JOURNAL OF APPLIED PHYSICS

VOLUME 95, NUMBER 11

1 JUNE 2004

Study of thiol-induced adhesion of stressed III–V semiconductor on wax using thin film elastic relaxation F. Bessueille, M. Kostrzewa, J.-L. Leclercq, and G. Greneta) LEOM, UMR CNRS 5512, Ecole Centrale de Lyon, 36 avenue Guy de Collongue, 69134 Ecully Cedex, France

共Received 15 October 2003; accepted 6 February 2004兲 In this paper we underline the role played by octadecylthiols as adhesion promoters on the elastic relaxation of a compressively-prestressed 共0.8%兲 In0.65Ga0.35As thin film stuck on Apiezon-W wax. Surface morphologies by Nomarski Optical Microscopy and Atomic Force Microscopy reveal drastic changes when octadecylthiols are involved in the sticking process. In the ‘‘no-thiol’’ case, the surface morphology displays closely joined regularly distributed undulations 共wavelength ⬇4.8 ␮m, height ⬇0.18 ␮m兲. On the contrary, in the ‘‘thiol’’ case, the wrinkling looks like a large-meshed wire lattice 关wire–wire distance ⬇共10– 40 ␮m兲, height ⬇0.4 ␮m兴. These thiol-induced changes in morphology are explained as due to an energy compromise between relaxing the film stress on the one hand and stressing the self-assembled monolayer organization at the wax/film interface on the other hand. © 2004 American Institute of Physics. 关DOI: 10.1063/1.1690486兴

I. INTRODUCTION

their surface self-organization. This effect is even strengthened by the fact that thiol solubility decreases as their chains lengthen. Our aim of this Letter is to study how thiols promote adhesion between a III–V semiconductor and a wax. The method used here is to compare the surface morphologies after the elastic relaxation of a compressively prestressed thin film stuck onto wax with and without surface functionalization by thiols. In brief, it is now well established that a compressively-prestressed thin film can release its energy by buckling when stuck onto a thick viscous layer like wax.12–16 In that case, the amplitude and periodicity of the surface undulations are known to be dependent on the elastic properties and compression rate of the thin film on the one hand and on the other hand on the viscous layer properties. In this Letter, we will highlight the fact that the way the thin layer is stuck onto the viscous layer is equally important, especially if promoters like SAMs are implicated.

Long-chain 共more than 11 methyl groups兲 alkanethiols, fluoroalkanethiols, or disulfides are well known for forming densely packed, well-organized monolayers on metal surfaces like gold or silver.1– 6 These organic self-assembled monolayers 共SAMs兲 have been used for years to modify surface properties because they can easily be tuned from hydrophobic to hydrophillic just by selectively changing their terminal groups.1 In particular, they are widely used to promote adhesion between semiconductors and technological waxes when dealing with ultrathin film mechanical handling. However, despite the technological importance of III–V semiconductors for optoelectronics, there have been relatively few studies concerning SAMs on these materials for any purposes.7–11 Thiol molecules have a head group that is able to interact strongly with specific sites on a single crystal surface. This reaction is of the chemisorption type and thus so strongly exothermic 共⬃40 kcal/mol兲 that all available sites on the crystal surface are occupied. This reaction can be done by getting rid of molecules already adsorbed on the surface, like impurities. This process is known as the ‘‘self-cleaning process.’’ 3,4 In addition to this head group, a thiol molecule contains a long hydrocarbon chain. When the thiol head groups are well-organized as a reconstructed overlayer on a single crystal surface, these long chains are usually close enough to interrelate via a van der Waals interaction 共⬃10 kcal/mol兲. That interaction makes them get up and organize themselves into assemblies lying along a common direction generally tilted from the surface normal. Finally, the nature of the end group, especially if polar may have some impact on the way the molecules autoassemble on the surface. At this point, it is worth noting that the longer the chains, the stronger the van der Waals interaction and thus the better

II. EXPERIMENTAL PROCEDURES

The initial semiconductor material is the heterostructure presented in Fig. 1共a兲. It has been grown by Solid Source Molecular Beam Epitaxy 共SSMBE兲 on an epiready InP共001兲 sacrificial substrate. It comprises a 500 nm thick pseudomorphic In0.53Ga0.47As layer acting both as an etch stop and buffer layer, a 200 nm thick InP layer, and finally the 30 nm thick In0.65Ga0.35As compressively prestressed 共0.8%兲 film whose elastic relaxation we are currently studying. The next experimental step is the key step we are focusing on, i.e., functionalizing of the surface before sticking it, upside down, onto a silicon host substrate. To successfully grow SAMs on its surface, the heterostructure surface has to be completely freed from any surface oxides. For this reason, the sample is first dipped into HF solution 共HF: 49%兲 for one minute and then rinsed with the thiol solution. What is more, a liquid meniscus is kept on top of the surface during all

a兲

Electronic mail: [email protected]

0021-8979/2004/95(11)/6098/5/$22.00

6098

© 2004 American Institute of Physics

Downloaded 02 Aug 2004 to 132.168.106.67. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

Bessueille et al.

J. Appl. Phys., Vol. 95, No. 11, 1 June 2004

FIG. 1. Key technological steps to obtain thin InGaAs membranes stuck onto a Si host substrate on black wax using thiols as adhesion promoters: 共a兲 Growth of a heterostructure 共including a compressively prestressed thin film兲 by Molecular Beam Epitaxy and surface functionalization; 共b兲 sticking upside down onto a Si host substrate covered with wax; and 共c兲 chemical etching of the initial substrate 共InP兲, buffer layer (In0.53Ga0.47As), and stopping layer 共InP兲 freeing the stressed In0.65Ga0.35As thin film.

transfers in air. Besides and prior to any uses, all thiol solutions have to be outgassed for 15 min by ultrasound. The samples are then immersed for at least three hours into a freshly prepared isopropanol solution containing octadecylthiols 关 CH3 (CH2 ) 17SH兴 at a dilution of 10 mM. This immersion time must be sufficient to allow both proper surface self-cleaning and good alkyl-chain self-organization, even if the initial chemisorption of the head group is a rather quick process 共3 min兲. After that, the heterostructure is washed first in pure isopropanol, then in pure ethanol, and finally dried. At this point, the heterostructure is ready to be stuck upside down onto the silicon host substrate 关Fig. 1共b兲兴 previously covered with a thick Apiezon-W wax layer. The Si host substrate is preheated up to 100 °C in order to obtain homogeneous spreading of the wax. The bonding process is achieved by pressing and cooling down to room temperature. The next step is the selective chemical etching of the sacrificial InP substrate performed with HCl:H2 O 共9 mol/l兲 and of the thick In0.53Ga0.47As etch-stop layer with FeCl3 :H2 O 共14%兲 solution. Finally, the progressive selective etching with HCl:H2 O 共9 mol/l兲 of the InP layer just above the thin strained In0.65Ga0.35As membrane is controlled by color changes 关Fig. 1共c兲兴, allowing compressive stress to be released. To gain some additional insight into the role played by the thiols at the interface between the wax and the In0.65Ga0.35As membrane, this latter 共after its elastic relaxation兲 is eventually

6099

FIG. 2. Nomarski optical microscopy images of a In0.65Ga0.35As thin film patterned into mesas stuck onto wax: 共a兲 without thiol functionalization; and 共b兲 with thiol functionalization. The mesa edges are along the 兵110其 directions.

etched away using FeCl3 :H2 O 共14%兲 solution allowing the bare wax surface morphology to be characterized. This chemical etching is known not to modify the wax surface. III. EXPERIMENTAL RESULTS

Surface morphologies were characterized using 共i兲 Nomarski optical microscopy throughout the technological process, and 共ii兲 Atomic Force Microscopy 共AFM兲 before 共thin film surface morphology兲 and after the thin film chemical etching by FeCl3 :H2 O 共14%兲 solution 共bare wax surface morphology兲. The AFM study was carried out using a Park Scientific Instruments microscope 共model CP兲 operating in contact mode. As a typical example, two large-scale Nomarski views of an In0.65Ga0.35As thin film patterned into mesas are reported in Fig. 2. These two images give an idea of the surface morphology with 关Fig. 2共a兲兴 and without 关Fig. 2共b兲兴 the thiols involved in the sticking process. But what is more important for the present study is that such a patterning allows the crystallographic orientations to be determined in a very simple way from the mesa edge alignment with the 兵110其 crystal cleaving directions. On the other hand, it is clear from both Figs. 2共a兲 and 2共b兲 that the thin film relaxation is strongly responsive to the occurrence of edges. For instance, near the mesa edges, the wrinkles stand perpendicular to the 兵110其 directions in contrast to what happens at the mesa centers, where they are orientated themselves along the 兵100其 directions. Moreover, in Fig. 2共a兲, the smallest mesas and the corners of the large ones appear to be nearly flat, as if completely elastically relaxed via an in-plane gliding pro-

Downloaded 02 Aug 2004 to 132.168.106.67. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

6100

J. Appl. Phys., Vol. 95, No. 11, 1 June 2004

Bessueille et al.

FIG. 3. Relaxation of In0.65Ga0.35As thin film without thiol functionalization. The 共a兲 Nomaski image (130 ␮ m⫻130 ␮ m) for the already elastically relaxed thin film; 共b兲 Nomaski image (130 ␮ m⫻130 ␮ m) for the bare wax surface after the etching of the already elastically relaxed thin film; 共c兲 AFM image (80 ␮ m⫻80 ␮ m) for the already elastically relaxed film surface; and 共d兲 AFM image (80 ␮ m⫻80 ␮ m) for the bare wax surface after the etching of the already elastically relaxed thin film.

cess. If we ignore these peculiar edge effects that are not the focus of this article, Figures 3 and 4 display more precise Nomarski and AFM images of large-mesa central areas without and with thiols promoting the sticking process, respectively. Images are taken either on the relaxed semiconductor surface 关Figs. 3共a兲, 3共c兲, 4共a兲, and 4共c兲兴 or on the bare wax surface 关Figs. 3共b兲, 3共d兲, 4共b兲, and 4共d兲兴 after the complete chemical etching of the former. First of all, it is worth mentioning that most of the morphology’s main features are visible with standard Nomarski microscopy and look very much like those observed with AFM. Second, in both Figs. 3 and 4, the morphologies at the thin film surface and at the wax

FIG. 4. The same as in Fig. 3, but with thiol functionalization.

FIG. 5. Detailed views (20 ␮ m⫻20 ␮ m兲 for 共a兲 the thin film surface without thiol functionalization; 共b兲 the bare wax surface without thiol functionalization; 共c兲 the thin film surface with thiol functionalization; and 共d兲 the bare wax surface with thiol functionalization. Some typical profile lines are drawn and reported below the corresponding images.

surface reflect the same general trends, even if the features are a little more precise and detailed on the wax surface images than on the film surface ones. Indeed, views 共b兲 and 共d兲 reveal more about the film/wax interface than do views 共a兲 and 共c兲 because they are recorded on the bare wax surface 共where the wax directly molds the interface morphology shaped by the film stress relaxation兲 instead of on the film surface 共where the interface morphology is smoothed by the film’s own thickness兲. In particular, when looking at Fig. 4共d兲, one can clearly see secondary features that either cut off the corners made by the main features when crossing each other, or that when too far away from such corners adopt the same orientation as the main features. Let us finally turn to our main point, e.g., the comparison between surface morphologies when using 共Fig. 3兲 or not using 共Fig. 4兲 thiols as adhesion promoters. In Fig. 3, the wrinkling resembles closely joined regularly distributed undulations while in Fig. 4, it forms a large-meshed wire lattice. To be more precise, four additional scaled-down (20⫻20 ␮ m2 ) AFM images dedicated to the shape and dimensions of these wrinkles are displayed in Fig. 5. In the ‘‘no-thiol’’ case, the mean undulation wavelength is the same, viz., 4.8 ␮m, for both the film surface 关Figs. 5共a兲兴 and the bare wax surface 关Fig. 5共b兲兴. The mean wrinkle height 共twice the wave amplitude兲 is around 0.18 ␮m on the thin film surface 关Figs. 5共a兲兴, but only around

Downloaded 02 Aug 2004 to 132.168.106.67. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

J. Appl. Phys., Vol. 95, No. 11, 1 June 2004

Bessueille et al.

6101

0.13 ␮m on the bare wax surface 关Fig. 5共b兲兴. The magnitude of experimental wavelength is what is expected from Ref. 12, i.e., 0.9 ␮m, even if five times greater, indicating the occurrence of slight edge-driven gliding and/or curvature relaxations. In the ‘‘no-thiol case,’’ the surface morphology presents no asymmetry between the waves along the two perpendicular 兵100其 directions. In the ‘‘thiol’’ case, first, the mean wire–wire distance is around nine times greater than in the ‘‘no-thiol’’ case and, second, it seems to be no longer exactly the same along the two 兵100其 crystallographic directions 关even if these are the same for a zinc-blende 共100兲 face兴. The wire–wire distance varies from the 共10– 40 ␮m兲 range along one 兵100其 direction to 共30– 60 ␮m兲 along the other one. However, in this instance, we can reasonably ascribe that slight asymmetry to a slight tilt of the sample with respect to its host substrate leading to slight differences in relaxation conditions. The mean wire height at film surface is around 0.4 ␮m, thus some two times greater than that observed in the ‘‘no-thiol’’ case. It decreases to around 0.06 ␮m on the wax surface. In this case, an accurate value is difficult to obtain because of a groove present on almost all the tops of the wax wrinkles, as if they had been torn off by some wrenching process. IV. DISCUSSION

In our opinion, the differences observed between the ‘‘thiol’’ and ‘‘no-thiol’’ cases can be related to the thiol selforganization at the III–V thin film/wax interface. The main points of the discussion are illustrated in Fig. 6. Note that this figure is clearly out of proportion because both the semiconductor thickness and the thiols are of nanometer scale while the wrinkle wavelengths are of micrometer scale. However, we believe that this tentative diagram can help to build up a picture of what happens at such an interface. As already said in the Introduction, the thiol heads find it energetically favorable to occupy all the available crystallographic sites on a single crystal surface. As a consequence, they organize themselves into a well-ordered lattice strongly correlated with the lattice of the III–V semiconductor 共100兲 face onto which they are chemisorbed. Moreover, the thiol long alkyl chains minimize their mutual van der Waals interaction energy by rising up and adopting a common parallel orientation slightly slanting from the surface normal. The III–V semiconductor surface is thus completely covered by thiols, but this covering is divided into domains that differ from each other by their tilt azimuth angle. Therefore, after adhering to the wax, thiols at the wax/semiconductor interface are 共i兲 strongly bonded by their head groups onto the semiconductor surface on one side, and 共ii兲 closely strapped by their rather long straightened-up chain in the wax on the other side, as schematized in Fig. 6共a兲. They thus act as pillars at the film/wax interface producing a considerable interface widening 共due to their long chain entanglement with wax兲 and interface strengthening 共due to their strong head chemisorption energy on the III–V surface兲. If such a thiolinterfaced film attempts to relax elastically by undulating in the way a film without thiols usually does 关Fig. 6共b兲兴, the thiol self-organization below would be strongly disturbed

FIG. 6. A schematic diagram for 共a兲 the thin film before any elastic relaxation showing how SAMs are entangled in the wax widening the wax/film interface; 共b兲 the thin film relaxation by undulations showing how SAMs react against the local curvature; 共c兲 the thin film relaxation by in-plane expansion forming wrinkles at boundaries between unperturbed SAM domains; 共d兲 the bare wax surface 共after the thin film removal by chemical etching兲 showing the fingerprint of the wrenching of the wax at the wrinkle top.

and thus highly stressed. In other words, such a system faces an energy dilemma: either it undulates and relaxes the film stress 共but concomitantly stresses the thiol self-organization兲 or it remains plane but does not allow any film relaxation. However, undulating is not the only way a compressed film can relax its stress: it can also relax via in-plane expansion, thus avoiding any curvature but producing a large material displacement that it will have to cope with by wrinkling, for example 关Fig. 6共c兲兴. It should be noted that such an in-plane expansion meets the requirements for thiol self-organization by altering only the intervals between them. There again, the system is expected to compromise between the energy gained from the in-plane relaxed domains and the energy lost in highly stressed wrinkles. In our view, the best locations for the latter could be the thiol domain boundaries themselves where, in any case, the self-organization is already broken. The previously stated assumptions are in good agreement with what is observed in the present experiment for the wrinkle orientation and shape. Let us start with the wrinkle orientation. Even if in both Figs. 3 and 4, the wrinkle mesh at the center of large mesas 共where there is no edge effect兲 is elongated toward 兵100其 directions, the phenomenon is much more obvious when thiols are involved in the sticking process 共Fig. 4兲. Normally, in the absence of any edge influence and of any adhesion promoters, III–V single crystal film

Downloaded 02 Aug 2004 to 132.168.106.67. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

6102

Bessueille et al.

J. Appl. Phys., Vol. 95, No. 11, 1 June 2004

favors16 undulations orientated along 兵100其 axes rather than along 兵110其 ones because of a slight positive elastic anisotropy parameter.17 Seeing that the thiols self-organization correlates with the III–V 共100兲 surface and strengthens its pattern, one can expect from the above arguments a reinforcement of the original natural propensity to orientate wrinkles along 兵100其 directions. 共Note that dislocations within III–V film cannot account for the wrinkles because they would lead to effects along 兵110其 directions.兲 Turning now to the wrinkle shape itself, we have found that the extra length involved in the wrinkles is compatible with what is expected from the in-plane expansion of the neighboring domains. Note that, even for as low a mismatch as 0.8%, the film’s extra-length in a wrinkle is so huge that it tears off the wax at the top of the wrinkles, as illustrated in Figs. 6共c兲 and 6共d兲. This accounts for the grooves present on almost all the tops of the wrinkles at the bare wax surface 关Fig. 5共d兲兴. This also implies a decrease in wrinkle height from 0.4 ␮ on the film surface to 0.06 ␮ on the wax surface. In this case, the free-standing wrinkles play the same role for a domain as free edges do for a mesa. It results that secondary features in Fig. 4 develop perpendicularly to the wrinkles as undulations or wrinkles do perpendicularly to edges in Fig. 2. But, the scale of the effects 共undulation amplitude and wavelength兲 is now that of an already partially-relaxed small domain instead of that of a fully stressed large mesa. V. CONCLUSION

To conclude, this approach using a compressively prestressed film to probe an interface could address a wider range of scientific tasks than that presented in the present study, i.e., the role played by thiol at the interface between a III–V semiconductor and Apiezon-W wax. As a matter of fact, the imprint left on a host substrate by a thin stressed film when relaxing is typical of the system as a whole, viz. the film, the host substrate, but also the interface they share. For example, in the case presented in this paper, the interface morphology clearly offsets the way the SAM selforganization competes with the film for an energy compromise in which the film is allowed to relax but only in patches

in order to allocate plane areas that are as large as possible for unstressed SAMs to settle. Finally, engineering an interface between a thin single crystal and a thick glass-like layer on a host substrate structured in such a way that it favors in-plane expansion rather than buckling 共like the one the thiol self-organization provides in this paper兲 could be a significant breakthrough in compliance substrate research. ACKNOWLEDGMENTS

This paper is a byproduct of a wider project dedicated to ‘‘compliant substrates for heteroepitaxy’’ partially supported by the ‘‘Re´gion Rhoˆne-Alpes’’ under Contracts No. 00815050 and No. 00815165. The authors thank G. Hollinger 共ECL-LEOM-CNRS 5512兲 and A. Danescu 共ECLLTDS-CNRS 5513兲 for fruitful discussions. F. Schreiber, Prog. Surf. Sci. 65, 151 共2000兲. Y. Xia and G. M. Whitesides, Angew. Chem., Int. Ed. 37, 550 共1998兲. 3 A. Ulman, Ultrathin Organic Films, 1st ed. 共Academic, San Diego, 1991兲. 4 E. B. Troughton, C. Bain, G. M. Whitesides, R. G. Nuzzo, D. L. Allara, and M. D. Porter, Langmuir 4, 365 共1988兲. 5 G. Liu, J. A. Rodriguez, J. Dvorak, J. Hrbek, and T. Jirsak, Surf. Sci. 505, 295 共2002兲. 6 A. Ulman, Chem. Rev. 共Washington, D.C.兲 96, 1533 共1996兲. 7 K. Remashan and K. N. Bhat, Thin Solid Films 342, 20 共1999兲. 8 T. Hou, C. M. Greenlief, S. W. Keller, L. Nelen, and J. F. Kauffman, Chem. Mater. 9, 3181 共1997兲. 9 N. Camillone, K. A. Khan, and R. M. Osgood, Jr., Surf. Sci. 453, 83 共2000兲. 10 K. Adlkofer, M. Tanaka, H. Hillebrandt, G. Wiegand, E. Sackmann, T. Bolom, R. Deutschmann, and G. Abstreiter, Appl. Phys. Lett. 76, 3313 共2000兲. 11 E. F. Duijs, F. Findeis, R. Deutschmann, M. Bichler, A. Zrenner, G. Abstreiter, K. Adlkofer, M. Tanaka, and E. Sackmann, Phys. Status Solidi B 224, 871 共2001兲. 12 N. Shridhar, D. J. Srolovitz, and Z. Suo, Appl. Phys. Lett. 78, 2482 共2001兲. 13 R. Huang and Z. Suo, Int. J. Solids Struct. 39, 1791 共2002兲. 14 R. Huang and Z. Suo, J. Appl. Phys. 91, 1135 共2002兲. 15 H. Yin, R. Huang, K. D. Hobart, Z. Suo, T. S. Kuan, C. K. Inoki, S. R. Shieh, T. S. Duffy, F. J. Kub, and J. C. Sturn, J. Appl. Phys. 91, 9716 共2002兲. 16 N. Mokni, A. Danescu, and F. Sidoroff, unpublished. 17 The elastic anisotropy parameter is defined as ␰ ⫽(C 12⫹C 44⫺C 11)/C 11 , where C 11 , C 12 , and C 44 are the thin film elastic parameters. In our case, C 11⫽0.9582⫻1011 N/m2 , C 12⫽0.4825⫻1011 N/m2 , C 44 ⫽0.4825⫻1011 N/m2 , and thus ␰ ⫽0.03. 1 2

Downloaded 02 Aug 2004 to 132.168.106.67. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp