- Email: [email protected]
Microcontact printing of indium metal using salt solution “ink”
Thin Solid Films 515 (2007) 6812 – 6816 www.elsevier.com/locate/tsf
Microcontact printing of indium metal using salt solution “ink” C.G. Allen a,⁎, J.C. Dorr a , A.A. Khandekar b , J.D. Beach a , I.C. Schick a , E.J. Schick a , R.T. Collins a , T.F. Kuech b b
a Physics Department, Colorado School of Mines, Golden, CO 80401, United States Department of Chemical Engineering, University of Wisconsin-Madison, Madison, WI 53706, United States
Received 31 August 2006; received in revised form 2 January 2007; accepted 13 February 2007 Available online 23 February 2007
Abstract Poly(dimethlysiloxane) stamps were made from Si masters fabricated using photolithography and anisotropic etching. GaCl3 and In(NO3)3 were microcontact printed onto Si substrates, creating arrays of micron size metal salt deposits. The In(NO3)3 deposits were further processed by annealing in an N2 :H2 (9:1) forming gas environment at 600 °C which converted the deposits into In metal. The ability to inexpensively pattern metal arrays on semiconductor surfaces has implications for ohmic contacts and, with additional processing, arrays of semiconductor crystallites for optoelectronic applications. © 2007 Elsevier B.V. All rights reserved. Keywords: Microcontact printing; Indium metal arrays; Anisotropic etching
1. Introduction Microcontact printing (μCP) has been widely researched [1–4] as an alternative, cost effective method for doing some of the processes that might otherwise be performed by photolithography or electron beam lithography. μCP transfers soluble materials on a “stamp” to a substrate. Poly(dimethlysiloxane) (PDMS) is a common stamp material. PDMS stamps can be cast using patterned silicon (Si) substrates [1,5,6], often called stamp masters, as templates. Stamp masters may be used multiple times, to make multiple PDMS stamps. A single stamp may be used for multiple microcontact printings. The μCP process eliminates the use of costly equipment and clean room environment after the initial master production. Features produced by this method have good resolution [7] and are not necessarily diffraction limited [5,8,9]. For these reasons, μCP is researched for applications where higher through-put and less expensive processing can significantly lower costs (e.g. semiconductor microelectronic processing and photovoltaics manufacturing).
⁎ Corresponding author. E-mail address: [email protected] (C.G. Allen). 0040-6090/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2007.02.070
Patterns of metals (Ag, Au, Cu) have previously been generated on various substrates using μCP [10]. In those approaches, the metal surface was patterned with selfassembled monolayers (SAMs) that acted as resists in selective wet etches. SAMs have also been used as templates that allow selective deposition on a substrate surface [6,8–11]. Nitrates have previously been deposited by a template method. Qin et al. [8] patterned a gold surface into grids of hydrophobic and hydrophilic regions. The sample was then submerged into an aqueous solution of potassium nitrate (KNO3). After drying, KNO3 deposits were present on hydrophilic regions. Zhong et al. [9] created deposits of metal (cobalt, nickel, and iron) nitrates on hydrophilic Si (Si with native oxide) substrates by patterning the surface with hydrophobic regions. These deposits were then converted to oxide and reduced to metal. These processes are dependent on the ability to develop inks that will form monolayers on the substrate surface which change the surface hydrophobicity. This can limit the range of suitable substrates. An alternative approach involving direct transfer of a solution from a PDMS stamp to a substrate has been explored. Chang et al. [12] described direct contact printing using metal-organic inks. Copper oxide was pyrolyzed from patterned precursor and then reduced to metal. Here, we present results of patterning directly from a PDMS stamp onto Si substrates with native
C.G. Allen et al. / Thin Solid Films 515 (2007) 6812–6816
Fig. 1. Schematic of procedure for producing the stamp master. (a) Initial substrate, (b) patterned with PR, (c) BOE etched and PR removed, (d) KOH etched, (e) oxide removed, (f ) PDMS formed and cured, (g) PDMS removed.
oxides, using aqueous metallic salt solutions as the ink. Si provides a convenient, clean, flat, test substrate and is a system where patterning of metal contacts could be important. It is reasonable to suspect this method could be extended to a large variety of both hydrophobic and hydrophilic substrates. 2. Experimental details Stamp masters were created from Si substrates with thick (∼ 600 nm), thermally grown oxides. A schematic of the entire process is shown in Fig. 1. The oxide surface was patterned using conventional photolithography procedures. Shipley 505A positive photoresist (PR), thinned 1:1 with ethylacetate, was spin coated on the smooth oxide layer to a thickness of 110 nm. G-line exposure occurred in a contact mask aligner using a
6813
photomask with a square array of holes on 60 μm spacings. Holes 13 μm in diameter were developed using Shipley MFCD-26. A “hard” bake at 110 °C was necessary for the PR to act as an etch barrier in a buffered oxide etch (BOE). A standard BOE [13] removed oxide in the exposed regions of the PR. The PR was then removed with acetone. The patterned samples were cleaned in 60 °C, H2O:NH4 OH:H2O2 (5:1:1) and H2O:HCl:H2 O2 (6:1:1), to remove any residual organic and ionic contaminants. The oxide layer acted as an etch mask in a solution of potassium hydroxide (KOH) at 80 °C. KOH is a selective, anisotropic etchant of Si, preferentially etching the (100) planes [14,15] while negligibly etching the (111) Si planes and the oxide. In regions where the silicon was exposed through the oxide mask, (100) planes were etched to completion creating inverted etch pit structures with ∼ 60° slopes. The etch undercuts the circular holes in the oxide creating inverted tetragons with nearly square bases and sides that meet at a point. The base size depends on the etched diameter of the oxide holes. The barrier oxide was then removed. The resulting structures have an aspect ratio, depth to width, of 0.87. This is within the ideal aspect ratio for PDMS stamping found by Delamarche et al. with features a factor of 10 smaller [16]. Fig. 2a is an optical micrograph of the patterned Si after the oxide was removed. PDMS stamps were formed using Sylgard 184 (Dow Corning, Midland, MI) elastomeric base and curing agent mixed 10:1, respectively, by weight. The mixture was first placed into an evacuated furnace (∼200 mbar) for about 30 min to release gases dissolved in the mixture. The degassed PDMS was then poured into a form with the patterned Si as the base. The stamp was again degassed and cured overnight (10-15 h) in vacuum at 70 °C. After cooling to room temperature, the stamp was carefully removed from the patterned silicon with tweezers (Fig. 2b). The features formed in the stamps were similar to
Fig. 2. (a) Optical micrograph (light-field) of an etched Si substrates used as a stamp-master. (b) Optical micrograph (dark-field) of a PDMS stamp (The image is focussed at the tips of the pyramids. The blur of the features and apparent triangular, instead of tetrahedral, shape is an artifact of dark field imaging in the unfocussed regions. Scanning the focal plane through the pyramids shows the correct tetrahedral structure.)
6814
C.G. Allen et al. / Thin Solid Films 515 (2007) 6812–6816
Fig. 3. Gallium chloride deposits prepared by μCP from a 0.10 M solution.
features prepared by Marzolin et al. in silica structures using a sol–gel method [17]. The PDMS stamps were affixed to a circular roller with a radius of 12.7 cm for smooth, uniform, and stable contact and separation between the stamp and substrate surface during the μCP. Solutions of gallium chloride (GaCl3 ) and indium nitrate (In (NO3 )3) were prepared as “inks” for μCP. A 0.10-M solution of GaCl3 was prepared by mixing solid anhydrous GaCl3 (Aldrich, St. Louis, MO) with water. The solution was swabbed on the stamp using a cotton tip. Enough solution was applied so that the surface was visually wet. Samples cut from (100) oriented silicon wafers were used as the substrate to receive the μCP. The initial contact of the wet stamp with a silicon substrate was seldom successful in producing reliable patterns. However, it was effective in removing excess solution so that an immediate second μCP on a second Si substrate resulted in well-defined arrays of deposits. The best results were achieved when contact
was maintained for a few seconds on the first substrate (∼ 3 s) and then the stamp was quickly transferred to a second substrate and contact was made for ∼ 1 s. The resulting patterns appeared dry even after immediate (b 30 s later) inspection with the optical microscope (Fig. 3). An In(NO3 )3 solution (∼ 0.1 M) was prepared by mixing 0.15 g of In(NO3)3·xH2O salt (K&K Laboratories, Beverly Hills, CA) with 27.5 mL of 9.5% (weight) nitric acid. The same μCP conditions were used as above and the results are shown in Fig. 4a. The sample with deposits made from the In(NO3)3 solution was annealed in a forming gas, N2:H2 (9:1), environment for 1 h. Gas flowed through a desiccant (Agilient, Palo Alto, CA) to remove moisture contamination. The gas then flowed to the inlet of the 5 cm quartz tube in a three zone tube furnace at a constant flow rate of a 2.5 L/min. The tube was equipped with a quartz sample holder and sliding quartz rod which allowed the sample to remain in the room temperature portion of the tube while the furnace ramped in temperature and purged. The sample was inserted into the heated portion of the tube and annealed at 600 °C for 1 h. The sample was then retracted into the room temperature portion of the tube. The gas flowed for 10 additional minutes to allow the sample to cool enough to be removed. An optical micrograph after the anneal is shown in Fig. 4b. X-ray diffraction (XRD) was used to verify that the above process resulted in metallic indium. Because of the small volume of material, no X-ray pattern could be distinguished with the actual μCP deposits. Samples which were processed in the same manner but began with a larger volume of the In(NO3)3 salt deposits were characterized. A dropper was used to generously coat a Si sample with the In(NO3)3 solution. The sample was dried in ambient air and annealed under the same conditions as the patterned sample. A 2 θ XRD pattern was recorded using the CuKα line of a Kritalloflex 81 XRD machine (Siemens, New York, NY), calibrated using the pattern from 99.99% pure indium (In) foil (Aesar, Ward Hill, MA). Fig. 5 shows a comparison of the
Fig. 4. (a) Indium nitrate deposits prepared by μCP onto a Si substrate. (b) Similar deposits after being annealed in forming gas.
C.G. Allen et al. / Thin Solid Films 515 (2007) 6812–6816
Fig. 5. The XRD pattern of annealed In(NO3)3 applied with a dropper to a (100) Si substrate (top) that, when compared to the standard, JCPDS 05-0642 (bottom), confirms the reduction of the nitrate into indium metal.
pattern from the annealed sample with that of peak locations and intensities given by an XRD standard [18]. The peak locations (32.96, 36.33, 39.17, 54.48) are in excellent agreement. There are some small differences in the relative intensities of the sample peaks compared to the powder pattern which could indicate a deviation from completely random order. Atomic Force Microscope (AFM) images of the deposits before and after the anneal were generated using a Multimode AFM-2 (Veeco, Santa Barbara, CA) and high resonance frequency tapping mode tips (NanoWorld AG, Switzerland) with radii of curvature less than 10 nm. The deposits formed from the In(NO3)3 solution (Fig. 6a) showed roughly circular deposits (similar results were obtained for GaCl3). The AFM image of a similar individual deposit after annealing (Fig. 6b) showed significant change in the deposit.
6815
metallic deposit. The deposit was instead comprised of many smaller In deposits. The sizes vary, but in general, few are larger than 200 nm in length. In an effort to find conditions which would lead to a single In deposit, we explored the effects of temperature. Annealing In(NO3)3 at temperatures lower than 600 °C did not completely reduce the In(NO3)3 to metal. This was evidenced by indium oxide peaks in the XRD pattern. Although the forming gas was purified in desiccant to remove water, we believe there was still sufficient oxidizing impurities (H2O or O2) to inhibit reduction to metal at lower temperature. At higher temperatures, we continued to see multiple deposits of the type shown in Fig. 6b. We suspect that reduction to metal is also being inhibited by incorporation of Si into the metal. Silicon oxides are very stable, thus difficult to reduce and easily form when trace amounts of water or oxygen are present. If an In/Si alloy “skin” formed on the deposits during annealing, it could very easily oxidize. This may also explain why many of the deposits in Fig. 6b appear to be in contact and still did not coalesce. Oxide skins present on each surface would prevent coalescence. Trace amounts of Si may be sufficient. One possible solution to this problem would be moving
3. Results and discussion The deposit arrays shown in Figs. 3 and 4 have deviation from exact 60 μm spacings. This may be attributed to deformation of the stamp during the stamping process and/or liquid mobility on the surface. An estimate of the deviation was made by calculating the transverse variation of the centers of the deposits in a row. The largest deviation found in Fig. 3 was 2.5 μm. AFM reveals that some residue is found around the main deposits (Fig. 6a). This is attributed to excessive pressure during the stamping, causing the stamp to deform and contact a larger region than desired. Decreasing the contact pressure led to less well defined patterns making a trade-off between residue and uniformity. A stamp with a greater aspect ratio may help to eliminate contact in regions between the points. With the present anisotropic etching process for producing the master, the aspect ratio is fixed. This approach, however, was selected to allow a simple test of stamping. Other etching processes (e.g. deep reactive ion etching) could be used to produce masters with greater aspect ratios and features which do not end in points. This may help minimize the deviations and residue between the deposits. After annealing and converting to metal, AFM images (Fig. 6b) showed that each individual deposit was not a single coalesced
Fig. 6. (a) AFM of an indium nitrate deposit. (b) AFM of a similar deposit after being reduced to indium.
6816
C.G. Allen et al. / Thin Solid Films 515 (2007) 6812–6816
to a higher anneal temperature where reduction to metal is more efficient. However, at higher temperatures, the solubility of Si in In also increases. The best approach is probably moving to lower temperatures combined with better gas purification to minimize Si alloying, while still allowing oxide reduction. Since the XRD indicates reduction to In metal, it would be interesting to measure the conductivity of the In. However, for both the stamped and dropper prepared samples, individual deposits of In were too small to reliably contact. There have been reports of strongly suppressed conductivity in copper features prepared by μCP [19]. The authors of that study suggest the cause may be silicon oligomers that are incorporated from the surface of the PDMS. While their process was quite different from the approach here, it is possible that PDMS contamination may influence the conductivity of the In deposits. This effect does not limit the usefulness of stamped metal features, because it has been shown to be minimized or eliminated by pretreatments of the stamps, such as, vitrification in a UV/ozone environment [20] or soaking in solvents [21]. One application for metallic deposits on silicon substrates was demonstrated by Beach et al. [22]. The authors converted arrays of metallic gallium into GaAs crystallites through anneals in arsine gas. These crystallites were then used as seeds and enlarged using selective organometallic chemical vapor deposition GaAs growth. Similarly, patterned metallic In deposits could be converted to, for example, InAs crystallites. In addition to functioning as seed crystals, arrays of optically active semiconductor crystallites such as GaAs and InAs on Si have potential applications for optoelectronic devices and photonic crystals, particularly if the sizes could be reduced to nanoscale regimes. Another potential application may be contacts to Si devices. By changing the initial photolithography of the masters, it seems conceivable that one may pattern lines as opposed to dots. 4. Conclusions Anisotropic KOH etching was used to make stamp masters. PDMS stamps were cast from the masters. μCP with metallic salt solutions (GaCl3 and In(NO3)3) was demonstrated to be an effective method for patterning Si substrates with arrays of micron size salt deposits. This method was a more direct approach than previous work that used μCP of SAMs for wet etching or for creating a templated surface of hydrophillic and hydrophobic regions. It is believed that a stamp with an aspect ratio greater than 0.9 is necessary to reduce unwanted deposition when attempting to pattern small features that are to be spaced by tens of microns. The deposits made using In(NO3)3 solution were annealed in forming gas to reduce the deposits to In. Si alloying and
subsequent oxidation is identified as a likely reason for the formation of many small deposits instead of a single, large coalesced deposit. A higher purity hydrogen environment may allow the reduction to occur at lower temperatures and decrease the alloying effect. The entire μCP process demonstrates a low cost method of patterning that is potentially useful in microelectronics processing. Acknowledgements The authors would like to thank Don Williamson for his help collecting XRD data and John Yarbrough, Jami Dashdorj, and Sudipta Bera for their discussions. This material is based on work supported by the National Science Foundation under Grant No. DMR-0103945 and Air Force Research Laboratory Award Nos. F29601-02-C-0222 and F29601-02-C-0223. References [1] A. Kumar, G. Whitesides, Appl. Phys. Lett. 63 (1993) 2002. [2] Y. Xia, M. Mirksich, E. Kim, G. Whitesides, J. Am. Chem. Soc. 117 (1995) 9576. [3] E. Delamarche, H. Schmid, A. Bietsch, N. Larsen, H. Rothuizen, B. Michel, H. Biebuyck, J. Phys. Chem., B 102 (1998) 3324. [4] B. Ridley, B. Nivi, B. Hubert, C. Bulthaup, E. Wilhelm, J. Jacobson, Appl. Phys. Lett. 63 (1993) 2002. [5] D. Wang, S. Thomas, K. Wang, Y. Xia, G. Whitesides, Appl. Phys. Lett. 70 (1997) 1593. [6] F.C.M.J.M van Delft, F.C. van den Heuvel, A.E. Kuiper, P.C. Thüne, J.W. Niemantsverdriet, Microelectron. Eng. 73 (2004) 202. [7] P. St. John, H. Craighead, Appl. Phys. Lett. 68 (1996) 1022. [8] D. Qin, Y. Xia, B. Xu, H. Yang, C. Zhu, G. Whitesides, Adv. Mater. 11 (1999) 1433. [9] Z. Zhong, B. Gates, Y. Xia, Langmuir 16 (2000) 10369. [10] Y. Xia, G. Whitesides, Annu. Rev. Mater. Res. 28 (1998) 153. [11] N. Jeon, K. Finni, K. Branshaw, R. Nuzzo, Langmuir 13 (1997) 3382. [12] N. Chang, J. Richardson, P. Clem, J. Hsu, Small 2 (2006) 75. [13] S. Campbell, The Science and Engineering of Microelectronic Fabrication, Oxford University Press, New York, 2001, p. 260. [14] I. Barycka, I. Zubel, Sens. Actuators, A, Phys. 48 (1995) 228. [15] M. Yun, V. Burrows, M. Kozicki, J. Vac. Sci. Technol., B 16 (1998) 2844. [16] E. Delamarche, H. Schmid, B. Michel, H. Biebuyck, Adv. Mater. 9 (1997) 741. [17] C. Marzolin, S. Smith, M. Prentiss, G. Whitesides, Adv. Mater. 10 (1998) 571. [18] Powder Diffraction File, JCPDS-International Centre for Diffraction Data, Newton Square, PA, 2002, Card 05-0642. [19] K. Felmet, Y.L. Loo, Y. Sun, Appl. Phys. Lett. 85 (2004) 3316. [20] K. Glasmästar, J. Gold, A.S. Andersson, D. Sutherland, B. Kasemo, Langmuir 19 (2003) 5475. [21] D. Graham, D. Price, B. Ratner, Langmuir 18 (2002) 1518. [22] J. Beach, C. Veauvy, R. Caputo, R. Collins, A. Khandekar, T. Kuech, C. Inoki, T. Kuan, R. Hollingsworth, Appl. Phys. Lett. 84 (2004) 5323.
Microcontact printing of indium metal using salt solution “ink” C.G. Allen a,⁎, J.C. Dorr a , A.A. Khandekar b , J.D. Beach a , I.C. Schick a , E.J. Schick a , R.T. Collins a , T.F. Kuech b b
a Physics Department, Colorado School of Mines, Golden, CO 80401, United States Department of Chemical Engineering, University of Wisconsin-Madison, Madison, WI 53706, United States
Received 31 August 2006; received in revised form 2 January 2007; accepted 13 February 2007 Available online 23 February 2007
Abstract Poly(dimethlysiloxane) stamps were made from Si masters fabricated using photolithography and anisotropic etching. GaCl3 and In(NO3)3 were microcontact printed onto Si substrates, creating arrays of micron size metal salt deposits. The In(NO3)3 deposits were further processed by annealing in an N2 :H2 (9:1) forming gas environment at 600 °C which converted the deposits into In metal. The ability to inexpensively pattern metal arrays on semiconductor surfaces has implications for ohmic contacts and, with additional processing, arrays of semiconductor crystallites for optoelectronic applications. © 2007 Elsevier B.V. All rights reserved. Keywords: Microcontact printing; Indium metal arrays; Anisotropic etching
1. Introduction Microcontact printing (μCP) has been widely researched [1–4] as an alternative, cost effective method for doing some of the processes that might otherwise be performed by photolithography or electron beam lithography. μCP transfers soluble materials on a “stamp” to a substrate. Poly(dimethlysiloxane) (PDMS) is a common stamp material. PDMS stamps can be cast using patterned silicon (Si) substrates [1,5,6], often called stamp masters, as templates. Stamp masters may be used multiple times, to make multiple PDMS stamps. A single stamp may be used for multiple microcontact printings. The μCP process eliminates the use of costly equipment and clean room environment after the initial master production. Features produced by this method have good resolution [7] and are not necessarily diffraction limited [5,8,9]. For these reasons, μCP is researched for applications where higher through-put and less expensive processing can significantly lower costs (e.g. semiconductor microelectronic processing and photovoltaics manufacturing).
⁎ Corresponding author. E-mail address: [email protected] (C.G. Allen). 0040-6090/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2007.02.070
Patterns of metals (Ag, Au, Cu) have previously been generated on various substrates using μCP [10]. In those approaches, the metal surface was patterned with selfassembled monolayers (SAMs) that acted as resists in selective wet etches. SAMs have also been used as templates that allow selective deposition on a substrate surface [6,8–11]. Nitrates have previously been deposited by a template method. Qin et al. [8] patterned a gold surface into grids of hydrophobic and hydrophilic regions. The sample was then submerged into an aqueous solution of potassium nitrate (KNO3). After drying, KNO3 deposits were present on hydrophilic regions. Zhong et al. [9] created deposits of metal (cobalt, nickel, and iron) nitrates on hydrophilic Si (Si with native oxide) substrates by patterning the surface with hydrophobic regions. These deposits were then converted to oxide and reduced to metal. These processes are dependent on the ability to develop inks that will form monolayers on the substrate surface which change the surface hydrophobicity. This can limit the range of suitable substrates. An alternative approach involving direct transfer of a solution from a PDMS stamp to a substrate has been explored. Chang et al. [12] described direct contact printing using metal-organic inks. Copper oxide was pyrolyzed from patterned precursor and then reduced to metal. Here, we present results of patterning directly from a PDMS stamp onto Si substrates with native
C.G. Allen et al. / Thin Solid Films 515 (2007) 6812–6816
Fig. 1. Schematic of procedure for producing the stamp master. (a) Initial substrate, (b) patterned with PR, (c) BOE etched and PR removed, (d) KOH etched, (e) oxide removed, (f ) PDMS formed and cured, (g) PDMS removed.
oxides, using aqueous metallic salt solutions as the ink. Si provides a convenient, clean, flat, test substrate and is a system where patterning of metal contacts could be important. It is reasonable to suspect this method could be extended to a large variety of both hydrophobic and hydrophilic substrates. 2. Experimental details Stamp masters were created from Si substrates with thick (∼ 600 nm), thermally grown oxides. A schematic of the entire process is shown in Fig. 1. The oxide surface was patterned using conventional photolithography procedures. Shipley 505A positive photoresist (PR), thinned 1:1 with ethylacetate, was spin coated on the smooth oxide layer to a thickness of 110 nm. G-line exposure occurred in a contact mask aligner using a
6813
photomask with a square array of holes on 60 μm spacings. Holes 13 μm in diameter were developed using Shipley MFCD-26. A “hard” bake at 110 °C was necessary for the PR to act as an etch barrier in a buffered oxide etch (BOE). A standard BOE [13] removed oxide in the exposed regions of the PR. The PR was then removed with acetone. The patterned samples were cleaned in 60 °C, H2O:NH4 OH:H2O2 (5:1:1) and H2O:HCl:H2 O2 (6:1:1), to remove any residual organic and ionic contaminants. The oxide layer acted as an etch mask in a solution of potassium hydroxide (KOH) at 80 °C. KOH is a selective, anisotropic etchant of Si, preferentially etching the (100) planes [14,15] while negligibly etching the (111) Si planes and the oxide. In regions where the silicon was exposed through the oxide mask, (100) planes were etched to completion creating inverted etch pit structures with ∼ 60° slopes. The etch undercuts the circular holes in the oxide creating inverted tetragons with nearly square bases and sides that meet at a point. The base size depends on the etched diameter of the oxide holes. The barrier oxide was then removed. The resulting structures have an aspect ratio, depth to width, of 0.87. This is within the ideal aspect ratio for PDMS stamping found by Delamarche et al. with features a factor of 10 smaller [16]. Fig. 2a is an optical micrograph of the patterned Si after the oxide was removed. PDMS stamps were formed using Sylgard 184 (Dow Corning, Midland, MI) elastomeric base and curing agent mixed 10:1, respectively, by weight. The mixture was first placed into an evacuated furnace (∼200 mbar) for about 30 min to release gases dissolved in the mixture. The degassed PDMS was then poured into a form with the patterned Si as the base. The stamp was again degassed and cured overnight (10-15 h) in vacuum at 70 °C. After cooling to room temperature, the stamp was carefully removed from the patterned silicon with tweezers (Fig. 2b). The features formed in the stamps were similar to
Fig. 2. (a) Optical micrograph (light-field) of an etched Si substrates used as a stamp-master. (b) Optical micrograph (dark-field) of a PDMS stamp (The image is focussed at the tips of the pyramids. The blur of the features and apparent triangular, instead of tetrahedral, shape is an artifact of dark field imaging in the unfocussed regions. Scanning the focal plane through the pyramids shows the correct tetrahedral structure.)
6814
C.G. Allen et al. / Thin Solid Films 515 (2007) 6812–6816
Fig. 3. Gallium chloride deposits prepared by μCP from a 0.10 M solution.
features prepared by Marzolin et al. in silica structures using a sol–gel method [17]. The PDMS stamps were affixed to a circular roller with a radius of 12.7 cm for smooth, uniform, and stable contact and separation between the stamp and substrate surface during the μCP. Solutions of gallium chloride (GaCl3 ) and indium nitrate (In (NO3 )3) were prepared as “inks” for μCP. A 0.10-M solution of GaCl3 was prepared by mixing solid anhydrous GaCl3 (Aldrich, St. Louis, MO) with water. The solution was swabbed on the stamp using a cotton tip. Enough solution was applied so that the surface was visually wet. Samples cut from (100) oriented silicon wafers were used as the substrate to receive the μCP. The initial contact of the wet stamp with a silicon substrate was seldom successful in producing reliable patterns. However, it was effective in removing excess solution so that an immediate second μCP on a second Si substrate resulted in well-defined arrays of deposits. The best results were achieved when contact
was maintained for a few seconds on the first substrate (∼ 3 s) and then the stamp was quickly transferred to a second substrate and contact was made for ∼ 1 s. The resulting patterns appeared dry even after immediate (b 30 s later) inspection with the optical microscope (Fig. 3). An In(NO3 )3 solution (∼ 0.1 M) was prepared by mixing 0.15 g of In(NO3)3·xH2O salt (K&K Laboratories, Beverly Hills, CA) with 27.5 mL of 9.5% (weight) nitric acid. The same μCP conditions were used as above and the results are shown in Fig. 4a. The sample with deposits made from the In(NO3)3 solution was annealed in a forming gas, N2:H2 (9:1), environment for 1 h. Gas flowed through a desiccant (Agilient, Palo Alto, CA) to remove moisture contamination. The gas then flowed to the inlet of the 5 cm quartz tube in a three zone tube furnace at a constant flow rate of a 2.5 L/min. The tube was equipped with a quartz sample holder and sliding quartz rod which allowed the sample to remain in the room temperature portion of the tube while the furnace ramped in temperature and purged. The sample was inserted into the heated portion of the tube and annealed at 600 °C for 1 h. The sample was then retracted into the room temperature portion of the tube. The gas flowed for 10 additional minutes to allow the sample to cool enough to be removed. An optical micrograph after the anneal is shown in Fig. 4b. X-ray diffraction (XRD) was used to verify that the above process resulted in metallic indium. Because of the small volume of material, no X-ray pattern could be distinguished with the actual μCP deposits. Samples which were processed in the same manner but began with a larger volume of the In(NO3)3 salt deposits were characterized. A dropper was used to generously coat a Si sample with the In(NO3)3 solution. The sample was dried in ambient air and annealed under the same conditions as the patterned sample. A 2 θ XRD pattern was recorded using the CuKα line of a Kritalloflex 81 XRD machine (Siemens, New York, NY), calibrated using the pattern from 99.99% pure indium (In) foil (Aesar, Ward Hill, MA). Fig. 5 shows a comparison of the
Fig. 4. (a) Indium nitrate deposits prepared by μCP onto a Si substrate. (b) Similar deposits after being annealed in forming gas.
C.G. Allen et al. / Thin Solid Films 515 (2007) 6812–6816
Fig. 5. The XRD pattern of annealed In(NO3)3 applied with a dropper to a (100) Si substrate (top) that, when compared to the standard, JCPDS 05-0642 (bottom), confirms the reduction of the nitrate into indium metal.
pattern from the annealed sample with that of peak locations and intensities given by an XRD standard [18]. The peak locations (32.96, 36.33, 39.17, 54.48) are in excellent agreement. There are some small differences in the relative intensities of the sample peaks compared to the powder pattern which could indicate a deviation from completely random order. Atomic Force Microscope (AFM) images of the deposits before and after the anneal were generated using a Multimode AFM-2 (Veeco, Santa Barbara, CA) and high resonance frequency tapping mode tips (NanoWorld AG, Switzerland) with radii of curvature less than 10 nm. The deposits formed from the In(NO3)3 solution (Fig. 6a) showed roughly circular deposits (similar results were obtained for GaCl3). The AFM image of a similar individual deposit after annealing (Fig. 6b) showed significant change in the deposit.
6815
metallic deposit. The deposit was instead comprised of many smaller In deposits. The sizes vary, but in general, few are larger than 200 nm in length. In an effort to find conditions which would lead to a single In deposit, we explored the effects of temperature. Annealing In(NO3)3 at temperatures lower than 600 °C did not completely reduce the In(NO3)3 to metal. This was evidenced by indium oxide peaks in the XRD pattern. Although the forming gas was purified in desiccant to remove water, we believe there was still sufficient oxidizing impurities (H2O or O2) to inhibit reduction to metal at lower temperature. At higher temperatures, we continued to see multiple deposits of the type shown in Fig. 6b. We suspect that reduction to metal is also being inhibited by incorporation of Si into the metal. Silicon oxides are very stable, thus difficult to reduce and easily form when trace amounts of water or oxygen are present. If an In/Si alloy “skin” formed on the deposits during annealing, it could very easily oxidize. This may also explain why many of the deposits in Fig. 6b appear to be in contact and still did not coalesce. Oxide skins present on each surface would prevent coalescence. Trace amounts of Si may be sufficient. One possible solution to this problem would be moving
3. Results and discussion The deposit arrays shown in Figs. 3 and 4 have deviation from exact 60 μm spacings. This may be attributed to deformation of the stamp during the stamping process and/or liquid mobility on the surface. An estimate of the deviation was made by calculating the transverse variation of the centers of the deposits in a row. The largest deviation found in Fig. 3 was 2.5 μm. AFM reveals that some residue is found around the main deposits (Fig. 6a). This is attributed to excessive pressure during the stamping, causing the stamp to deform and contact a larger region than desired. Decreasing the contact pressure led to less well defined patterns making a trade-off between residue and uniformity. A stamp with a greater aspect ratio may help to eliminate contact in regions between the points. With the present anisotropic etching process for producing the master, the aspect ratio is fixed. This approach, however, was selected to allow a simple test of stamping. Other etching processes (e.g. deep reactive ion etching) could be used to produce masters with greater aspect ratios and features which do not end in points. This may help minimize the deviations and residue between the deposits. After annealing and converting to metal, AFM images (Fig. 6b) showed that each individual deposit was not a single coalesced
Fig. 6. (a) AFM of an indium nitrate deposit. (b) AFM of a similar deposit after being reduced to indium.
6816
C.G. Allen et al. / Thin Solid Films 515 (2007) 6812–6816
to a higher anneal temperature where reduction to metal is more efficient. However, at higher temperatures, the solubility of Si in In also increases. The best approach is probably moving to lower temperatures combined with better gas purification to minimize Si alloying, while still allowing oxide reduction. Since the XRD indicates reduction to In metal, it would be interesting to measure the conductivity of the In. However, for both the stamped and dropper prepared samples, individual deposits of In were too small to reliably contact. There have been reports of strongly suppressed conductivity in copper features prepared by μCP [19]. The authors of that study suggest the cause may be silicon oligomers that are incorporated from the surface of the PDMS. While their process was quite different from the approach here, it is possible that PDMS contamination may influence the conductivity of the In deposits. This effect does not limit the usefulness of stamped metal features, because it has been shown to be minimized or eliminated by pretreatments of the stamps, such as, vitrification in a UV/ozone environment [20] or soaking in solvents [21]. One application for metallic deposits on silicon substrates was demonstrated by Beach et al. [22]. The authors converted arrays of metallic gallium into GaAs crystallites through anneals in arsine gas. These crystallites were then used as seeds and enlarged using selective organometallic chemical vapor deposition GaAs growth. Similarly, patterned metallic In deposits could be converted to, for example, InAs crystallites. In addition to functioning as seed crystals, arrays of optically active semiconductor crystallites such as GaAs and InAs on Si have potential applications for optoelectronic devices and photonic crystals, particularly if the sizes could be reduced to nanoscale regimes. Another potential application may be contacts to Si devices. By changing the initial photolithography of the masters, it seems conceivable that one may pattern lines as opposed to dots. 4. Conclusions Anisotropic KOH etching was used to make stamp masters. PDMS stamps were cast from the masters. μCP with metallic salt solutions (GaCl3 and In(NO3)3) was demonstrated to be an effective method for patterning Si substrates with arrays of micron size salt deposits. This method was a more direct approach than previous work that used μCP of SAMs for wet etching or for creating a templated surface of hydrophillic and hydrophobic regions. It is believed that a stamp with an aspect ratio greater than 0.9 is necessary to reduce unwanted deposition when attempting to pattern small features that are to be spaced by tens of microns. The deposits made using In(NO3)3 solution were annealed in forming gas to reduce the deposits to In. Si alloying and
subsequent oxidation is identified as a likely reason for the formation of many small deposits instead of a single, large coalesced deposit. A higher purity hydrogen environment may allow the reduction to occur at lower temperatures and decrease the alloying effect. The entire μCP process demonstrates a low cost method of patterning that is potentially useful in microelectronics processing. Acknowledgements The authors would like to thank Don Williamson for his help collecting XRD data and John Yarbrough, Jami Dashdorj, and Sudipta Bera for their discussions. This material is based on work supported by the National Science Foundation under Grant No. DMR-0103945 and Air Force Research Laboratory Award Nos. F29601-02-C-0222 and F29601-02-C-0223. References [1] A. Kumar, G. Whitesides, Appl. Phys. Lett. 63 (1993) 2002. [2] Y. Xia, M. Mirksich, E. Kim, G. Whitesides, J. Am. Chem. Soc. 117 (1995) 9576. [3] E. Delamarche, H. Schmid, A. Bietsch, N. Larsen, H. Rothuizen, B. Michel, H. Biebuyck, J. Phys. Chem., B 102 (1998) 3324. [4] B. Ridley, B. Nivi, B. Hubert, C. Bulthaup, E. Wilhelm, J. Jacobson, Appl. Phys. Lett. 63 (1993) 2002. [5] D. Wang, S. Thomas, K. Wang, Y. Xia, G. Whitesides, Appl. Phys. Lett. 70 (1997) 1593. [6] F.C.M.J.M van Delft, F.C. van den Heuvel, A.E. Kuiper, P.C. Thüne, J.W. Niemantsverdriet, Microelectron. Eng. 73 (2004) 202. [7] P. St. John, H. Craighead, Appl. Phys. Lett. 68 (1996) 1022. [8] D. Qin, Y. Xia, B. Xu, H. Yang, C. Zhu, G. Whitesides, Adv. Mater. 11 (1999) 1433. [9] Z. Zhong, B. Gates, Y. Xia, Langmuir 16 (2000) 10369. [10] Y. Xia, G. Whitesides, Annu. Rev. Mater. Res. 28 (1998) 153. [11] N. Jeon, K. Finni, K. Branshaw, R. Nuzzo, Langmuir 13 (1997) 3382. [12] N. Chang, J. Richardson, P. Clem, J. Hsu, Small 2 (2006) 75. [13] S. Campbell, The Science and Engineering of Microelectronic Fabrication, Oxford University Press, New York, 2001, p. 260. [14] I. Barycka, I. Zubel, Sens. Actuators, A, Phys. 48 (1995) 228. [15] M. Yun, V. Burrows, M. Kozicki, J. Vac. Sci. Technol., B 16 (1998) 2844. [16] E. Delamarche, H. Schmid, B. Michel, H. Biebuyck, Adv. Mater. 9 (1997) 741. [17] C. Marzolin, S. Smith, M. Prentiss, G. Whitesides, Adv. Mater. 10 (1998) 571. [18] Powder Diffraction File, JCPDS-International Centre for Diffraction Data, Newton Square, PA, 2002, Card 05-0642. [19] K. Felmet, Y.L. Loo, Y. Sun, Appl. Phys. Lett. 85 (2004) 3316. [20] K. Glasmästar, J. Gold, A.S. Andersson, D. Sutherland, B. Kasemo, Langmuir 19 (2003) 5475. [21] D. Graham, D. Price, B. Ratner, Langmuir 18 (2002) 1518. [22] J. Beach, C. Veauvy, R. Caputo, R. Collins, A. Khandekar, T. Kuech, C. Inoki, T. Kuan, R. Hollingsworth, Appl. Phys. Lett. 84 (2004) 5323.