- Email: [email protected]
Electrical conduction in 10–20 nm thick polycrystalline tin oxide thin films deposited by chemical vapor deposition
Thin Solid Films 517 (2009) 2953–2958
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / t s f
Electrical conduction in 10–20 nm thick polycrystalline tin oxide thin films deposited by chemical vapor deposition Yuji Matsui a,⁎,1, Yuichi Yamamoto b a b
New Products Development Center, AGC Automotive Japan/Asia, Asahi Glass Co., Ltd. 426-1, Sumita, Aikawa-Machi, Aiko-Gun, KANAGAWA 243-0301 Japan Research Center, Asahi Glass Co., Ltd. 1150 Hazawa-cho, Kanagawa-ku, Yokohama, Kanagawa, 221-8755, Japan
a r t i c l e
i n f o
Article history: Received 16 May 2006 Received in revised form 28 October 2008 Accepted 4 November 2008 Available online 12 November 2008 Keywords: Chemical Vapor Deposition (CVD) Electrical properties and measurements Grain boundary Tin oxide
a b s t r a c t Electrical properties were studied for chemical vapor deposited fluorine doped tin oxide films that were less than 20 nm thick. The electrical properties of the coatings were found to be affected by the type of additive alcohol used in the deposition process. Conductivity was superior for ethanol or isopropyl alcohol (IPA) compared to methanol. Hall effect measurements showed that mobility and carrier concentration were best for IPA, less for ethanol, and least for methanol. Influence of carrier scattering factors to electrical properties was speculated. Potential barrier for carrier scattering at grain boundaries was estimated to be lower in an IPAadded film compared to methanol-added films. Experimental results suggested electrical properties were influenced by size and density of tin oxide micro-grains. It was concluded that interconnections between the micro-grains increased mobility and carrier concentration of very thin films. © 2008 Elsevier B.V. All rights reserved.
Atmospheric pressure Chemical Vapor Deposition (AP-CVD) is widely used for tin oxide coating on glass substrates [1–5]. It offers low manufacturing costs and high productivity. Many studies have been examined AP-CVD deposited tin oxide transparent conductive oxide (TCO) films deposited through the hydrolysis of tin chloride, following the Eq. (1) [3–6].
ticated way [10]. These studies clarified that the thickness dependence of the resistivity is attributable to scattering of free carriers at planar potential walls at grain boundaries as well as scattering at external surfaces (surface scattering). At 10−20 nm film thickness, grain size decreases to a comparable size with the mean free path of carriers. The mean free path can be experimentally determined using Sondheimer's relation, which is expressed in the following Eq. (2) [11].
SnCl4 þ 2H2 O→SnO2 þ 4HCl
ρd = ρ∞ d + 3=8ρ∞ l∞ ð1−pÞ;
1. Introduction
ð1Þ
High sheet resistance, very thin tin oxide films have been used for pen type touch tablets in front of LCD displays [7]. The device consists of a TCO film deposited glass laminated with a transparent plastic film. Since high transmission is necessary for TCO films for this application, film thickness should be as thin as possible to reduce optical absorption and reflection by the film. Therefore, the thickness is usually limited to less than 20 nm [8]. Uniformity of sheet resistance across the coating is a crucial requirement. But the uniformity is liable to be deteriorated, because electrical resistivity of thin films can increase drastically at lower film thickness, e.g. 20 nm. The increase of resistivity with decreasing film thickness (the size effect), was studied theoretically by Mayadas and Shatzkes [9]. Another model describes the effect as carrier scattering effects in more sophis-
⁎ Corresponding author. E-mail addresses: [email protected] (Y. Matsui), [email protected] (Y. Yamamoto). 1 Presently, Industrial Glass Div., AGC Flat Glass Japan/Asia, Asahi Glass Co., Ltd., 8th Floor, Shinyurakucho Bldg, 1-12-1 Yurakucho, Chiyoda-ku, Tokyo, 100-8405 Japan. 0040-6090/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2008.11.047
ð2Þ
where l∞, ρ∞, d and p respectively indicate the mean free path, resistivity in sufficiently thick films, film thickness, and elastic refractive coefficient of the free carrier at the film surface. The mean free path is reported to be 20−50 nm in thick copper films [12]. Since the mean free path is closely related to mobility, the relation between mobility and film thickness is worth noticing. The influence of grain size and potential barrier at grain boundaries to mobility was studied by Seto [13]. The study is generally applied to estimate the potential barrier height in heavily doped polycrystalline materials. Effective mobility, μeff, is expressed in the following Eq. (3). −1=2 μ eff = Dq 2πm⁎ kT expð−Eb =kT Þ;
ð3Þ
where D, q, m⁎, k, T and Eb respectively represent the grain size, electronic charge, effective mass of the free carrier, Boltzmann constant, sample temperature for mobility measurements and the potential barrier height. For the case that mobility is independent of grain size, Vancea et al. explained that carrier concentration is influenced by grain boundaries
2954
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
2. Experimental details
Fig. 1. Schematic diagram of the horizontal tunnel-type AP-CVD furnace.
and that mobility is exclusively influenced by back ground scattering (scattering by defects in crystallites and phonons) [12]. Carrier concentration changes as a function of grain size according to the following Eq. (4). n∞ = nb T ⁎ðl∞=D∞Þ for l∞ ND∞ ;
ð4Þ
where n∞, nb, T ⁎, l∞ and D∞ respectively represent the carrier concentration in thick films, in bulk material, the quantum mechanical transmission coefficient at the grain boundary, the mean free path of carrier and grain size in thick films. Another type of carrier scattering, surface scattering, can be understood as inelastic scattering of the electron wave function at a film surface. The relative magnitude of surface roughness to the de Broglie wavelength can be a standard to estimate contribution of the scatting to electrical conduction. The de Broglie wavelength of free electrons is defined by the following Eq. (5) [14]. λD = 2π=kF ;
ð5Þ
where λD and kF respectively represent the de Broglie wavelength and Fermi wave number. This study is intended to elucidate influence of carrier scattering factors on electrical properties of 10−20 nm thickness tin oxide films. Crystalline properties and electrical properties of methanol-, ethanol-, and IPA-added films were studied and compared. In this study, X-ray diffraction (XRD) was used to estimate changes in crystalline properties of tin oxide micro-grains. Sheet resistance was measured using the four-point probe method. Changes in mobility and carrier concentration were measured by using Hall effect measurements in relation to the type and amount of added alcohol. The potential barrier at the grain boundary was estimated from the temperature dependence of Hall mobility. Results of our study showed that grain boundary scattering eliminated carrier transport, while the surface scattering was of negligible impact on the carrier transport in very thin films.
An in-house-developed horizontal tunnel type AP-CVD furnace was used to deposit fluorine-doped tin oxide films through the reaction of the Eq. (1). Fig. 1 illustrates schematic diagram of the CVD furnace. Glass substrates were carried through a heating zone, a deposition zone and a cooling zone of the furnace successively. In the deposition zone, the glass substrates passed under a gas feeder, the underneath face of which was placed parallel to the glass surface. The feeder has slit-like gas nozzles that supply vapor of SnCl4, H2O and alcohol. Typical mean gas velocity was 50 cm/s. Hydrogen fluoride (HF) was used to dope fluorine into oxide crystallites. Vaporization of SnCl4, H2O and alcohol were done by a bubbling method. Vaporized materials were transported to the gas feeder through stainless steel tubes. Nitrogen was used as the dilution gas for source materials. The molar supply rate of each material is calculable from vapor pressure and flow-rate of the nitrogen bubbling gas. The substrate temperature for tin oxide deposition was 540 °C. Three types of alcohol were used as additives to the reaction of the Eq. (1): methanol, ethanol and IPA. Common soda-lime glass, 1.1 mm thick and 300 mm × 350 mm size, was used as a substrate. The 350 mm side was parallel to the glass transportation direction in the CVD furnace. A 50 nm thick silica layer was deposited before tin oxide deposition to impede the migration of alkali metal ions from glass into the tin oxide layer. Optical transmission measurements were done to determine the film thickness via the following procedures, because common stylus apparatuses for thickness measurements could not be used for 10–20 nm films. First, a numerical simulation for the transmission spectrum of the Glass (1.1 mm)/SiO2 (50 nm)/SnO2 system was carried out as a function of tin oxide film thickness using in-house-developed optical simulation software. A relationship between maximum transmission and film thickness was determined from the simulation. Next, the transmission spectrum of the actual Glass (1.1 mm)/SiO2 (50 nm)/SnO2 system was measured using a spectrometer (UV3100-PC double-monochrometer; Shimadzu Corp.). A relationship between maximum transmission and tin chloride supply was inferred from these measurements. Based on these two results, calibration between the tin chloride supply and film thickness of 5–20 nm was obtained, which was useful for deposition of 10–15 nm films. The changes of crystal phases of 10 nm films in relation to the added methanol amount were identified by using a Rigaku Rint2000 grazing-angle XRD. Cu-kα radiation operated at 50 kV–280 mA was used as a X-ray source. The incident angle was maintained at 0.5°. Molar ratios of H2O and HF to SnCl4 for measured films were 293 and 4.4, respectively. Molar ratios of methanol (CH3OH) to SnCl4 for measured tin oxide films are listed in Table 1. Resistivity was estimated in relation to the film thickness by measuring sheet resistance at room temperature using a Loresta IP MCP-T250 (Mitsubishi Chemical) four-point probe. The measurement was done at 12 points by 25 mm increments along the 350 mm side of the glass. The average of 12 measured values was designated as the
Table 1 CH3OH/SnCl4 molar ratios, XRD intensity I(hkl) and approximated volume fractions of grains F(hkl) for 10 nm tin oxide films CH3OH/SnCl4
Diffraction intensity
Molar ratio
I(110)
I(101)
I(211)
F(110)
F(101)
F(211)
0 6.7 59 98 201 SnO2 Cassiterite
791 516 477 559 539 100
268 109 106 117 110 75
231 147 145 152 164 57
0.61 0.67 0.66 0.68 0.66 0.43
0.21 0.14 0.14 0.14 0.14 0.32
0.18 0.19 0.20 0.18 0.20 0.25
Calculated F(hkl)
F(hkl) was defined by the Eq. (6). Data of cassiterite (PDF #41-1445) are listed for reference.
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
2955
Table 2 Type of added alcohol, film thickness and molar ratios of source materials for tin oxide films used for Hall effect measurements Alcohol
Thickness
H2O/SnCl4
HF/SnCl4
Alcohol/SnCl4
Methanol
11 nm 11 nm 920 nm 11 nm 11 nm 13 nm
620 310 150 293 311 145
3.9 3.9 4.0 3.9 4.2 4.0
124, 248, 310, 620 248, 620 0.7, 1.0, 1.6 5, 16, 53, 127, 320 3, 11, 57, 132, 286 7, 25, 41, 86, 101
Ethanol IPA
sheet resistance. Measured films were deposited under approximately equivalent material supply conditions. H2O/SnCl4, HF/SnCl4 and alcohol/SnCl4 molar ratios were controlled to 295 ± 15, 5.0 ± 1.0 and 155 ± 25, respectively. Carrier concentration and Hall mobility were measured in relation to the added alcohol amount for 11 nm or 13 nm films, using the van der Pauw method with an ACCENTTM HL5500 Hall effect measurement system. Square 1 cm2 samples from the center part of the glass substrate were used for the measurement. Thick 920 nm methanoladded films were measured for comparison. Temperature dependence of Hall mobility on 11 nm films was measured from 73 K to 833 K for two kinds of methanol-added films and an IPA-added film. The potential barrier height at the grain boundary was calculated by applying the Eq. (3). Type of added alcohol, film thickness and molar ratios of source materials for tin oxide films used for Hall effect measurements are listed in Table 2. 3. Results Fig. 2 shows optical transmission spectra of 1.1 mm thick glass coated with tin oxide films. Film thickness was 10 nm or 15 nm. The spectrum of bare 1.1 mm thick glass is shown for comparison. Transmission decrease from the 10 nm film was less than one percent at 400 nm wavelength. However the decrease from the 15 nm film was more than three percent. The 5 nm of thickness increase caused a significant decrease in optical transmission. Fig. 3 shows the XRD pattern of the film deposited without alcohol addition. The film shows peaks corresponding to the (110), (101), (200), (211), (220), (301) and (310) planes of tin oxide. The (110), (101) and (211) planes showed higher diffraction intensity than other
Fig. 2. Optical transmission spectra of glass substrates coated with methanol-added tin oxide films. Thickness of the SiO2 undercoating, 50 nm. Solid and broken lines correspond to substrates with 10 nm and 15 nm tin oxide films, respectively. H2O/SnCl4 and HF/SnCl4 molar ratios for film deposition, 311 and 5.9, respectively. CH3OH/SnCl4 molar ratios, 434 and 414 respectively for 10 nm and 15 nm films. Blank circles correspond to the spectrum of 1.1 mm thick bare glass.
Fig. 3. XRD pattern of a 10 nm tin oxide film deposited without alcohol addition. H2O/ SnCl4 and HF/SnCl4 molar ratios for film deposition, 294 and 4.4, respectively.
planes. Approximated volume fractions of grains oriented to a certain plane (hkl) were calculated using Eq. (6): FðhklÞ ¼IðhklÞ = Ið110Þ þ Ið101Þ þ Ið211Þ ;
ð6Þ
where I(110), I(101) and I(211) are the intensities corresponding respectively to diffraction planes (110), (101) and (211). The subscript (hkl) can be (110), (101) or (211). Table 1 lists I(hkl) and F(hkl) of tin oxide films in relation to CH3OH / SnCl4 molar ratios. Data of cassiterite (PDF 41-1445) [15] are listed for comparison. F(hkl) values of methanoladded films are almost unchanged by CH3OH/SnCl4 molar ratios. F(110) values of films are higher than that of cassiterite. To the contrary, F(101) and F(211) values of films are lower than those of cassiterite. These results mean (110) is a predominant growth direction of films, and this type of film growth does not change by methanol addition. Fig. 4 shows change in electrical resistivity in relation to the film thickness. Although resistivity increase with decreasing thickness was
Fig. 4. Changes in resistivity in relation to the tin oxide film thickness; methanol — (Δ), ethanol — (●), and IPA — (□) added films. H2O/SnCl4, alcohol/SnCl4 and HF/SnCl4 molar ratios for film deposition were controlled to 292 ± 15, 5.0 ± 1.0 and 155 ± 25, respectively. Lines are to guide the eye.
2956
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
a common phenomenon, methanol-added films showed a distinctive increase compared to ethanol-added or IPA-added films. At 10 nm, resistivity of the methanol-added film was more than five times of the value at 18 nm, while the degree of the increase was less than two times in ethanol- or IPA-added films. Figs. 5, 6 and 7 respectively show changes in resistivity, carrier concentration and Hall mobility of 11 nm, 13 nm and 920 nm films in relation to the alcohol/SnCl4 molar ratio. In Fig. 5, resistivity of very thin methanol-added films decreased with increasing methanol addition, whereas it increased from ethanol or IPA addition. A minimum point appeared in the IPA-added films deposited with a H2O/SnCl4 molar ratio of 145, while it was not apparent in the films deposited with a H2O/SnCl4 molar ratio of 311. Lowest bulk resistivity measured was 2.2 × 10− 3 Ωcm, 9.4 × 10− 4 Ωcm and 7.6× 10− 4 Ωcm for methanol-, ethanol-, and IPAadded films, respectively. Resistivity of 920 nm methanol-added films was lower than 11 nm films by one order of magnitude. As shown in Fig. 6, carrier concentration was not equivalent in the three types of very thin films. The measured carrier concentration was in the order of methanol- b ethanol- b IPA- added films. The highest carrier concentration measured was 3.2 × 1020 cm− 3, 5.8 × 1020 cm− 3 and 6.4 × 1020 cm− 3, respectively in methanol-, ethanol-, and IPAadded films. IPA-added films deposited at a H2O/SnCl4 molar ratio of 145 showed a peak at an alcohol/SnCl4 molar ratio of 25. Carrier concentration of 920 nm methanol-added films was approximately same magnitude with the 11 nm methanol-added films. As shown in Fig. 7, Hall mobility of 11 nm methanol-added films changed in a contrasting manner to those of ethanol-, or IPA-added films. Mobility increased from methanol addition, whereas it decreased from ethanol or IPA addition. The highest mobility measured was 8.9, 12.2 and 13.3 cm2/vs, respectively for methanol-, ethanol-, and IPA-
Fig. 5. Changes in resistivity in relation to alcohol/SnCl4 molar ratios for tin oxide films; methanol-added films deposited at H2O/SnCl4 molar ratios of 620 (▲) and 310 (Δ), ethanol-added films (●), and IPA-added films deposited at H2O/SnCl4 molar ratios of 311 (□) and 145 (■). Plots of 920 nm methanol-added films are shown for comparison. Type of added alcohol, film thickness and molar ratios of source materials are listed in Table 2. Lines are to guide the eye.
Fig. 6. Changes in carrier concentration in relation to alcohol/SnCl4 molar ratios for tin oxide films. Symbols used for plots are common with those used in Fig. 5. Type of added alcohol, film thickness and molar ratios of source materials are listed in Table 2. Lines are to guide the eye.
Fig. 7. Changes in Hall mobility in relation to alcohol/SnCl4 molar ratios for tin oxide films. Symbols used for plots are common with those used in Fig. 5. Type of added alcohol, film thickness and molar ratios of source materials are listed in Table 2. Hall mobility as a function of temperature was measured for films marked as A, B and C. Lines are to guide the eye.
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
added films. Mobility of 920 nm methanol-added films was more than 30 cm2/vs, and decreased with increasing methanol addition. Hall mobility as a function of temperature was measured for methanol-added and IPA-added films which are marked as A, B and C in Fig. 7. The relations between calculated μT1/2 values and 1/kT were studied for these three kinds of films, where μ, T and k respectively represent Hall mobility, temperature for Hall effect measurements and Boltzmann constant. The relation between Ln(μT1/2) and (kT)− 1 are shown in Fig. 8. Linear dependence between Ln(μT1/2) and (kT)− 1 could be seen in the lower temperature region between 90 K and 169 K. The potential barrier height, Eb, was calculated in the temperature region by applying the Eq. (3). The calculated value was 59 meV, 49 meV and 47 meV for films A, B and C respectively. These values are more than four times the reported values of 10 meV for thick tin oxide films [16]. 4. Discussion 4.1. Resistivity in relation to added alcohol types According to the theory advanced by Petritz, the resistivity of polycrystalline films is generally written as a linear summation of resistivity in the grain boundary layers and the single crystal-like crystallites, as expressed in the following Eq. (7). ρ = ρ G + ρS ;
ð7Þ
in this equation, ρ, ρG and ρS respectively indicate resistivity of poly crystalline films, grain boundary layers and single crystal-like crystallites. In this study, methanol-added very thin films were polycrystalline films. Crystalline properties of films did not change with CH3OH/ SnCl4 molar ratios as summarized in Table 1. Thus decrease of resistivity in 11 nm methanol-added films shown in Fig. 5 could be attributed to change in the ρG component, not in the ρS component. Since resistivity decreased in the order methanol- N ethanol- N IPAadded films, the absolute value of the ρG component would decrease in the same order.
2957
4.2. Changes in carrier concentration According to the theory expressed by the Eq. (4), the effective carrier concentration would be decreased by the grain boundary scattering. The theory is inapplicable to very thin films of this study. It is worth noticing that carrier concentration of IPA-added films showed a maximum point as it was shown in Fig. 6. The potential barrier at the grain boundary is generally understood as an electron depletion layer that originated from electron trapping states located at the grain surface [17]. The measured carrier concentration is an average value of the concentrations in depletion layers and that in bulk tin oxide crystallites. Development of interconnections between grains might decrease the trapping states and increase carrier concentration. On the other hand, volume fraction of the depletion layer increases with increasing grain density. These two effects would work simultaneously with increasing IPA addition. The maximum point could appear from a summation of these two effects. In the lower water supply condition (H2O/SnCl4 molar ratio of 145), the carrier concentration peak became apparent at this condition. The interconnection would not be as highly developed as the higher water supply conditions (H2O/SnCl4 molar ratio of 311). 4.3. Mobility change in relation to potential barrier at grain boundaries The estimated potential barrier in methanol-added films, marked as “A” and “B” in Fig. 6, was 59 meV to 49 meV, respectively. Meanwhile, the estimated potential barrier in the IPA-added film, marked as “C” in Fig. 7, was 47 meV. The potential barrier presumably decreased by development of interconnection between micro-grains. On the other hand, mobility increased by increasing methanol addition. Conversely, it decreased with increasing ethanol or IPA addition as shown in Fig. 7. This phenomenon can be attributed to shrinking of grain size, the factor D in the Eq. (3). The potential barrier height was found to be more than four times of that in thick films. This result is consistent with a reported result that the grain boundary contribution to film resistivity becomes smaller in films with thickness greater than 350 nm [18]. In thick films, mobility would be decided by intra-grain carrier scatterings. The fact that mobility in 920 nm films decreased with methanol addition would be attributable to increase of intra-grain scattering centers, such as defects in tin oxide crystallites. 4.4. Carrier scattering at the film surface The de Broglie wavelengths of free electrons were calculated using the Eq. (5), assuming a Drude-type free carrier for kF. Calculated values for methanol-, ethanol- or IPA- added films are 3.0 nm, 2.4 nm and 2.4 nm, respectively, at their highest measured carrier concentrations. On the other hand, the surface roughness of the above mentioned films would be more than 1 nm. The de Broglie wavelength is comparable size with the surface roughness. Free carriers arriving at the film surface would be diffused on reflection. Hence, a coherent reflection of the electron wave function will not occur at the film surface. Its contribution to electrical conduction would be of little influence in very thin films. 5. Conclusion
Fig. 8. Relations between calculated μT1/2 values and 1/kT, where μ, T and k respectively represent Hall mobility, sample temperature for Hall effect measurements and Boltzmann constant. Samples A, B and C correspond to two kinds of methanol-added films and an IPA-added film shown in Fig. 7, respectively. Symbols, experimental results; lines, fits with the Eq. (3).
Electrical conduction was studied for tin oxide films of less than 20 nm thickness that were deposited by chemical vapor deposition using tin chloride as a source material. Methanol addition on deposition did not change crystalline orientations of tin oxide micro-grains. Resistivity of ethanol- or IPA-added films was lower than half the value of the methanol- added film at 10 nm film thickness. The carrier concentration in films was speculated to be influenced by two kinds of effects, development of interconnections between micro-grains and
2958
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
increase in volume fraction of grain boundary layers. The estimated potential barrier in an IPA- added film was lower than that in methanoladded films. Experimental results suggested a dominant influence of carrier scatterings at the grain boundary. Development of interconnections between tin oxide micro-grains would be an essential factor for carrier transport in very thin films. The contribution of the carrier scattering at the film surface was of negligible influence for electrical conduction. Acknowledgements The authors would like to thank to Mr. Goto and Dr. Odaka of Asahi Glass Co., Ltd. for valuable discussions and encouragement. References [1] [2] [3] [4]
J.M. Blocher Jr., Thin Solid Films 77 (1981) 51. R.J. McCurdy, Thin Solid Films 351 (1999) 66. R.N. Ghoshtagore, J. Electrochem. Soc. 125 (1978) 110. G.N. Advani, A.G. Jordan, C.H.P. Lupis, R.L. Longini, Thin Solid Films 62 (1979) 361.
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]
[18]
M. Mizuhashi, Y. Goto, K. Adachi, Jpn. J. Appl. Phys. 27 (1988) 2053. G. Sanon, R. Rup, A. Mansingh, Thin Solid Films 190 (1990) 287. N. Kikuchi, E. Kusano, E. Kishio, A. Kinbara, Vacuum 66 (2002) 365. K. Inoue; Y. Akao, S. Takaki, Y. Hayashi, T. Fujiwara, U.S. Patent No. 6329044, 11 Dec. 2001. A.F. Mayadas, M. Shatzkes, Phys. Rev. B (1970) 1382. H. Maron, M. Ritterband, M. Eizenberg, Thin Solid Films 510 (2006) 62. E.H. Sondheimer, Phys. Rev. 80 (1950) 401. J. Vancea, H. Hoffmann, K. Kastner, Thin Solid Films 121 (1984) 201. J.Y. Seto, J. Appl. Phys. 46 (1975) 5247. C.M. Wolfe, N. Holonyak Jr., G.E. Stillman, Physical Properties of Semiconductors, Prentice Hall, Englewood Cliffs, New Jersey, 1989, p. 30. W. Baur, Acta Cryst. 9 (1956) 515. K. Sato, K. Adachi, Y. Hayashi, M. Mizuhashi, 20th IEEE Photovoltaic Specialists Conference Proc., LasVegas, 1988, p. 267. A. Klein, in: D.S. Ginley, M.E. Hawley, D.C. Paine, D.H. Blank, S.K. Streiffer (Eds.), Transport and Microstructural Phenomena in Oxide Electronics, U.S.A. April 16 – 20, 2001, Materials Research Society Symposium Proceedings, vol. 666, 2001, p. F1.10.1. A.S. Gilmore, A. Al-Kaoud, V. Kaydonov, T.R. Ohno, in: D.S. Ginley, M.E. Hawley, D.C. Paine, D.H. Blank, S.K. Streiffer (Eds.), Transport and Microstructural Phenomena in Oxide Electronics, U.S.A. April 16–20, 2001, Materials Research Society Symposium Proceedings, vol. 666, 2001, F3.10.1.
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / t s f
Electrical conduction in 10–20 nm thick polycrystalline tin oxide thin films deposited by chemical vapor deposition Yuji Matsui a,⁎,1, Yuichi Yamamoto b a b
New Products Development Center, AGC Automotive Japan/Asia, Asahi Glass Co., Ltd. 426-1, Sumita, Aikawa-Machi, Aiko-Gun, KANAGAWA 243-0301 Japan Research Center, Asahi Glass Co., Ltd. 1150 Hazawa-cho, Kanagawa-ku, Yokohama, Kanagawa, 221-8755, Japan
a r t i c l e
i n f o
Article history: Received 16 May 2006 Received in revised form 28 October 2008 Accepted 4 November 2008 Available online 12 November 2008 Keywords: Chemical Vapor Deposition (CVD) Electrical properties and measurements Grain boundary Tin oxide
a b s t r a c t Electrical properties were studied for chemical vapor deposited fluorine doped tin oxide films that were less than 20 nm thick. The electrical properties of the coatings were found to be affected by the type of additive alcohol used in the deposition process. Conductivity was superior for ethanol or isopropyl alcohol (IPA) compared to methanol. Hall effect measurements showed that mobility and carrier concentration were best for IPA, less for ethanol, and least for methanol. Influence of carrier scattering factors to electrical properties was speculated. Potential barrier for carrier scattering at grain boundaries was estimated to be lower in an IPAadded film compared to methanol-added films. Experimental results suggested electrical properties were influenced by size and density of tin oxide micro-grains. It was concluded that interconnections between the micro-grains increased mobility and carrier concentration of very thin films. © 2008 Elsevier B.V. All rights reserved.
Atmospheric pressure Chemical Vapor Deposition (AP-CVD) is widely used for tin oxide coating on glass substrates [1–5]. It offers low manufacturing costs and high productivity. Many studies have been examined AP-CVD deposited tin oxide transparent conductive oxide (TCO) films deposited through the hydrolysis of tin chloride, following the Eq. (1) [3–6].
ticated way [10]. These studies clarified that the thickness dependence of the resistivity is attributable to scattering of free carriers at planar potential walls at grain boundaries as well as scattering at external surfaces (surface scattering). At 10−20 nm film thickness, grain size decreases to a comparable size with the mean free path of carriers. The mean free path can be experimentally determined using Sondheimer's relation, which is expressed in the following Eq. (2) [11].
SnCl4 þ 2H2 O→SnO2 þ 4HCl
ρd = ρ∞ d + 3=8ρ∞ l∞ ð1−pÞ;
1. Introduction
ð1Þ
High sheet resistance, very thin tin oxide films have been used for pen type touch tablets in front of LCD displays [7]. The device consists of a TCO film deposited glass laminated with a transparent plastic film. Since high transmission is necessary for TCO films for this application, film thickness should be as thin as possible to reduce optical absorption and reflection by the film. Therefore, the thickness is usually limited to less than 20 nm [8]. Uniformity of sheet resistance across the coating is a crucial requirement. But the uniformity is liable to be deteriorated, because electrical resistivity of thin films can increase drastically at lower film thickness, e.g. 20 nm. The increase of resistivity with decreasing film thickness (the size effect), was studied theoretically by Mayadas and Shatzkes [9]. Another model describes the effect as carrier scattering effects in more sophis-
⁎ Corresponding author. E-mail addresses: [email protected] (Y. Matsui), [email protected] (Y. Yamamoto). 1 Presently, Industrial Glass Div., AGC Flat Glass Japan/Asia, Asahi Glass Co., Ltd., 8th Floor, Shinyurakucho Bldg, 1-12-1 Yurakucho, Chiyoda-ku, Tokyo, 100-8405 Japan. 0040-6090/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2008.11.047
ð2Þ
where l∞, ρ∞, d and p respectively indicate the mean free path, resistivity in sufficiently thick films, film thickness, and elastic refractive coefficient of the free carrier at the film surface. The mean free path is reported to be 20−50 nm in thick copper films [12]. Since the mean free path is closely related to mobility, the relation between mobility and film thickness is worth noticing. The influence of grain size and potential barrier at grain boundaries to mobility was studied by Seto [13]. The study is generally applied to estimate the potential barrier height in heavily doped polycrystalline materials. Effective mobility, μeff, is expressed in the following Eq. (3). −1=2 μ eff = Dq 2πm⁎ kT expð−Eb =kT Þ;
ð3Þ
where D, q, m⁎, k, T and Eb respectively represent the grain size, electronic charge, effective mass of the free carrier, Boltzmann constant, sample temperature for mobility measurements and the potential barrier height. For the case that mobility is independent of grain size, Vancea et al. explained that carrier concentration is influenced by grain boundaries
2954
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
2. Experimental details
Fig. 1. Schematic diagram of the horizontal tunnel-type AP-CVD furnace.
and that mobility is exclusively influenced by back ground scattering (scattering by defects in crystallites and phonons) [12]. Carrier concentration changes as a function of grain size according to the following Eq. (4). n∞ = nb T ⁎ðl∞=D∞Þ for l∞ ND∞ ;
ð4Þ
where n∞, nb, T ⁎, l∞ and D∞ respectively represent the carrier concentration in thick films, in bulk material, the quantum mechanical transmission coefficient at the grain boundary, the mean free path of carrier and grain size in thick films. Another type of carrier scattering, surface scattering, can be understood as inelastic scattering of the electron wave function at a film surface. The relative magnitude of surface roughness to the de Broglie wavelength can be a standard to estimate contribution of the scatting to electrical conduction. The de Broglie wavelength of free electrons is defined by the following Eq. (5) [14]. λD = 2π=kF ;
ð5Þ
where λD and kF respectively represent the de Broglie wavelength and Fermi wave number. This study is intended to elucidate influence of carrier scattering factors on electrical properties of 10−20 nm thickness tin oxide films. Crystalline properties and electrical properties of methanol-, ethanol-, and IPA-added films were studied and compared. In this study, X-ray diffraction (XRD) was used to estimate changes in crystalline properties of tin oxide micro-grains. Sheet resistance was measured using the four-point probe method. Changes in mobility and carrier concentration were measured by using Hall effect measurements in relation to the type and amount of added alcohol. The potential barrier at the grain boundary was estimated from the temperature dependence of Hall mobility. Results of our study showed that grain boundary scattering eliminated carrier transport, while the surface scattering was of negligible impact on the carrier transport in very thin films.
An in-house-developed horizontal tunnel type AP-CVD furnace was used to deposit fluorine-doped tin oxide films through the reaction of the Eq. (1). Fig. 1 illustrates schematic diagram of the CVD furnace. Glass substrates were carried through a heating zone, a deposition zone and a cooling zone of the furnace successively. In the deposition zone, the glass substrates passed under a gas feeder, the underneath face of which was placed parallel to the glass surface. The feeder has slit-like gas nozzles that supply vapor of SnCl4, H2O and alcohol. Typical mean gas velocity was 50 cm/s. Hydrogen fluoride (HF) was used to dope fluorine into oxide crystallites. Vaporization of SnCl4, H2O and alcohol were done by a bubbling method. Vaporized materials were transported to the gas feeder through stainless steel tubes. Nitrogen was used as the dilution gas for source materials. The molar supply rate of each material is calculable from vapor pressure and flow-rate of the nitrogen bubbling gas. The substrate temperature for tin oxide deposition was 540 °C. Three types of alcohol were used as additives to the reaction of the Eq. (1): methanol, ethanol and IPA. Common soda-lime glass, 1.1 mm thick and 300 mm × 350 mm size, was used as a substrate. The 350 mm side was parallel to the glass transportation direction in the CVD furnace. A 50 nm thick silica layer was deposited before tin oxide deposition to impede the migration of alkali metal ions from glass into the tin oxide layer. Optical transmission measurements were done to determine the film thickness via the following procedures, because common stylus apparatuses for thickness measurements could not be used for 10–20 nm films. First, a numerical simulation for the transmission spectrum of the Glass (1.1 mm)/SiO2 (50 nm)/SnO2 system was carried out as a function of tin oxide film thickness using in-house-developed optical simulation software. A relationship between maximum transmission and film thickness was determined from the simulation. Next, the transmission spectrum of the actual Glass (1.1 mm)/SiO2 (50 nm)/SnO2 system was measured using a spectrometer (UV3100-PC double-monochrometer; Shimadzu Corp.). A relationship between maximum transmission and tin chloride supply was inferred from these measurements. Based on these two results, calibration between the tin chloride supply and film thickness of 5–20 nm was obtained, which was useful for deposition of 10–15 nm films. The changes of crystal phases of 10 nm films in relation to the added methanol amount were identified by using a Rigaku Rint2000 grazing-angle XRD. Cu-kα radiation operated at 50 kV–280 mA was used as a X-ray source. The incident angle was maintained at 0.5°. Molar ratios of H2O and HF to SnCl4 for measured films were 293 and 4.4, respectively. Molar ratios of methanol (CH3OH) to SnCl4 for measured tin oxide films are listed in Table 1. Resistivity was estimated in relation to the film thickness by measuring sheet resistance at room temperature using a Loresta IP MCP-T250 (Mitsubishi Chemical) four-point probe. The measurement was done at 12 points by 25 mm increments along the 350 mm side of the glass. The average of 12 measured values was designated as the
Table 1 CH3OH/SnCl4 molar ratios, XRD intensity I(hkl) and approximated volume fractions of grains F(hkl) for 10 nm tin oxide films CH3OH/SnCl4
Diffraction intensity
Molar ratio
I(110)
I(101)
I(211)
F(110)
F(101)
F(211)
0 6.7 59 98 201 SnO2 Cassiterite
791 516 477 559 539 100
268 109 106 117 110 75
231 147 145 152 164 57
0.61 0.67 0.66 0.68 0.66 0.43
0.21 0.14 0.14 0.14 0.14 0.32
0.18 0.19 0.20 0.18 0.20 0.25
Calculated F(hkl)
F(hkl) was defined by the Eq. (6). Data of cassiterite (PDF #41-1445) are listed for reference.
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
2955
Table 2 Type of added alcohol, film thickness and molar ratios of source materials for tin oxide films used for Hall effect measurements Alcohol
Thickness
H2O/SnCl4
HF/SnCl4
Alcohol/SnCl4
Methanol
11 nm 11 nm 920 nm 11 nm 11 nm 13 nm
620 310 150 293 311 145
3.9 3.9 4.0 3.9 4.2 4.0
124, 248, 310, 620 248, 620 0.7, 1.0, 1.6 5, 16, 53, 127, 320 3, 11, 57, 132, 286 7, 25, 41, 86, 101
Ethanol IPA
sheet resistance. Measured films were deposited under approximately equivalent material supply conditions. H2O/SnCl4, HF/SnCl4 and alcohol/SnCl4 molar ratios were controlled to 295 ± 15, 5.0 ± 1.0 and 155 ± 25, respectively. Carrier concentration and Hall mobility were measured in relation to the added alcohol amount for 11 nm or 13 nm films, using the van der Pauw method with an ACCENTTM HL5500 Hall effect measurement system. Square 1 cm2 samples from the center part of the glass substrate were used for the measurement. Thick 920 nm methanoladded films were measured for comparison. Temperature dependence of Hall mobility on 11 nm films was measured from 73 K to 833 K for two kinds of methanol-added films and an IPA-added film. The potential barrier height at the grain boundary was calculated by applying the Eq. (3). Type of added alcohol, film thickness and molar ratios of source materials for tin oxide films used for Hall effect measurements are listed in Table 2. 3. Results Fig. 2 shows optical transmission spectra of 1.1 mm thick glass coated with tin oxide films. Film thickness was 10 nm or 15 nm. The spectrum of bare 1.1 mm thick glass is shown for comparison. Transmission decrease from the 10 nm film was less than one percent at 400 nm wavelength. However the decrease from the 15 nm film was more than three percent. The 5 nm of thickness increase caused a significant decrease in optical transmission. Fig. 3 shows the XRD pattern of the film deposited without alcohol addition. The film shows peaks corresponding to the (110), (101), (200), (211), (220), (301) and (310) planes of tin oxide. The (110), (101) and (211) planes showed higher diffraction intensity than other
Fig. 2. Optical transmission spectra of glass substrates coated with methanol-added tin oxide films. Thickness of the SiO2 undercoating, 50 nm. Solid and broken lines correspond to substrates with 10 nm and 15 nm tin oxide films, respectively. H2O/SnCl4 and HF/SnCl4 molar ratios for film deposition, 311 and 5.9, respectively. CH3OH/SnCl4 molar ratios, 434 and 414 respectively for 10 nm and 15 nm films. Blank circles correspond to the spectrum of 1.1 mm thick bare glass.
Fig. 3. XRD pattern of a 10 nm tin oxide film deposited without alcohol addition. H2O/ SnCl4 and HF/SnCl4 molar ratios for film deposition, 294 and 4.4, respectively.
planes. Approximated volume fractions of grains oriented to a certain plane (hkl) were calculated using Eq. (6): FðhklÞ ¼IðhklÞ = Ið110Þ þ Ið101Þ þ Ið211Þ ;
ð6Þ
where I(110), I(101) and I(211) are the intensities corresponding respectively to diffraction planes (110), (101) and (211). The subscript (hkl) can be (110), (101) or (211). Table 1 lists I(hkl) and F(hkl) of tin oxide films in relation to CH3OH / SnCl4 molar ratios. Data of cassiterite (PDF 41-1445) [15] are listed for comparison. F(hkl) values of methanoladded films are almost unchanged by CH3OH/SnCl4 molar ratios. F(110) values of films are higher than that of cassiterite. To the contrary, F(101) and F(211) values of films are lower than those of cassiterite. These results mean (110) is a predominant growth direction of films, and this type of film growth does not change by methanol addition. Fig. 4 shows change in electrical resistivity in relation to the film thickness. Although resistivity increase with decreasing thickness was
Fig. 4. Changes in resistivity in relation to the tin oxide film thickness; methanol — (Δ), ethanol — (●), and IPA — (□) added films. H2O/SnCl4, alcohol/SnCl4 and HF/SnCl4 molar ratios for film deposition were controlled to 292 ± 15, 5.0 ± 1.0 and 155 ± 25, respectively. Lines are to guide the eye.
2956
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
a common phenomenon, methanol-added films showed a distinctive increase compared to ethanol-added or IPA-added films. At 10 nm, resistivity of the methanol-added film was more than five times of the value at 18 nm, while the degree of the increase was less than two times in ethanol- or IPA-added films. Figs. 5, 6 and 7 respectively show changes in resistivity, carrier concentration and Hall mobility of 11 nm, 13 nm and 920 nm films in relation to the alcohol/SnCl4 molar ratio. In Fig. 5, resistivity of very thin methanol-added films decreased with increasing methanol addition, whereas it increased from ethanol or IPA addition. A minimum point appeared in the IPA-added films deposited with a H2O/SnCl4 molar ratio of 145, while it was not apparent in the films deposited with a H2O/SnCl4 molar ratio of 311. Lowest bulk resistivity measured was 2.2 × 10− 3 Ωcm, 9.4 × 10− 4 Ωcm and 7.6× 10− 4 Ωcm for methanol-, ethanol-, and IPAadded films, respectively. Resistivity of 920 nm methanol-added films was lower than 11 nm films by one order of magnitude. As shown in Fig. 6, carrier concentration was not equivalent in the three types of very thin films. The measured carrier concentration was in the order of methanol- b ethanol- b IPA- added films. The highest carrier concentration measured was 3.2 × 1020 cm− 3, 5.8 × 1020 cm− 3 and 6.4 × 1020 cm− 3, respectively in methanol-, ethanol-, and IPAadded films. IPA-added films deposited at a H2O/SnCl4 molar ratio of 145 showed a peak at an alcohol/SnCl4 molar ratio of 25. Carrier concentration of 920 nm methanol-added films was approximately same magnitude with the 11 nm methanol-added films. As shown in Fig. 7, Hall mobility of 11 nm methanol-added films changed in a contrasting manner to those of ethanol-, or IPA-added films. Mobility increased from methanol addition, whereas it decreased from ethanol or IPA addition. The highest mobility measured was 8.9, 12.2 and 13.3 cm2/vs, respectively for methanol-, ethanol-, and IPA-
Fig. 5. Changes in resistivity in relation to alcohol/SnCl4 molar ratios for tin oxide films; methanol-added films deposited at H2O/SnCl4 molar ratios of 620 (▲) and 310 (Δ), ethanol-added films (●), and IPA-added films deposited at H2O/SnCl4 molar ratios of 311 (□) and 145 (■). Plots of 920 nm methanol-added films are shown for comparison. Type of added alcohol, film thickness and molar ratios of source materials are listed in Table 2. Lines are to guide the eye.
Fig. 6. Changes in carrier concentration in relation to alcohol/SnCl4 molar ratios for tin oxide films. Symbols used for plots are common with those used in Fig. 5. Type of added alcohol, film thickness and molar ratios of source materials are listed in Table 2. Lines are to guide the eye.
Fig. 7. Changes in Hall mobility in relation to alcohol/SnCl4 molar ratios for tin oxide films. Symbols used for plots are common with those used in Fig. 5. Type of added alcohol, film thickness and molar ratios of source materials are listed in Table 2. Hall mobility as a function of temperature was measured for films marked as A, B and C. Lines are to guide the eye.
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
added films. Mobility of 920 nm methanol-added films was more than 30 cm2/vs, and decreased with increasing methanol addition. Hall mobility as a function of temperature was measured for methanol-added and IPA-added films which are marked as A, B and C in Fig. 7. The relations between calculated μT1/2 values and 1/kT were studied for these three kinds of films, where μ, T and k respectively represent Hall mobility, temperature for Hall effect measurements and Boltzmann constant. The relation between Ln(μT1/2) and (kT)− 1 are shown in Fig. 8. Linear dependence between Ln(μT1/2) and (kT)− 1 could be seen in the lower temperature region between 90 K and 169 K. The potential barrier height, Eb, was calculated in the temperature region by applying the Eq. (3). The calculated value was 59 meV, 49 meV and 47 meV for films A, B and C respectively. These values are more than four times the reported values of 10 meV for thick tin oxide films [16]. 4. Discussion 4.1. Resistivity in relation to added alcohol types According to the theory advanced by Petritz, the resistivity of polycrystalline films is generally written as a linear summation of resistivity in the grain boundary layers and the single crystal-like crystallites, as expressed in the following Eq. (7). ρ = ρ G + ρS ;
ð7Þ
in this equation, ρ, ρG and ρS respectively indicate resistivity of poly crystalline films, grain boundary layers and single crystal-like crystallites. In this study, methanol-added very thin films were polycrystalline films. Crystalline properties of films did not change with CH3OH/ SnCl4 molar ratios as summarized in Table 1. Thus decrease of resistivity in 11 nm methanol-added films shown in Fig. 5 could be attributed to change in the ρG component, not in the ρS component. Since resistivity decreased in the order methanol- N ethanol- N IPAadded films, the absolute value of the ρG component would decrease in the same order.
2957
4.2. Changes in carrier concentration According to the theory expressed by the Eq. (4), the effective carrier concentration would be decreased by the grain boundary scattering. The theory is inapplicable to very thin films of this study. It is worth noticing that carrier concentration of IPA-added films showed a maximum point as it was shown in Fig. 6. The potential barrier at the grain boundary is generally understood as an electron depletion layer that originated from electron trapping states located at the grain surface [17]. The measured carrier concentration is an average value of the concentrations in depletion layers and that in bulk tin oxide crystallites. Development of interconnections between grains might decrease the trapping states and increase carrier concentration. On the other hand, volume fraction of the depletion layer increases with increasing grain density. These two effects would work simultaneously with increasing IPA addition. The maximum point could appear from a summation of these two effects. In the lower water supply condition (H2O/SnCl4 molar ratio of 145), the carrier concentration peak became apparent at this condition. The interconnection would not be as highly developed as the higher water supply conditions (H2O/SnCl4 molar ratio of 311). 4.3. Mobility change in relation to potential barrier at grain boundaries The estimated potential barrier in methanol-added films, marked as “A” and “B” in Fig. 6, was 59 meV to 49 meV, respectively. Meanwhile, the estimated potential barrier in the IPA-added film, marked as “C” in Fig. 7, was 47 meV. The potential barrier presumably decreased by development of interconnection between micro-grains. On the other hand, mobility increased by increasing methanol addition. Conversely, it decreased with increasing ethanol or IPA addition as shown in Fig. 7. This phenomenon can be attributed to shrinking of grain size, the factor D in the Eq. (3). The potential barrier height was found to be more than four times of that in thick films. This result is consistent with a reported result that the grain boundary contribution to film resistivity becomes smaller in films with thickness greater than 350 nm [18]. In thick films, mobility would be decided by intra-grain carrier scatterings. The fact that mobility in 920 nm films decreased with methanol addition would be attributable to increase of intra-grain scattering centers, such as defects in tin oxide crystallites. 4.4. Carrier scattering at the film surface The de Broglie wavelengths of free electrons were calculated using the Eq. (5), assuming a Drude-type free carrier for kF. Calculated values for methanol-, ethanol- or IPA- added films are 3.0 nm, 2.4 nm and 2.4 nm, respectively, at their highest measured carrier concentrations. On the other hand, the surface roughness of the above mentioned films would be more than 1 nm. The de Broglie wavelength is comparable size with the surface roughness. Free carriers arriving at the film surface would be diffused on reflection. Hence, a coherent reflection of the electron wave function will not occur at the film surface. Its contribution to electrical conduction would be of little influence in very thin films. 5. Conclusion
Fig. 8. Relations between calculated μT1/2 values and 1/kT, where μ, T and k respectively represent Hall mobility, sample temperature for Hall effect measurements and Boltzmann constant. Samples A, B and C correspond to two kinds of methanol-added films and an IPA-added film shown in Fig. 7, respectively. Symbols, experimental results; lines, fits with the Eq. (3).
Electrical conduction was studied for tin oxide films of less than 20 nm thickness that were deposited by chemical vapor deposition using tin chloride as a source material. Methanol addition on deposition did not change crystalline orientations of tin oxide micro-grains. Resistivity of ethanol- or IPA-added films was lower than half the value of the methanol- added film at 10 nm film thickness. The carrier concentration in films was speculated to be influenced by two kinds of effects, development of interconnections between micro-grains and
2958
Y. Matsui, Y. Yamamoto / Thin Solid Films 517 (2009) 2953–2958
increase in volume fraction of grain boundary layers. The estimated potential barrier in an IPA- added film was lower than that in methanoladded films. Experimental results suggested a dominant influence of carrier scatterings at the grain boundary. Development of interconnections between tin oxide micro-grains would be an essential factor for carrier transport in very thin films. The contribution of the carrier scattering at the film surface was of negligible influence for electrical conduction. Acknowledgements The authors would like to thank to Mr. Goto and Dr. Odaka of Asahi Glass Co., Ltd. for valuable discussions and encouragement. References [1] [2] [3] [4]
J.M. Blocher Jr., Thin Solid Films 77 (1981) 51. R.J. McCurdy, Thin Solid Films 351 (1999) 66. R.N. Ghoshtagore, J. Electrochem. Soc. 125 (1978) 110. G.N. Advani, A.G. Jordan, C.H.P. Lupis, R.L. Longini, Thin Solid Films 62 (1979) 361.
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]
[18]
M. Mizuhashi, Y. Goto, K. Adachi, Jpn. J. Appl. Phys. 27 (1988) 2053. G. Sanon, R. Rup, A. Mansingh, Thin Solid Films 190 (1990) 287. N. Kikuchi, E. Kusano, E. Kishio, A. Kinbara, Vacuum 66 (2002) 365. K. Inoue; Y. Akao, S. Takaki, Y. Hayashi, T. Fujiwara, U.S. Patent No. 6329044, 11 Dec. 2001. A.F. Mayadas, M. Shatzkes, Phys. Rev. B (1970) 1382. H. Maron, M. Ritterband, M. Eizenberg, Thin Solid Films 510 (2006) 62. E.H. Sondheimer, Phys. Rev. 80 (1950) 401. J. Vancea, H. Hoffmann, K. Kastner, Thin Solid Films 121 (1984) 201. J.Y. Seto, J. Appl. Phys. 46 (1975) 5247. C.M. Wolfe, N. Holonyak Jr., G.E. Stillman, Physical Properties of Semiconductors, Prentice Hall, Englewood Cliffs, New Jersey, 1989, p. 30. W. Baur, Acta Cryst. 9 (1956) 515. K. Sato, K. Adachi, Y. Hayashi, M. Mizuhashi, 20th IEEE Photovoltaic Specialists Conference Proc., LasVegas, 1988, p. 267. A. Klein, in: D.S. Ginley, M.E. Hawley, D.C. Paine, D.H. Blank, S.K. Streiffer (Eds.), Transport and Microstructural Phenomena in Oxide Electronics, U.S.A. April 16 – 20, 2001, Materials Research Society Symposium Proceedings, vol. 666, 2001, p. F1.10.1. A.S. Gilmore, A. Al-Kaoud, V. Kaydonov, T.R. Ohno, in: D.S. Ginley, M.E. Hawley, D.C. Paine, D.H. Blank, S.K. Streiffer (Eds.), Transport and Microstructural Phenomena in Oxide Electronics, U.S.A. April 16–20, 2001, Materials Research Society Symposium Proceedings, vol. 666, 2001, F3.10.1.