Processes in silicon deposition by hot-wire chemical vapor deposition

Current Opinion in Solid State and Materials Science 6 (2002) 465–470

Processes in silicon deposition by hot-wire chemical vapor deposition P.A.T.T. van Veenendaal*, R.E.I. Schropp Debye Institute of Physics of Devices, Utrecht University, P.O. Box 80,000, 3508 TA Utrecht, The Netherlands Accepted 16 August 2002

Abstract Hot-wire chemical vapor deposition is a rapidly developing CVD technique for the deposition of silicon thin films and silicon alloys and may become a competitor of the plasma-enhanced (PE) CVD method due to significant advantages such as high deposition rate, efficient source gas utilization, lack of ion bombardment, and low equipment cost. Little is known, however, about the mechanisms for catalytic decomposition of the source gases, gas phase reactions at commonly used pressures, and the growth reactions. In this article, the differences in the reactions at various filament materials are discussed and it is shown that the subsequent reactions in the gas phase and reactions contributing to film growth can be substantially different from those in PE-CVD, due to the lack of energetic electrons and ions. Further work is necessary to identify the role of each precursor for the deposition of amorphous and microcrystalline films.  2002 Elsevier Science Ltd. All rights reserved. Keywords: Hot-wire CVD; Silicon; Filament material; Gas decomposition; Radicals; Gas phase reactions; Film growth

1. Introduction A promising new method to deposit silicon thin films is the hot-wire chemical vapor deposition technique (hot-wire CVD). It was first introduced in 1979 and patented as ‘thermal CVD’ by Wiesmann and co-workers [1,2]. Due to disappointing results Wiesmann and his team ceased to work on the subject. In 1985, Matsumura [3] introduced ‘thermal CVD’, described by Wiesmann, as ‘catalytic CVD’ (Cat-CVD) for the deposition of fluorinated amorphous silicon. Mahan et al. [4] later introduced the term hot-wire chemical vapor deposition (hot-wire CVD). In hot-wire CVD, the reactant gases, e.g. silane (SiH 4 ) and hydrogen (H 2 ), are catalytically decomposed at the surface of a hot filament, with filament temperatures T fil in the range from 1500 to 2000 8C. The main advantages of hot-wire CVD over PE-CVD, which is the commonly used technique to deposit thin silicon films in industry, are (i) absence of ion bombardment, (ii) high deposition rate, (iii) low equipment cost and (iv) high gas utilization (up to 80% [*5]). Possible issues in hot-wire CVD are the control of the substrate temperature and aging of the filaments. With the hot-wire CVD technique, it is possible to deposit a wide variety of silicon morphologies and alloys, *Corresponding author. Tel.: 131-30-253-2345; fax: 131-30-2543165. E-mail address: [email protected] (P.A.T.T. van Veenendaal).

e.g. a-Si:H, mc-Si:H, het-Si:H, poly-Si:H, and SiN x , depending on the filament temperature, pressure, gas flows, and substrate temperature. At Utrecht University, the main focus in hot-wire CVD is on the deposition of a-Si:H [6], poly-Si:H [7] and SiNx [8]. A cross-sectional view of a hot-wire deposition chamber is shown in Fig. 1. This paper will provide an overview of studies regarding the different steps in the deposition and growth of silicon

Fig. 1. Cross-sectional view of hot-wire deposition chamber.

1359-0286 / 02 / $ – see front matter  2002 Elsevier Science Ltd. All rights reserved. PII: S1359-0286( 02 )00104-3

466

P. A.T.T. van Veenendaal, R.E.I. Schropp / Current Opinion in Solid State and Materials Science 6 (2002) 465–470

thin films in hot-wire CVD. After a description of the filament materials used, gas decomposition, evaporated radicals, gas phase reactions and film growth will be described.

one collision with the hot filament (ad ) has been determined by Honda et al. [*5]. This value is derived from the number of silane molecules colliding with the catalyzer surface and the number of deposited silicon atoms. The number of colliding silane molecules per unit time, G, is described as:

2. Filament materials

1 G 5 ]nvp Dfil L 4

Presently, the most frequently used filament materials are tungsten (W) and tantalum (Ta). Apart from these materials, Matsumura [9] reported on the use of molybdenum (Mo), vanadium (V) and platinum (Pt) as filament materials. The main conclusions of this study, with respect to the filament material, were that the film properties are dependent on the filament temperature, but independent of the filament material.. More recently, Duan et al. [10] and Veenendaal et al. [11] used rhenium (Re) as the filament material. Van Veenendaal et al. [11] showed that, except for filament temperature T fil .1950 8C, the crystals in polycrystalline silicon are oriented in the (220) direction. Finally, Morrison et al. [12] reported on the deposition of microcrystalline silicon using graphite as the ¨ catalyzer. Bruhne et al. [13] also reported on the use of graphite for the deposition of microcrystalline silicon with (220) orientation only. The layers deposited contained a considerable amount of carbon.

3. Decomposition at the filament surface In 1988, Doyle et al. [14] performed one of the first investigations on the decomposition of silane at a hot filament, made of tungsten. This investigation showed a linear relationship between the film growth rate (G) and the decomposition efficiency (ad ). In 1991, Horbach et al. [15] found a relationship similar to that found by Doyle et al. [14], but in a higher temperature range. At a filament temperature T fil ,1800 8C the logarithms of both the decomposition coefficient and the growth rate are proportional to 1 /T fil . At T fil .1800 8C they both saturate. This saturation is explained by complete silane decomposition. The dissociation of silane on a hot tungsten surface has been investigated by Tonokura et al. [16]. From this study, it followed that silane dissociates to give H and Si atoms through the following successive surface dissociation reactions: SiH 4 →SiH 3 1H→SiH 2 12H→SiH13H→Si1 4H. It was found that the activation energy for Si atom production is about 234 kJ / mol for the tungsten filament. This activation energy is much lower than the energy required for direct bond breaking that would be needed for the removal of Si from the metal surface. For example, the bond dissociation energy of Si from the tungsten (100) surface is estimated to be about 535 kJ / mol [17]. This low activation energy leads to the conclusion that the process is catalytic of nature. The decomposition probability of one SiH 4 molecule by

(1)

where n, v, Dfil and L are the density of silane molecules, the mean thermal velocity of the molecules, the diameter and length of the filament, respectively. The number of deposited silicon atoms per unit time, G, is described by: G 5 r NAdp D

E f(x) dx /M

Si

(2)

Here, r, NA , D, f(x) and MSi are the density of a-Si:H, Avogadro’s number, the diameter of the reactor tube, the measured deposition rate and the atomic weight of silicon, respectively. The efficiency of gas use for silane, L, is thus given by: G L5] F

(3)

where F is the number of supplied silane molecules per unit time. The number of collisions on the catalyzer surface by one molecule, A, is described by: G A5] (4) F It was found that a silane molecule collides 0.2–2 times with the catalyzer surface. The derived relation between the gas use efficiency L and the silane decomposition probability ad is: 1 2 L 5s1 2 add 4

(5)

The silane decomposition probability at a filament temperature T fil of 2000 8C is about 40%. Another important observation reported by Inoue et al. [**18] is the fact that the dilution of the silane gas with hydrogen does not change the signal intensities of Si, SiH 2 and SiH 3 . It seems that H 2 has no effect on the catalytic decomposition processes of SiH 4 , although hydrogen itself is dissociated at the filament as well, producing atomic hydrogen [19]. Van der Werf et al. [20] showed a decrease in filament temperature upon exposure of the filament to hydrogen. This decrease is explained by the dissociation of hydrogen on the filament surface. Furthermore, it was suggested that the filament is covered by a silicon-rich silicide, which has also been reported by van Veenendaal and co-workers [21,22]. The results of these studies showed an increase in silicon content in the near-surface region of a tungsten filament with time, while the silicon content on a tantalum filament saturates rather quickly (see Fig. 2). Presently, it is suggested [20] that the extent of coverage of the filament with Si primarily affects the

P. A.T.T. van Veenendaal, R.E.I. Schropp / Current Opinion in Solid State and Materials Science 6 (2002) 465–470

Fig. 2. Silicon content in the near-surface region of the filament ( rSi /rM ) as a function of deposition time (t dep ), for different filament materials (M). The lines are guides to the eye.

catalytic dissociation of H 2 molecules and that the reduced primary dissociation of H 2 in SiH 4 / H 2 mixtures promotes the deposition of microcrystalline silicon with a preferred (220) orientation. Thin films with a dominant (220) orientation of the crystals are indeed obtained at tungsten filament temperatures at which silicide formation of the tungsten occurs and where it is now suspected that hardly any H 2 dissociation occurs. Indeed, for Ta filaments, (220) oriented material is reported [23] for filament temperatures below the threshold temperature for H 2 dissociation [20]. Furthermore, the use of graphite filaments is reported to always lead to (220) oriented material [13], while it is known from the field of hot filament deposition of diamond that graphite does not efficiently dissociate H 2 [24,25]. It is thus concluded that both atomic H production and SiH 4 depletion (which prevents the annihilation of atomic H [26]) promote microcrystallinity, however with a random orientation, while a limited atomic H production promotes (220) oriented material. Two processes are suggested as an explanation for these significant differences in silicide formation on W and Ta: (1) the catalytic dissociation of the reactant gases at a tantalum surface is different from that at a tungsten surface, and (2) the formation of a silicide-like alloy is inhibited more on the surface of the tantalum filament than on a tungsten filament.

4. Evaporated radicals As was shown previously, for tungsten filaments at sufficiently high filament temperature, the silane is fully cracked into one Si and four H atoms. Only at temperatures below 1700 K, SiH 2 and SiH 3 could be detected [**18]. It is suggested that in this temperature regime, a Si / W alloy is formed on the filament [27]. It is presumed that this alloy affects the decomposition of silane at the filament surface and virtually blocks the decomposition of

467

H 2 . Matsumura [22] also found that at filament temperatures above 1700 K, the major species desorbed from the filament is the Si atom. The maximum production of Si atoms is observed at about 1800 K. The latter results were obtained with W, Mo and Ta filaments. The Si atom is the only major species above T fil 51700 K for all three filaments. The temperature dependence below T fil 51700 K is large and different for these three filaments. Activation energies for Si atom desorption from the filament below 1700 K are found to be (251663), (96625) and (71620) kJ / mol for Mo, Ta and W filaments, respectively [**18]. This difference in activation energies again indicates that the decomposition of SiH 4 on the hot filament is not caused by thermal decomposition, but by catalytic reactions at the filament surface.

5. Gas phase reactions At low pressures (,5 mbar), the Si and H atoms that come from the filament thermally diffuse to the substrate [*5], with only few or no gas phase reactions. Duan et al. [28] reported on single photon ionization mass spectrometry measurements at 1.8310 22 mbar at W filament temperatures of 1950 8C. The major silicon containing gas species detected is elemental Si itself, along with minor contributions of SiH 3 and Si 2 H x . However, these pressures are several orders lower than the pressures used during practical silicon deposition, e.g. the pressure used during deposition is in the order of 0.1 mbar. At higher pressures (.5 mbar), the silicon atom is highly reactive. It can abstract an H atom from silane, resulting in SiH and SiH 3 , or it can insert into a Si–H bond [29]. Molenbroek [29] described three possible insertion reactions, namely: Si 1 SiH 4 → SiH 1 SiH 3

(6)

Si 1 SiH 4 → HSiSiH *3

(7)

Si 1 SiH 4 → 2SiH 2

(8)

Because the first and third reactions are endothermic, they are unlikely to occur. HSiSiH 3 is formed through an exothermic reaction and will thus be the most probable species to exist. The formation reaction of HSiSiH 3 has been the subject of ab-initio molecular orbital calculations by Sakai et al. [30]. According to these calculations, triplet Si atoms as well as singlet Si atoms react with SiH 4 to yield HSiSiH 3 . Since Si atoms produced at the hot filament should be triplets, the formation reaction is spin forbidden if HSiSiH 3 is in singlet manifold. There are two possibilities: (i) singlet HSiSiH 3 is produced by the nonadiabatic reaction pathway from Si( 3 P)1SiH 4 triplet surface, or (ii) triplet HSiSiH 3 is generated but it is relaxed to singlet manifold by collisions with a third body. It follows,

468

P. A.T.T. van Veenendaal, R.E.I. Schropp / Current Opinion in Solid State and Materials Science 6 (2002) 465–470

that HSiSiH 3 is unstable and that it will react with SiH 4 in the gas phase. There are three possible reactions, namely [**18]: HSiSiH 3 1 SiH 4 ( 1 M) → Si 3 H 8 ( 1 M)

(9)

HSiSiH 3 ( 1 M) → H 2 SiSiH 2 ( 1 M)

(10)

HSiSiH 3 1 SiH 4 → SiH 2 1 Si 2 H 6

(11)

In reactions (9) and (10), M stands for a third body (e.g. an atom or molecule). Up to now, no Si 3 H 8 has been detected. Therefore, Inoue et al. [**18] suggest that reaction (9) can be neglected. However, Molenbroek [29] suggests that abstracting H 2 from the Si 3 H 8 molecule forms Si 3 H 6 . The reaction product of reaction (10), H 2 SiSiH 2 , is a rather stable closed shell molecule and it has been expected to be an important precursor species for the film growth. In the experiments described by Inoue et al. [**18], the most prominent species detected is Si 2 H 6 . They suggested that SiH 2 , produced via reaction (11), further reacts with SiH 4 , according to: SiH 2 1 SiH 4 ( 1 M) → Si 2 H 6 ( 1 M)

(12)

The presence of atomic hydrogen in the reactor, results in the occurrence of the following reaction: H 1 SiH 4 → SiH 3 1 H 2

(13)

The SiH 3 species does not react with SiH 4 and the only gas phase reaction of SiH 3 is self-recombination. Gallagher [*31] also proposed a gas phase growth reaction, in which Si atoms react with silane: Si 1 SiH 4 → Si 2 H *4 → 2SiH 2 , SiH 1 SiH 3 , Si 2 H 2 1 H 2 , Si 2 H 4

(14)

where Si 2 H 4* is an unstable intermediate. Some of the reaction products will react with silane to produce more stable silanes, such as Si 2 H 6 and Si 3 H 6 . Goodwin [32] pointed out, that the reaction yielding Si 2 H 2 1H 2 is energetically favored, since it is an exothermic one by 110 kJ / mol. The relative contribution of Si 2 H 2 , being a closedshell molecule, to film growth is however still very uncertain as the reaction probability is unknown. From a theoretical point of view, the main gas phase reaction species are thus: SiH 3 , Si 2 H 6 , Si 3 H 6 and H 2 SiSiH 2 . It is expected that the detection of the actual gas phase reaction species will take place in the near future.

same. The film surface will be mostly H covered, with approximately the same ratio R between the amount of dangling (Si–) and hydrogen passivated bonds (Si–H) as in the vapor. There, this ratio is n radical /n silane . This follows from the fact that the most frequent gas–surface collisions are with SiH 4 , since silane is the most abundant gas in the reactor. This dynamic equilibrium reaction is given by Si 2 1 SiH 4 ↔Si–H 1 SiH 3

(15)

and leads to R ? n radical /n silane . Typical values for n radical / 24 23 n silane are 10 –10 , so a similar value is expected for R. This leads to dangling bonds at an average spacing of 30–100 surface sites or 5–15 nm. In order to grow a compact film it is necessary to overcome the tendency for the incident radicals to strike film-surface peaks more frequently than valleys. This requires not only radical diffusion over distances in excess of dangling bond separations, but also an increased affinity for settling onto valleys. Following radical-film Si–Si bonding, H 2 evolution occurs from the reaction: Si–H 1 Si–H → Si–Si 1 H 2

(16)

within the top few atomic layers of the film. One very fundamental difference between radio frequent (RF) PE-CVD and hot-wire CVD was however disregarded by Gallagher, namely the absence of ions in hotwire CVD. These ions are very important in RF-PE-CVD. They are essential in creating a dense a-Si:H network [33], have a significant contribution to the growth rate [33] and determine the properties of thin film silicon to a large extent [34,35]. A way to compare the two deposition techniques is to study the surface reaction probability, b. A number of groups have reported on values of b for different species and these values are listed in Table 1. In the hot-wire CVD case, the lower value was determined for the deposition of a-Si:H, while the higher value was found for mc-Si:H deposition. The difference in b between RF-PE-CVD and hot-wire CVD can have two causes [**42]: (i) a different radical responsible for the growth or (ii) a changing reactivity of the surface. The first effect follows from the abundance of atomic H present in the reactor, leading to stripping of SiH x . Furthermore, the pressure in PE-CVD is higher, resulting in more gas phase reactions. The latter effect is related to the presence of a physisorbed hydrogen layer at the surface of the growing Table 1 Reported values of surface reaction probability b

6. Film growth Although Matsumura [9] concluded that the deposition process in hot-wire CVD is very different from that in both the conventional thermal CVD as well as in PE-CVD, Gallagher [*31] is convinced that the models used in PE-CVD as well as in hot-wire CVD should be about the

Species

b

RF-PE-CVD Si SiH SiH 2 SiH 3 Hot-wire CVD

1 [37] 1 [37] 0.660.2 [38] 0.1–0.4 [39–41] 0.29–0.54 [**42]

P. A.T.T. van Veenendaal, R.E.I. Schropp / Current Opinion in Solid State and Materials Science 6 (2002) 465–470 Table 2 Reactions of Si, H and SiH 3 on the surface a (17a) (17b) (18a) (18b) (18c)

Si1SiH (s) →SiSiH (s) 5SiH (s) 12d.b. H1SiH (s) →H 2 1d.b. SiH 3 1d.b.→SiH 3(s) Si1d.b.→Si (s) 13d.b. H1d.b.→SiH (s)

a

Subscript (s) refers to a radical bonded to Si in the film. Abbreviation d.b. stands for dangling bond.

silicon layer, as suggested by Matsuda et al. [40]. Radicals arriving at the surface will recombine with the physisorbed hydrogen, resulting in a high surface reaction probability. The large contributions of ions in RF-PE-CVD [**42] will remove this layer, resulting in a lower b. Because there are no ions present in the gas phase during hot-wire CVD, it is likely that such a physisorbed hydrogen layer exists. This layer will then be responsible for the high surface reaction probability in hot-wire CVD. Molenbroek [29] investigated the reactions that can occur at the surface. These reactions are listed in Table 2. Reaction (17a) and (17b) involve the reactions of Si and H with a hydrogen passivated surface, while reactions (18a) and (18b) deal with the direct reaction of SiH 3 , Si and H with a dangling bond. Reactions (17a) and (18b) result in the creation of two new dangling bonds. This means that whether Si reacts with a dangling bond or a Si–H bond, two new dangling are created in the process. If there are enough dangling bonds formed, it is possible for them to react with each other to form Si–Si bonds. Hydrogen atoms will either abstract a H to create a dangling bond (17b), or passivate a dangling bond (18c). A dangling bond can also be passivated by SiH 3 , via reaction (18a). Nozaki et al. [36] performed laser induced fluorescence (LIF) measurements and found that Si is the film growth precursor at low pressures (10 22 mbar). At higher pressures, gas phase reactions lead to a different growth precursor. Two of the proposed candidates are disilene (Si 2 H 4 ) and SiH 3 . In summary, the most likely growth precursors are Si, SiH 3 and Si 2 H 4 , depending on the pressure and filament temperature.

7. Conclusions An overview of studies regarding the different steps in the deposition and growth of silicon thin films in hot-wire CVD has been given. Studies on the decomposition of silane on the filament showed that the process is catalytic of nature. Also, it was shown that the growth rate is proportional to the decomposition probability. This decomposition probability of silane is |40% at a filament temperature of 2000 8C. At filament temperature above 1800 8C, silane is decomposed into Si and 4H. At lower

469

filament temperatures, also SiH 2 and SiH 3 have been detected as decomposition products and below 1600 8C, H 2 is no longer decomposed at tungsten filaments. The types of primary species that are produced at the wire seem to affect the structure of resulting microcrystalline silicon layers. The dominant gas phase reactions are the reaction of Si and H with silane, resulting in SiH 3 , Si 2 H 6 , Si 3 H 6 and H 2 SiSiH 2 . The precursors dominating the film growth are Si, SiH 3 and Si 2 H 4 , depending on pressure and filament temperature.

Acknowledgements This work was supported by the Netherlands Organization for Energy and Environment (NOVEM). We thank all our co-workers for the preparation of numerous samples and for their help with the many characterizations and analysis.

References Papers of particular interest, published within the annual period of review, have been highlighted as: * of special interest; ** of outstanding interest. [1] Wiesmann HJ, Ghosh AK, McMahon T, Strongin M. J Appl Phys 1979;50:3752. [2] Wiesmann HJ. US patent 4 237 150, December 2, 1980. [3] Matusmura H, Tachibana H. Appl Phys Lett 1985;47:833. [4] Mahan AH, Carapella J, Nelson BP, Crandall RS, Balberg I. J Appl Phys 1991;69:6728. [*5] Honda N, Masuda A, Matsumura H. J Non-Cryst Solids 2000;266– 269:100, These authors showed that the decomposition probability of one SiH 4 molecule by one collision with the catalyzer is about 40% at a catalyzer temperature of 2000 8C. [6] Feenstra KF. PhD thesis, Utrecht University, 1998. [7] van Veenendaal PATT. PhD thesis, Utrecht University, 2002. [8] Stannowski B. PhD thesis, Utrecht University, 2002. [9] Matsumura H. J Appl Phys 1989;65(11):4396. [10] Duan HL, Zaharias GA, Bent SF. Mater Res Soc Symp Proc 2001;664:A3.1.1. [11] van Veenendaal PATT, van der Werf CMH, Rath JK, Schropp REI. J Non-Cryst Solids 2002;299–302:1184. [12] Morrison S, Madan A. Proceedings of 17th EPVSEC. 2002: in press. ¨ ¨ [13] Bruhne K, Schubert MB, Kohler C, Werner JH. Thin Solid Films 2001;395:163. [14] Doyle J, Robertson R, Lin GH, He MZ, Gallagher A. J Appl Phys 1988;64(6):3215. [15] Horbach C, Beyer W, Wagner H. J Non-Cryst Solids 1991;137– 138:661. [16] Tonokura K, Inoue K, Koshi M. Thin Solid Films 2002: in press. [17] Wang Y, Pollard R. J Electrochem Soc 1995;142:1712. [**18] Inoue K, Tange S, Tonokure K, Koshi M. Thin Solid Films 2001;395:42, These authors gave an overview of reactions that are possible to occur during the deposition of silicon by hot-wire CVD. [19] Sommer M, Smith FW. J Mater Res 1990;511:2433. [20] van der Werf CHM, van Veenendaal PATT, van Veen MK, Hard-

470

[21]

[22] [23] [24] [25] [26] [27] [28] [29] [30] [**31]

P. A.T.T. van Veenendaal, R.E.I. Schropp / Current Opinion in Solid State and Materials Science 6 (2002) 465–470 eman AJ, Rusche MYS, Rath JK, Schropp REI. In: Extended abstracts of the 2nd international conference on Cat-CVD process, Denver, CO, 2002, Abstract no. 5. van Veenendaal PATT, Rath JK, Gijzeman OLJ, Schropp REI. Polycrystalline semiconductors. VI. Materials, technologies, and large area electronics. In: Bonnaud O, Mohammed-Brahim T, Strunk HP, Werner JH, editors, Solid state phenomena, vol. 53, Uettikon am See, Switzerland: Scitech, 2001, pp. 80–1. van Veenendaal PATT, Gijzeman OLJ, Rath JK, Schropp REI. Thin Solid Films 2001;395:194. Fonrodana M, Soler D, Asensi JM, Bertomeu J, Andreu J. J Non-Cryst Solids 2002;229–302:14. Sommer M, Smith FW. J Mater Res 1990;511:2433. Zeiler E, Schwarz S, Rosiwal SM, Singer RF. Mater Sci Eng A 2002;335:236. Kondo M, Fukawa M, Guo L, Matsuda A. J Non-Cryst Solids 2000;266–269:84. Matsumura H. Jpn J Appl Phys 1998;37:3175. Duan HL, Zaharias GA, Bent SF. Thin Solid Films 2001;395:36. Molenbroek EC. PhD thesis, University of Colorado, 1995. Sakai S, Deisz J, Gordon MS. J Phys Chem 1989;93:1888. Gallagher A. Thin Solid Films 2001;395:25, This author indicated a

[32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [**42]

number of research issues with respect to the growth chemistry of silicon deposited by hot-wire CVD. Goodwin DG. Mater Res Soc Symp Proc 1999;557:79. Hamers EAG, van Sark WGJHM, Bezemer J, Meiling H, van der Weg WF. J Non-Cryst Solids 1998;26:205. Roca i Cabaroccas P, Morin P, Chu V, Conde J, Liu JZ, Park HR, Wagner S. J Appl Phys 1991;69:2942. Matsuda A, Kumagai K, Tanaka K. Jpn J Appl Phys 1983;22:L-34. Nozaki Y, Kongo K, Miyazaki T, Kitazoe M, Horii K, Umemoto H, Masuda A, Matsumura H. J Appl Phys 2000;88:5437. Ho P, Breiland WG, Buss RJ. J Chem Phys 1989;91:2627. Hertl M, Jolly J. J Phys D 2000;33:381. Doughty DA, Doyle JR, Lin GH, Gallagher A. J Appl Phys 1990;67:6220. Matsuda A, Goto T. Mater Res Soc Symp Proc 1990;164:3. Kessels WMM, van der Sanden MCM, Steverens RJ, Schram DC. J Appl Phys 2000;87:3313. Hamers EAG, Fontcuberta i Morral A, Niikura C, Brenot R, Roca i Cabaroccas P. J Appl Phys 2000;88:3674, These authors reported on the surface reaction probabilities for the deposition of a-Si:H by hot-wire CVD and the fact that the deposition of mc-Si:H in hot-wire CVD and PE-CVD occurs via different growth precursors.