1. Introduction

Optical communication has provided the exclusive means for carrying high capacity data over long distances for over three decades [1]. As modern data storage and computing rely increasingly on high-rate sharing of information, optics-based techniques steadily penetrate towards rack-level, board-level and even chip-level communications [2,3]. The future growth of both computation and communication depends, to a large extent, on the successful integration of optical communication system functionalities alongside electronic integrated circuits on the silicon material platform [2,3]. Hence, the realization of photonic devices on silicon, or silicon photonics, is a research area of much interest and significance.

While the silicon-on-insulator (SOI) material platform is generally favorable for making passive devices such as waveguides, interferometers and resonators, the properties of silicon raise several challenges to the implementation of active photonic devices [2,3]. For example, the indirect semiconductor bandgap of silicon renders the generation and amplification of light though stimulated emission highly inefficient, and its lack of a Pockels effect makes electro-optic modulation challenging [4]. Much effort has been dedicated in recent years to overcoming these deficiencies, leading to all-silicon modulators [5,6] and even light sources [7]. Nevertheless, state-of-the-art silicon-photonic light sources, amplifiers, modulators and detectors rely on the hybrid integration of additional electro-optic materials, primarily InP-based semiconductors, on top of SOI waveguides [8–16].

The foremost challenge in the fabrication of hybrid-silicon photonic devices is to bond wafers of dissimilar materials. Unfortunately, the direct epitaxial growth of standard GaAs-based and InP-based materials on silicon substrates is difficult, primarily due to the mismatch in lattice constants and in thermal expansion coefficients [17]. Hybrid silicon-devices reported to-date make use of either direct bonding, or an intermediate thermosetting adhesive polymer layer such as benzocyclobutene (BCB) based polymers [18]. Annealing temperatures of hybrid-photonic devices are restricted to 350 °C or lower. Therefore, direct bonding processes often rely on oxygen plasma treatment of both substrates [19], as the preferred approach when modest annealing temperatures are required. Initially, hydrogen bonds form between the two oxidized surfaces, and subsequent annealing then leads to covalent bonds being formed across the interface between SOI and InP. In BCB-mediated bonding, a diluted oligomer solution is spin-coated onto the InP sample. The solvent is then evaporated, the InP and SOI samples are pressed together and the adhesive is cured [18]. Although both methods are successfully employed in the fabrication of hybrid silicon-InP electro-optic devices, they are not without drawbacks. Low-temperature direct bonding suffers from out-gassing of by-products such as hydrogen or water which can lead to local de-bonding. It also does not tolerate even modest levels of contamination and surface roughness. BCB-based bonding results in a relatively thick interface which can hinder the coupling of light and the thermal conductivity across the interface.

We report herein an alternative procedure for the bonding of silicon to InP and the bonding of two silicon wafers to each other. Our process relies on the deposition of self-assembled monolayers (SAMs) of an organic material on either one or both surfaces to be bonded [20–22]. The monolayer-forming molecule consists of a hydrocarbon chain with a terminal functional group that bonds to the surface of the substrate, while the other terminus bears chemical functionality that remains free-standing on the substrate surface following the deposition process. The free-standing group is modified by controlled in situ chemistry so that it can bond to a SAM on the opposite substrate or with the opposite substrate itself.

Compared with direct bonding processes [18], SAM-assisted bonding is carried out at lower temperatures (120-150 °C), is potentially free of outgassing, and provides the flexibility of adjusting the surface chemistry to accommodate a variety of materials. The < 5 nm-thick interface is much thinner than that of BCB bonding and should not disrupt the transfer of light between the substrates or the electrical conductivity across the interface [23]. At the same time, the few-nm interface may relax the surface roughness requirements that are imposed by direct bonding [24]. SAM-assisted bonding of two silicon wafers was previously reported [25]. We report herein a modification of this methodology which takes advantage of controlled surface chemistry on well-defined functionalized monolayers and extends this bonding paradigm to non-silicon, electro-optically active materials. The work below extends a brief earlier report [26].

2. Monolayer-assisted bonding of silicon-to-InP and silicon-to-silicon

The silicon samples used in this study were taken from n-type, 1-10 Ω⋅cm, (1 0 0) oriented commercially-available wafers. The root-mean-squared micro-roughness of the polished wafer surfaces was verified by atomic-force microscopy to be on the order of 0.3 nm. The sizes of the samples were approximately 1 × 1 cm². The samples were thoroughly cleaned in organic solvents and piranha (H₂SO₄/H₂O₂) solution to remove organic contaminants (see appendix for further details on all experimental procedures).

The monolayer-forming molecules are shown in Fig. 1(a) . They consist of a chain of n CH₂ units, with two terminal functional groups. In most experiments n = 9 was used, although successful bonding was achieved using n = 14 as well. The length of the molecular chain is easily varied based on synthesis using commercially-available materials [27,28]. The functional group that anchors onto the silicon native oxide surface was trichlorosilane (SiCl₃) [27–30] (see appendix for further detail). The other end of the molecule is thioacetate (see Fig. 1(a)), which is modified in later stages of the bonding process. The process of self-assembly is illustrated in Fig. 1(b). The silicon samples are placed in a solution of the monolayer-forming molecule in dicyclohexyl for an hour at room temperature. The trichlorosilane groups attach to the oxidized silicon surfaces, and form an ordered, well-packed array of molecular chains with thioacetate groups on the surface [27–30].

 

Fig. 1 (a) The monolayer-forming molecule is a polymethylene chain, terminated by a trichlorosilane group on one end and a thioacetate group on the other end. (b) Illustration of monolayer self-assembly on an oxidized silicon surface (see also appendix).

Download Full Size | PDF

The samples are then cleaned to remove physisorbed residues of the monolayer-deposition solution. The assembly of monolayers is readily evident by changes in the surface wetting of the substrate wherein the hydrophilic surface of oxidized silicon becomes relatively hydrophobic following SAM deposition. The water contact angle of the monolayer-coated surface is 74° (adv) /70° (rec) for n = 9, and 80° (adv) /78° (rec) for n = 14, and the thickness of the resulting monolayer (by ellipsometry) is 1.8 nm and 2.4 nm for the two chain lengths, respectively [27–30]. Next, the free-standing surface thioacetate groups are cleaved to create an array of thiol (S-H) groups for bonding (Fig. 2 , see appendix). Acetyl removal is monitored by infra-red spectroscopy based on the disappearance of the characteristic C = O double bond signal [30,31].

 

Fig. 2 Illustration of the hydrolysis reaction for cleaving the free-standing thioacetate end-group of self-assembled monolayers, leaving a thiol-decorated surface (see also appendix).

Download Full Size | PDF

Subsequent processing depends on the specific substrates to be bonded and the two bonding paradigms that have been pursued. In one case, we have extended the known self-assembly from solution of alkylthiols onto InP [32,33] to the bonding of a thiol-decorated silicon substrate to InP, as illustrated in Fig. 3(a) . Following a thorough solvent clean and immersion in HF, an oxide-free InP surface was brought in contact with the thiol-functionalized-SAM-modified silicon samples (see appendix). The size of the commercially available InP (100) samples was typically 8 × 8 mm², and the root-mean-squared micro-roughness of their surfaces was smaller than 0.2 nm. Bonding of InP to thiol-SAM-coated silicon was achieved at 120-150 °C for 12-24 hours in a bonding press. Figure 3(b) shows a bonded pair of samples, and Fig. 3(c) presents a scanning electron microscope (SEM) image of the cross-section of the bonding interface (made using a focused ion beam). Analysis of high-resolution SEM images of such cross-sections provided an upper limit of 6 nm for the thickness of the bonding interface (which includes both the silicon oxide and the SAM).

 

Fig. 3 (a) Bonding a thiol-bearing silicon sample to an InP sample. (b) Bonded InP and silicon samples. (c) Scanning electron microscope cross-section of bonded samples of InP and silicon, following cleaving and focused ion beam processing.

Download Full Size | PDF

The quality of the bonding was evaluated using infra-red microscopy. Figure 4 shows an image of the bonding interface between a silicon sample and an InP sample (a different sample from those shown in Figs. 3(b) and 3(c)). The InP piece was cut out of the edge of the wafer, so that a region on the right-hand side could not be bonded (see figure). Interferometric rings are evident in the area that is not bonded, whereas the bonded region is free of such rings. The image illustrates the formation of a continuous bonded interface in the flat part of the sample. InP samples taken from the center of the wafer exhibited a uniform bonding across their entire surface.

 

Fig. 4 Top-view, infra-red microscopy image of an InP sample and a silicon sample bonded together. Rings appear in a corner region that was taken from the edge of the InP wafer and could not be bonded. An illustration of the partially bonded samples is shown to the right of the microscopy image.

Download Full Size | PDF

A second kind of process bonds thiol-SAM modified silicon wafers to a second monolayer-treated surface. In this case, the thiol groups on a monolayer surface are converted into disulfides using iodine in methanol (Fig. 5(a) , appendix) [34]. The bonding of a thiol-SAM-modified silicon wafer to such a disulfide-bearing surface is based on a disulfide exchange reaction in which the intra-SAM S-S bonds of the disulfide surface are opened by reaction with a thiol-decorated monolayer to form new disulfide bonds across the interface (see Fig. 5(b)) [25,31]. Figure 5(c) shows a pair of silicon samples bonded using this process.

 

Fig. 5 (a) In situ disulfide-SAM formation (see also appendix). (b) Disulfide exchange bonding a thiol-bearing sample to a disulfide bearing sample. (c) Two silicon samples bonded using the above procedure.

Download Full Size | PDF

The above bonding experiments were carried out without the benefit of a clean-room environment, leading to some variation in the pull-test strengths of silicon-to-silicon bonded samples. The strengths of 20% of the samples were in the range of 2-5 MPa, whereas the strengths of about 50% of the samples were on the order of 1 MPa or lower. Remarkably, virtually all samples were strong enough to withstand rough handling, in spite of the unfavorable preparation conditions. As a control experiment, the pull test strengths of silicon samples that were directly bonded after wet cleaning by organic solvents and piranha solution [35] at these relatively modest temperatures were always below 2 MPa. We also attempted silicon to silicon bonding with thiol-terminated SAMs on both surfaces, and with disulfide-terminated SAMs on both surfaces (analogous to [25]). In these two cases we could not get pull test strengths above 1 MPa. In spite of the variability in results, the disulfide-to-thiol SAM-assisted bonding did provide statistically higher pull test strengths. Future work will evaluate the strengths of samples bonded in a clean room environment. We suggest that the relative hydrophobicity of the SAM treated surface minimizes surface contamination and combines with the modest flexibility introduced by the monolayer chains to provide a robust, readily generalizable process.

The pull test strengths of InP bonded to silicon samples reached 2 MPa. The higher end of the obtained strength values exceeds those in the literature [36]. Here too, virtually all bonded samples provided handling strength. Further work is necessary to quantify the strength of this bonding in a controlled environment. Also, the effects on bonding strength of the chain length of the monolayer-forming molecule, the use of various combinations of thiols and disulfides, and the bonding temperature, are still under study. In particular, pull test strengths of InP and silicon samples, directly bonded using oxygen plasma as a control experiment following the work of Pasquariello and Hjort [37], did reach over 3 MPa.

3. Conclusions

In summary, we have proposed and demonstrated SAM-based wafer bonding between a monolayer modified surface and a second, bare, substrate and bonding between two monolayer-modified surfaces. The application of SAMs to the low-temperature bonding of InP to silicon is particularly important for its potential application in hybrid photonic devices. We suggest that bonding based on SAMs provides several significant potential advantages: a) the functional groups at the termini of the monolayer-forming molecules can be adjusted to accommodate a variety of materials; b) relatively low temperatures are used; c) the disulfide exchange-based variant of the process is free of outgassing; and d) the few nm-thin bonding interface could mitigate the surface flatness requirement of direct bonding without disrupting light transfer. Ongoing work is aimed at the fabrication of hybrid-silicon photonic devices based on these bonding principles. The robustness and stability of the bonding interface during operation of active devices remains to be evaluated.

Appendix: experimental procedures

Preparation of silicon samples: Samples were cleaned in chloroform, acetone and ethanol, blown dry by a flow of filtered nitrogen, and immersed in a mixture of H₂SO₄ and H₂O₂ (7:3 ratio) for 20 minutes at 80 °C to remove organic contaminants. The samples were rinsed in deionized water and blown dry by filtered nitrogen.

Monolayer self-assembly on oxidized silicon surfaces: The monolayer-forming molecules are prepared by literature procedures [27,28] and mixed with dicyclohexyl (20 μL:10 mL) in a dried test tube. The samples are put in this solution for 1 hour at room temperature, after which they are rinsed in chloroform, blown dry under a filtered nitrogen stream, subjected to ultrasonic cleaning in chloroform for 15 minutes, and blown dry again under a filtered nitrogen stream. The samples are then cleaned in n-hexane at 80 °C for 6 minutes, rinsed with room-temperature n-hexane and blown dry with nitrogen.

Cleaving of thioacetate groups: The samples are immersed overnight in a mixture of hydrochloric acid in methanol (1:9 ratio) at 80 °C, after which they are cleaned using the same procedure that was applied following monolayer self-assembly.

Conversion of thiol terminated SAMs into disulfide terminated SAMs [31]: The samples are placed in a solution of iodine (30 mg) in methanol (2 mL) for 15 minutes at room temperature, and subsequently cleaned as above.

Preparation of InP samples: Samples were cleaned in chloroform, acetone and ethanol, blown dry by a flow of filtered nitrogen, and immersed in a solution of 4% aqueous hydrofluoric acid for 10 minutes to remove the native oxide layer. Next, the samples were rinsed three times in deionized water and blown dry with nitrogen. No further treatment of the surface was needed. Following the treatment, the InP was quickly pressed against the silicon sample for bonding in order to avoid oxide re-growth when exposed to air.

Bonding: The bonding took place in a stainless steel mechanical fixture. The wafers to be bonded are pressed against each other inside the fixture and placed in an oven for 24 hours. The temperatures used were 120 - 150 °C. The fixture is closed with screws until the screw head is flush against the metal without applying further pressure. Due to the heat in the oven, the pressure is assumed to be higher than the weight applied.

Pull-Test: A commercial adhesion tester, Defelsko Positest AT-A, was used. The bonded samples were glued to a metal holder on one side and to a transparent window from the other side (see Fig. 6 ). Care was taken to prevent the epoxy from reaching the bonding interface and affecting the measurement. The strength measurements had an uncertainty of ±0.5 MPa.

 

Fig. 6 A schematic illustration of the pull-test setup. The bonded wafers are glued to a metallic holder from one side and to a transparent window from the other side using slow cure epoxy.

Download Full Size | PDF

Acknowledgments

The work was supported in part by the Israeli Science Foundation (ISF) under grants 635-10 and 1646-08; by the 'TERA SANTA' MAGNET consortium of the Office of the Chief Scientist, the Israeli Ministry of Industry, Trade and Labor; and by the Edward and Judith Steinberg Chair in Nanotechnology, Bar-Ilan University.

References and links

1. G. P. Agrawal, Fiber-Optic Communication Systems (John Wiley, 2002).

2. L. Pavesi and D. J. Lockwood, eds., Silicon Photonics (Springer-Verlag, 2004).

3. G. T. Reed, Silicon-Photonics: The State of the Art (John Wiley, 2008)

4. R. A. Soref and B. R. Bennett, “Electrooptical effects in silicon,” IEEE J. Quantum Electron. 23(1), 123–129 (1987). [CrossRef]  

5. Q. Xu, B. Schmidt, S. Pradhan, and M. Lipson, “Micrometre-scale silicon electro-optic modulator,” Nature 435(7040), 325–327 (2005). [CrossRef]   [PubMed]  

6. D. J. Thomson, F. Y. Gardes, J. Fedeli, S. Zlatanovic, Y. Hu, B. P. P. Kuo, E. Myslivets, N. Alic, S. Radic, G. Z. Mashanovich, and G. T. Reed, “High data rate silicon optical modulator with self-aligned fabrication process,” paper OM2E.3 in technical digest of Optical Fiber Communication 2012 (OFC2012) Conference, Los Angeles, CA (Optical Society of America, 2012).

7. H. Rong, R. Jones, A. Liu, O. Cohen, D. Hak, A. W. Fang, and M. J. Paniccia, “A continuous-wave Raman silicon laser,” Nature 433(7027), 725–728 (2005). [CrossRef]   [PubMed]  

8. A. W. Fang, H. Park, O. Cohen, R. Jones, M. J. Paniccia, and J. E. Bowers, “Electrically pumped hybrid AlGaInAs-silicon evanescent laser,” Opt. Express 14(20), 9203–9210 (2006). [CrossRef]   [PubMed]  

9. H. Park, A. W. Fang, O. Cohen, R. Jones, M. J. Paniccia, and J. E. Bowers, “A hybrid AlGaInAs–silicon evanescent amplifier,” IEEE Photon. Technol. Lett. 19(4), 230–232 (2007). [CrossRef]  

10. J. Van Campenhout, P. Rojo Romeo, P. Regreny, C. Seassal, D. Van Thourhout, S. Verstuyft, L. Di Cioccio, J. M. Fedeli, C. Lagahe, and R. Baets, “Electrically pumped InP-based microdisk lasers integrated with a nanophotonic silicon-on-insulator waveguide circuit,” Opt. Express 15(11), 6744–6749 (2007). [CrossRef]   [PubMed]  

11. H. Park, A. W. Fang, R. Jones, O. Cohen, O. Raday, M. N. Sysak, M. J. Paniccia, and J. E. Bowers, “A hybrid AlGaInAs-silicon evanescent waveguide photodetector,” Opt. Express 15(10), 6044–6052 (2007). [CrossRef]   [PubMed]  

12. Y. H. Kuo, H. W. Chen, and J. E. Bowers, “High speed hybrid silicon evanescent electroabsorption modulator,” Opt. Express 16(13), 9936–9941 (2008). [CrossRef]   [PubMed]  

13. H. W. Chen, Y. H. Kuo, and J. E. Bowers, “A Hybrid silicon–AlGaInAs phase modulator,” IEEE Photon. Technol. Lett. 20(23), 1920–1922 (2008). [CrossRef]  

14. H. W. Chen, Y. H. Kuo, and J. E. Bowers, “High speed hybrid silicon evanescent Mach-Zehnder modulator and switch,” Opt. Express 16(25), 20571–20576 (2008). [CrossRef]   [PubMed]  

15. X. K. Sun, A. Zadok, M. J. Shearn, K. A. Diest, A. Ghaffari, H. A. Atwater, A. Scherer, and A. Yariv, “Electrically pumped hybrid evanescent Si/InGaAsP lasers,” Opt. Lett. 34(9), 1345–1347 (2009). [CrossRef]   [PubMed]  

16. Y. Tang, J. Peters, and J. E. Bowers, “1.3μm hybrid silicon electroabsorption modulator with bandwidth beyond 67 GHz,” paper PDP5A.5 in technical digest of Optical Fiber Communication 2012 (OFC2012) Conference, Los Angeles, CA (Optical Society of America, 2012).

17. H. Kroemer, T.-Y. Liu, and P. M. Petroff, “GaAs on Si and related systems: problems and prospects,” J. Cryst. Growth 95(1–4), 96–102 (1989). [CrossRef]  

18. D. Liang, G. Roelkens, R. Baets, and J. E. Bowers, “Hybrid integrated platforms for silicon photonics,” Materials 3(3), 1782–1802 (2010). [CrossRef]  

19. D. Pasquariello, C. Hedlund, and K. Hjort, “Oxidation and induced damages in oxygen plasma in situ wafer bonding,” J. Electrochem. Soc. 147(7), 2699–2703 (2000). [CrossRef]  

20. J. Sagiv, “Organized monolayers by adsorption. 1. Formation and structure of oleophobic mixed monolayers on solid surfaces,” J. Am. Chem. Soc. 102(1), 92–98 (1980). [CrossRef]  

21. S. Onclin, B. J. Ravoo, and D. N. Reinhoudt, “Engineering silicon oxide surfaces using self-assembled monolayers,” Angew. Chem. Int. Ed. Engl. 44(39), 6282–6304 (2005). [CrossRef]   [PubMed]  

22. A. Ulman, “Formation and structure of self-assembled monolayers,” Chem. Rev. 96(4), 1533–1554 (1996). [CrossRef]   [PubMed]  

23. W. Wang, T. Lee, and M. A. Reed, “Mechanism of electron conduction in self-assembled alkanethiol monolayer devices,” Phys. Rev. B 68(3), 035416 (2003). [CrossRef]  

24. W. Y. Zhang, J. P. Labukas, S. Tatic-Lucic, L. Larson, T. Bannuru, R. P. Vinci, and G. S. Ferguson, “Novel room-temperature first-level packaging process for microscale devices,” Sens. Actuators A Phys. 123–124, 646–654 (2005). [CrossRef]  

25. G. Kräuter, Y. Bluhm, C. Batz-Sohn, and U. Gösele, “The joining of parallel plates via organic monolayers: chemical reactions in a spatially confined system,” Adv. Mater. (Deerfield Beach Fla.) 11(12), 1035–1038 (1999). [CrossRef]  

26. V. Artel, I. Bakish, T. Kraus, M. Shubely, Y. Ben-Ezra, E. Shekel, S. Zach, A. Zadok, and C. N. Sukenik, “Low temperature wafer bonding of silicon to InP and silicon to LiNbO₃ using self-assembled monolayers,” paper OM3E.4 in technical digest of Optical Fiber Communication 2012 (OFC2012) Conference, Los Angeles, CA (Optical Society of America, 2012).

27. N. Balachander and C. N. Sukenik, “Monolayer transformation by nucleophilic substitution: applications to the Creation of new monolayer assemblies,” Langmuir 6(11), 1621–1627 (1990). [CrossRef]  

28. S. Meth and C. N. Sukenik, “Siloxane-anchored thin films on silicon dioxide-modified stainless steel,” Thin Solid Films 425(1-2), 49–58 (2003). [CrossRef]  

29. R. J. Collins and C. N. Sukenik, “Sulfonate-functionalized, siloxane-anchored, self-assembled monolayers,” Langmuir 11(6), 2322–2324 (1995). [CrossRef]  

30. I. Aped, Y. Mazuz, and C. N. Sukenik, “Variations in the structure and reactivity of thioester functionalized self-assembled monolayers and their use for controlled surface modification,” Beilstein J. Nanotechnol. 3, 213–220 (2012). [CrossRef]   [PubMed]  

31. V. Artel, I. Aped, R. Cohen, and C. N. Sukenik, “Controlled formation of thiol and disulfide interfaces” (submitted for publication).

32. Y. Gu, Z. Lin, R. A. Butera, V. S. Smentkowski, and D. H. Waldeck, “Preparation of self-assembled monolayers on InP,” Langmuir 11(6), 1849–1851 (1995). [CrossRef]  

33. H. Lim, C. Carraro, R. Maboudian, M. W. Pruessner, and R. Ghodssi, “Chemical and thermal stability of alkanethiol and sulfur passivated InP(100),” Langmuir 20(3), 743–747 (2004). [CrossRef]   [PubMed]  

34. H. F. Gilbert, “Thiol/disulfide exchange equilibria and disulfide bond stability,” Methods Enzymol. 251, 8–28 (1995). [CrossRef]   [PubMed]  

35. U. Gösele and Q.-Y. Tong, “Semiconductor wafer bonding,” Annu. Rev. Mater. Sci. 28(1), 215–241 (1998). [CrossRef]  

36. M. M. R. Howlader, T. Watanabe, and T. Suga, “Characterization of the bonding strength and interface current of p-Si/n-InP wafers bonded by surface activated bonding method at room temperature,” J. Appl. Phys. 91(5), 3062–3066 (2002). [CrossRef]  

37. D. Pasquariello and K. Hjort, “Plasma-assisted InP-to-Si low temperature wafer bonding,” IEEE J. Sel. Top. Quantum Electron. 8(1), 118–131 (2002). [CrossRef]