Chemical Anchoring and Photolithography - American Chemical Society

May 11, 2005 - Department of Chemical and Biomolecular Engineering, Korea AdVanced Institute of Science and Technology,. 373-1 Guseong-dong, Yuseong-g...
0 downloads 12 Views 257KB Size
10584

J. Phys. Chem. B 2005, 109, 10584-10589

Patterning of Single-Wall Carbon Nanotubes via a Combined Technique (Chemical Anchoring and Photolithography) on Patterned Substrates Myung-Sup Jung,†,‡ Sung-Ouk Jung,‡ Dae-Hwan Jung,† Young Koan Ko,† Yong Wan Jin,‡ Jongmin Kim,‡ and Hee-Tae Jung*,† Department of Chemical and Biomolecular Engineering, Korea AdVanced Institute of Science and Technology, 373-1 Guseong-dong, Yuseong-gu, Daejeon 305-701, Korea, and Materials and DeVices Research Center, Samsung AdVanced Institute of Technology, P.O. Box 111, Suwon 440-600, Korea ReceiVed: February 16, 2005; In Final Form: April 7, 2005

Single-walled carbon nanotubes (SWNTs) have been chemically attached with high density onto a patterned substrate. To form the SWNT pattern, the substrate was treated with acid-labile group protected amine, and an amine prepattern was formed using a photolithographic process with a novel polymeric photoacid generator (PAG). The polymeric PAG contains a triphenylsulfonium salt on its backbone and was synthesized to obtain a PAG with enhanced efficiency and ease of spin-coating onto the amine-modified glass substrate. The SWNT monolayer pattern was then formed through the amidation reaction between the carboxylic acid groups of carboxylated SWNTs (ca-SWNTs) and the prepatterned amino groups. A high-density multilayer was fabricated via further repeated reaction between the carboxylic acid groups of the ca-SWNTs and the amino groups of the linker with the aid of a condensation agent. The formation of covalent amide bonding was confirmed by X-ray photoelectron spectroscopy (XPS) analysis. Scanning electron microscopy and UV-vis-near-IR results show that the patterned SWNT films have uniform coverage with high surface density. Unlike previously reported patterned SWNT arrays, this ca-SWNT patterned layer has high surface density and excellent surface adhesion due to its direct chemical bonding to the substrate.

1. Introduction Carbon nanotubes (CNTs) have attracted much attention in recent years because of their remarkable physical and chemical properties. Their potential applications include flat panel displays, logic gates, sensors, and energy storage, which all require that the CNTs be patterned at defined positions with large-scale control of location and orientation.1-3 Many efforts have been made to pattern CNTs on a large scale. According to previous reports, CNT patterns on substrates have been fabricated with various selective growth and selfassembly methods. To prepare for the selective growth of CNTs, various technologies such as plasma patterning3 as well as softlithographic4 and photolithographic methods2,5-7 have been applied to the prepatterning of metal catalysts. By growing multiwalled carbon nanotubes (MWNTs) on these prepatterned metal catalysts, well-defined MWNT patterns have been produced with high resolution and density. However, the patterned growth method is difficult to apply for SWNT patterning and always requires the use of high processing temperatures. As a result, the application of this method has been limited. Selfassembly methods have also been applied to the patterning of SWNTs and have proved to be especially powerful for patterning at low temperatures. Several self-assembly methods, including adsorption onto prepatterned substrates8-11 as well as chemical assembly onto various surfaces,12-14 have been used to pattern SWNT films. Although this method has great potential for the modification and functionalization of surfaces, the adhesion * Corresponding author. Telphone: +82-42-869-3931. Fax: +82-42869-3910. E-mail: [email protected]. † Korea Advanced Institute of Science and Technology. ‡ Samsung Advanced Institute of Technology.

between the resulting SWNT patterns and substrates is not sufficient for practical applications, which results in low surface density. In this paper we report a novel fabrication method for patterning of SWNTs with high density and excellent surface adhesion by the covalent attachment of ca-SWNTs onto an amine prepatterned glass with the aid of a condensation agent. To prepare the amine prepatterned glass, a glass plate was treated with aminosilane compound and acid-labile group protected amine, and the acid-labile group was then selectively deprotected with a photolithographic process. We also show that a novel polymeric PAG with triphenylsulfonium salt on its backbone has the enhanced ease of spin coating onto the acid-labile group modified glass substrate. The SWNT monolayer pattern was formed by the reaction between the carboxylic acid groups of the ca-SWNTs and the prepatterned amino groups, and a multilayer pattern with high density was fabricated via further repeated reactions between the carboxylic acid groups of the ca-SWNTs and the amino groups of the linker with the aid of the condensation agent. The variation of the surface density of the ca-SWNT multilayer film with respect to the number of reaction cycles was monitored. The resulting ca-SWNT patterned layer was found to have high density and excellent surface adhesion due to its direct chemical bonding with the substrate surface. 2. Experimental Sections 2.1. Synthesis of the Polymeric Photo Acid Generator. Scheme 1 has described the synthesis of the monomer and the subsequent polymerization of the polymeric PAG. A mixture of 2,6-dimethylanisole (0.213 mol) and 4,4′-hydroxyphenylsul-

10.1021/jp0508103 CCC: $30.25 © 2005 American Chemical Society Published on Web 05/11/2005

SWNT Patterns via Combined Technique on Substrates

J. Phys. Chem. B, Vol. 109, No. 21, 2005 10585

SCHEME 1: Synthetic Procedure for the Polymeric PAG

foxide (0.213 mol) was gradually added to a stirred solution of phosphorus pentaoxide (0.1074 mol) in 250 mL of methanesulfonic acid at 0 °C and allowed to react overnight at room temperature. The reaction mixture was slowly poured into ice water, and sticky precipitates were formed. After removing water, the precipitates were dissolved in an excess mixture of sodium chloride aqueous solution (2 N) and ethanol. A white powder was slowly precipitated with the evaporation of ethanol and collected by filtration. The product was purified by recrystallization from ethanol to afford bis(4-hydroxyphenyl)(4-methoxy-3,5-dimethylphenyl)sulfonium chloride (yield 54%). Bis(4-hydroxyphenyl)(4-methoxy-3,5-dimethylphenyl)sulfonium chloride (0.026 mol) was dissolved in 26 mL of N,N′dimethylacetamide, and then triethylamine (0.077 mol) was added. After cooling to 0 °C, sebacoyl chloride (0.013 mol) and terephthaloyl chloride (0.013 mol) in 26 mL of heptane were added to the reaction mixture and allowed to react overnight at room temperature. After the separation and removal of the top layer of heptane from the mixture, a N,N′-dimethylacetamide layer was added dropwise to diethyl ether (400 mL). The precipitate was filtered and dried under vacuum. The powder was resolubilized in methanol (20 mL) and reprecipitated in water. The crude product was filtered again and dried under vacuum. The product (polymeric PAG in chloride salt form, 0.01 mol) was dissolved in 10 mL of methanol and added to sodium p-toluenesulfonate (0.03 mol). After 3 h reaction, the solution was concentrated by removing methanol (7 mL) and precipitated with water. The precipitate was filtered and dried to obtain the p-toluenesulfonic acid salt of the polymeric PAG. 2.2. Surface Treatment of the Glass Plates by Chemical Self-Assembly. The fabrication scheme is illustrated in Scheme 2a. Each glass plate was immersed in a cleaning solution (1 L of 95% ethanol, 12 mL of water, 120 g of NaOH), rinsed several times with water, and air-dried. The glass plate was then rinsed with 95% ethanol and immersed in 0.1% (v/v) (3-aminopropyl)triethoxysilane in ethanol for 5 min at room temperature, rinsed three times with 95% ethanol, and dried in a vacuum oven at 120 °C for 20 min. The dried glass substrate was placed under an argon gas atmosphere for 12 h, immersed in N,N-dimethylformamide (DMF), and washed with dichloromethane to fix the amino groups on the solid matrix of the glass plate.

SCHEME 2: Schematic Illustration of the Fabrication of the Amine Prepattern with Photolithographic Process. (a) Chemical Attachment of the Linker onto the AmineTreated Substrate; (b) Photolithographic Process with Polymeric PAG

A 0.5 mL aliquot of N-methyl-2-pyrrolidone solution containing 0.1 mM 6-(N-tert-butoxycarbonylamino)caproic acid and 0.2 mM N,N-diisopropylethylamine (DIEA) was mixed with NMP solution containing 0.12 mM O-(7-azabenzotriazol-1-yl)N,N,N′,N′-tetramethyluronium hexafluorophosphate (HATU) for 5 min. The surface-treated glass plate was immersed therein and stirred at 60 °C for 2 h. The 6-(N-tert-butoxycarbonylamino)caproic acid reacts with the amino groups attached to the glass plate. The unreacted amine groups were then capped with acetyl groups by incubating the glass plate in acetic anhydride/ pyridine (1:3 (v/v)) solution for 1 h. 3. Results and Discussion As shown in Scheme 2b, a photolithographic process was utilized to fabricate the amine prepattern. Conventional chemical-amplification type photoresists could be used for the selective

10586 J. Phys. Chem. B, Vol. 109, No. 21, 2005

Jung et al.

Figure 1. Scanned fluorescence images of amine prepatterned glass plates treated with FITC fluorescent dye. The amine prepatterns were fabricated with various PAGs: (a) monomeric PAG; (b) monomeric PAG with surfactant; (c) polymeric PAG. The average spot intensity is (a) 9100, (b) 22 300, and (c) 22 300. The average spot intensity of b was calculated for the only well-defined spot.

SCHEME 3: Monolayer and Multilayer Deposition of ca-SWNTs onto the Prepatterned Substrate

deprotection of the acid-labile groups (t-BOC). However, these types of photoresists produce insufficient photogenerated acid to remove the acid-labile groups from the glass surface and do not easily migrate from the polymer matrix to the surface, which means that more PAG is required. However above a certain amount of PAG, PAG domains form that scatter light and interfere with the formation of the fine pattern. Further, the acids that are generated can diffuse in the opposite direction from the protection group when the glass substrate is heated on a hot plate, thereby reducing the PAG’s efficiency. Besides, in the case of using the PAG alone, formation of thin film by spin coating is not possible. Therefore, we synthesized a novel polymeric PAG for use in the amine prepatterning in order to obtain enhanced efficiency and ease of spin coating onto the amine-modified slide glass. The chemical identification and purity of the polymeric PAG was determined with 1H NMR spectroscopy (Bruker UltraShield 300). The peaks at 1.2-1.6 and 2.2 ppm were found to be due to the protons of the methylene groups in the sebacoyl moiety; those at 2.4 ppm are due to the protons of methyl groups in the phenyl sulfonium p-toluenesulfonate moiety; those at 3.5-3.8 ppm are due to the proton in the methoxy group of the phenyl sulfonium moiety; those at 6.7-8.3 ppm are due to the protons of the phenyl group. The polymeric PAG has a molecular weight approximately Mw ∼ 3600 with a polydispersity ∼ 2.5, as determined with GPC (PL-GPC, Polymer Laboratory) analysis at 80 °C. The efficiency of the polymeric PAG was evaluated by comparison with that of a monomeric PAG, bis(4-hydroxyphenyl)(4-methoxy-3,5-dimethylphenyl)sulfonium p-toluenesulfonate. The solutions of monomeric PAG were prepared in N-methyl2-pyrrolidone with and without surfactant. Each PAG solution

was spin-coated onto the substrate with attached acid-labile protecting groups (t-BOC), as described in the Experimental Section. The glass plates patterned with each PAG were treated with a fluorescent dye (fluorescein-5-isothiocyanate, FITC) that couples with the patterned amino groups. Figure 1 shows scanned fluorescent images of the patterned glass substrates; the detection of the fluorescence signals was performed with a fluorescence scanner (Scanarray 5000, GSI Lumonics). The fluorescence signal was barely visible for the monomeric PAG, due to its poor coating formation on the substrate (Figure 1a). Although the fluorescence of the patterned image was somewhat improved when a surfactant was added to the monomeric PAG solution, the fluorescence signals of some square patterns were still unclear due to imperfect PAG coating (Figure 1b). However, the fluorescence is clearly evident for the polymeric PAG (Figure 1c), indicating its ease of coating formation. These results clearly show that the polymeric PAG has much higher efficiency for selective surface modification (amine prepatterning) than the monomeric PAG. The polymeric PAG was dissolved in N-methyl-2-pyrrolidone to obtain a solution with a final concentration of 30 wt % (w/ v) and spin-coated onto the surface of the substrate with attached acid-labile protecting groups (t-BOC). The coated glass plate was prebaked for 2 min at 80 °C to evaporate the solvent and to form a polymeric PAG layer on the substrate. The substrate was then irradiated with deep UV light (λ ) 248 nm) through a chrome-patterned photomask using a UV mask aligner (ORIEL). After an exposure of 300 mJ/cm2, a postexposure baking process was carried out at 100 °C for 2 min. As a result of these processes, acid is generated from the polymeric PAG in the exposed regions and the t-BOC groups become separated from the linker. The glass plate was then developed using an

SWNT Patterns via Combined Technique on Substrates

J. Phys. Chem. B, Vol. 109, No. 21, 2005 10587

Figure 2. XP spectrum of the N 1s region for a ca-SWNT multilayer constructed with ODA linkers.

aqueous tetramethylammonium hydroxide solution (2.38%), washed with ethanol, and dried at 40 °C for 15 min to remove the PAG and the acid-labile groups from the exposed region. Thus a positive amine pattern was formed on the glass plate. The scheme for SWNT monolayer deposition onto the prepatterned substrate is illustrated in Scheme 3. The as-prepared SWNTs (AP-SWNTs, CarboLex Inc.) were purified and shortened using a method described previously in some detail.15,16 A 2 mg amount of ca-SWNTs was dispersed in 150 mL of DMF by sonication for 2 h. A 0.2 g amount of DIEA was added dropwise to the dispersion of the ca-SWNTs, followed by stirring for 10 min. A 50 mL aliquot of DMF dissolved in 0.4 g of HATU was then added to the dispersion of the ca-SWNTs, followed by stirring for 20 min. The glass substrate was vertically immersed in the solution, and the reaction was allowed to continue at room temperature for 6 h. The prepatterned amine groups on the substrate react with the carboxylic acid groups of the ca-SWNTs in the presence of the coupling agent, and the ca-SWNT pattern forms by amide bonding. After the glass substrate was taken out of the solution, it was washed three times, each for 15 min, in a beaker containing 200 mL of DMF. The glass substrate was then washed three times, each for 15 min, in a beaker containing 200 mL of methylene chloride. The glass substrate was then dried at 40 °C for 15 min. As shown in Scheme 3, the glass substrate on which the caSWNT monolayer formed was then allowed to react in 200 mL of DMF containing 4 g of 4,4′-oxydianiline (ODA) at room temperature for 2 h. The carboxylic acid groups of the ca-SWNT layer that are activated by the coupling agent, HATU, react with amine groups to form amide bonds. This reaction converts the carboxyl groups on the surface of the SWNT layer to amine groups. After the glass substrate was taken out of the solution, it was washed three times, each for 15 min, in a beaker containing 200 mL of DMF. Subsequently, the glass substrate was washed three times, each for 15 min, in a beaker containing 200 mL of methylene chloride, resulting in a glass substrate on which ODA has reacted with the carboxyl groups of the patterned monolayer of ca-SWNTs. The deposition of further ca-SWNTs and washings were then carried out with the method described above to produce a bilayer structure of carbon nanotubes on the substrate. The amino groups of the diamine compounds on the monolayer react with additional ca-SWNTs in the presence of the coupling agent, and more ca-SWNTs are thus deposited on the carbon nanotubes monolayer. By repeating these processes of forming a diamine monolayer and subsequently reacting the diamine monolayer with additional ca-SWNTs, the ca-SWNTs can be deposited

Figure 3. Variation of the UV-vis-near-IR transmittance spectra of the ca-SWNT multilayer films on a glass plate with the number of reaction cycles.

layer by layer, and finally a patterned multilayer of carbon nanotubes is formed. To verify if the ODA serves as a chemical linker of the caSWNTs, the ca-SWNT multilayer was analyzed by X-ray photoelectron spectroscopy (XPS, V. G. ESCALAB MK II). Figure 2 shows the XP N 1s spectrum. The N 1s peak could be deconvoluted into two component peaks. The peak at 400.2 eV was attributed to nitrogen atoms of amine, and the peak at 401.4 eV, to nitrogen atoms of amide. The amine N 1s peak at 400.2 eV may be due to the remaining amino group on one side of ODA. However, the presence of the amide N 1s peak at 401.4 eV is evidence that ODA played an important role in the formation of the SWNT multilayer. Thus the multilayer is held together by amide bonding derived from the amino groups of ODA and the carboxylic acid groups of the ca-SWNTs. UV-vis-near-IR spectroscopy (JASCO V-560) was used to investigate the quantity of deposited ca-SWNTs in each reaction cycle. As shown in Figure 3, the transmittance of the glass plate in the visible light region is decreased with the number of reaction cycles, indicating that the quantity of deposited caSWNTs on the glass plate was increased with repeated reaction cycles. This result shows that the end group of ca-SWNTs deposited substrate was converted carboxylic acid to amino group via amide bond with ODA and the modified end group is chemically assembled with ca-SWNTs in each reaction step. Thus the diamine compound, ODA, plays a decisive role in the multilayer deposition as a chemical linker. The variation of the surface density of the SWNT multilayer film with the number of successive reaction cycles was monitored with scanning electron microscopy (SEM, Philips XL30SFEG). The growth of the multilayer results in the variation of the surface density on the glass substrate. The SEM image of the sample treated with one reaction cycle shows that its substrate is sparsely covered with randomly oriented SWNT ropes of 400-600 nm in length (Figure 4a). Compared to the sparsely covered surfaces after the initial reaction cycle, the SWNT multilayers formed after three or more reaction cycles exhibit uniform surface coverage over large areas (Figure 4b,c). Close inspection of the SWNT multilayer formed after five reaction cycles (Figure 4c) reveals that the SWNT ropes are piled up to a height of approximately four to five layers, with an appearance similar to that of a microphotograph of the cellulose fibers of paper. The overlapping morphology of the multilayer is consistent with ca-SWNT accumulation occurring via alternating reactions in which ca-SWNTs react with ODA-

10588 J. Phys. Chem. B, Vol. 109, No. 21, 2005

Jung et al.

Figure 4. SEM images of the multilayers for various reaction cycles: (a) one reaction cycle; (b) three reaction cycles; (c) five reaction cycles.

Figure 5. SEM images of the patterned ca-SWNT multilayer on an amine-modified slide glass, where the bright line regions are covered with ca-SWNTs: (a) square array (each square area is 80 µm2); (b) line patterns (line width is 1-4 µm); (c) higher magnification image of the 3 µm line. The needlelike fragments are SWNT ropes.

SWNT ropes already attached to the amine-modified surface; after reaction with ODA, these ca-SWNTs form a new layer for the next reaction cycle. The SEM images confirm that the proposed fabrication technique based on diamine linkers produces films with superb uniformity and high surface density over large areas. Thus a ca-SWNT multilayer is selectively formed on the prepatterned substrate. No optical contrast was observed in the absence of ca-SWNT layers on the prepatterned glass substrate prepared with t-BOC protection groups and PAG. However, after attaching the ca-SWNTs chemically to the patterned regions, regular bright line arrays were clearly seen in the SEM images. All SEM images shown in Figure 5 were obtained after five reaction cycles. Figure 5a shows an SEM image of a pattern in which 80 × 80 µm2 squares were arranged. Figure 5b shows a SEM image of line and space patterns with a line width of 1-4 µm. Figure 5c shows a magnified line image of Figure 5b. The magnified line pattern has a clear contrast, which indicates the high density of the micropatterned SWNT array. ca-SWNT ropes were not detected on the region protected with t-BOC groups, indicating that the shielding provided by the t-BOC protection groups effectively prevented the amine groups of the substrate from reacting with the carboxylic acid groups of the ca-SWNTs. These results indicate that the polymeric PAG effectively deprotects the t-BOC groups of the UV-exposed region without causing any damage to the unexposed region. Thus, well-defined micropatterns of ca-SWNTs were easily and successfully fabricated with the polymeric PAG and the photolithographic process proposed in this article. 4. Conclusion We have fabricated a high-density ca-SWNT pattern by using a photolithographic process and a selective chemical reaction.

To fabricate the ca-SWNT pattern, a glass substrate was treated with acid-labile group protected amine, and then an amine prepattern was formed using a photolithographic process with a novel polymeric PAG. The polymeric PAG has triphenylsulfonium salt moieties on its polymer backbone and moderate molecular weight. In contrast to photoresists or monomeric PAGs, the polymeric PAG is more useful for simple and efficient selective surface modification due to its improved spincoating properties. A ca-SWNT pattern with monolayer structure was formed by the reaction between the carboxylic acid groups of the ca-SWNTs and the prepatterned amine groups on the substrate. A high-density ca-SWNT pattern with multilayer structure was fabricated via further repeated reactions between the carboxylic acid groups of the ca-SWNTs and the amino groups of the linker with the aid of a condensation agent. This approach enables the formation of micropatterned ca-SWNT arrays on solid substrates at mild temperatures. Also, this patterning method provides strong adhesion between the SWNTs and the substrates, resulting in high surface density structures on various substrates, which is critical to their practical applications such as solar cells and batteries, flat panel displays, transistors, chemical and biological sensors, and semiconductor devices. Acknowledgment. This work was supported by the KOSEF through the Young Scientist Program, CUPS, and the Brain Korea 21 program. References and Notes (1) (a) Dresselhaus, M. S.; Dresselhaus, G.; Avouris, P. Carbon Nanotubes; Springer: Berlin, 2001. (b) Harris, P. J. F. Carbon Nanotubes and Related Structure; Cambridge University Press: Cambridge, U.K., 1999. (c) Ebbesen, T. W. Carbon Nanotubes; CRC Press: Boca Raton, FL, 1997. (d) Ajayan, P. M. Chem. ReV. 1999, 99, 1787 (e) Rao, C. N. R.; Satishkumar, A. G.; Nath, M. ChemPhysChem 2001, 2, 78.

SWNT Patterns via Combined Technique on Substrates (2) Ren, Z. F.; Huang, Z. P.; Wang, D. Z.; Wen, J. G.; Xu, J. W.; Wang, J. H.; Calvet, L. E.; Chen, J.; Klemic, J. F.; Reed, M. A. Appl. Phys. Lett. 1999, 75, 1086. (3) Chen, Q.; Dai, L. Appl. Phys. Lett. 1999, 76, 2719. (4) Huang, S.; Mau, A. W. H.; Turney, T. W.; White, P. A.; Dai, L. J. Phys. Chem. B 2000, 104, 2193. (5) Huang, S.; Mau, A. H. W. Appl. Phys. Lett. 2003, 82, 796. (6) Yang, Y.; Huang, S.; He, H.; Mau, A. W. H.; Dai, L. J. Am. Chem. Soc. 1999, 121, 10832. (7) Huang, S.; Dai, L.; Mau, A. W. H. AdV. Mater. 2002, 14, 1140. (8) Burghard, M.; Duesberg, G.; Philipp, G.; Murster, J.; Roth, S. AdV. Mater. 1998, 10, 584. (9) Shimoda, H.; Oh, S. J.; Geng, H. Z.; Walker, R. J.; Zhang, X. B.; McNeil, L. E.; Zhou, O. AdV. Mater. 2002, 14, 899. (10) Liu, J.; Casavant, M. J.; Cox, M.; Walters, D. A.; Boul, P.; Lu W.; Rimberg, A. J.; Smith, K. A.; Colbert, D. T.; Smally, R. E. Chem. Phys. Lett. 1999, 303, 125.

J. Phys. Chem. B, Vol. 109, No. 21, 2005 10589 (11) Oh, S. J.; Cheng, Y.; Zhang, J.; Shimoda, H.; Zhou, O. Appl. Phys. Lett. 2003, 82, 2521. (12) Wu, B.; Zhang, J.; Wei, Z.; Cai, S.; Liu, Z. J. Phys. Chem. B 2001, 105, 5075. (13) Diao, P.; Liu, Z.; Wu, B.; Nan, X.; Zhang, J.; Wei, Z. ChemPhysChem 2002, 898. (14) Rao, S. G.; Huang, L.; Setyawan, W.; Hong, S. Nature 2003, 425, 36. (15) Liu, J.; Rinzler, A. G. R.; Dai, H.; Hafner, J. H.; Bradly, R. K.; Boul, P. J.; Lu, A.; Iverson, T.; Shelimov, K.; Huffman, C. B.; Rodriguezmarcias, F.; Shon, Y. S.; Lee, T. R.; Colbert, D. T.; Smally, R. E. Science 1998, 280, 1253. (16) Ringler, A. G.; Liu, J.; Dai, H.; Nikolaev, P.; Huffman, C. B.; Rodriguez-marcias, F. J.; Boul, P. J.; Lu, A. H.; Heymann, D.; Colbert, D. T.; Lee, R. S.; Fischer, J. E.; Rao, A. M.; Eklund, P. C.; Smally, R. E. Appl. Phys. A 1998, 67, 29.