www.elsevier.com/locate/procedia E-MRS Spring meeting 2009, Symposium B Development of atmospheric pressure CVD processes for high- quality transparent conductive oxides Ariël de Graaf a , Joop van Deelen a, *, Paul Poodt a , Ton van Mol b , Karel Spee b , Frank Grob a , Ando Kuypers a a TNO Science and Industry, De Rondom 1, 5616 AP Eindhoven, The Netherlands b TNO Holst Centre, High Tech Campus 31, 5656 AE Eindhoven, The Netherlands Abstract For the past decade TNO has been involved in the research and development of atmospheric pressure CVD (APCVD) and plasma enhanced CVD (PECVD) processes for deposition of transparent conductive oxides (TCO), such as tin oxide and zinc oxide. It is shown that by combining precursor development, fundamental gas phase and surface chemistry studies, process optimization and modeling-based reactor design, the demanding product requirements and cost issues of different types of thin film PV can be met. Our studies on the APCVD deposition of SnO 2 :F reveal the influence of different types of precursors and process conditions on the transmittance, morphology and conductance of the film. It is shown that a high transmittance (80%) and low resistivity (4.0⋅10 -4 Ω⋅cm) film can be obtained in combination with an intrinsic surface structure that enhances the light trapping effect. Keywords: transparent conductive oxides; tin oxide; zinc oxide; atmospheric pressure chemical vapor deposition; plasma-enhanced chemical vapor deposition; gas phase chemistry; surface chemistry; process optimization; surface morphology 1. Introduction Cost-effective, thin film PV requires large scale deposition of transparent conductive oxides (TCO). Fluor-doped tin oxide is an excellent TCO and its main application is shifting from low-e glass to photovoltaics [1]. It can be applied by various methods, such as atomic layer deposition [2], plasma assisted chemical vapor deposition [3], atmospheric chemical vapor deposition (APCVD) [1] and spray pyrolysis [4]. APCVD is of high interest, as it gives the opportunity for low cost, large scale production, which is evident from the presently available km 2 -scale production capacity in the glass industry. TNO has been active in R&D of TCO processing for over a decade. In contrast to low-e coatings, application of TCO in PV has high demands on both the transparency and the conductivity. High conductivities can be obtained by increasing the carrier concentration using a dopant. However, * Corresponding author. Tel.: +31-(0)40-2650785; fax: +31-(0)40-2650850. E-mail address: [email protected]. c 2010 Published by Elsevier Ltd Received 1 June 2009; received in revised form 1 December 2009; accepted 20 December 2009 Energy Procedia 2 (2010) 41–48 www.elsevier.com/locate/procedia 1876-6102 c 2010 Published by Elsevier Ltd doi:10.1016/j.egypro.2010.07.008
Transcript
Available online at www.sciencedirect.com
Energy Procedia 00 (2009) 000–000
Energy
Procedia www.elsevier.com/locate/procedia
E-MRS Spring meeting 2009, Symposium B
Development of atmospheric pressure CVD processes for high-quality transparent conductive oxides
Ariël de Graaf a, Joop van Deelen a,*, Paul Poodt a, Ton van Mol b, Karel Spee b, Frank Grob a, Ando Kuypers a
aTNO Science and Industry, De Rondom 1, 5616 AP Eindhoven, The Netherlands bTNO Holst Centre, High Tech Campus 31, 5656 AE Eindhoven, The Netherlands
Keywords: transparent conductive oxides; tin oxide; zinc oxide; atmospheric pressure chemical vapor deposition; plasma-enhanced chemical vapor deposition; gas phase chemistry; surface chemistry; process optimization; surface morphology
1. Introduction
Cost-effective, thin film PV requires large scale deposition of transparent conductive oxides (TCO). Fluor-doped tin oxide is an excellent TCO and its main application is shifting from low-e glass to photovoltaics [1]. It can be applied by various methods, such as atomic layer deposition [2], plasma assisted chemical vapor deposition [3], atmospheric chemical vapor deposition (APCVD) [1] and spray pyrolysis [4]. APCVD is of high interest, as it gives the opportunity for low cost, large scale production, which is evident from the presently available km2-scale production capacity in the glass industry. TNO has been active in R&D of TCO processing for over a decade. In contrast to low-e coatings, application of TCO in PV has high demands on both the transparency and the
conductivity. High conductivities can be obtained by increasing the carrier concentration using a dopant. However,
incorporation of dopants usually leads to an increased absorption of light in the infrared range. Optimizing the TCO for solar cell applications is therefore in most cases a trade-off between transparency and conductivity. The surface
morphology is another important parameter to consider, as it can improve the light scattering properties [5-7]. In case of high scattering by the TCO, thinner layers can be used for the active material in the solar cell [8].
The commercially available Asahi U-type glass is a SnO2:F coated glass that is often used as a benchmark, since
it combines good conductivity with fair transparency over a wide wavelength range [7]. However, it should be noted that for the use in silicon solar cells, especially the transparency in the wavelength range up to 1100 nm is important, because it corresponds with the band gap of Si. Light with a longer wavelength is not absorbed by Si and, therefore, does not contribute to the cell’s current density. For CuInSe2 solar cells, the cut-off wavelength is higher than 1200 nm, demanding an even broader absorption window.
The shift from low-e applications towards PV requires a substantial change in product characteristics. Process
control is the key to product improvement. This is supported by several lab-scale investigations that clearly demonstrate the impact of deposition parameters on the resulting layer characteristics [9–11]. At TNO, lab research is combined with fundamental understanding of the process and in-house upscaling to industrially relevant proportions. The present paper gives a concise overview of related activities within TNO.
2. Research at TNO
Recently, the glass industry indicated a serious lack of fundamental understanding in the deposition processes [12]. In response to this, Sandia labs and TNO have commenced a study in which a continuously stirred tank CVD reactor (Figure 1) was utilized to determine the species in the gas phase during SnO2 deposition from monobutyltin trichloride (MBTC) and water (Figure 2) [13]. Simultaneously, a stagnant point flow reactor was used for determination of the deposition rate (Figure 3). The deposition rate data together with knowledge of the species formed in the gas phase were used as input for the construction of a CVD model with which different hypothetical growth mechanisms can be studied. The CVD model includes both gas phase and surface reaction kinetics. Five different models were evaluated in which MBTC is preadsorbed on the surface (model A), MBTC reacts with adsorbed oxygen (model B) or adsorbed OH (model C), MBTC reacts with water in the gas phase to form a complex followed by a reaction with adsorbed oxygen (model D) or adsorbed OH (model E). As shown in Figure 4, the
Figure 1: Schematic representation of continuously stirred tank reactor for studying the gas phase chemistry during SnO2:F deposition.
42 A. de Graaf et al. / Energy Procedia 2 (2010) 41–48
model is able to reproduce the trends observed in the experiments quite accurately. Based on the modeling and additional studies [14], model D could be pinpointed as most likely deposition mechanism for the deposition with MBTC as precursor. Various aspects however, such as the influence of the water concentration, have not been fully investigated yet [15].
Another issue that has not been addressed in growth rate modeling studies, is morphology. Morphology is
depending on a wide variety of growth factors, such as deposition temperature, flow rates of the gases, tin precursor used, substrate material, etc. For SnO2 deposition many different tin precursors can be used. Tin precursors with simple alkyl and/or chloride groups are cheap and therefore preferred in large-scale applications. Figure 5 shows SEM images of SnO2 layers produced with different tin precursors [1]. From these images it becomes clear that the highest surface roughness is obtained by using tin tetrachloride (TTC). The morphology resembles that of the Asahi U-type, which is also deposited with TTC. In contrast, tetramethyltin (TMT) yields much smoother layers.
16
14
12
10
8
6
4
2
0
Mol
e Fr
actio
n *1
03
600580560540520500Temperature (C)
MBTC HCl C2H4 H2O SnCl2 CO SnCl4
Figure 2: Mole fraction of different molecules in the gas phase as function of temperature in the continuously stirred tank CVD reactor related to the decomposition of monobutyltin trichloride (MBTC) in the presence of water.
Figure 3: Stagnant point flow reactor equipped with induction heating.
A. de Graaf et al. / Energy Procedia 2 (2010) 41–48 43
The morphology differences result in different transmittance characteristics for layers grown with various tin
precursors, as shown in Figure 6. The transmittance of the layer deposited with TMT shows fringes, because the layer has a smooth surface. A more pronounced native structure decreases the fringes and increases the proportion of diffused light. This translates into a higher current density, when used in solar cells (see Table 1). Asahi U-type is used as a benchmark. It can be seen that, although the current density is similar, the voltage is still somewhat lower for the cell made with Asahi U-type. This is probably due to a difference in surface morphology, which may influence the deposition characteristics and as a result also the material quality and connected solar cell output. It should be noted that the conductivities of the TCO layers are in the same range.
Figure 4: Growth rate versus temperature for various monobutyltin trichloride (MBTC) concentrations. The lines represent different deposition models [13].
a b
c d
a b
c d
Figure 5: SEM images of SnO2:F deposited with different precursors: a. TMT, b. MBTC, c. TTC, d. Asahi U-type. The RMS surface roughness measured on an area of 5 × 5 µm2 is as follows: a. 7 nm, b. 22 nm, c. 39 nm, d. 40 nm.
44 A. de Graaf et al. / Energy Procedia 2 (2010) 41–48
Author name / Energy Procedia 00 (2009) 000–000
Figure 7 shows a comparison between the total and diffuse transmittance of the benchmark and two different layers produced at TNO under different deposition conditions. The conductivity of the SnO2:F made at TNO is over 20% higher than that of Asahi U-type. In addition, the transparency is higher for wavelengths up to 1100 nm, whereas for wavelengths above 1200 nm Asahi U-type TCO displays a higher transparency. This is attributed to its lower carrier concentration (cf. Table 1), as determined with a Hall measurement system based on the Van der Pauw method. However, as explained before, the light in the wavelength range above 1200 nm is not used by the solar cell. For this reason, it is expected that Si cells made with SnO2:F produced by TNO can deliver a higher current density, if the deposition parameters of the Si layers are optimized for this particular substrate.
As mentioned before, the light scattering is of vital importance for thin film cell performance. Enhanced light
scattering offers the opportunity for thinner layers. Not only does this reduce deposition time and costs, it also results in higher cell efficiencies, because the chance on recombination is decreased for thinner layers. It has been modeled that for a 500 nm thin film cell, the current density can be doubled when changing from a flat TCO to a structured TCO [7]. For optimization of the light scattering not only the scattering percentage itself is of importance, but also the angular dependence of the diffused light should be taken into account.
Table 1: Solar cell parameters of a-Si:H p-i-n solar cells prepared on four different tin oxide layers
TMT MBTC TTC Asahi U-type Open circuit voltage [V] 0.763 0.761 0.761 0.792 Short circuit current [mA/cm2] 12.59 13.09 15.73 15.80 Fill factor 0.583 0.672 0.688 0.696 Efficiency [%] 5.60 6.69 8.24 8.71 Resistivity (TCO) [10-4 Ω⋅cm] 5.4 6.9 4.0 7.8
Carrier concentration (TCO) [1020 cm-3] 3.9 4.8 5.0 2.2 Figure 8 shows four morphologies obtained with APCVD for SnO2:F layers using TTC as precursor. As can be
seen, to a certain extent the morphology can be tuned by changing the process parameters. Important process parameters are the substrate temperature, the belt speed, the flow rates of the different precursors and the type of precursors used. The change in morphology will also affect the diffuse transmittance characteristics of the TCO, which is clearly seen in Figure 7 (type I and type II). In Figure 9 it is shown that the grain size can be tuned as well and that the grain size is rather constant throughout
the layer. In the deposition process, however, the grains develop as the layer thickness increases. It appears that the
500 1000 1500 2000 25000.0
0.2
0.4
0.6
0.8
1.0
Tran
smis
sion
T
Wavelength [nm]
TMT Total T MBTC Total T TTC Total T TMT Diffuse T MBTC Diffuse T TTC Diffuse T
Tran
smitt
ance
500 1000 1500 2000 25000.0
0.2
0.4
0.6
0.8
1.0
Tran
smis
sion
T
Wavelength [nm]
TMT Total T MBTC Total T TTC Total T TMT Diffuse T MBTC Diffuse T TTC Diffuse T
Tran
smitt
ance
Figure 6: Total and diffuse transmittance as function of wavelength for SnO2:F deposited with different precursors.
A. de Graaf et al. / Energy Procedia 2 (2010) 41–48 45
initial layer morphology, which in turn influences the morphology of the final layer, is mainly determined by the initial nucleation density [16]. By tuning the nucleation circumstances, morphologies as depicted in Figure 8 and 9 can be obtained.
The surface chemistry is of vital importance for both the deposition rate and the morphology. This topic is currently investigated in a reactor that enables gas phase mixing at atmospheric pressure and deposition at extremely low pressure. In this way, the deposition can be slowed down to the level that species on the surface can be studied by means of XPS.
The deposition process, i.e. decomposition of precursors and incorporation of the fragments in the growing layer,
0
0,2
0,4
0,6
0,8
1
0 500 1000 1500 2000 2500Wavelength [nm]
Transm
ittancee [-]
Asahi U-type
TNO type I
TNO type II
Figure 7: The total and diffuse transmittance as function of wavelength for layers made with tin tetrachloride at two different deposition conditions (type I and II) and for the Asahi U-type.
Figure 8: SEM images of various surface morphologies obtained with APCVD for SnO2:F layers using TTC as precursor.
46 A. de Graaf et al. / Energy Procedia 2 (2010) 41–48
requires substantial energy. Therefore, APCVD is usually performed at elevated temperatures. For temperature-sensitive substrates other techniques, such as plasma, can be used to deliver the required energy to the precursor molecules, without damaging the surface. Plasma-enhanced CVD is widely used in low-pressure processes. TNO is presently working on atmospheric pressure plasma-enhanced CVD (AP-PECVD) for the deposition of metal oxides [19]. Figure 10 shows an example of the reactor used for studying the deposition of zinc oxide by means of AP-PECVD.
It is expected that in the near future fundamental understanding of the deposition processes will increase the
capabilities of tuning the CVD process both for thermal and plasma-enhanced CVD. Already, CVD modeling [20] is giving valuable input for new reactor designs that minimize the cost of deposition, without compromising the material quality. This will create optimized layer characteristics at lower prices per square meter.
3. Conclusions
In this paper some TNO activities concerning TCO deposition have been reviewed. By integration of TNO’s experience on the full width of the CVD spectrum, technological solutions for obtaining optimal electrical conduction in combination with the right surface morphology and superior transmittance can be realized. With the combination of precursor development [21], process optimization, fundamental gas phase and surface chemistry studies, and modeling-based reactor design, the product requirements and cost issues for different types of thin film PV can be met.
Figure 9: SEM images showing cross sections of SnO2:F layers with various grain sizes. The layer thickness is about 1 µm for all samples.
Figure 10: The AP-PECVD reactor used for studying the deposition of zinc oxide at low substrate temperatures.
A. de Graaf et al. / Energy Procedia 2 (2010) 41–48 47
[1] T. van Mol, F. Grob, K. Spee, K. van der Werf, R. Schropp, “Comparison of Photovoltaic Performance of SnO2:F Coated Substrates made using APCVD with Different Sn Precursors”, Proceedings of EUROCVD-14, 2003.
[2] J. Sundqvist, J. Lu, M. Ottosson, A. Harsta, “Growth of SnO2 thin films by atomic layer deposition and chemical vapour deposition: A comparative study”, Thin Solid Films 514, 2006, pp. 63-68.
[3] M. Losurdo, D. Barreca, P. Capezzuto, G. Bruno, E. Tondello, “Interrelation between nanostructure and optical properties of oxide thin films by spectroscopic ellipsometry”, Surf. Coat. Technol. 151-152, 2002, pp. 2-8.
[4] S. Aukkaravittayapun, N. Wontida, T. Kasecwatin, S. Charojrochkul, K. Unnanon, P. Chindaudom, “Large scale F-doped SnO2 coating on glass by spray pyrolysis”, Thin Solid Films 496, 2006, pp. 117-120.
[5] A.M.B. van Mol, G. R. Alcott, M. D. Allendorf, “Tin oxide precursor chemistry and its link to coating properties”, American Ceramic Soc. Bull. 84, 2005, pp. 37-41.
[6] J. van Deelen, F. Grob, K. Spee, A. Kuypers, “High quality SnO2:F produced by moving beld APCVD”, Proceeding 23rd European Photovoltaic Solar Energy Conference 2008, 2510-2512.
[7] J. Krc, M. Zeman, F. Smole, M. Topic, “Optical modelling of thin film silicon solar cells deposited on textured substrates”, Thin Solid Films 451-452, 2004, pp. 298-302.
[8] T. Oyama, M. Kambe, N. Taneda, K. Masumo, “Requirements for TCO substrate in Si-based thin film solar cells – toward tandem“, Mater. Res. Soc. Symp. Proc. 1101, 2008, pp. 1101-KK02-01.
[9] D. Davazoglou, “Optical properties of SnO2 thin films grown by atmospheric pressure chemical vapour deposition oxidizing SnCl4”, Thin Solid Films 302, 1997, pp. 204-213.
[10] P. Rajaram, Y. C. Goswami, S. Rajagopalan, “Optical and structural properties of SnO2 films grown by a low-cost CVD technique”, Materials Letters 54, 2002, pp. 158-163.
[11] J. Jeong S.P. Choi, K.J. Hong, Y.T. O H.J. Song, J.B. Koo, I.H. Lee, J.S. Park, D.C. Shin, “Atomic scale faceting and its effect on the grain size distribution of SnO2 thin films during deposition”, Mater. Sci. Eng. B 110, 2004, pp. 240-242.
[12] M. D. Allendorf, “Research needs for coatings on glass. Summary of the US Department of Energy roadmapping workshop”, Thin Solid Films 392, 2001, pp. 155-163.
[13] A.M.B. van Mol, Y. Chae, A. H. McDaniel, M.D. Allendorf, “Chemical vapor deposition of tin oxide: fundamentals and applications”, Thin Solid Films 502, 2006, pp. 72-78.
[14] Y. Chae, W. G. Houf, A. H. McDaniel, M. D. Allendorf, “Models for the chemical vapor deposition of tin oxide from monobutyltin trichloride”, J. Electrochem. Soc. 153, 2006, pp. C309-C317.
[15] C. Moralez, H. Juarez, T. Diaz, Matsumoto, E. Rosendo, G. Garcia, M. Rubin, F. Mora, M. Pacio, A. Garcia, “Low temperature SnO2 films deposited by APCVD”, Microelectronics Journal 39, 2008, pp. 586-588.
[16] Y. Matsui, M. Mitsuhashi, Y. Yamamoto, S. Higashi, “Influence of alcohol on grain growth of tin oxide in chemical vapor deposition”, Thin Solid Films 515, 2007, pp. 2854-2859.
[17] H. Huang, O.K. Atan, Y.C. Lee, M. S. Tse, “Preparation and characterisation of nanocrystalline SnO2 thin films by PECVD”, J. Crystal Growth 288, 2006, pp. 70-74.
[18] I. Volintiru, M. Creatore, M.C.M. van de Sanden, “In situ spectroscopic ellipsometry growth studies on the Al-doped ZnO films deposited by remote plasma-enhanced metalorganic chemical vapor deposition”, J. Appl. Phys. 103, 2008, pp. 033704.
[19] P. Poodt, H. Winands, S. Kouijzer, P. Grauls, P. Blom, M. Theelen, J. van Deelen, A. de Graaf, “Atmospheric pressure plasma enhanced chemical vapor deposition for flexible solar cells”, To be published in Proceedings of CAPPSA 2009.
[20] M. Li, J. F. Sopko, J. W. McCamy, “Computational fluid dynamic modeling of tin oxide deposition in an impinging chemical vapor deposition reactor”, Thin Solid Films 515, 2006, pp. 1400-1410.
[21] A.M.B. van Mol, J.P.A.M. Driessen, J. L. Linden, M.H.J.M. de Croon, C.I.M.A. Spee, J.C. Schouten, Vapour pressures of precursors for the CVD of titanium nitride and tin oxide, Chem. Vap. Deposition 7, 2001, pp. 101-104.
48 A. de Graaf et al. / Energy Procedia 2 (2010) 41–48
