Amorphous and Microcrystalline Silicon

Part of the Springer Handbooks book series (SPRINGERHAND)


Processes used to grow hydrogenated amorphous silicon (a-Si : H) and microcrystalline silicon (μc-Si : H) from SiH4 and H2 ∕ SiH4 glow discharge plasmas are reviewed. Differences and similarities between growth reactions of a-Si : H and μc-Si : H in a plasma and on a film-growing surface are discussed, and the process of nucleus formation followed by epitaxial-like crystal growth is explained as being unique to μc-Si : H. The application of a reaction used to determine the dangling-bond defect density in the resulting a-Si : H and μc-Si : H films is emphasized, since it can provide clues about how to improve the optoelectronic properties of those materials for device applications, especially thin-film silicon-based solar cells. Material issues related to the realization of low-cost and high-efficiency solar cells are described, and finally recent progress in this area is reviewed.

Hydrogenated amorphous silicon (a-Si:H) and microcrystalline silicon (μc-Si:H) are recognized as being useful materials for constructing devices related to optoelectronics, such as solar cells, thin-film transistors, etc. [25.1, 25.2]. Several methods have been proposed for the preparation of device-grade a-Si:H and μc-Si:H. These include: reactive sputtering of a crystalline silicon target with Ar+H2 plasma [25.3]; mercury-sensitized photochemical vapor deposition (CVD ) utilizing a decomposition reaction of silane (SiH4) molecules with photoexcited Hg (Hg*) [25.4]; a direct photo CVD method where high-energy photons from a Xe-resonance lamp or a low-pressure Hg lamp are used for the direct excitation of SiH4 molecules to excited electronic states [25.5, 25.6]; a hot-wire CVD method for decomposing SiH4 by means of catalytic reactions on a heated metal surface [25.7]; and a plasma-enhanced CVD method (PECVD ). The PECVD method is the most widely used of these due to its ability to consistently prepare uniform, high-quality materials on a large-area substrate.

In this chapter, the PECVD method is highlighted, details regarding the processes used to grow a-Si:H and μc-Si:H from reactive plasmas are explained, and the determination reaction (which is used to obtain the dangling-bond defect density in the films: one of the most important structural properties that influences device performance) is interpreted in order to obtain clues about how to control the optoelectronic properties of those materials for device applications.

25.1 Reactions in SiH4 and SiH4 ∕ H2 Plasmas

The initial event required for the growth of a-Si:H and μc-Si:H is the decomposition of the source gas material in SiH4 or SiH4/H2 glow discharge plasma. Figure 25.1 shows a schematic of the dissociation pathway of SiH4 and H2, during which the molecules are excited to higher electronic states due to inelastic collisions with high-energy electrons in the plasma [25.8]. As the electrons in the plasma usually have a wide variety of energies, from zero to several tens of electron volts (eV), ground-state electrons of source gas molecules are excited into their electronic excited states almost simultaneously due to inelastic collisions with energetic electrons. Excited electronic states of complicated molecules like SiH4 are usually dissociating states, from which dissociation occurs spontaneously to SiH3, SiH2, SiH, Si, H2 and H, as shown in Fig. 25.1, depending on the stereochemical structure of the excited state. Hydrogen molecules are also decomposed to atomic hydrogen. Excitation of ground-state electron to vacuum-state gives rise to ionization events, generating new electrons and ions, which maintains the plasma.
Fig. 25.1

Schematic showing the dissociation of SiH4 and H2 molecules to a variety of chemical species in the plasma via excited electronic states. The electron energy distribution function in the plasma is also shown

Reactive species produced in the plasma also experience secondary reactions, mostly with parent SiH4 and H2 molecules, as shown in Fig. 25.2 , resulting in a steady state. Reaction-rate constants for each reaction are summarized in the literature [25.9]. Steady-state densities of reactive species are basically determined by the balance between their generation and annihilation rate. Therefore, highly reactive species such as SiH2, SiH and Si (short-lifetime species) have much smaller densities than SiH3 in the steady-state plasma, although the generation rates of those species are not very different from that of SiH3 showing low reactivity along with SiH4 and H2 (long-lifetime species).
Fig. 25.2

Representative secondary reactions of the chemical species produced in the plasma with SiH4 and H2 molecules. Their reaction rate constants are available in the literature

Steady-state densities of reactive species have been measured using various gas-phase diagnostic techniques [25.10, 25.11, 25.12, 25.13, 25.14, 25.15, 25.16], such as optical emission spectroscopy (OES) [25.10], laser-induced fluorescence (LIF) [25.12], infrared laser absorption spectroscopy (IRLAS) [25.13], and ultraviolet light absorption spectroscopy (UVLAS) [25.16]. Figure 25.3 shows the steady state number densities of the chemical species, including both emissive and ionic species, in the SiH4 and SiH4 ∕ H2 plasmas used to prepare device-grade a-Si:H and μc-Si:H. It is clear from Fig. 25.3 that the SiH3 radical is the dominant chemical species in the growth of both a-Si:H and μc-Si:H, although the density ratio of short-lifetime species to SiH3 changes depending on the conditions used to generate the plasma. For instance, when high electric power is supplied to the plasma (high electron density) under low SiH4 flow rate conditions, SiH4 is rapidly dissociated and then depleted, giving rise to reduced probabilities of gas-phase reactions of short lifetime species with SiH4 molecules. This leads to an increased contribution of short-lifetime species to the film growth, which causes a deterioration of the structural properties of the resulting films.
Fig. 25.3

Number densities of chemical species in realistic steady-state plasmas measured or predicted by various diagnostic techniques

The steady-state density of atomic hydrogen (H) varies widely in the plasma, as shown in Fig. 25.3 . This is mainly due to the change in the hydrogen dilution ratio R (H2/SiH4) in the source gas material, i. e., the density of atomic hydrogen increases with increasing R. Noting the fact that more μc-Si:H is formed with increasing R at constant electron density in the plasma and constant substrate (surface) temperature, it is clear that atomic hydrogen plays an important role in μc-Si:H growth [25.17] although SiH3 is the dominant film precursor for both a-Si:H and μc-Si:H growth [25.18].

25.2 Film Growth on a Surface

25.2.1 Growth of a-Si:H

Upon reaching a film-growing surface, some of the SiH3 is reflected at the surface, while the remaining SiH3 is adsorbed on the surface. Some adsorbed SiH3 molecules interacts with the surface hydrogen at its landing site, forming SiH4 and leaving dangling bonds on the surface. This is known as growth-site formation. Other adsorbed SiH3 radicals begin to diffuse across the surface to find the site containing the dangling bonds, whereupon Si − Si bond formation (film growth) occurs, as shown schematically in Fig. 25.4. This surface reaction scheme for film growth has been proposed on the basis of two experimental results [25.18].
Fig. 25.4

Schematic of the surface growth of a-Si:H. The large spheres and small spheres represent Si and H, respectively

Figure 25.5 shows the general concept of the surface-reaction process. Some the flux of SiH3 is reflected off the surface (the proportion of molecules reflected is given by the reflection probability). The remaining SiH3 is adsorbed onto the surface and it changes its form as follows:
  1. 1.

    SiH3 abstracts bonded H from the surface, forming SiH4, or two of the SiH3 radicals interact on the surface, forming Si2H6 (with a recombination probability γ);

  2. 2.

    Surface-diffusing SiH3 sticks to the site containing the dangling bond, forming Si − Si bond (with a sticking probability s).

The total loss probability (β )  is given by the sum of recombination probability and the sticking probability (γ + s), and the reflection probability is therefore 1 − β. Among the reaction probabilities mentioned above, the loss probability has been measured using a grid method (GM ) as well as a step-coverage method (SCM ). Figures 25.6a and 25.6b show the loss probabilities obtained by GM and SCM, as well as the sticking probability predicted from the film deposition rate as a function of the substrate temperature, respectively.
Fig. 25.5

General concepts behind the surface reactions of incoming SiH3 radicals

Fig. 25.6

Loss probability β (a) of SiH3 radicals reaching the film-growing surface, as measured using the grid method and the step-coverage method, and the deposition rate (corresponding to the sticking probability s for SiH3 )  (b) as a function of substrate temperature during the growth of a-Si:H

The loss probability β, related to the reflectivity (1 − β), is simply dependent on the nature of the site where SiH3 from the plasma lands. Therefore, the temperature-independent 1 − β seen in Fig. 25.6 suggests that almost all of the surface sites are covered with bonded H over the whole temperature range used in the experiment, although a few dangling bonds are created thermally above 350C, as shown by in situ infrared reflection absorption spectroscopy (IRRAS) [25.19]. On the other hand, SiH3 diffusing across the surface is easily captured by even just a few dangling bonds, because a mobile species on the surface can find a specific site at a distant location from the landing site.

The temperature-independent behavior of 1 − β and the temperature-dependent behavior of s observed in Figs. 25.6a and 25.6b have been perfectly reproduced by theoretical simulations based on the surface-reaction scheme shown in Fig. 25.4, assuming reasonable activation energies for the surface diffusion of SiH3, the abstraction reaction of H with SiH3, the saturation of the site containing the dangling bond with SiH3, and thermal H-removal processes [25.20].

25.2.2 Growth of μc-Si:H

Atomic hydrogen reaching the film-growing surface plays an important role in the growth of μc-Si:H. This has been confirmed in the μc-Si:H-formation map drawing on the radio frequency (RF ) power density/hydrogen dilution ratio plane as shown in Fig. 25.7. As seen in the figure, large-crystallite μc-Si:H is prepared under high hydrogen dilution conditions, indicating the importance of atomic H in the growth of μc-Si:H. Figure 25.8 shows the crystalline volume fraction (determined by the x-ray diffraction peak area) in the resulting μc-Si:H as a function of the substrate temperature during film growth for three different hydrogen dilution ratios [25.17]. The crystalline volume fraction increases with increasing substrate temperature, reaching a maximum at around 350C, and then suddenly drops to zero above 500C. The lack of μc-Si:H formation above 500C suggests that surface hydrogen coverage is a requirement for crystallite formation in the resulting film [25.17].
Fig. 25.7

Crystal size (light: δ )  and volume  (dark: Xc )  of microcrystallites in the resulting films, mapped out on the RF power density/hydrogen dilution ratio plane

Fig. 25.8

Volume (Xc) of microcrystallites in the resulting films plotted against the substrate temperature during film growth

Figure 25.9 shows the negative effect of ionic species impinging on the film-growing surface on the formation of μc-Si:H, studied using a triode reactor, as shown in the inset of the figure [25.21]. The Raman crystallinity Ic/Ia, defined as the ratio of the peak intensity from the crystalline phase at around 520 cm−1 to that from the amorphous phase at 480 cm−1 in the Raman scattering spectrum, deteriorates as the energy of the ionic species impinging on the film-growing surface (controlled by the bias voltage applied to the mesh electrode) is increased.
Fig. 25.9

Raman crystallinity in the resulting μc-Si:H as a function of the energy of the ions impinging on the film-growing surface (which is controlled by the application of the positive bias voltage to the mesh electrode during film growth)

The surface reaction behavior of SiH3 reaching the film-growing surface has also been investigated for μc-Si:H growth using GM and SCM, in contrast to a-Si:H growth [25.18]. Figures 25.10a and 25.10b show the loss probability (β) and the deposition rate (corresponding to the sticking probability s) of SiH3 reaching the film-growing surface as a function of the substrate temperature for the growth of both μc-Si:H and a-Si:H. Unlike in the case of a-Si:H growth, the loss probability β shows a significant dependence on substrate temperature and the sticking probability s shows no temperature dependence in the case of μc-Si:H growth.
Fig. 25.10

Loss probability β and deposition rate as a function of substrate temperature for μc-Si:H growth in comparison to those for a-Si:H growth

Based on these experimental results, the following properties of the formation process of μc-Si:H can be specified:
  1. 1.

    The film precursor is SiH3, the same as in the case of a-Si:H growth.

  2. 2.

    Atomic hydrogen reaching the film-growing surface plays an important role in the formation of μc-Si:H.

  3. 3.

    The film becomes amorphous when the substrate temperature is higher than 500C.

  4. 4.

    High-energy ions impinging on the surface result in crystallinity deterioration.

  5. 5.

    The surface loss probability shows a temperature dependence whereas the sticking probability does not show a temperature dependence in the case of μc-Si:H growth.


Growth Models for μc-Si:H

In an attempt to explain the specific phenomena observed during the formation of μc-Si:H, three models have been proposed:
  1. 1.

    The surface diffusion model [25.17]

  2. 2.

    The etching model [25.22]

  3. 3.

    The chemical annealing model [25.23].

The surface diffusion model is depicted schematically in Fig. 25.11. Here, a high atomic H flux from the plasma results in full bonded hydrogen surface coverage and also local heating through hydrogen exchange reactions on the film-growing surface. These two actions enhance the surface diffusion of film precursors (SiH3). As a consequence, the SiH3 adsorbed on the surface can find energetically favorable sites, leading to the formation of an atomically ordered structure (nucleus formation). After the formation of the nucleus, epitaxial-like crystal growth takes place with enhanced surface diffusion of SiH3 [25.17, 25.24].
Fig. 25.11

Surface diffusion model for μc-Si:H formation. The large spheres and small spheres represent Si and H, respectively

The etching model has been proposed due to the observation that the rate of film growth decreases with increasing hydrogen dilution ratio R. The concept behind the etching model is shown schematically in Fig. 25.12. Atomic H reaching the film-growing surface breaks Si − Si bonds, preferentially weak bonds involved in the amorphous network structure, leading to the removal of Si atoms weakly bonded to other Si atoms. This site is replaced with a new film precursor SiH3, creating a rigid, strong Si − Si bond, which gives rise to an ordered structure [25.22, 25.24].
Fig. 25.12

Etching model for μc-Si:H formation

The chemical annealing model has been proposed in order to explain the observation that crystal formation is observed during hydrogen plasma treatment; growth occurs layer-by-layer via an alternating sequence of thin amorphous film growth and hydrogen plasma treatment. Several monolayers of amorphous silicon are deposited, and these layers are exposed to hydrogen atoms produced in the hydrogen plasma. This procedure is repeated several tens of times in order to fabricate the proper thickness to be able to evaluate the film structure. The absence of any significant reduction in film thickness during the hydrogen plasma treatment is difficult to explain using the etching model, and so the chemical annealing model was proposed, as schematically shown in Fig. 25.13. During the hydrogen plasma treatment, many atomic hydrogens permeate through the subsurface (the growth zone), giving rise to the crystallization of an amorphous network through the formation of a flexible network without any significant removal of Si atoms [25.23, 25.24].
Fig. 25.13

Chemical annealing model for μc-Si:H formation

These three models have been carefully examined, and the merits and drawbacks of each model have been discussed [25.24, 25.25].

More microscopic observations have recently been reported, based on the use of in situ diagnostic techniques, and a detailed mechanism for the formation process of μc-Si:H has been proposed.

Formation of the Nucleus

Figure 25.14 shows the evolution in surface roughness during film growth, as obtained by spectroscopic ellipsometry (SE ), for three hydrogen dilution ratios R of 0, 10 and 20 [25.26]. As is seen in the figure, after the formation of an island, the enforced coalescence of islands takes place, which results in a smooth surface under μc-Si:H-growth conditions (R = 20). After the smooth surface is obtained (formation of the nucleus is confirmed at this point in time), surface roughness is then enhanced due to an orientation-dependent crystal growth rate [25.27]. As soon as the smooth surface appears, a particular surface absorption band is observed in the infrared absorption spectrum, as measured using the in situ attenuated total reflection technique (ATR) during film growth [25.27, 25.28]. Figure 25.15 shows the surface infrared absorption spectrum, showing the presence of specific bands at 1897 cm−1 and 1937 cm−1 together with the usual Si − H x surface and bulk absorption bands (which occur between 2000 cm−1 and 2150 cm−1). This new absorption band is assigned to the SiH2(Sid) complex, which is also sketched in Fig. 25.15 . Note that the number density (absorption intensity) of the SiH2(Sid) complex is found to be proportional to the magnitude of the internal stress in the film just before these complexes appear. A nucleation model has been proposed based on the experimental facts mentioned above. The enforced island coalescence due to the enhanced surface diffusion of SiH3 gives rise to an internal stress involving many strained Si − Si bonds in the amorphous incubation layer [25.27]. Atomic hydrogen attacks the strained Si − Si bonds, forming SiH2(Sid) complexes on the film-growing surface. These complexes provide structural flexibility, which enables structural order to be obtained via successive Si − SiH3 bond formation at this site; in other words it acts as a prenucleation site on the film-growing surface.
Fig. 25.14

Evolution in surface roughness, as measured using spectroscopic ellipsometry, during film growth for three different Rs

Fig. 25.15

Surface infrared absorption spectrum from the film just before nucleus formation, showing the appearance of the Si − H2(Sid) complex whose structure is also shown

Epitaxial-Like Crystal Growth

Figure 25.16 shows a cross-sectional transmission electron microscope (TEM ) image of typical μc-Si:H films deposited on a glass substrate [25.29]. In the figure, epitaxial-like crystal growth is clearly observed to occur from the nucleus. It is a well-known fact that epitaxial crystal growth occurs only when the surface diffusion length of the film precursor is sufficiently long. In order to investigate the origin of the enhanced surface diffusion of SiH3 during the formation of μc-Si:H, an isotope labeling experiment has been carried out [25.30]. D2 was used as the source gas material (D2 ∕ SiH4) instead of H2 ∕ SiH4 during film growth under constant substrate temperature conditions, and the number densities of both D and H incorporated into the resulting film are measured by infrared absorption spectroscopy in order to estimate the degree of H-to-D exchange reactions, which mostly occur on the film-growing surface (and the first layers below).
Fig. 25.16

Cross-sectional transmission electron microscope image of typical μc-Si:H deposited on a glass substrate

Figure 25.17 shows the Raman crystallinity Ic ∕ Ia plotted against the number density ratio of the D/H incorporated in the resulting films. As is clearly from the figure, μc-Si:H is obtained only when D/H exceeds a critical value, indicating that the D/H exchange reaction is required to some extent for the formation of μc-Si:H. The D/H exchange reaction involves two steps:
  1. 1.

    Atomic D reaching the film-growing surface abstracts bonded H on the surface (the Eley-Rideal reaction), forming HD and providing a Si dangling bond site;

  2. 2.

    A recombination reaction occurs between the dangling bond and another atomic D adsorbed onto the surface (including the subsurface), forming a Si − D bond.

These abstraction and recombination reactions are known to be strong exothermal reactions, with 1.4 eV and 3.1 eV of energy released, respectively. Considering the highly exothermal nature of the D/H exchange reaction, and the existence of a threshold value for the degree of D/H exchange reactions required for the formation of μc-Si:H, as shown in Fig. 25.17 , local heating on the film-growing surface is believed to play a major role in the enhanced surface diffusion of the film precursor (SiH3) during both nucleus formation (strained bond formation and SiH2(Sid)-complex formation) and the epitaxial-like crystal growth associated with μc-Si:H film growth [25.17, 25.24, 25.28, 25.31].
Fig. 25.17

Raman crystallinity plotted against the number density ratio of D/H incorporated into the resulting films prepared from D2/SiH4 plasmas

25.3 Defect Density Determination for a-Si:H and μc-Si:H

One of the most important structural properties of a-Si:H and μc-Si:H for device applications is their dangling bond defect densities, because each dangling bond creates a localized deep state in the band gap of the material, which acts as a recombination center for photoexcited electrons and holes, although the free carrier mobility is another important property in semiconductors.

25.3.1 Dangling Bond Defects

Figure 25.18 shows the dangling bond defect density in the resulting a-Si:H as a function of substrate temperature. This dependency of the dangling bond density on the substrate temperature has been explained by taking into account the steady-state dangling-bond density on the film-growing surface. The structural properties of the resulting film are generally largely determined by the steady-state surface properties during the thin film growth process, because the surface formed at any given instant is incorporated into the bulk in the next instant due to the successive layering nature of the film growth [25.32, 25.33]. The steady state number density of surface dangling bonds is determined by the balance between the rate of generation of dangling bonds and their annihilation rate.
Fig. 25.18

Dangling bond defect density in the resulting a-Si:H films as a function of the substrate temperature during film growth

At low substrate temperatures, surface dangling bonds are produced by abstraction reactions of surface bonded H atoms with SiH3; the reaction rate for this is almost independent of substrate temperature, since the reaction rate involves two rates, i. e., the adsorption rate of SiH3 on the surface (negative temperature dependence) and dissociation rate of bonded H with SiH3 (positive temperature dependent). The surface dangling bond is saturated by SiH3 diffusing across the surface (this is dangling bond annihilation), which exhibits a slower rate at low temperatures due to the slower surface diffusion of SiH3, which results in high dangling bond density on the steady-state film-growing surface. As a consequence, the dangling-bond density in a-Si:H grown at low substrate temperatures can be as high as 1019 cm−3. When the substrate temperature is increased during film growth, the steady state number density of surface dangling bonds is reduced drastically due to the combination of the temperature-independent H-abstraction reaction and the thermally enhanced surface diffusion of SiH3, which gives rise to a minimum defect density of 1015 cm−3 in the resulting a-Si:H at a substrate temperature of \(\approx{\mathrm{250}}\,{\mathrm{{}^{\circ}\mathrm{C}}}\). The increased dangling bond density of the a-Si:H prepared at substrate temperatures higher than 350C is explained by the increased rate of generation of steady-state dangling bonds due to the addition of a new term in the generation rate of these bonds associated with the thermal removal of surface bonded H, although its annihilation rate strictly increases with increasing substrate temperature [25.33, 25.34].

We note here that a remarkable increase in the contribution from short-lifetime species such as SiH2, SiH and Si, which show high reactivity and no diffusion on the film-growing surface, is observed upon the depletion of SiH4-parent molecules, which causes an increase in the number density of dangling bond defects in the resulting films through an enhancement of the dangling bond generation rate and a reduction of the dangling bond annihilation rate on the film-growing surface [25.20, 25.32].

Based on our understanding of the defect density determination reaction during film growth, several trials have attempted to control the defect density in a-Si:H. The steady state defect density on the film-growing surface could be reduced when the growth rate was much faster than the thermal H removal rate in the substrate temperature range above 350C, where the steady state defect density is mainly determined by the thermal H removal process. Figure 25.19 shows the number density of dangling bond defects in a-Si:H films plotted against their growth rate. As expected, the defect density in a-Si:H shows no growth- rate dependence when the film is prepared at substrate temperatures lower than 300C, whereas the defect density monotonically decreases with increasing growth rate when the film is deposited at 400C and 450C, and a defect density of 1014 cm−3 has been demonstrated [25.35].
Fig. 25.19

Dangling bond defect density in a-Si:H films plotted against their deposition rates for three different substrate temperatures

25.3.2 Dangling Bond Defect Density in μc-Si:H

Figure 25.20 shows typical dangling bond defect densities in μc-Si:H as a function of substrate temperature together with those in its a-Si:H counterpart [25.33, 25.35, 25.36]. As seen from the figure, the substrate temperature-dependent dangling bond defect density in μc-Si:H in the temperature range above 300C shows a similar trend to that for a-Si:H, indicating that the defect determination reaction on the film-growing surface is identical for both μc-Si:H and a-Si:H growth in this temperature range; in other words, the increase in defect density with increasing substrate temperature is controlled by the thermal H removal process on the film-growing surface. On the other hand, the number density of defects in μc-Si:H prepared at low substrate temperatures shows much lower values than those in a-Si:H. Considering that the defect density in the resulting film is largely determined by the defect density on the steady-state film-growing surface, the much lower defect density in μc-Si:H is caused by an increase in the defect annihilation rate (through an enhanced surface diffusion of SiH3 film precursors) on the film-growing surface due to the local heating from H-exchange reactions that occur during the course of μc-Si:H film growth.
Fig. 25.20

Dangling bond defect density in μc-Si:H films as a function of substrate temperature during film growth in comparison to that in a-Si:H

25.4 Device Applications

a-Si:H and μc-Si:H are highly promising materials applicable to electronic or optoelectronic thin-film devices such as thin-film transistors (TFT s), position sensors, color sensors, solar cells, etc. [25.1, 25.2].

A thin-film transistor array with a-Si:H active layer has been developed for switching devices in liquid crystal displays (LCD), and large-area LCDs more than 50 in across have already been made commercially available based on this technology. A laser crystallization technique has been developed in order to increase the carrier mobility of a-Si:H-based TFTs and thus reduce the active area of the transistors in the LCD. LCDs with a-Si:H or laser-crystallized thin-film Si-based transistor arrays are widely used as flat panel displays in televisions and monitors, where they are in competition with plasma display panels (PDP s) and other flat panel display systems. If the carrier mobility in as-deposited μc-Si:H is drastically improved through the enhanced control of the film growth process, μc-Si:H will be widely used not only for thin-film transistors but also for signal-scanning devices such as charge-coupled devices monolithically arranged in the periphery of the LCD.

Thin-film Si-based solar cells has also been widely expected to provide low-cost photovoltaics. Actually, a-Si:H-based solar cells have already been widely used in pocket calculators, and now large-area solar cells are being developed for electricity generation. One big advantage of using a-Si:H for solar cell applications is its large optical absorption coefficient compared to single-crystalline or polycrystalline silicon counterparts, showing indirect optical transition properties, and so a thickness of less than 1 μm is enough to absorb sufficient sunlight for electricity generation when using a-Si:H-based solar cells. Low-temperature processes using PECVD are also advantageous in terms of reducing the cost of producing a-Si:H-based solar cells. However, there is a well-known phenomenon that occurs in a-Si:H called photo-induced degradation [25.37], where the conversion efficiency of a-Si:H-based solar cells, usually ≈ 10%, is degraded to less than 8% after prolonged exposure to light. Recombinations photogenerated electrons and holes are believed to trigger this photoinduced degradation in a-Si:H; therefore, increasing the field in the a-Si:H-based solar cell has proved an effective way to reduce the degradation. In fact, 0.1 μm-thick a-Si:H-based solar cells do not show any photoinduced degradation, although the conversion efficiency naturally deteriorates due to the reduced absorption of sunlight. The fabrication of a tandem-type stacked solar cell structure consisting of a top cell with a thin a-Si:H layer and a bottom cell containing narrow-gap materials such as a-SiGe : H, μc-Si:H and μc-SiGe:H has been proposed as a promising way to overcome photoinduced degradation and so to achieve high conversion efficiency in thin-film silicon-based solar cells. Initially, a-SiGe:H was adopted for the bottom cell material [25.38]; however, this material also shows severe photoinduced degradation. Recently, μc-Si:H or μc-SiGe:H have been proposed as promising candidates for bottom cell materials, because these materials do not exhibit photoinduced degradation [25.39, 25.40]. In this proposal, high rates of growth of those materials are crucial to the low-cost fabrication of tandem-type solar cells, since μc-Si:H and μc-SiGe:H undergo largely indirect optical transitions. Therefore, urgent material issues for the realization of low-cost/high-efficiency thin-film silicon-based solar cells include the need to improve the photoinduced stability of high-quality a-Si:H and the need to achieve a high rate of growth of device-grade μc-Si:H or μc-SiGe:H.

25.5 Recent Progress in Material Issues Related to Thin-Film Silicon Solar Cells

25.5.1 Controlling Photoinduced Degradation in a-Si:H

A relationship between the degree of photoinduced degradation and the dihydride bonding (Si − H2) density has been reported in a-Si:H prepared under a variety of deposition conditions, where the substrate temperature, plasma-excitation frequency, gas-flow rate, hydrogen dilution ratio, working pressure, power density, etc. have all been varied [25.41]. Figure 25.21 shows the degree of photoinduced degradation, defined as the difference in the fill factors of photo-I–V characteristics of Ni-a-Si:H Schottky diode before and after light soaking plotted against the Si − H2 density in a-Si:H film, as measured by infrared absorption spectroscopy. Furthermore, it has been suggested from mass spectrometric results that the Si − H2 density in the resulting a-Si:H is strongly increased by the contributions from higher silane-related species (HSRS ) such as Si4H9 when the substrate temperature is kept constant. The contribution of these HSRS to the film precursor (SiH3) has been theoretically analyzed using a couple of gas-phase reaction-rate equations [25.42]. This analysis predicted that the contribution ratio is a complex function of the electron temperature in the plasma, the electron density in the plasma, the gas temperature, the hydrogen dilution ratio R, and the gas residence time during film growth. Based on our understanding of the network structure (Si − H2 bonding configuration) responsible for the photoinduced degradation in a-Si:H and the chemical species (HSRS) responsible during film growth (shown schematically in Fig. 25.22), a guiding principle for obtaining highly stabilized a-Si:H has been proposed [25.42, 25.43].
Fig. 25.21

Relationship between the degree of photoinduced degradation and the dihydride (Si − H2) content in a-Si:H prepared under various deposition conditions. The star symbol represents for a-Si:H prepared under conditions of a reduced ratio of contributions from higher silane-related species versus the SiH3 contribution at a substrate temperature of 250C

Fig. 25.22

Surface reaction image of the incorporation of Si − H2 bonds into the resulting a-Si:H due to the contributions from higher silane-related species such as Si4H9

By following the guiding principle, a-Si:H with minimized Si − H2 density in the network has been prepared by adjusting the plasma parameters during film growth under high growth rate conditions, and an a-Si:H-based solar cell showing a stable conversion efficiency of 8.2% (Fig. 25.23) has been fabricated at a high growth rate of 2 nm ∕ s [25.41].
Fig. 25.23

Photo-I–V characteristics of a stable n–i–p a-Si:H-based solar cell fabricated at a high growth rate of 2 nm ∕ s

Moreover, a-Si:H containing a Si − H2 density of 0% has been successfully prepared using a triode reactor at substrate temperatures as low as 250C by making use of the difference in the gas phase diffusion coefficients of SiH3 (light) and HSRS (heavy) during film growth. High efficiency stable p–i–n a-Si:H based solar cells (efficiency after light soaking of 10.2% as shown in Fig. 25.24) have been fabricated using the triode reactor, although the growth rate is as low as 0.1 nm ∕ s [25.44].
Fig. 25.24

Photo-I-V characteristics of a stable p–i–n a-Si:H-based solar cell fabricated using a triode reactor

25.5.2 High Growth Rates of Device-Grade μc-Si:H

In the past, conventional high hydrogen dilution methods performed at relatively low working gas pressures (several tens of mTorr) have been used to obtain device-grade μc-Si:H (with low dangling bond defect densities of \(\approx{\mathrm{10^{16}}}/{\mathrm{cm^{3}}}\)) [25.17]. Recently, a simple concept for preparing device-grade μc-Si:H at a high growth rate has been proposed, known as the high-pressure depletion (HPD ) method [25.21, 25.45]. The production rate of the film precursor SiH3 in the plasma, which is proportional to the growth rate of μc-Si:H, is determined by the product of the electron density and the number density of SiH4. To increase the production rate of SiH3 in the plasma, a high power density (radio frequency, RF, or very high frequency, VHF) and high partial pressures are needed, because the electron density is basically a function of the power density applied to the plasma and the number density of SiH4 is proportional to the partial pressure. However, hydrogen atoms (the chemical species responsible during the formation of μc-Si:H) are strongly scavenged by SiH4 molecules during their transportation from their production area to the film-growing surface. In order to enhance the survival of the hydrogen atoms, it has been suggested that the SiH4 molecules should be depleted by applying high power density to the plasma. As conditions of high total pressure are also useful for decreasing the effects of ion bombardment during film growth through the reduction of the electron temperature in the plasma, high working pressures along with SiH4 depletion conditions (HPD) have recently been popularly adopted for the high-rate growth of device-grade μc-Si:H. It has also been suggested that short lifetime species such as SiH2, SiH, Si are scavenged by a large number of H2 molecules present under HPD conditions, although the reaction rate of short lifetime species with H2 is lower than that with SiH4 [25.9], by two order of magnitudes.

Figure 25.25 shows the number density of dangling bond defects, as measured by electron spin resonance (ESR ), in the resulting μc-Si:H as a function of growth rate [25.46]. When the conventional low pressure regime is used for the growth of μc-Si:H, the defect density increases exponentially with increasing growth rate, as seen in the figure (see A). This is caused by both an increase in the ion bombardment (due to the high power density) and an increase in the contributions from short-lifetime chemical species such as SiH2, SiH and Si due to the reduced ability of those species to react with SiH4 and H2 (due to the low pressure and SiH4 depletion) during film growth. On the other hand, the slope in Fig. 25.25 becomes shallower when HPD conditions are used during the growth of μc-Si:H, as shown in the figure (see B), which illustrates the usefulness of the HPD method for obtaining high-quality μc-Si:H at high growth rates.
Fig. 25.25

Relationships between the dangling bond defect density in μc-Si:H films and their deposition rates for (A) μc-Si:H prepared using the conventional low-pressure regime and for (B) μc-Si:H prepared using the high-pressure depletion (HPD) method. The star symbol represents μc-Si:H prepared under HPD conditions with the novel cathode design

The validity of the HPD method has also been demonstrated during the fabrication of μc-Si:H-based solar cells. Figure 25.26 shows the photo-I–V characteristics of a μc-Si:H-based p–i–n single-junction solar cell prepared at a high growth rate of 2 nm ∕ s using the HPD method, which exhibits a reasonably high conversion efficiency of 11.8% [25.47].
Fig. 25.26

Photo-I–V characteristics of a p–i–n μc-Si:H-based solar cell fabricated at reasonably high growth rate of 2 nm ∕ s under HPD conditions

However, HPD conditions require that the spacing between the cathode and anode is reduced in the conventional capacitively coupled plasma reactor, giving rise to nonuniform plasma production and nonuniform film growth. To overcome this problem encountered when using HPD conditions, the structure of the cathode surface has been designed to produce uniform plasma production even in a large-area parallel plate electrode configuration. Figure 25.27 shows the structure of this novel design of cathode. A multitude of holes (hollows) with interconnecting slots are arranged on the cathode surface, which cause strong coupling between the high-density plasmas produced in each hole where source gas injection is performed. Using this new type of cathode, quite high growth rates (more than 8 nm ∕ s) have been obtained along with reasonably low defect densities in the resulting μc-Si:H, as shown by star symbol in Fig. 25.25 [25.48].
Fig. 25.27

Photograph of the surface structure on the cathode designed for the production of high-density/uniform plasmas

25.6 Summary

In this chapter, the processes involved in the growth of a-Si:H and μc-Si:H from SiH4 and SiH4 ∕ H2 plasma have been interpreted in detail. The defect density determination reaction that takes place on the film-growing surface was discussed in order to obtain clues that may lead to enhanced optoelectronic properties in those materials. The recent status of work done in the fields of thin-film transistors and solar cells was reviewed, as these are the main device applications of those materials. Recent progress in resolving material issues related to solar cell applications were also described.

Finally, we note here that the concepts used in and our understanding of the film growth process mentioned here are widely applicable to other processes, especially processes where thin films are grown from reactive plasmas.


  1. 25.1
    W.E. Spear, P.G. LeComber: Solid State Commun. 17, 1193 (1975)CrossRefGoogle Scholar
  2. 25.2
    C.R. Wronski, D.E. Carlson, R.E. Daniel: Appl. Phys. Lett. 29, 602 (1976)CrossRefGoogle Scholar
  3. 25.3
    T. Moustakas: Semiconductors and Semimetals, Vol. 21A (Academic, New York 1984) p. 55Google Scholar
  4. 25.4
    T. Saito, S. Muramatsu, T. Shimada, M. Migitaka: Appl. Phys. Lett. 42, 678 (1983)CrossRefGoogle Scholar
  5. 25.5
    Y. Mishima, M. Hirose, Y. Osaka, K. Nagamine, Y. Ashida, K. Isogaya: Jpn. J. Appl. Phys. 22, L46 (1983)CrossRefGoogle Scholar
  6. 25.6
    T. Fuyuki, K.Y. Du, S. Okamoto, S. Yasuda, T. Kimoto, M. Yoshimoto, H. Matsunami: J. Appl. Phys. 64, 2380 (1988)CrossRefGoogle Scholar
  7. 25.7
    A.H. Mahan, B.P. Nelson, S. Salamon, R.S. Crandall: Mater. Res. Soc. Proc. 219, 673 (1991)CrossRefGoogle Scholar
  8. 25.8
    M. Tsuda, S. Oikawa, K. Saito: J. Chem. Phys. 91, 6822 (1989)CrossRefGoogle Scholar
  9. 25.9
    J. Perrin, O. Leroy, M.C. Bordage: Contrib. Plasma Phys. 36, 3 (1996)CrossRefGoogle Scholar
  10. 25.10
    A. Matsuda, K. Nakagawa, K. Tanaka, M. Matsumura, S. Yamasaki, H. Okushi, S. Iizima: J. Non-Cryst. Solids 35-36, 183 (1980)CrossRefGoogle Scholar
  11. 25.11
    A. Matsuda, K. Tanaka: Thin Solid Films 92, 171 (1982)CrossRefGoogle Scholar
  12. 25.12
    Y. Matsumi, T. Hayashi, H. Yoshikawa, S. Komiya: J. Vac. Sci. Technol. A 4, 1786 (1986)CrossRefGoogle Scholar
  13. 25.13
    N. Itabashi, N. Nishiwaki, M. Magane, T. Goto, A. Matsuda, C. Yamada, E. Hirota: Jpn. J. Appl. Phys. 29, 585 (1990)CrossRefGoogle Scholar
  14. 25.14
    N. Itabashi, N. Nishiwaki, M. Magane, S. Saito, T. Goto, A. Matsuda, C. Yamada, E. Hirota: Jpn. J. Appl. Phys. 29, L505 (1990)CrossRefGoogle Scholar
  15. 25.15
    K. Tachibana, T. Mukai, H. Harima: Jpn. J. Appl. Phys. 30, L1208 (1991)CrossRefGoogle Scholar
  16. 25.16
    A. Kono, N. Koike, H. Nomura, T. Goto: Jpn. J. Appl. Phys. 34, 307 (1995)CrossRefGoogle Scholar
  17. 25.17
    A. Matsuda: J. Non-Cryst. Solids 59-60, 767 (1983)CrossRefGoogle Scholar
  18. 25.18
    A. Matsuda, T. Goto: Mater. Res. Soc. Proc. 164, 3 (1990)CrossRefGoogle Scholar
  19. 25.19
    Y. Toyoshima, K. Arai, A. Matsuda, K. Tanaka: J. Non-Cryst. Solids 137-138, 765 (1991)CrossRefGoogle Scholar
  20. 25.20
    J.L. Guizot, K. Nomoto, A. Matsuda: Surf. Sci. 244, 22 (1991)CrossRefGoogle Scholar
  21. 25.21
    M. Kondo, M. Fukawa, L. Guo, A. Matsuda: J. Non-Cryst. Solids 266-269, 84 (2000)CrossRefGoogle Scholar
  22. 25.22
    C.C. Tsai, G.B. Anderson, R. Thompson, B. Wacker: J. Non-Cryst. Solids 114, 151 (1989)CrossRefGoogle Scholar
  23. 25.23
    K. Nakamura, K. Yoshida, S. Takeoka, I. Shimizu: Jpn. J. Appl. Phys. 34, 442 (1995)CrossRefGoogle Scholar
  24. 25.24
    A. Matsuda: Thin Solid Films 337, 1 (1999)CrossRefGoogle Scholar
  25. 25.25
    K. Saito, M. Kondo, M. Fukawa, T. Nishimiya, W. Futako, I. Shimizu, A. Matsuda: Res. Soc. Proc. Mater. 507, 843 (1998)CrossRefGoogle Scholar
  26. 25.26
    J. Koh, Y. Lee, H. Fujiwara, C.R. Wronski, R.W. Collins: Appl. Phys. Lett. 73, 1526 (1998)CrossRefGoogle Scholar
  27. 25.27
    H. Fujiwara, M. Kondo, A. Matsuda: Surf. Sci. 497, 333 (2002)CrossRefGoogle Scholar
  28. 25.28
    H. Fujiwara, Y. Toyoshima, M. Kondo, A. Matsuda: J. Non-Cryst. Solids 266-269, 38 (2000)CrossRefGoogle Scholar
  29. 25.29
    H. Fujiwara, M. Kondo, A. Matsuda: Phys. Rev. B 63, 115306 (2001)CrossRefGoogle Scholar
  30. 25.30
    S. Suzuki, M. Kondo, A. Matsuda: J. Non-Cryst. Solids 299-302, 93 (2002)CrossRefGoogle Scholar
  31. 25.31
    H. Fujiwara, M. Kondo, A. Matsuda: Jpn. J. Appl. Phys. 41, 2821 (2002)CrossRefGoogle Scholar
  32. 25.32
    A. Matsuda, K. Nomoto, Y. Takeuchi, A. Suzuki, A. Yuuki, J. Perrin: Surf. Sci. 227, 50 (1990)CrossRefGoogle Scholar
  33. 25.33
    G. Ganguly, A. Matsuda: Phys. Rev. B 47, 3361 (1993)Google Scholar
  34. 25.34
    G. Ganguly, A. Matsuda: J. Non-Cryst. Solids 164-166, 31 (1993)CrossRefGoogle Scholar
  35. 25.35
    G. Ganguly, A. Matsuda: Jpn. J. Appl. Phys. 31, L1269 (1992)CrossRefGoogle Scholar
  36. 25.36
    Y. Nasuno, M. Kondo, A. Matsuda: Tech. Digest of PVSEC-12. Jeju (2001) 791Google Scholar
  37. 25.37
    D.L. Staebler, C.R. Wronski: Appl. Phys. Lett. 28, 671 (1977)Google Scholar
  38. 25.38
    K. Saito, M. Sano, K. Matsuda, T. Kondo, T. Nishimoto, K. Ogawa, I. Kajita: Proc. WCPEC-2 Vienna (1998) p. 351Google Scholar
  39. 25.39
    J. Meier, P. Torres, R. Platz, S. Dubail, U. Kroll, J.A. Anna Selvan, N. Pellaton Vaucher, Ch. Hof, D. Fischer, H. Keppner, A. Shah, K.-D. Ufort, P. Giannoules: Mater. Res. Soc. Proc. 420, 3 (1996)CrossRefGoogle Scholar
  40. 25.40
    K. Yamammoto, M. Yoshimi, T. Suzuki, Y. Tawada, Y. Okamoto, A. Nakajima: Mater. Res. Soc. Proc. 507, 131 (1998)CrossRefGoogle Scholar
  41. 25.41
    T. Nishimoto, M. Takai, H. Miyahara, M. Kondo, A. Matsuda: J. Non-Cryst. Solids 299-302, 1116 (2002)CrossRefGoogle Scholar
  42. 25.42
    M. Takai, T. Nishimoto, T. Takagi, M. Kondo, A. Matsuda: J. Non-Cryst. Solids 266-269, 90 (2000)CrossRefGoogle Scholar
  43. 25.43
    M. Takai, T. Nishimoto, M. Kondo, A. Matsuda: Appl. Phys. Lett. 77, 2828 (2000)CrossRefGoogle Scholar
  44. 25.44
    T. Matsui, K. Maejima, A. Bidiville, H. Sai, T. Koida, T. Suezaki, M. Matsumoto, K. Saito, I. Yoshida, M. Kondo: Proc. WCPEC-6 Kyoto (2014), 2TuO.6.2Google Scholar
  45. 25.45
    L. Guo, M. Kondo, M. Fukawa, K. Saito, A. Matsuda: Jpn. J. Appl. Phys. 37, L1116 (1998)CrossRefGoogle Scholar
  46. 25.46
    M. Kondo, T. Nishimoto, M. Takai, S. Suzuki, Y. Nasuno, A. Matsuda: Tech. Digest of PVSEC-12 Jeju (2001) 41Google Scholar
  47. 25.47
    H. Sai, K. Maejima, T. Matsui, T. Koida, M. Kondo, S. Nakao, Y. Takeuchi, H. Katayama, I. Yoshida: Proc. WCPEC-6 Kyoto (2014), 2WeO.14.1Google Scholar
  48. 25.48
    C. Niikura, M. Kondo, A. Matsuda: Proc. WCPEC-3 Osaka (2003), 5P-D4-03Google Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  1. 1.Dept. of Systems InnovationOsaka UniversityOsakaJapan

Personalised recommendations