**4.2 Hydrogen elimination process during film growth**

312 Solar Cells – Thin-Film Technologies

method with the same i-layer thickness. While further optimization is necessary to achieve higher stabilized efficiency, the result demonstrates the low degradation ratio of the a-Si:H solar cell with improving the stability of the i-layer itself, which is one of the essential

It is observed that the films grown by the triode system contain very low hydrogen concentrations, namely Si-H2 bond densities. Those values change with the distance between the mesh and the substrate where the lowest hydrogen concentration is observed at the largest distance between the mesh and the substrate. In this section, we will discuss the possible mechanism for the reduction of Si-H and Si-H2 bond densities in the triode deposition system.

Fig. 10. Thermal effusion of hydrogen from the a-Si:H films deposited at 110 o

Hydrogen elimination takes place both in a film growth state and in a post annealing state when a substrate temperature is high. To distinguish it in our case, at first, the thermal annealing tests were performed on the a-Si:H films prepared at the low substrate temperature of 110 oC using the diode system. The as-deposited films contain the large initial hydrogen concentrations (*C*H) of c.a. 27 at.%. After the growth, the individual film was kept in the deposition chamber and was annealed for 30 minutes at the certain temperature. The result is shown in figure 10 [Shimizu et al., 2007]. One can see that the hydrogen concentration is reduced at the high annealing temperatures. On the other hand, at the temperature of 250 oC, which is the substrate temperature used in our triode deposition system, no *C*H reduction takes place at least from the bulk. The result shows that under the substrate temperature of 250 oC, the hydrogen elimination process takes place

the sum of the Si-H and the Si-H2 bond densities [Shimizu et al., 2007].

during the film growth, i.e., most likely with gas reactions.

C. The *C*H is

solutions to obtain a stable a-Si:H solar cell.

**4.1 Hydrogen elimination process – post annealing** 

**4. Hydrogen elimination process** 

The possible hydrogen elimination processes during the a-Si:H film growth are the following and are schematically shown in figure 11.


Fig. 11. Schematic of the hydrogen elimination processes during the growth of a-Si:H.

a. Hydrogen abstraction reaction by an atomic hydrogen

Atomic hydrogen exists in an silane plasma [e.g., Matsuda, 2004]. It reacts with a bonded hydrogen of a film and forms H2 molecule, resulting in a hydrogen elimination. The probability of this reaction should be proportional to the flux of atomic hydrogen. In a silane plasma, generated radicals and ions collide with SiH4 molecule of which density is high in the gas phase. When the atomic hydrogen reacts with SiH4, SiH3 radical and H2 molecule are generated at the rate constant of ~ 3×10-12 cm3/s [Kushner, 1988; Perrin et al., 1996]:

$$\rm H + \rm SiH\_4 \rightarrow \rm SiH\_3 + H\_2. \tag{1}$$

The stable H2 molecule does not contribute to the abstraction of the bonded hydrogen. In the triode system, basically no atomic hydrogen is generated but only the collisions take place in the region between the mesh and the substrate, indicating that the density of atomic hydrogen near the substrate is low. Therefore, it is natural to say that the hydrogen elimination process is not dominated by atomic hydrogen in the triode system.

#### b. Spontaneous thermal desorption of surface hydrogen

The hydrogen desorption process from Si-H bond has been studied elsewhere [Toyoshima et al., 1991]. The activation energy of this reaction is estimated as 2 - 3 eV, and the reaction takes place only in the temperature range higher than 400 oC [Beyer & Wagner, 1983]. Therefore, it is unlikely that the spontaneous hydrogen desorption takes place under the substrate temperature of 250 oC as in our case.

Fabrication of the Hydrogenated

radical.

(*m*SinH2n+1) and that of a SiH3 radical (*m*SiH3) is

observed in the triode system [Shimizu et al., 2007].

**4.4 Effect of growth rate** 

Amorphous Silicon Films Exhibiting High Stability Against Light Soaking 315

First of all, the diffusion coefficient of a SiH3 radical is larger than that of a higher-ordered silane radical due to the difference of their mass. The ratio of the mass of a SinH2n+1 radical

Therefore, the diffusion coefficient of a SiH3 radical is *n*-times larger than that of a SinH2n+1

Second, the lifetime of a SiH3 radical is especially long in a SiH4 gas phase. The density of SiH4 molecule under the condition used in this study (0.1 Torr, 250 oC) is ~ 1015 cm-3. The density of SiH3 radicals is the highest among the other generated species, and the value is ~ 1012 cm-3 in an RF silane plasma [Matsuda, 2004]. In the case of VHF plasma, as in our case, the value changes due to a high electron density and a low electron temperature effects. Estimating from the deposition rate, SiH3 density can be one order of magnitude higher than that in an RF plasma, but in any cases, the density is still very low with respect to that of SiH4 molecule. Therefore, most of the generated species collide with SiH4. Here, the SiH3

 SiH3 + SiH4 SiH4 + SiH3. (7) After all, the diffusion length of SiH3 radical is very long according to equations (4) and (5). Under a certain gas flow rate condition, the radicals having small diffusion lengths are pumped out, but a large diffusion length species can still reach to the substrate. Thus, larger the distance between the mesh and the substrate, stronger the diffusion length effect. Therefore, we would like to propose that in the triode configuration, a long lifetime SiH3 radical mainly contribute to the film growth than that in a diode system. If a SiH3 radical sticks to a film surface, a cross-linking reaction takes place because the configuration of a five or six members of ring is easily formed with the SiH3. On the other hand, when a higher-ordered silane radical sticks to the surface, not all of the Si atoms are located in such a configuration due to its steric-hindrance. Thus, the cross-linking reaction can take place partially, resulting in the remaining of hydrogen in the film. Once hydrogen is incorporated into the bulk of a film, it cannot be thermally eliminated at 250 oC as shown in figure 10. Indeed, under the low SiH4 flow rate condition, the higher Si-H and Si-H2 bond densities are

In the case of the triode system, the growth rate is low compared to that observed in a conventional diode system. In our VHF plasma case, the growth rate observed with the diode system is 7.3 Å/s, and that observed in the triode system is 0.7 Å/s at *d*ms = 1 cm, and is 0.2 Å/s at *d*ms = 4 cm. When the growth rate is low, hydrogen concentration of the resulting film can be low when thermal desorption from the surface and the bulk are the dominant hydrogen elimination processes. However, such elimination processes at the

To confirm the effect of the growth rate furthermore, the experiments were performed with installing a second mesh as described in section 3.1.2. Installing the second mesh reduces the growth rate drastically. As shown in Table 1, the observed growth rate with the double mesh at the VHF power of 10 W is c.a. 0.1 Å/s and that with the single mesh at the same VHF power is 0.8 Å/s. Under those conditions, however, almost the same Si-H and Si-H2 bond densities are observed. Since the VHF power is fixed at the same value, the densities of the generated

substrate temperature of 250 oC are unlikely as discussed in the previous sections.

radical does not disappear due to the collision, resulting in its long lifetime:

*m*SinH2n+1 / *m*SiH3 ≈ *n.* (6)

#### c. Hydrogen abstraction reaction by a SiH3 radical

It is reported that the dominant deposition precursor for a-Si:H growth is a SiH3 radical [Matsuda, 2004]. When a SiH3 radical reaches to a growing surface, it physisorbs to one of the surface hydrogen atoms under a certain probability [Perrin et al., 1989; Matsuda et al., 1990].

$$\mathfrak{a} \triangleq \mathrm{SiH} \, + \, \mathrm{SiH}\_3 \, \blackrightarrow \, \mathfrak{s} \, \mathrm{SiHSiH}\_3 \,. \tag{2}$$

The physisorbed SiH3 radical diffuses on the surface with changing the physisorption spot, and it finally captures one of the surface hydrogen, forming SiH4 and it leaves from the surface. As a result, hydrogen is abstracted and a surface Si dangling bond is created. When another surface diffusing SiH3 radical reaches to the surface dangling bond, it is chemisorbed and a Si-Si bond is formed [Perrin et al., 1989; Matsuda et al., 1990]. Note that, if this dangling bond is not terminated with some radicals, defect density is increased in the resulting film. As one can see, as long as a SiH3 radical is supplied to the dangling bond site after the hydrogen elimination by another SiH3, the density of the surface hydrogen does not decrease through the process since the chemisorped SiH3 also contains hydrogen.

#### d. Hydrogen elimination process through a cross-linking reaction

A cross-linking hydrogen elimination reaction takes place with a pair of Si-H bonds facing each other. Hydrogen is eliminated with forming a H2 molecule and a Si-Si bond [Matsuda & Tanaka 1986; Perrin et al., 1989],

$$\mathbf{H} \equiv \mathbf{Si}\text{-}\mathbf{H} \; + \; \mathbf{H}\text{-}\mathbf{Si} \equiv \mathbf{\Rightarrow} \; \mathbf{H}\text{-}\mathbf{Si} \; \mathbf{=} + \; \mathbf{H}\_2 \; \text{.} \tag{3}$$

Different from the other cases (a)–(c), a dangling bond does not remain after this reaction, therefore, an atomic hydrogen or a SiH3 radical which contains hydrogen atom(s) does not stick to this site, resulting in a reduction of hydrogen concentration. As described above, a Si-Si bond is formed through this process, therefore, before the reaction, the both Si atoms should be located at the configuration of five or six members of ring to form a stable Si-Si bond. On the other hand, if two Si atoms are located apart from each other where the free energy of the resulting Si-Si bond including its surrounding structures after the reaction is higher than that of before, it is unlikely that the cross-linking reaction takes place. Such an inhabitation can occur when a higher-ordered silane radical sticks to a growing surface as discussed in the next section.

#### **4.3 Effect of deposition precursors on hydrogen elimination**

For the growth of a-Si:H film, a SiH3 radical is a dominant species. On the other hand, higher-ordered silane radicals are also generated in a plasma through insertion reactions of SiH2 [Takai et al., 2000]. It has been reported that when the higher-ordered silane radicals such as Si4H9 are incorporated into the film, Si-H2 bond density increases [Takai et al., 2000; Nishimoto et al., 2002]. Therefore, the flux of those species toward the substrate determines the property of the resulting film. The diffusion length (*L*) of a species is

$$L = (D\,\tau)^{1/2},\tag{4}$$

$$D \ll 1/m.\tag{5}$$

where *D* is a diffusion coefficient, *t* is a lifetime and *m* is a mass of a species.

It is reported that the dominant deposition precursor for a-Si:H growth is a SiH3 radical [Matsuda, 2004]. When a SiH3 radical reaches to a growing surface, it physisorbs to one of the surface hydrogen atoms under a certain probability [Perrin et al., 1989; Matsuda et al., 1990].

The physisorbed SiH3 radical diffuses on the surface with changing the physisorption spot, and it finally captures one of the surface hydrogen, forming SiH4 and it leaves from the surface. As a result, hydrogen is abstracted and a surface Si dangling bond is created. When another surface diffusing SiH3 radical reaches to the surface dangling bond, it is chemisorbed and a Si-Si bond is formed [Perrin et al., 1989; Matsuda et al., 1990]. Note that, if this dangling bond is not terminated with some radicals, defect density is increased in the resulting film. As one can see, as long as a SiH3 radical is supplied to the dangling bond site after the hydrogen elimination by another SiH3, the density of the surface hydrogen does not decrease through the process since the chemisorped SiH3 also contains hydrogen.

A cross-linking hydrogen elimination reaction takes place with a pair of Si-H bonds facing each other. Hydrogen is eliminated with forming a H2 molecule and a Si-Si bond [Matsuda

Different from the other cases (a)–(c), a dangling bond does not remain after this reaction, therefore, an atomic hydrogen or a SiH3 radical which contains hydrogen atom(s) does not stick to this site, resulting in a reduction of hydrogen concentration. As described above, a Si-Si bond is formed through this process, therefore, before the reaction, the both Si atoms should be located at the configuration of five or six members of ring to form a stable Si-Si bond. On the other hand, if two Si atoms are located apart from each other where the free energy of the resulting Si-Si bond including its surrounding structures after the reaction is higher than that of before, it is unlikely that the cross-linking reaction takes place. Such an inhabitation can occur when a higher-ordered silane radical sticks to a growing surface as

For the growth of a-Si:H film, a SiH3 radical is a dominant species. On the other hand, higher-ordered silane radicals are also generated in a plasma through insertion reactions of SiH2 [Takai et al., 2000]. It has been reported that when the higher-ordered silane radicals such as Si4H9 are incorporated into the film, Si-H2 bond density increases [Takai et al., 2000; Nishimoto et al., 2002]. Therefore, the flux of those species toward the substrate determines

 *D* ∝ 1/*m*. (5)

≡ SiH + SiH3 ≡ SiHSiH3. (2)

≡ Si-H + H-Si ≡ ≡ Si-Si ≡ + H2. (3)

)1/2, (4)

c. Hydrogen abstraction reaction by a SiH3 radical

d. Hydrogen elimination process through a cross-linking reaction

**4.3 Effect of deposition precursors on hydrogen elimination** 

 *L* = (*D*

the property of the resulting film. The diffusion length (*L*) of a species is

where *D* is a diffusion coefficient, *t* is a lifetime and *m* is a mass of a species.

& Tanaka 1986; Perrin et al., 1989],

discussed in the next section.

First of all, the diffusion coefficient of a SiH3 radical is larger than that of a higher-ordered silane radical due to the difference of their mass. The ratio of the mass of a SinH2n+1 radical (*m*SinH2n+1) and that of a SiH3 radical (*m*SiH3) is

$$
\mu\mathfrak{m}\_{\text{SimF2n}\ast 1} \int \mathfrak{m}\_{\text{S\"H-E\"}} \approx \mathfrak{m}.\tag{6}
$$

Therefore, the diffusion coefficient of a SiH3 radical is *n*-times larger than that of a SinH2n+1 radical.

Second, the lifetime of a SiH3 radical is especially long in a SiH4 gas phase. The density of SiH4 molecule under the condition used in this study (0.1 Torr, 250 oC) is ~ 1015 cm-3. The density of SiH3 radicals is the highest among the other generated species, and the value is ~ 1012 cm-3 in an RF silane plasma [Matsuda, 2004]. In the case of VHF plasma, as in our case, the value changes due to a high electron density and a low electron temperature effects. Estimating from the deposition rate, SiH3 density can be one order of magnitude higher than that in an RF plasma, but in any cases, the density is still very low with respect to that of SiH4 molecule. Therefore, most of the generated species collide with SiH4. Here, the SiH3 radical does not disappear due to the collision, resulting in its long lifetime:

$$\rm SiH\_3 + SiH\_4 \twoheadrightarrow SiH\_4 + SiH\_3. \tag{7}$$

After all, the diffusion length of SiH3 radical is very long according to equations (4) and (5). Under a certain gas flow rate condition, the radicals having small diffusion lengths are pumped out, but a large diffusion length species can still reach to the substrate. Thus, larger the distance between the mesh and the substrate, stronger the diffusion length effect. Therefore, we would like to propose that in the triode configuration, a long lifetime SiH3 radical mainly contribute to the film growth than that in a diode system. If a SiH3 radical sticks to a film surface, a cross-linking reaction takes place because the configuration of a five or six members of ring is easily formed with the SiH3. On the other hand, when a higher-ordered silane radical sticks to the surface, not all of the Si atoms are located in such a configuration due to its steric-hindrance. Thus, the cross-linking reaction can take place partially, resulting in the remaining of hydrogen in the film. Once hydrogen is incorporated into the bulk of a film, it cannot be thermally eliminated at 250 oC as shown in figure 10. Indeed, under the low SiH4 flow rate condition, the higher Si-H and Si-H2 bond densities are observed in the triode system [Shimizu et al., 2007].

#### **4.4 Effect of growth rate**

In the case of the triode system, the growth rate is low compared to that observed in a conventional diode system. In our VHF plasma case, the growth rate observed with the diode system is 7.3 Å/s, and that observed in the triode system is 0.7 Å/s at *d*ms = 1 cm, and is 0.2 Å/s at *d*ms = 4 cm. When the growth rate is low, hydrogen concentration of the resulting film can be low when thermal desorption from the surface and the bulk are the dominant hydrogen elimination processes. However, such elimination processes at the substrate temperature of 250 oC are unlikely as discussed in the previous sections.

To confirm the effect of the growth rate furthermore, the experiments were performed with installing a second mesh as described in section 3.1.2. Installing the second mesh reduces the growth rate drastically. As shown in Table 1, the observed growth rate with the double mesh at the VHF power of 10 W is c.a. 0.1 Å/s and that with the single mesh at the same VHF power is 0.8 Å/s. Under those conditions, however, almost the same Si-H and Si-H2 bond densities are observed. Since the VHF power is fixed at the same value, the densities of the generated

Fabrication of the Hydrogenated

**7. Acknowledgement** 

28, pp. 671-673.

Vol. 63, pp 2532-2551.

**8. References** 

fabricate stable a-Si:H films and related solar cells.

Technology Development Organization (NEDO), Japan.

potential. Solar Energy, Vol. 74, pp. 181-192.

Phys. Rev. B, Vol. 45, pp. 13367-13377.

Phys., Vol. 60, pp. 2351-2356.

Solids. Vol. 338–340, pp. 1–12.

**6. Conclusions** 

Amorphous Silicon Films Exhibiting High Stability Against Light Soaking 317

Stable a-Si:H films against light soaking are prepared with adopting a triode deposition method where a mesh is placed between a cathode and a substrate. The resulting films contain very low Si-H and Si-H2 bond densities compared with those observed in the films prepared by a conventional diode electrode method at the same substrate temperature. The hydrogen reduction effect is higher when the distance between the mesh and the substrate is increased. The films exhibit low initial defect densities and high photosensitivities. After the light soaking, high stabilities are observed in the films prepared by the triode system. The high stabilities of the films are also confirmed with the device configurations. It is most likely that the density of the precursors that reach to the growing surface is different each other in the triode and the diode systems. Control of gas phase condition is one of the key issues to

The authors acknowledge research support from the New Energy and Industrial

Beyer, W. & Wagner, H. (1983). The role of hydrogen in a-Si:H - Results of evolution and

Borrello, D., Vallat-Sauvain, E., Bailat, J., Kroll, U., Meier, J., Benagli, S., Marmelo, M.,

amorphous silicon photovoltaic devices. WIPO Patent: WO/2011/033072. Carlson, D. E. & Wronski, C. R. (1976). Amorphous silicon solar cell. Appl. Phys. Lett. Vol.

Drevillon, B. & Toulemonde, M. (1985). Hydrogen content of amorphous silicon films deposited in a multipole plasma. J. Appl. Phys., Vol. 58, pp. 535-540. Green, M. A. (2003). Crystalline and thin-film silicon solar cells: state of the art and future

Kushner, M. J. (1988). A model for the discharge kinetics and plasma chemistry during

Langford, A. A., Fleet, M. L., Nelson, B. P., Lanford, W. A. & Maley, N. (1992). Infrared

Matsuda, A. & Tanaka, K. (1986). Investigation of the growth kinetics of glow-discharge

Matsuda, A., Nomoto, K., Takeuchi, Y., Suzuki, A., Yuuki, A. & Perrin, J. (1990).

Matsuda, A. (2004). Microcrystalline silicon. Growth and device application. J. Non-Cryst.

Müller, J., Rech B., Springer, J. & Vanecek, M. (2004). TCO and light trapping in silicon thin

hydrogenated amorphous silicon. Surf. Sci., Vol. 227, pp. 50-56.

film solar cells. Solar Energy, Vol. 77, pp. 917–930.

Monteduro, G., Hoetzel, J., Steinhauser & J., Lucie, C. (2011). High-efficiency

plasma enhanced chemical vapor deposition of amorphous silicon. J. Appl. Phys.,

absorption strength and hydrogen content of hydrogenated amorphous silicon.

hydrogenated amorphous silicon using a radical separation technique. J. Appl.

Temperature dependence of the sticking and loss probabilities of silyl radicals on

annealing studies. J. Non-Cryst Solids, Vol. 59-60, pp. 161-168.

radicals and ions in the plasma are basically the same in the both cases. When the VHF power is reduced to 2 W with a single mesh, the observed growth rate is 0.2 Å/s which is the similar value observed at the VHF power of 10 W with the double mesh (0.1 Å/s). On the other hand, the observed Si-H and Si-H2 bond densities are different each other where lower values are observed in the low power case in which less higher-ordered silane radicals are produced due to a low electron temperature effect [Matsuda, 2004]. The results indicate that the gas phase condition is very important for determining a hydrogen concentration in the resulting film. Note that, when a plasma is unstable even in the triode case due to the lack of electrical matching, the observed Si-H and Si-H2 densities are higher than the expected values shown in figure 3 (the higher value date are not presented).

#### **5. Prospects for the future applications**

The properties and the stabilities of the a-Si:H films prepared by the triode deposition system have been demonstrated in this study. The quality of the film is very good, and it exhibits very high stability against light soaking. Although the triode method reduces a growth rate of a film due to its configuration, which is a disadvantage for mass productions, we used this system to study the fundamental features. Several results indicate that the control of the gas phase condition is one of the essential factors to obtain a stable a-Si:H film. Based on this knowledge, one could establish the alternative fabrication methods which can produce the preferable gas phase condition for a stable a-Si:H fabrication.

In our result, the degree of degradation correlates well with Si-H2 bond density in the film, which also corresponds to the former works. Beside the possibility of micro-void structure formation, one can propose the existence of chain-like Si-Si structures when the film contains large Si-H2 bond density. Since it is a flexible structure, it can cause instability against light soaking. In figure 12 the *FF* (= *FF*ini – *FF*deg) of the Schottky diode is plotted against the Si-H2 bond density. It is a re-plot of figure 8 in a semi-log scale. The extrapolated line shows that the *FF* value is zero at the Si-H2 bond density of c.a. 1.3×1019 cm-3 (≈ 0.03 at.%). Although it is a hypothesis, the correlation indicates the guideline for the fabrication of stable a-Si:H films.

Fig. 12. Light-induced change in the fill-factor (*FF = FF*ini*-FF*deg) of the Schottky diode as a function of the Si-H2 bond density (re-plot of fig. 8 in a semi-log scale).
