Growth of Ge Homoepitaxial Films by Metal-Organic Chemical Vapor Deposition Using t-C4H9GeH3

Ge homoepitaxial films are grown at low growth temperature of 320◦C by metal-organic chemical vapor deposition (MOCVD) using tertiarybutylgermane (t-C4H9GeH3). We also performed ab initio calculations in order to reveal the chemical reaction for the epitaxial growth. As the result, it was revealed that the t-C4H9GeH3 was most likely decomposed into germane (GeH4) and isobutene [CH2=C(CH3)2] through the β-hydrogen elimination. We considered that this chemical nature allowed the growth temperature as low as that obtained by GeH4 precursor with sufficiently suppressed C impurity incorporation. © The Author(s) 2015. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0191505jss] All rights reserved.

Ge has higher electron and hole mobilities than Si. 1 Thus, it is considered to be one of the promising post scaling materials. [2][3][4][5] Moreover, we believe that Ge homoepitaxial technique is also essential to realize Ge 1-x Sn x heteroepitaxial growth. A simultaneous supply with Sn precursor such as tetraethyl tin [(C 2 H 5 ) 4 Sn] enables to grow the Ge 1-x Sn x heteroepitaxial films. 6,7 Ge 1-x Sn x can be also used as a channel material and a stressor for Ge MOSFETs. 8,9 Moreover, it is also used for the optical devices because the band structure changes from indirect to direct as Sn concentration increased more than approximately 10 at.%. [10][11][12][13][14] We grew the Ge homoepitaxial films by metal-organic chemical vapor deposition (MOCVD) because MOCVD films can be grown at the low substrate temperature. [15][16][17] This feature also contributes low thermal damages to devices and high Sn incorporations in Ge 1-x Sn x films. 18 In our previous study, we reported the growth of Ge homoepitaxial films on Ge(001) substrates by MOCVD using tertiarybutylgermane (t-C 4 H 9 GeH 3 ) precursor. 19 In this paper, we achieved to grow the Ge homoepitaxial films at the low substrate temperature, and also investigated the precursor decomposition path by ab initio calculations for the better scientific understanding of the growth mechanism.
In the experiment, we used t-C 4 H 9 GeH 3 shown in Fig. 1 as a Ge precursor. This precursor has high vapor pressure appropriate for CVD process, and is expected to be decomposed at low substrate temperature. [20][21][22][23] Furthermore, it is extremely safe without any explosive or pyrophoric natures, and never react with H 2 O. Thus, our selected precursor has much safer than other metal-hydride CVD precursors such as GeH 4 and Ge 2 H 6 . [24][25][26][27][28] We believe the safe precursor is appropriate for the scientific experiment and even for the commercial mass production. We used a Ge(001) wafer as a substrate. After chemical cleaning using NH 4 OH, HCl, and HF, the residual surface native oxide was removed by thermal treatment at 450 • C in the CVD chamber. A precursor supply was controlled by the N 2 carrier gas flow rate, bottle temperature and pressure. In this study, the substrate temperature was varied from 320 to 380 • C. The deposition pressure and duration were 30 torr and 120 min, respectively. The injection rate for the Ge precursor was estimated to be 3.5 × 10 −5 mol/min. The deposited films were observed by scanning electron microscopy (SEM; Hitachi High Technologies SU-70) and transmission electron microscopy (TEM; JEOL JEM-2100). The impurities in the films such as carbon (C) and oxygen (O) were evaluated by X-ray photoelectron spectroscopy (XPS; Thermo Fisher Scientific K-Alpha). The ab initio calculation was carried out by using Gaussian software (HULINKS Inc. Gaussian 09).
First of all, we achieved to grow the Ge homoepitaxial films at the low substrate temperature of 320 • C. Figure 2a shows the crosssectional transmission electron microscopy (XTEM) image. The film thickness was approximately 30 nm. We evaluated the epitaxy by the transmission electron diffraction patterns from the Ge deposited film and the Ge substrate. As the result, both patterns showed the same crystalline orientations, as shown in Figs. 2b and 2c. Thus, we concluded that we have successfully grown the Ge homoepitaxial films on the Ge(001) substrate by MOCVD at low temperature of 320 • C. Furthermore, we evaluated the Ge homoepitaxial films at 340, 360 and 380 • C. The thickness was approximately 80, 245 and 320 nm, respectively. The surface morphologies of the films observed in the large area by SEM were almost the same quality as TEM observation shown in Fig. 2a. We found some defects in the epitaxial film even in the XTEM image. This may imply the defect density is relatively high, probably more than 10 6 /cm 2 . We believe we need further optimization for the surface cleaning of the Ge substrate prior to the epitaxial growth in order to reduce the defect.
We also investigated the C and O impurities in the Ge homoepitaxial film grown at 320 • C by XPS, and showed in Fig. 3a and 3b, respectively. After the surface was etched to the depth of approximately 20 nm using Ar sputtering, the C and O peaks which are due to the air exposure of samples were completely disappeared, and the peak shapes became similar to those obtained in the Ge substrate (after 50 nm Ar sputtering). Here, the broad peak around 534 eV in Fig. 3b is considered to be due to the Ge L 3 M 23 M 23 Auger electron. 29 The film grown at low temperature tends to include more impurities than that grown at high temperature. 30 However, in spite of using MOCVD precursor, our epitaxial film has no obvious impurities even at low substrate temperature of 320 • C.  Furthermore, we discuss the reaction mechanisms of MOCVD epitaxial growth using t-C 4 H 9 GeH 3 precursor. We calculated the decomposition energies of t-C 4 H 9 GeH 3 by ab initio calculations using Becke3LYP (B3LYP) density functional theory with 3-21G basis set. The calculated chemical reaction paths are decompositions from t-C 4 H 9 GeH 3 into GeH 2 and CH 3 CH(CH 3 ) 2 , GeH 3 and CH 3 C(CH 3 ) 2 , and GeH 4 and CH 2 =C(CH 3 ) 2 . The reaction energies were calculated to be approximately 170, 330 and 150 kJ/mol, respectively. The lowest reaction energy path seemed to be the decomposition into GeH 4 and CH 2 =C(CH 3 ) 2 (isobutene) through β-hydrogen eliminations. We considered that the growth species are GeH 4 molecules, and CH 2 =C(CH 3 ) 2 is pumped out without any chemical reactions. The low growth temperature, as low as that obtained with GeH 4 , should be achieved through the GeH 4 generation by the β-hydrogen eliminations, although the t-C 4 H 9 GeH 3 is stable enough at the room temperature. Moreover, we considered that CH 2 =C(CH 3 ) 2 generated  through the β-hydrogen eliminations is stable enough to prevent from more chemical reactions. So, we considered this is the reason why the C impurity incorporation was sufficiently suppressed.
We also evaluated the activation energies of the epitaxial growth using t-C 4 H 9 GeH 3 . Figure 4 shows the Arrhenius plot between 320 and 380 • C. Here, the growth at 380 • C may be under the supply limited conditions. Therefore, we calculated the activation energy between 320 and 360 • C. As the result, the activation energy using t-C 4 H 9 GeH 3 was approximately 160 kJ/mol. In the literature, it is reported that the activation energy using GeH 4 is approximately 140 kJ/mol. 31 So, we could reveal that the activation energy using t-C 4 H 9 GeH 3 was almost as low as that using GeH 4 . This result is consistent with the simulation results by ab initio calculation. The growth rate at 310 • C is calculated as approximately 8.0 nm/h from the activation energy, which cannot be recognized as practical. Therefore, we conclude the epitaxial temperature should be practically more than 320 • C under the present conditions. Further effort should be needed to lower the epitaxial temperature such as controlling the atmospheric pressure or changing the carrier gas during the growth.
In conclusion, we demonstrated the experimental results and discussed the mechanism of the low temperature Ge homoepitaxial growth using novel safe precursor, t-C 4 H 9 GeH 3 . We have achieved to grow the Ge homoepitaxial films at low temperature of 320 • C and the deposited epitaxial films had no obvious contaminations such as C and O, at least which should be less than the XPS detection limit of approximately 1 at.%. Moreover, we discussed the growth mechanism based on the ab initio calculations. As the result, we found that t-C 4 H 9 GeH 3 decomposes into GeH 4 and CH 2 =C(CH 3 ) 2 through the β-hydrogen eliminations. We considered that GeH 4 as a product of the β-hydrogen elimination contributes the low temperature growth, and CH 2 =C(CH 3 ) 2 contributes the low C impurity incorporation. We have also shown that the activation energy using t-C 4 H 9 GeH 3 is almost the same as that using GeH 4 . This result is consistent with the simulation results by ab initio calculation. So we believe that the Ge MOCVD using t-C 4 H 9 GeH 3 is more appropriate than traditional GeH 4 and Ge 2 H 6 for the safe epitaxial growth applicable for both the scientific experiment and commercial mass production.