- Email: [email protected]
Si superlattice into Si substrate
Applied Surface Science 385 (2016) 42–46
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
The reduction of critical H implantation dose for ion cut by incorporating B-doped SiGe/Si superlattice into Si substrate Zhongying Xue a,∗ , Da Chen b , Pengfei Jia a , Xing Wei a , Zengfeng Di a , Miao Zhang a,∗ a State Key Laboratory of Functional Materials for Informatics, Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, 865 Changning Road, Shanghai, 200050, China b Department of Microelectronic Science and Engineering, Ningbo University, Ningbo 315211, China
a r t i c l e
i n f o
Article history: Received 13 December 2015 Received in revised form 10 May 2016 Accepted 10 May 2016 Available online 13 May 2016 Keywords: H implantation Low dose Ion cut B-doped superlattice layer
a b s t r a c t An approach to achieve Si or SiGe film exfoliation with as low as 3 × 1016 /cm2 H implantation dose was investigated. Two intrinsic Si0.75 Ge0.25 /Si samples, merged with B-doped Si0.75 Ge0.25 layer and Bdoped Si0.75 Ge0.25 /Si superlattice (SL) layer respectively, were used to study the formation of crack after 3 × 1016 /cm2 H implantation and annealing. For the sample into which B doped Si0.75 Ge0.25 layer is incorporated, only few discrete cracks are observed along both sides of the B doped Si0.75 Ge0.25 layer; on the contrary, a continuous (100) oriented crack is formed in the B-doped Si0.75 Ge0.25 /Si SL layer, which means ion cut can be achieved using this material with 3 × 1016 /cm2 H implantation. As the SIMS profiles confirm that hydrogen tends to be trapped at B-doped SiGe/Si interface, the formation of continuous crack in SL layer can be ascribed to the more efficient hydrogen trapping by the multiple B-doped SiGe/Si interfaces. © 2016 Elsevier B.V. All rights reserved.
1. Introduction As a well known technique, ion cut process based on H implantation and wafer bonding has been employed to manufacture commercial silicon-on-insulator (SOI). Recently, ion cut technique has also been applied to fabricate SiGe-on-insulator (SGOI) and Germanium-on-insulator (GOI) [1–3]. SGOI and GOI materials, excelling in carrier mobility and having such advantages inherent to “on-insulator” structures as low parasitic capacitance, low leakage current and immunity to short channel effects, have been considered to be prospective channel materials for high performance Metal Oxide Semiconductor Field Effect Transistors (MOSFETs) in future technology nodes. However, due to the high stopping power of Ge atoms and the low mobility of point defects within the collision cascades, Ge is more likely to become amorphous than Si under ion implantation [4], which hampers the application of SGOI and GOI in IC industry. Therefore, developing a novel H implantation technique or considerably reducing the H implantation fluence becomes a big challenge to make high quality SGOI and GOI by ion cut technique. Utilizing plasma hydrogenation technique replacing conventional H implantation has been proposed to facilitate ion cut
∗ Corresponding authors. E-mail addresses: [email protected] (Z. Xue), [email protected] (M. Zhang). http://dx.doi.org/10.1016/j.apsusc.2016.05.047 0169-4332/© 2016 Elsevier B.V. All rights reserved.
process with lower radiation damage. In these approaches, different H trapping centers such as B-doped Si [5], Sb-doped Si [6] or strained SiGe [7] were merged to capture H efficiently. However, plasma hydrogenation is a thermal process that triggers the simultaneous occurrence of surface blistering, which, in turn, impedes the subsequent bonding process. On the contrary, conventional H implantation combined with high efficiency H trapping center may achieve ion cut with a prominently decreased H implantation dose. Recently, we have demonstrated an approach to create SGOI with half of typical H implantation fluence by inserting between Si0.75 Ge0.25 epitaxial layer and Si substrate a B doped Si0.70 Ge0.30 layer as a trapping center [8]. However, in order to trap sufficient H to form continuous cracking, the B doping concentration should be as high as 2 × 1019 /cm3 . In this work, we present a modified method to produce SGOI with 3 × 1016 /cm2 H fluence by combining B doped SiGe/Si superlattice (SL) with substrate, where the B concentration could dwindle to 8 × 1018 /cm3 .
2. Experimental Two SiGe heterostructures were epitaxially grown on p-type 8inch Si (100) substrates in a commercial reduced pressure chemical vapor deposition (RPCVD) system. Before deposition was carried out, Si substrates were cleaned in the standard RCA-1 and RCA2 solutions and rinsed in deionized water. After a Si wafer was loaded into the chamber, in-situ bake in H2 was performed to
Z. Xue et al. / Applied Surface Science 385 (2016) 42–46
43
Fig. 1. The sample schematic structures and the SEM micrographs after H ions implantation and annealing process. (a) schematic structure of sample A and process flow; (b) schematic structure of sample B and process flow; (c) SEM micrograph of sample A; (d) SEM micrograph of sample B.
remove any remaining oxide, and then SiGe heterostructure was deposited. As schematically shown by Fig. 1(a) and (b), the first SiGe heterostructure (sample A) comprises a 140 nm Si0.75 Ge0.25 layer, a 100 nm Si layer, a 15 nm B-doped Si0.75 Ge0.25 layer and Si substrate; the second sample (sample B) keeps a similar structure with sample A except that the 15 nm B-doped Si0.75 Ge0.25 layer is replaced by a 15 nm B-doped Si0.75 Ge0.25 /Si SL (three periods of 3 nm Si0.75 Ge0.25 /2 nm Si). For both samples, the B concentration of 15 nm interlayer is about 8 × 1018 /cm3 . The Ge compositions and B distributions are measured by SIMS and presented in supplementary information (Fig. S1). The purpose of inserting 100 nm Si layer is to block doped B diffusing into top 140 nm SiGe layer during the subsequent annealing process. What’s more, if the 140 nm SiGe/100 nm Si dual layers are transferred onto a SiO2 /Si handle wafer, the 100 nm Si layer can be taken as an etch sacrifice layer to achieve SGOI [9,10]. Both of the samples were implanted by 26 keV H+ ions with the dose of 3 × 1016 /cm2 and annealed at 600 ◦ C for 0.5 h in N2 atmosphere. Afterwards, the surface topographies of the samples were characterized by scanning electron microscope (SEM) and atomic force microscope (AFM), the microstructures of the heterostructures were observed by transmission electron microscopy (TEM), and the depth profiles of H were measured via secondary ion mass spectrometry (SIMS). 3. Results and discussion Fig. 1(c) and (d) shows the SEM micrographs of sample A and B, respectively. Although the H ions implantation and annealing process are virtually identical for both samples, the SEM images indicate the striking dissimilarity in the results. For sample A, only a few unruptured blisters, approximately 3 m in diameter, have
developed due to the low H ions implantation dose, whereas the blister quantities of sample B increase. Further, besides unruptured blisters, a considerable number of large craters (exfoliated blisters), basically 3–4 m in diameter, are observed in sample B, which means layer transfer would be achieved if a handle wafer was bonded [11]. Although doped B elements and stress between SiGe/Si play essential roles in the formation and expansion of blisters [5,8,12–15], considering the almost identical B doping concentration and Ge content in both samples, the low dose H implantation inducing blister rupturing in sample B should be ascribed to the more efficient hydrogen trapping by the SiGe/Si SL structure. Because most of the blisters’ heights fall into the range of several nanometers to tens of nanometers, hardly observable under SEM, AFM images were taken to identify them further, as shown by Fig. 2. In order to compare sample A and B, the measurement area of Fig. 2(b) is selected carefully to exclude ruptured blisters. Fig. 2(a) and (b) compared, it can be found that the blister density of sample A is slightly higher than that of sample B and the average blister size in sample A is smaller than that in sample B. The section analyses (demonstrated by white curves) reveal that for sample A, in the view range, the average blister height is no more than 7 nm, the maximum being 18 nm; for sample B, the blister height averages about 12 nm, maximum 50 nm. All of these suggest that more hydrogen were trapped in sample B and filled into microcavities, which, consequently, leads up to blisters expanding, even rupturing [16–18]. Fig. 2(c) shows some deep pits (ruptured blisters) of sample B, the section analysis suggests that the depth of the pit is about 240 nm, consistent with the total thickness of the 140 nm Si0.75 Ge0.25 and 100 nm Si, manifesting the film splitting occurs in SL layer instead of in the implantation range. As the color scale of Fig. 2(c) is 450 nm, significantly larger than that of Fig. 2(b), the
44
Z. Xue et al. / Applied Surface Science 385 (2016) 42–46
Fig. 3. Cross-sectional TEM micrograph and corresponding H distribution of sample A and sample B after H implantation and annealing. (a) sample A, (b) sample B. The inset of (b) shows the high resolution TEM micrograph of Si interlayer.
Fig. 2. The AFM images and section analyses of sample A and sample B; (a) sample A; (b) sample B excluding ruptured blisters; (c) sample B including ruptured blisters.
unruptured blisters in sample B cannot be observed in Fig. 2(c), but many fragments induced by blister rupturing are found scattered on the surface. The thermal evolution of implanted H ions was analyzed by using cross-sectional TEM micrographs and H SIMS in this work. As shown by Fig. 3(a) and (b), after H implantation and annealing, a dense band of defects due to H implantation are observed from both TEM micrographs. In Fig. 3(a), some dispersed platelets can be spotted from the defects band, vaguely though. In addition, some discrete cracks are formed along both sides of the B-doped Si0.75 Ge0.25 layer. Conversely, no platelet is discovered in the defects band in Fig. 3(b), but a continuous crack, parallel to the Si interlayer, is formed. From the high resolution TEM image of Si interlayer (inset of Fig. 3(b)), we can deduce that the continuous crack keeps (100) orientation. The differences between two TEM images suggest that compared to B-doped Si0.75 Ge0.25 layer, B-doped SL layer could promote H trapping more efficiently, as evidenced by the H SIMS data plotted and overlapped on TEM images. In Fig. 3(a), two H concentration peaks located at depth of 242 nm and 258 nm are found, coinciding with the locations of
Z. Xue et al. / Applied Surface Science 385 (2016) 42–46
45
Fig. 4. Process flow for ion cut by low dose H implantation process. (a) H implanted wafer bonds to SiO2 /Si handle wafer; (b) delamination of the wafer pair at the B-doped SiGe/Si interface; (c) cross-sectional TEM micrograph of as-cut sample; (d) cross-sectional TEM micrograph of SGOI.
separate cracks, and the H concentrations are 2.4 × 1020 /cm3 and 3.0 × 1020 /cm3 , respectively. In Fig. 3(b), the H concentration peak corresponding to the continuous crack is as high as 1.6 × 1021 /cm3 , remarkably higher than those in Fig. 3(a). It has been reported by literature that the critical Si layer exfoliation concentration is about 1 × 1021 /cm3 [13,19]. In sample B, the higher H concentration more than satisfies the film exfoliation condition, thus bringing a continuous crack into existence.
For sample A, since the Si0.75 Ge0.25 lattice is larger than Si, the B doped Si0.75 Ge0.25 interlayer possesses an in-plane compressive strain (Fig. S2), meanwhile, the Si0.75 Ge0.25 /Si interface is in a state of shear strain. As reported by literature [5,15], compressive strain and interface shear play important roles in crack formation. Furthermore, the doped B elements can not only facilitate the H migration and trapping, but also lower the energy barrier for the breaking of atomic bonds [12]. Therefore, the phenomenon of H accumulation and cracks emergence at B doped SiGe/Si interface
46
Z. Xue et al. / Applied Surface Science 385 (2016) 42–46
can be attributed to the joint action of compressive strain in SiGe layer, shear strain at interface and B doping: (1) Vacancies are inclined to agglomerate in regions of in-plane compressive strain [5]; (2) As B doping and interface shear can facilitate the H migration and trapping [14], more H atoms tend to be trapped at B doped SiGe/Si interface; (3) Because of the interaction of H with vacancy agglomerates, (100)-oriented platelets or H-filled bubbles tend to be formed at SiGe/Si interface; (4) Once (100)-oriented platelets are formed, the shear strain can facilitate cracking along the interface during annealing process [12]. As the B doping concentration and interface stress of sample A is not high enough, the numbers of platelets or H-filled bubbles are insufficient, which, consequently, gives rise to the discrete cracks. Considering almost identical B doping concentration, Ge content and SiGe strain degree of both samples (Fig. S2 and S3), we may account for the formation of continuous crack in sample B as below: Because each B-doped SiGe interface acts as an H trapping layer, the multiple B-doped SiGe/Si interfaces trap much more H within SL layers, as confirmed by H SIMS profiles. Correspondingly, the quantities of platelets or Hfilled bubbles increase substantially, which induces the formation of continuous crack. After H implantation, plasma activation and wafer bonding, the intrinsic Si0.75 Ge0.25 layer, Si layer and parts of SL layer of sample B are transferred onto a SiO2 /Si substrate, as schematically shown by Fig. 4(a) and (b). Fig. 4(c) shows the TEM image of as-cut sample, it clearly demonstrates that the film exfoliates along one of the SL interfaces: in some areas, along middle Si0.75 Ge0.25 /Si interface, in other areas, along upper Si0.75 Ge0.25 /Si interface. The mechanism of film preferentially exfoliating along SiGe/Si interface has been analyzed above: the shear stress between SiGe/Si and the supersaturated hydrogen accumulated in the B-doped SiGe layer facilitate the formation of crack along SiGe/Si interface. After etching the residual Si0.75 Ge0.25 /Si SL and 100 nm Si interlayer by HNO3 :HAc:HF etchant and TMAH solution respectively, SGOI was obtained, as shown in Fig. 4(d). 4. Conclusions In summary, we have presented an approach to achieve Si or SiGe film exfoliation by adopting half of the typical H fluence required for conventional ion cut process. It is found that the Bdoped SiGe/Si SL is a high efficiency H trapping center due to its multiple interfaces, so film splitting could occur within the SL layer using 3 × 1016 /cm2 H fluence. As the continuous crack emerges within B-doped SL layer instead of within the H ion implantation range, the present approach can not only decrease the critical H ion dose remarkably, but also confine the exfoliated film to a precise thickness.
Acknowledgments This work was financially supported from Creative Research Groups of National Natural Science Foundation of China (No. 61321492), National Natural Science Foundation of China under Grant Nos. 61176001, 51222211, 61274136, Chinese Academy of Sciences (CAS) International Collaboration and Innovation Program on High Mobility Materials Engineering. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.apsusc.2016.05. 047. References [1] C. Yu, C. Lee, C. Lin, C. Liu, Appl. Phys. Lett. 89 (2006) 101913. [2] C. Deguet, IEEE Int. SOI Conf. 2004 (2004). [3] G. Taraschi, A.J. Pitera, E.A. Fitzgerald, Solid State Electron. 48 (2004) 1297–1305. [4] G. Impellizzeri, S. Mirabella, M.G. Grimaldi, Appl. Phys. A—Mater. 103 (2011) 323–328. [5] L. Shao, Y. Lin, J.G. Swadener, J.K. Lee, Q.X. Jia, Y.Q. Wang, M. Nastasi, P.E. Thompson, N.D. Theodore, T.L. Alford, J.W. Mayer, P. Chen, S.S. Lau, Appl. Phys. Lett. 88 (2006) 021901. [6] L. Shao, Y. Lin, J.G. Swadener, J.K. Lee, Q.X. Jia, Y.Q. Wang, M. Nastasi, P.E. Thompson, N.D. Theodore, T.L. Alford, J.W. Mayer, P. Chen, S.S. Lau, Appl. Phys. Lett. 87 (2005) 251907. [7] Z.F. Di, Y.Q. Wang, M. Nastasi, F. Rossi, L. Shao, P.E. Thompson, Appl. Phys. Lett. 93 (2008) 254104. [8] D. Chen, M. Zhang, S. Liu, Y. Wang, M. Nastasi, Z. Xue, X. Wang, Z. Di, Appl. Phys. Lett. 103 (2013) 142102. [9] G. Taraschi, A.J. Pitera, L.M. McGill, Z.Y. Cheng, M.J.L. Lee, T.A. Langdo, E.A. Fitzgerald, J. Electrochem. Soc. 151 (2004) G47–G56. [10] G. Taraschi, T.A. Langdo, M.T. Currie, E.A. Fitzgerald, D.A. Antoniadis, J. Vac, Sci. Technol. B 20 (2002) 725–727. [11] M.K. Weldon, V.E. Marsico, Y.J. Chabal, A. Agarwal, D.J. Eaglesham, J. Sapjeta, W.L. Brown, D.C. Jacobson, Y. Caudano, S.B. Christman, E.E. Chaban, J. Vac, Sci. Technol. B 15 (1997) 1065–1073. [12] D. Chen, Z.Y. Xue, G. Wang, Q.L. Guo, L.J. Lin, M. Zhang, S. Liu, X. Wei, Appl. Phys. Exp. 7 (2014) 061302. [13] A.J. Pitera, E.A. Fitzgerald, J. Appl. Phys. 97 (2005) 104511. [14] S.W. Lee, C.A. Chueh, H.T. Chang, J. Electrochem. Soc. 156 (2009) H921–H924. [15] J.K. Lee, R.D. Averitt, M. Nastasi, J. Appl Phys. 96 (2004) 7045–7051. [16] X. Lu, N.W. Cheung, M.D. Strathman, P.K. Chu, B. Doyle, Appl. Phys. Lett. 71 (1997) 1804–1806. [17] L.J. Huang, Q.Y. Tong, Y.L. Chao, T.H. Lee, T. Martini, U. Gosele, Appl. Phys. Lett. 74 (1999) 982–984. [18] C. Coupeau, G. Parry, J. Colin, M.L. David, J. Labanowski, J. Grilhe, Appl. Phys. Lett. 103 (2013) 031908. [19] T. Ichimiya, A. Furuichi, Int. J. Appl. Radiat. Isot. 19 (1968) 573–578.
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
The reduction of critical H implantation dose for ion cut by incorporating B-doped SiGe/Si superlattice into Si substrate Zhongying Xue a,∗ , Da Chen b , Pengfei Jia a , Xing Wei a , Zengfeng Di a , Miao Zhang a,∗ a State Key Laboratory of Functional Materials for Informatics, Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, 865 Changning Road, Shanghai, 200050, China b Department of Microelectronic Science and Engineering, Ningbo University, Ningbo 315211, China
a r t i c l e
i n f o
Article history: Received 13 December 2015 Received in revised form 10 May 2016 Accepted 10 May 2016 Available online 13 May 2016 Keywords: H implantation Low dose Ion cut B-doped superlattice layer
a b s t r a c t An approach to achieve Si or SiGe film exfoliation with as low as 3 × 1016 /cm2 H implantation dose was investigated. Two intrinsic Si0.75 Ge0.25 /Si samples, merged with B-doped Si0.75 Ge0.25 layer and Bdoped Si0.75 Ge0.25 /Si superlattice (SL) layer respectively, were used to study the formation of crack after 3 × 1016 /cm2 H implantation and annealing. For the sample into which B doped Si0.75 Ge0.25 layer is incorporated, only few discrete cracks are observed along both sides of the B doped Si0.75 Ge0.25 layer; on the contrary, a continuous (100) oriented crack is formed in the B-doped Si0.75 Ge0.25 /Si SL layer, which means ion cut can be achieved using this material with 3 × 1016 /cm2 H implantation. As the SIMS profiles confirm that hydrogen tends to be trapped at B-doped SiGe/Si interface, the formation of continuous crack in SL layer can be ascribed to the more efficient hydrogen trapping by the multiple B-doped SiGe/Si interfaces. © 2016 Elsevier B.V. All rights reserved.
1. Introduction As a well known technique, ion cut process based on H implantation and wafer bonding has been employed to manufacture commercial silicon-on-insulator (SOI). Recently, ion cut technique has also been applied to fabricate SiGe-on-insulator (SGOI) and Germanium-on-insulator (GOI) [1–3]. SGOI and GOI materials, excelling in carrier mobility and having such advantages inherent to “on-insulator” structures as low parasitic capacitance, low leakage current and immunity to short channel effects, have been considered to be prospective channel materials for high performance Metal Oxide Semiconductor Field Effect Transistors (MOSFETs) in future technology nodes. However, due to the high stopping power of Ge atoms and the low mobility of point defects within the collision cascades, Ge is more likely to become amorphous than Si under ion implantation [4], which hampers the application of SGOI and GOI in IC industry. Therefore, developing a novel H implantation technique or considerably reducing the H implantation fluence becomes a big challenge to make high quality SGOI and GOI by ion cut technique. Utilizing plasma hydrogenation technique replacing conventional H implantation has been proposed to facilitate ion cut
∗ Corresponding authors. E-mail addresses: [email protected] (Z. Xue), [email protected] (M. Zhang). http://dx.doi.org/10.1016/j.apsusc.2016.05.047 0169-4332/© 2016 Elsevier B.V. All rights reserved.
process with lower radiation damage. In these approaches, different H trapping centers such as B-doped Si [5], Sb-doped Si [6] or strained SiGe [7] were merged to capture H efficiently. However, plasma hydrogenation is a thermal process that triggers the simultaneous occurrence of surface blistering, which, in turn, impedes the subsequent bonding process. On the contrary, conventional H implantation combined with high efficiency H trapping center may achieve ion cut with a prominently decreased H implantation dose. Recently, we have demonstrated an approach to create SGOI with half of typical H implantation fluence by inserting between Si0.75 Ge0.25 epitaxial layer and Si substrate a B doped Si0.70 Ge0.30 layer as a trapping center [8]. However, in order to trap sufficient H to form continuous cracking, the B doping concentration should be as high as 2 × 1019 /cm3 . In this work, we present a modified method to produce SGOI with 3 × 1016 /cm2 H fluence by combining B doped SiGe/Si superlattice (SL) with substrate, where the B concentration could dwindle to 8 × 1018 /cm3 .
2. Experimental Two SiGe heterostructures were epitaxially grown on p-type 8inch Si (100) substrates in a commercial reduced pressure chemical vapor deposition (RPCVD) system. Before deposition was carried out, Si substrates were cleaned in the standard RCA-1 and RCA2 solutions and rinsed in deionized water. After a Si wafer was loaded into the chamber, in-situ bake in H2 was performed to
Z. Xue et al. / Applied Surface Science 385 (2016) 42–46
43
Fig. 1. The sample schematic structures and the SEM micrographs after H ions implantation and annealing process. (a) schematic structure of sample A and process flow; (b) schematic structure of sample B and process flow; (c) SEM micrograph of sample A; (d) SEM micrograph of sample B.
remove any remaining oxide, and then SiGe heterostructure was deposited. As schematically shown by Fig. 1(a) and (b), the first SiGe heterostructure (sample A) comprises a 140 nm Si0.75 Ge0.25 layer, a 100 nm Si layer, a 15 nm B-doped Si0.75 Ge0.25 layer and Si substrate; the second sample (sample B) keeps a similar structure with sample A except that the 15 nm B-doped Si0.75 Ge0.25 layer is replaced by a 15 nm B-doped Si0.75 Ge0.25 /Si SL (three periods of 3 nm Si0.75 Ge0.25 /2 nm Si). For both samples, the B concentration of 15 nm interlayer is about 8 × 1018 /cm3 . The Ge compositions and B distributions are measured by SIMS and presented in supplementary information (Fig. S1). The purpose of inserting 100 nm Si layer is to block doped B diffusing into top 140 nm SiGe layer during the subsequent annealing process. What’s more, if the 140 nm SiGe/100 nm Si dual layers are transferred onto a SiO2 /Si handle wafer, the 100 nm Si layer can be taken as an etch sacrifice layer to achieve SGOI [9,10]. Both of the samples were implanted by 26 keV H+ ions with the dose of 3 × 1016 /cm2 and annealed at 600 ◦ C for 0.5 h in N2 atmosphere. Afterwards, the surface topographies of the samples were characterized by scanning electron microscope (SEM) and atomic force microscope (AFM), the microstructures of the heterostructures were observed by transmission electron microscopy (TEM), and the depth profiles of H were measured via secondary ion mass spectrometry (SIMS). 3. Results and discussion Fig. 1(c) and (d) shows the SEM micrographs of sample A and B, respectively. Although the H ions implantation and annealing process are virtually identical for both samples, the SEM images indicate the striking dissimilarity in the results. For sample A, only a few unruptured blisters, approximately 3 m in diameter, have
developed due to the low H ions implantation dose, whereas the blister quantities of sample B increase. Further, besides unruptured blisters, a considerable number of large craters (exfoliated blisters), basically 3–4 m in diameter, are observed in sample B, which means layer transfer would be achieved if a handle wafer was bonded [11]. Although doped B elements and stress between SiGe/Si play essential roles in the formation and expansion of blisters [5,8,12–15], considering the almost identical B doping concentration and Ge content in both samples, the low dose H implantation inducing blister rupturing in sample B should be ascribed to the more efficient hydrogen trapping by the SiGe/Si SL structure. Because most of the blisters’ heights fall into the range of several nanometers to tens of nanometers, hardly observable under SEM, AFM images were taken to identify them further, as shown by Fig. 2. In order to compare sample A and B, the measurement area of Fig. 2(b) is selected carefully to exclude ruptured blisters. Fig. 2(a) and (b) compared, it can be found that the blister density of sample A is slightly higher than that of sample B and the average blister size in sample A is smaller than that in sample B. The section analyses (demonstrated by white curves) reveal that for sample A, in the view range, the average blister height is no more than 7 nm, the maximum being 18 nm; for sample B, the blister height averages about 12 nm, maximum 50 nm. All of these suggest that more hydrogen were trapped in sample B and filled into microcavities, which, consequently, leads up to blisters expanding, even rupturing [16–18]. Fig. 2(c) shows some deep pits (ruptured blisters) of sample B, the section analysis suggests that the depth of the pit is about 240 nm, consistent with the total thickness of the 140 nm Si0.75 Ge0.25 and 100 nm Si, manifesting the film splitting occurs in SL layer instead of in the implantation range. As the color scale of Fig. 2(c) is 450 nm, significantly larger than that of Fig. 2(b), the
44
Z. Xue et al. / Applied Surface Science 385 (2016) 42–46
Fig. 3. Cross-sectional TEM micrograph and corresponding H distribution of sample A and sample B after H implantation and annealing. (a) sample A, (b) sample B. The inset of (b) shows the high resolution TEM micrograph of Si interlayer.
Fig. 2. The AFM images and section analyses of sample A and sample B; (a) sample A; (b) sample B excluding ruptured blisters; (c) sample B including ruptured blisters.
unruptured blisters in sample B cannot be observed in Fig. 2(c), but many fragments induced by blister rupturing are found scattered on the surface. The thermal evolution of implanted H ions was analyzed by using cross-sectional TEM micrographs and H SIMS in this work. As shown by Fig. 3(a) and (b), after H implantation and annealing, a dense band of defects due to H implantation are observed from both TEM micrographs. In Fig. 3(a), some dispersed platelets can be spotted from the defects band, vaguely though. In addition, some discrete cracks are formed along both sides of the B-doped Si0.75 Ge0.25 layer. Conversely, no platelet is discovered in the defects band in Fig. 3(b), but a continuous crack, parallel to the Si interlayer, is formed. From the high resolution TEM image of Si interlayer (inset of Fig. 3(b)), we can deduce that the continuous crack keeps (100) orientation. The differences between two TEM images suggest that compared to B-doped Si0.75 Ge0.25 layer, B-doped SL layer could promote H trapping more efficiently, as evidenced by the H SIMS data plotted and overlapped on TEM images. In Fig. 3(a), two H concentration peaks located at depth of 242 nm and 258 nm are found, coinciding with the locations of
Z. Xue et al. / Applied Surface Science 385 (2016) 42–46
45
Fig. 4. Process flow for ion cut by low dose H implantation process. (a) H implanted wafer bonds to SiO2 /Si handle wafer; (b) delamination of the wafer pair at the B-doped SiGe/Si interface; (c) cross-sectional TEM micrograph of as-cut sample; (d) cross-sectional TEM micrograph of SGOI.
separate cracks, and the H concentrations are 2.4 × 1020 /cm3 and 3.0 × 1020 /cm3 , respectively. In Fig. 3(b), the H concentration peak corresponding to the continuous crack is as high as 1.6 × 1021 /cm3 , remarkably higher than those in Fig. 3(a). It has been reported by literature that the critical Si layer exfoliation concentration is about 1 × 1021 /cm3 [13,19]. In sample B, the higher H concentration more than satisfies the film exfoliation condition, thus bringing a continuous crack into existence.
For sample A, since the Si0.75 Ge0.25 lattice is larger than Si, the B doped Si0.75 Ge0.25 interlayer possesses an in-plane compressive strain (Fig. S2), meanwhile, the Si0.75 Ge0.25 /Si interface is in a state of shear strain. As reported by literature [5,15], compressive strain and interface shear play important roles in crack formation. Furthermore, the doped B elements can not only facilitate the H migration and trapping, but also lower the energy barrier for the breaking of atomic bonds [12]. Therefore, the phenomenon of H accumulation and cracks emergence at B doped SiGe/Si interface
46
Z. Xue et al. / Applied Surface Science 385 (2016) 42–46
can be attributed to the joint action of compressive strain in SiGe layer, shear strain at interface and B doping: (1) Vacancies are inclined to agglomerate in regions of in-plane compressive strain [5]; (2) As B doping and interface shear can facilitate the H migration and trapping [14], more H atoms tend to be trapped at B doped SiGe/Si interface; (3) Because of the interaction of H with vacancy agglomerates, (100)-oriented platelets or H-filled bubbles tend to be formed at SiGe/Si interface; (4) Once (100)-oriented platelets are formed, the shear strain can facilitate cracking along the interface during annealing process [12]. As the B doping concentration and interface stress of sample A is not high enough, the numbers of platelets or H-filled bubbles are insufficient, which, consequently, gives rise to the discrete cracks. Considering almost identical B doping concentration, Ge content and SiGe strain degree of both samples (Fig. S2 and S3), we may account for the formation of continuous crack in sample B as below: Because each B-doped SiGe interface acts as an H trapping layer, the multiple B-doped SiGe/Si interfaces trap much more H within SL layers, as confirmed by H SIMS profiles. Correspondingly, the quantities of platelets or Hfilled bubbles increase substantially, which induces the formation of continuous crack. After H implantation, plasma activation and wafer bonding, the intrinsic Si0.75 Ge0.25 layer, Si layer and parts of SL layer of sample B are transferred onto a SiO2 /Si substrate, as schematically shown by Fig. 4(a) and (b). Fig. 4(c) shows the TEM image of as-cut sample, it clearly demonstrates that the film exfoliates along one of the SL interfaces: in some areas, along middle Si0.75 Ge0.25 /Si interface, in other areas, along upper Si0.75 Ge0.25 /Si interface. The mechanism of film preferentially exfoliating along SiGe/Si interface has been analyzed above: the shear stress between SiGe/Si and the supersaturated hydrogen accumulated in the B-doped SiGe layer facilitate the formation of crack along SiGe/Si interface. After etching the residual Si0.75 Ge0.25 /Si SL and 100 nm Si interlayer by HNO3 :HAc:HF etchant and TMAH solution respectively, SGOI was obtained, as shown in Fig. 4(d). 4. Conclusions In summary, we have presented an approach to achieve Si or SiGe film exfoliation by adopting half of the typical H fluence required for conventional ion cut process. It is found that the Bdoped SiGe/Si SL is a high efficiency H trapping center due to its multiple interfaces, so film splitting could occur within the SL layer using 3 × 1016 /cm2 H fluence. As the continuous crack emerges within B-doped SL layer instead of within the H ion implantation range, the present approach can not only decrease the critical H ion dose remarkably, but also confine the exfoliated film to a precise thickness.
Acknowledgments This work was financially supported from Creative Research Groups of National Natural Science Foundation of China (No. 61321492), National Natural Science Foundation of China under Grant Nos. 61176001, 51222211, 61274136, Chinese Academy of Sciences (CAS) International Collaboration and Innovation Program on High Mobility Materials Engineering. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.apsusc.2016.05. 047. References [1] C. Yu, C. Lee, C. Lin, C. Liu, Appl. Phys. Lett. 89 (2006) 101913. [2] C. Deguet, IEEE Int. SOI Conf. 2004 (2004). [3] G. Taraschi, A.J. Pitera, E.A. Fitzgerald, Solid State Electron. 48 (2004) 1297–1305. [4] G. Impellizzeri, S. Mirabella, M.G. Grimaldi, Appl. Phys. A—Mater. 103 (2011) 323–328. [5] L. Shao, Y. Lin, J.G. Swadener, J.K. Lee, Q.X. Jia, Y.Q. Wang, M. Nastasi, P.E. Thompson, N.D. Theodore, T.L. Alford, J.W. Mayer, P. Chen, S.S. Lau, Appl. Phys. Lett. 88 (2006) 021901. [6] L. Shao, Y. Lin, J.G. Swadener, J.K. Lee, Q.X. Jia, Y.Q. Wang, M. Nastasi, P.E. Thompson, N.D. Theodore, T.L. Alford, J.W. Mayer, P. Chen, S.S. Lau, Appl. Phys. Lett. 87 (2005) 251907. [7] Z.F. Di, Y.Q. Wang, M. Nastasi, F. Rossi, L. Shao, P.E. Thompson, Appl. Phys. Lett. 93 (2008) 254104. [8] D. Chen, M. Zhang, S. Liu, Y. Wang, M. Nastasi, Z. Xue, X. Wang, Z. Di, Appl. Phys. Lett. 103 (2013) 142102. [9] G. Taraschi, A.J. Pitera, L.M. McGill, Z.Y. Cheng, M.J.L. Lee, T.A. Langdo, E.A. Fitzgerald, J. Electrochem. Soc. 151 (2004) G47–G56. [10] G. Taraschi, T.A. Langdo, M.T. Currie, E.A. Fitzgerald, D.A. Antoniadis, J. Vac, Sci. Technol. B 20 (2002) 725–727. [11] M.K. Weldon, V.E. Marsico, Y.J. Chabal, A. Agarwal, D.J. Eaglesham, J. Sapjeta, W.L. Brown, D.C. Jacobson, Y. Caudano, S.B. Christman, E.E. Chaban, J. Vac, Sci. Technol. B 15 (1997) 1065–1073. [12] D. Chen, Z.Y. Xue, G. Wang, Q.L. Guo, L.J. Lin, M. Zhang, S. Liu, X. Wei, Appl. Phys. Exp. 7 (2014) 061302. [13] A.J. Pitera, E.A. Fitzgerald, J. Appl. Phys. 97 (2005) 104511. [14] S.W. Lee, C.A. Chueh, H.T. Chang, J. Electrochem. Soc. 156 (2009) H921–H924. [15] J.K. Lee, R.D. Averitt, M. Nastasi, J. Appl Phys. 96 (2004) 7045–7051. [16] X. Lu, N.W. Cheung, M.D. Strathman, P.K. Chu, B. Doyle, Appl. Phys. Lett. 71 (1997) 1804–1806. [17] L.J. Huang, Q.Y. Tong, Y.L. Chao, T.H. Lee, T. Martini, U. Gosele, Appl. Phys. Lett. 74 (1999) 982–984. [18] C. Coupeau, G. Parry, J. Colin, M.L. David, J. Labanowski, J. Grilhe, Appl. Phys. Lett. 103 (2013) 031908. [19] T. Ichimiya, A. Furuichi, Int. J. Appl. Radiat. Isot. 19 (1968) 573–578.