The valence band discontinuity is determined to be 0.48 eV, while the conduction band discontinuity is ?0.1 eV. EC and EV are the conduction and valence bands, EF is the Fermi level, and EC and EV are the conduction and valence band discontinuities. FIG. 2. J-V Plot for a p-type sample bonded and annealed in a N2 environment. In region 1 ?2 to ?0.3 V , the current density is limited by series resistance effects. In region 2 ?0.3 to 0 V , the current density has an exponential relation to the applied voltage. In region 3 0–2 V , the current density has a weak dependence on the applied voltage. FIG. 3. Extracted barrier heights for measured devices. The dashed line indicates the average barrier height of 0.59 eV. 222111-2 McKay et al. Appl. Phys. Lett. 90, 222111 2007 Downloaded 10 Feb 2009 to 128.187.0.164. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp band discontinuity EV is determined to be 0.48±.04 eV at room temperature. The corresponding conduction band discontinuity is determined using the differences in the band gaps of the two materials to be ?0.1 eV, indicating type-II alignment for the heterojunction Fig. 1 b . Due to the lattice mismatch between InGaAs and Si, a large number of dangling bonds are expected at the heterointerface. In the bonding process adopted in our approach, most of these are passivated by hydrogen due to the HF treatment prior to bonding. Previous work has shown that a nearly ideal interface, free from significant charge trapping, can be fabricated between InGaAs and Si even in the presence of these defects.13 While the presence of trapped charge at the interface will affect the current density measurements by increasing the ideality factor in Eq. 1 , this does not significantly affect the measurement of the valence band discontinuity. It is known that the presence of hydrogen at the interface can lead to significant changes in the relative band alignment of the Si/SiO2 heterojunction, and similar impact might be present in our bonded interface.14 More work will have to be done to determine the effects of interfacial hydrogen on the band discontinuities. In summary, we fabricated wafer bonded p-type InGaAs/Si heterojunctions to measure the hole transport properties of the interface. Thermionic emission-diffusion theory was used to extract barrier height information from I-V measurements. A valence band discontinuity of 0.48 eV and a conduction band discontinuity of ?0.1 eV indicated a type-II alignment for the wafer bonded heterojunction fabricated by the above process. InGaAs/Si heterojunctions will play an important role in optoelectronic applications, especially photodetectors operating in the near infrared wavelengths. 1Pallab Bhattacharya, Semiconductor Optoelectronic Devices, 2nd ed. Prentice Hall, Upper Saddle River, 1997 p. 549. 2F. E. Ejeckam, C. L. Chua, Z. H. Zhu, Y. H. Lo, M. Hong, and R. Bhat, Appl. Phys. Lett. 67, 3936 1995 . 3A. R. Hawkins, T. E. Reynolds, D. R. England, D. I. Babic, M. J. Mondry, K. Streubel, and J. E. Bowers, Appl. Phys. Lett. 68, 3692 1996 . 4Y. Kang, P. Mages, A. R. Clawson, S. S. Lau, Y. H. Lo, P. K. L. Yu, A. Pauchard, Z. Zhu, and Y. Zhou, Appl. Phys. Lett. 79, 970 2001 . 5B. F. Levine, C. J. Pinzone, S. Hui, C. A. King, R. E. Leibenguth, D. R. Zolnowski, D. V. Lang, H. W. Krautter, and M. Geva, Appl. Phys. Lett. 75, 2141 1999 . 6P. Mages, Ph.D. thesis, University of California, San Diego, 2003. 7A. Black, A. R. Hawkins, N. M. Margalit, D. I. Babic, A. L. Holmes, Y. L. Chang, P. Abraham, J. E. Bowers, and E. L. Hu, IEEE J. Quantum Electron. 3, 943 1997 . 8Q.-Y. Tong and U. Gosele, Seminconductor Wafer Bonding Wiley, New York, 1999 Chap. 4 pp. 49–54. 9S. M. Sze, Physics of Semiconductor Devices 2nd ed. Wiley, New York, 1981 Chap. 2 and 5. 10R. J. Schuelke and M. S. Lundstrom, Solid-State Electron. 27, 1111 1984 . 11H. Wada, Y. Ogawa, and T. Kamijoh, Appl. Phys. Lett. 62, 738 1993 . 12Y. C. Zhou, Z. H. Zhu, D. Crouse, and Y. H. Lo, Appl. Phys. Lett. 73, 2337 1998 . 13B. F. Levine, A. R. Hawkins, S. Hiu, B. J. Tseng, C. A. King, L. A. Gruezke, R. W. Johnson, D. R. Zolnowski, and J. E. Bowers, Appl. Phys. Lett. 70, 2449 1997 . 14P. Perfetti, C. Quaresima, C. Coluzza, C. Fortunato, and G. Margaritondo, Phys. Rev. Lett. 57, 2065 1986 . 222111-3 McKay et al. Appl. Phys. Lett. 90, 222111 2007 Downloaded 10 Feb 2009 to 128.187.0.164. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp