Study on effect of process and structure parameters on SiNxHy growth by in-line PECVD

Study on effect of process and structure parameters on SiNxHy growth by in-line PECVD

Solar Energy 198 (2020) 469–478 Contents lists available at ScienceDirect Solar Energy journal homepage: www.elsevier.com/locate/solener Study on e...

3MB Sizes 0 Downloads 7 Views

Solar Energy 198 (2020) 469–478

Contents lists available at ScienceDirect

Solar Energy journal homepage: www.elsevier.com/locate/solener

Study on effect of process and structure parameters on SiNxHy growth by inline PECVD

T



Yujin Caoa,1, Jicheng Zhoua,1, , Yaqing Rena, Wei Xua, Wenfeng Liub, Xianwu Caib, Baoxing Zhaoc a

School of Energy Science and Engineering, Central South University, Changsha 410083, China Hunan Red Solar Photoelectricity Science and Technology Co., Ltd, Changsha 410005, China c Suzhou Talesun Solar Technologies Co., Ltd., Changshu 215542. China b

A R T I C LE I N FO

A B S T R A C T

Keywords: In-line PECVD SiNxHy film COMSOL Multi-field coupling simulation Solar cell

This paper builds a physical model by using the finite element method on COMSOL simulation platform to simulate the in-line PECVD process. The in-line PECVD simulation model couples the flow field, thermal field, chemical reaction field and plasma field and is verified through experiments. In addition, a new simulation strategy is proposed to solve the problem of dynamic coating silicon nitride film. Through this simulation method, process parameters and structural parameters of the in-line PECVD equipment are optimized. Effects of microwave tube position, microwave shield size, total gas flow, pressure and temperature on main coating reaction region is studied. The results have shown that the position of microwave tube, the total gas flow and temperature have greater influence on the region of reaction zone, the concentration distribution of reactants and the thickness of SiNxHy film. Through optimizing process parameters and structural parameters, the coating rate of SiNxHy film can be increased from 0.0607 nm/s to 0.15 nm/s and the deposition thickness can be increased from 6.78 nm to 15.59 nm. The molar concentration of SiNxHy particles in the reaction region has grown by more than 23.5%. And the relative intensity of reaction field inside PECVD chamber has increased by 1.3–2.3, which is calculated by molar concentration ratio of SiNxHy particles. This paper provides a reference for the optimization of in-line PECVD equipment.

1. Introduction With the development of new energy technologies, photovoltaic technology is becoming more and more prominent and important. (Kabir et al., 2018; Liu et al., 2018; Yuan et al., 2015). In the field of photovoltaics, in-line PECVD equipment is widely used such PERC solar cells and heterojunction solar cells. (Dullweber and Schmidt, 2016; Green, 2015; Ruan et al., 2019). Plasma enhanced chemical vapor deposition (PECVD) method is often used to coat a silicon nitride film on the solar cell surface as an antireflection film and passivation layer. As an anti-reflection film, silicon nitride film can reduce reflectance so that more sunlight is absorbed by the solar cell (Doshi et al., 1997). As a passivation layer, it can provide very low surface recombination velocities and reduce charge carrier recombination on the surface of the silicon wafer (El amrani et al., 2008; Hezel and Schörner, 1981; Moschner et al., 2004; Soppe et al., 2005). Therefore, coating a silicon nitride film on both sides of the silicon wafer is a necessary process for preparing a high-efficiency solar cell. In addition, in-line PECVD

equipment has many advantages, such as low process temperatures, low voltage and high throughput (Aberle, 2001; Caquineau and Despax, 1997; Doshi et al., 1997; Soppe et al., 2005; Strobel et al., 2009). As for in-line PECVD reactor, there are flow fields, thermal fields, chemical reaction fields and plasma fields in the process chamber, which involves a series of complex processes, such as plasma chemical reaction process, convection and diffusion propagation, thermal effect, gas phase chemical reactions, surface heterogeneous reactions and surface growth reactions (Armaou and Christofides, 1999; Dollet et al., 1995; Kushner, 1988, 1992). The quality of silicon nitride film is often affected by parameters such as temperature, pressure, total gas flow and gas flow ratio(Bavafa et al., 2008; Crose et al., 2017, 2018; Sansonnens et al., 2003; Sobbia et al., 2005; Wei et al., 2006; Wu et al., 2015; Xia et al., 2016). However, it can be found that most of the researches are focused on small PECVD chamber, which are oftentimes applied in laboratories. The parameters such as reactor structure and process conditions are very different for the in-line PECVD equipment on the production line. In addition, the influence of different parameters, such as total reaction



Corresponding author. E-mail address: [email protected] (J. Zhou). 1 These authors contributed equally. https://doi.org/10.1016/j.solener.2020.01.054 Received 14 June 2019; Received in revised form 27 September 2019; Accepted 20 January 2020 0038-092X/ © 2020 International Solar Energy Society. Published by Elsevier Ltd. All rights reserved.

Solar Energy 198 (2020) 469–478

Y. Cao, et al.

gas flow, reaction gas ratio, pressure and temperature, on coating silicon nitride film has different or even opposite results in some researches. Hence, this study is aimed at in-line PECVD equipment, which is applied in industrial production. Besides, most studies consider the influence of process parameters on the coating film, but the structural parameters of the in-line PECVD equipment should also be taken into consideration (Bavafa et al., 2008; Kim et al., 2005; Wu et al., 2015; Xiang et al., 2019). In this paper, the research on the in-line PECVD equipment mainly uses simulation experiments. Simulation experiment is based on finite element method in the platform of COMSOL Multiphysics 5.3a. In the case library of COMSOL, SiHx films are grown on the surface of silicon wafers by using the method of chemical vapor deposition (CVD), which is worthy of reference. In addition, the ionization and deposition processes of different substances in PECVD systems is studied by some researchers who are also using COMSOL simulation platform and provide valuable reference (Bouherine et al., 2016; Kumar et al., 2019; Li et al., 2016). Based on the COMSOL simulation platform, material definition, multi-physics setting and boundary layer definition are performed in the physical model. In order to ensure the validity of the physical model, the deposition thickness is compared with the simulated deposition thickness to verify the correctness of the in-line PECVD physical model. In addition, the thickness of the silicon nitride film, the minority carrier lifetime and the implied open circuit voltage (iVoc) of the solar cell are tested by experiment. In the coating process, the silicon wafer is transported into the in-line PECVD reactor at a speed of about 30 mm/s. Dynamic grids increases the difficulty of simulation research and makes simulation calculations difficult to converge for dynamic problems. In order to solve this problem, a new simulation strategy is proposed and focuses on the impact of optimized process parameters and structural parameters on the reaction field within the in-line PECVD chamber. Through analysis, the reaction intensity and reaction time are both proportional to the thickness of the silicon nitride film. The reaction efficiency on the surface of the silicon wafer will be improved by locking the moving silicon wafer in a wider and stronger reaction area and the thickness of the silicon nitride film will be thicker. Similarly, the reaction intensity around the silicon wafer surface is stronger, and the deposited thickness of the silicon nitride film on the silicon wafer substrate is also thicker. Hence, research goal of this paper is to enlarge the range of reaction zone and the intensity of reaction field inside the in-line PECVD chamber by adjusting the structure parameters and process parameters. The coating reaction rate will also be increased in this way. Finally, through simulation studies, it is found that the position of the microwave tube inside the in-line PECVD has a great influence on the reaction area, affecting the thickness of film growth, which may be ignored by many researchers. In addition, in the study of process parameters, the total reaction gas flow and chamber temperature have a more prominent effect on the reaction zone and intensity of reaction field. This paper will provide a reference for the optimization of in-line PECVD equipment.

Fig. 1. The image of in-line PECVD reactor and simulated model. (a) Side view of in-line PECVD reactor. (b) Simplified model of an in-line PECVD reactor. (c) Unstructured grid of the simulated area of PECVD reactor.

work is carried out and a two-dimensional physical model is established which is shown in Fig. 1(b). In the in-line PECVD, each U-shaped microwave shield will correspond to a microwave tube (arrow 2), which will emit microwaves (Liehr and Dieguez-Campo, 2005; Soppe et al., 2005). Ammonia gas (arrow 3) is ejected from the center of the Ushaped microwave shield, and silane (arrow 4) is ejected from the ports on both sides of the U-shaped microwave shield. In the plasma zone, ammonia gas is sprayed into the microwave region and excited into a plasma state. Thereafter, ammonia and silane are subjected to precursor reaction and gas phase reaction in the reaction zone. Eventually, the reaction product falling on the silicon wafer (arrow 5) will produce a surface reaction of silicon nitride film growth. The area between the Ushaped microwave shield and the heating plate (arrow 6) is the main reaction area in the research. The reacted gas is pumped away from both sides of the chamber (arrow 7). Only a small portion of the gas flows into the area below the heating plate. Therefore, the area under the heating plate is neglected. Based on the actual production process parameters of the in-line PECVD, the internal temperature of reactor is maintained at about 450℃ and the reaction chamber is stabilized at about 25 Pa. The total flow rate of the reaction gas is 650 sccm, and the gas flow of silane and ammonia are 200 sccm and 450 sccm respectively. Nitrogen acts as a carrier gas whose maximum flow rate can reach 120 L/min. For large-sized in-line PECVD equipment, the differences between the non-structural mesh and the structural mesh are small (Crose et al., 2018). In this research, non-structural mesh is built in COMSOL Multiphysics 5.3a which is shown in Fig. 1(c). In order to make the simulation results more accurate, this study further encrypts the grid at the boundary of the microwave tube and the silicon wafer.

2. Simulation study 2.1. Modeling and meshing The physical model is built based on the size of in-line PECVD equipment. The internal reactor structure of the in-line PECVD is complex, as is shown in Fig. 1(a). Many complex structures are distributed inside the in-line PECVD reactor. According to the internal characteristics of the in-line PECVD reactor structure in Fig. 1(a), four U-shaped microwave shields (arrow 1) arranged equidistantly on the upper cover plate and the bottom of the chamber is a heating plate. The wafer carrier enters the reactor for the coating process. According to the characteristics of the in-line PECVD reactor structure, the simplified

2.2. Mathematical model formulas In COMSOL Multiphysics 5.3a, integrals are set for reaction area and the surface of wafer. In this way, the variation of the deposited 470

Solar Energy 198 (2020) 469–478

Y. Cao, et al.

understand the chemical mechanism of silicon nitride film, the gas phase and surface growth mechanisms which are finally selected to represent the major reactions in coating process are presented in Tables 2 and 3. At the same time, based on the plasma field and chemical reaction field in COMSOL Multiphysics 5.3a, the mechanism of the above reaction is related to the following description equation. In the plasma field, the electron motion in the reaction zone consists of drift motion and diffusion motion. The electron drift diffusion equation and flux are shown in the following equation:

thickness and the molar mass and molar concentration of the reactants are made clearer. In addition, the multi-physics fields in the in-line PECVD are also investigated. For the heat and flow field inside the chamber, the internal heating mode of the in-line PECVD is mainly heated by the resistance wire. Additionally, there are many inlet and outlet ports inside the in-line PECVD, including the inlet and outlet of silane and ammonia of the in-line PECVD. The reaction gas is mainly laminar in the in-line PECVD, which ensures that the internal gas flow can form a relatively stable flow field (Ding et al., 2011; Howling et al., 2012). The transportation process of the reaction gas inside the reactor mainly follows the control of mass conservation, momentum conservation, and the basic law of energy conservation. The mass conservation equation is as follows: ⎯→ ⎯ ∂ρ + ∇ (ρυ ) = 0 ∂t

(1)

Here, υ is the overall velocity vector, ρ is the overall density, and t is time. The Navier-Stokes momentum is given by (2)

where p is the static pressure and SM is the generalized momentum source. The inside of the PECVD reactor is at a low pressure and the gas density is different at standard atmospheric pressure. The effect of gravity on momentum is relatively small, which will be negligible in the calculations. Energy conservation can be considered by the following equation:

∂p ∂ (ρE ) + SE + ∇ ·(ρEυ) = ∇ ·(λ∇T + τυ) + ∂t ∂t

q = h (T − Tout )

(5)

(7)

→ ∂ (ρYi ) υ Yi ) = −∇ · Ji + Ri + ∇ ·(ρ→ ∂t

(8)

→ ∇T Ji = −ρDi ∇Yi − DiT T

(9)

P RT

Q



x i Mi

(10) → Among the above parameters, ρ , Yi and Ji are the total density of heavy particles, heavy particle mass fraction and diffusion vector of the group i . The x i is molar fraction of the group i . Mi is the molar mass. In addition, the electron energy distribution function is in accordance with the Maxwellian distribution. Here, the rate constant k ∗f of the bombardment dissociation reaction can be determined by the following equations.

Here, λ is the fluid heat transfer coefficient, T is the control temperature, and SE is the energy term, wherein the energy term is the mean internal heat source. E is the total energy, including gas internal energy, pressure energy and kinetic energy. In the above energy equation, the thermal field and the flow field are coupled. In addition to the above control equations, solid heat transfer inside the in-line PECVD reactor should also be considered. The solid heat transfer inside the chamber is mainly related to heat conduction, convective heat transfer existing on the wall surface. There is a heat exchange between the internal reaction gas and the internal structure. The expression is as follows: (4)

∇Γe = −ne μe E − De ∇ne

ρ=

(3)

∂ (ρcT ) = ∇ ·(λs ∇T ) + ST ∂t

(6)

where ne , μe , De , R e are considered to be the electron number density, electron mobility, electron diffusion coefficient and electron generation rate. The transport of heavy particles in the reactor follows the equation below. From the aspect of flow field coupling, the effect of chemical reactions on the gas density in the in-line PECVD reactor is considered.

⎯→ ⎯

∂ (ρυ) + ρ (υ ·∇) υ = −∇p + ∇τ + SM ∂t

∂ne + ∇Γe = R e ∂t

i=0

2e me

k ∗f = α

∫0



εσj (ε ) f (ε ) dε

(11)

where α is the probability of dissociation reaction, ε is the electron energy, σj is the collision cross section, me is the electron mass and f is electron energy distribution function. The surface reaction rate is determined by the following formula: K

K

vf

vr

qi = kf , i ∏ ck ki − kr , i ∏ ck ki k=1

k=1

(12)

vij

Among them, λs is the solid heat transfer coefficient. c is the specific heat capacity and ST is the internal heat source. T is the control temperature and Tout is the ambient temperature. In addition to the heat flow field, the microwave tube emits microwaves outward to form a plasma field in the in-line PECVD reactor, and then the reaction gas is excited into plasma state (Liehr and Dieguez-Campo, 2005). The process mechanism of silicon nitride film growth by in-line PECVD can be divided into the following three aspects: plasma excitation reaction, gas phase reaction and growth reaction on the wafer surface. The plasma discharge reaction is shown in Table 1. Additionally, the internal chemical reaction of the in-line PECVD reactor is also very complicated, because at least 40 kinds of chemical reactions are simultaneously produced on the silicon nitride film coating process (Dollet et al., 1995; Kushner, 1988, 1992). By referring to the papers of relevant researches, the following 23 major chemical reactants are oftentimes considered in the chemical reaction field study (Bavafa et al., 2008; Xia et al., 2016). These reactants are mainly: H2, H, SiH4, SiH3, SiH2, Si2H6, SiH4+, NH3, NH2, NH, NH4+, NH3+, NH2+, SiNH3, SiNH4, SiNH5, SiN2H4, SiN2H5 and SiN2H6. In order to better

γi ⎞ ∏ σ j ⎛ 1 ⎞ 8RT kf , i = ⎜⎛ ⎟ m 1 γ 2 − i ⎠ (Γtot ) ⎝ 4 ⎠ πMk ⎝

(13)

wherein ck is the molar concentration of the reactant k , and kf , kr represent the rate of the positive and negative reaction respectively.kf is represented by the formula (13). Γtot is the reaction surface concentration, m is the reaction order, T is the surface reaction temperature, R is the gas constant, Mk is the mean molecular mass, and γi is the sticking coefficient. The concentration of surface species follows the following equation:

dΓ = dt

N

∑ qi Δσi i=1

(14)

where Γ , qi , Δσi are considered to be the surface site concentration, reaction rate for reaction i and the change in site occupancy number for reaction i. The surface reaction material reaction equation is as follows.

Rsurf , k dZk = dt Γtot 471

(15)

Solar Energy 198 (2020) 469–478

Y. Cao, et al.

Table 1 Plasma discharge reactions taken into account in simulated model. Reaction

Rate constant (cm3 s )

Δε (V )

Reference

e + NH3 → e + NH3 e + NH3 → 2e + NH3∗

a a

0 0.42

(Kushner, 1992) (Kushner, 1992; Xia et al., 2016)

a

10.2

(Kushner, 1992; Xia et al., 2016)

a a a 8.34 × 10−9 8.74 × 10−9

3.9 5.6 8.6 0 0.27

(Dollet et al., 1995; Kushner, 1992; Yousfi and Benabdessadok, 1996) (Dollet et al., 1995; Kushner, 1992; Yousfi and Benabdessadok, 1996) (Dollet et al., 1995; Kushner, 1992; Yousfi and Benabdessadok, 1996) (Kushner, 1988) (Kushner, 1988; Xia et al., 2016)

e e e e e e

+ + + + + +

NH3 → e + NH3+ NH3 → e + NH + H2 NH3 → e + NH2 + H NH3 → e + NH + 2H SiH4 → e + SiH4 SiH4 → 2e + SiH4∗

e + SiH4 → e + SiH4+ e + SiH4 → e + SiH3 + H e + SiH4 → e + SiH2 + 2H a

a

12.9

(Kushner, 1988; Xia et al., 2016)

1.59 × 10−10 1.87 × 10−11

8 8

(Dollet et al., 1995) (Dollet et al., 1995)

Rate coefficient is obtained by convolving the electron energy distribution with the cross section from the indicated references.

Table 2 The main chemical reaction course of gas phase (Bavafa et al., 2008).

sk =

Reaction rate (cm3 mol·s )

H + NH2 → NH3 NH2 + NH2 → NH + NH3 H + SiH4 → H2 + SiH3 H + Si2 H6 → SiH3 + SiH4 SiH2 + SiH4 → Si2 H6 SiH3 + SiH3 → SiH2 + SiH4 SiH4 + NH → SiNH3 + H2 NH2 + SiH3 → SiNH3 + H2 NH2 + SiH3 → SiH2 + NH3 SiNH3 + H → SiNH4 SiNH3 + NH2 → SiN2 H5 SiNH4 + H → SiNH3 + H2 SiNH4 + H → SiH2 + NH3 SiNH4 + NH2 → SiN2 H4 + H2 SiN2 H4 + H → SiN2 H5 SiN2 H5 + H → SiN2 H4 + H2 SiN2 H5 + H → SiN2 H6

1 × 10−13 1.4 × 10−12 3.2 × 10−12 2.7 × 10−12 6.7 × 10−11 1.5 × 10−10 10−11 7.8 × 10−11 2.1 × 10−11 3 × 10−10 10−10 3 × 10−10 3.5 × 10−11 9.8 × 10−11 3 × 10−10 3 × 10−10 2.5 × 10−12

where V is the volume of reactor, ωk is the mass fraction, Rsurf , k, l is the rate expression on the surface, Al is the surface area, Mf , l is the inward mass flux of the surface l and sk is the surface rate expression for each species which comes from summing the surface reaction rates multiplied by their stoichiometric coefficients over all surface reactions. 3. Verification In order to ensure the correctness of the physical model, the physical model is verified by experiments. Five coating experiments are performed in the in-line PECVD. As is shown in Fig. 2, a 4 × 6 wafer carrier (1046 mm × 741 mm) is transported into the in-line PECVD reactor at the speed of 35 mm/s for coating process. The coating time in process chamber is about 50 s. After each experiment, six silicon wafer samples (A1, A6, B3, C4, D1, and D6) are selected for testing. In total, 30 silicon wafer samples (156 mm × 156 mm) are selected for testing. Five test points are selected on each silicon wafer sample. The sample silicon wafer is P-type single crystal silicon, which is made by LONGi. And the silicon wafer is polished on both sides. The size of the silicon wafer is 156.75 × 156.75 ± 0.25 mm, the diameter is 210 ± 0.25 mm, and the thickness is 180 + 20/−10 μm. The bulk resistivity is 1 Ω·cm. Silicon nitride film is coated on both sides of the silicon wafer. In this experiment, 150 sets of experimental data are collected. The SiNxHy deposition thickness and refractive index are tested by ellipsometer (SE400S, SENTECH, Germany). In addition, the minority carrier lifetime before and after sintering and iVoc are tested (WCT-120, Sinton Instruments, USA). Also, the relationship between these parameters and film thickness is investigated. The sintering process is completed in a sintering furnace (Safire, Despatch, USA) with a peak sintering temperature of 780 °C and the speed of conveyor belt is 240 in/min. The thickness of the silicon nitride film is used as a verification index to verify the correctness of the simulation model. In Fig. 3(a), the deposition thickness of the silicon nitride film on the silicon wafer is simulated. The abscissa x = 0, which represents the center of the silicon wafer. And the simulated thickness and experimental thickness of the

(16)

Here, Zk is the site fraction (dimensionless), Rsurf , k is the surface velocity expression and Γtot is the total surface site concentration. For surface deposits, the surface material deposition thicknss is represented by the following formula:

Rsurf , k Mk dhk = dt ρk

(17)

In the in-line PECVD reactor, the mass fraction of internal particles is based on the following formula:



d (ωk ) = VRk + dt

Mf =

∑ hl Al Rsurf ,k,l Mk − ωk ∑ hl Al Mf ,l l

(18)

l

∑ Mk sk

(19)

k=1

(20)

i=1

Reaction

∑ Zk = 1

∑ vki qi

Table 3 Surface reactions of silicon nitride film growth. Reaction

Sticking coefficient γi

Reference

SiH3 → Si (b) + SiH4 + H2 Si2 H5 → Si (b) + Si2 H6 + H2 SiNH3 → SiN (s ) + H2 SiN2 H4 → SiN2 (s ) + H2 SiN2 H5 → SiN (s ) + SiNH5 + H2 SiN2 H5 → SiN2 (s ) + SiN2 H6 + H2

0.2 0.1 0.02 0.02 0.016 0.01

(Bavafa et al., 2008; Caquineau, 1996; Dollet et al., 1995) (Caquineau, 1996; Dollet et al., 1995; Xia et al., 2016) (Caquineau, 1996) (Caquineau, 1996) (Caquineau, 1996) (Caquineau, 1996; Dollet et al., 1995; Xia et al., 2016)

472

Solar Energy 198 (2020) 469–478

Y. Cao, et al.

Fig. 2. Illustrative diagram of testing silicon wafers and testing points.

silicon nitride film at the test point are compared. For the experimental data, the five fixed points on the silicon wafer are tested. The data of test points 1, 2, and 3 on 30 wafer samples are counted and sorted. Since the simulation model is two-dimensional, in the Y direction, only the three experimental points (tested point 1, 2 and 3) can be compared. In order to better illustrate the correctness of the model, the range of the experimental thickness of the silicon nitride film is shown in Fig. 3(b). It can be seen from Fig. 3(b) that the simulation data is in good agreement with the experimental data, which is within the range of experimental thickness. In addition, the thickness and refractive index distribution of the silicon nitride film deposited on the silicon carrier in the Y-direction are also shown in Fig. 3(c). It can be found that the film thickness is in the range of 94–100 nm and the refractive index is also in the range of 2.0–2.2. Therefore, the simulated thickness is also within the range of experimental test. Besides, since the deposition thickness is different, it is tested whether minority carrier lifetime and iVoc will be affected, which is shown in Fig. 3(d). In Fig. 3(d), the maximum variation of iVoc is approximately 0.01 V. This shows that the change in silicon nitride thickness has less effect on iVoc. And there is also no obvious linear relationship between minority carrier lifetime and the thickness of silicon nitride film. In addition to simulating the deposition thickness, the simulation experiment also obtained the internal flow field of in-line PECVD equipment. It is observed from Fig. 3(e) that the gas flow rate on both sides of the silicon wafer is faster. The reactive groups on both sides of the silicon wafer are more likely to be blown away, resulting in a relatively low deposition thickness on both sides of the simulated film. Besides, the data monitoring line (white line) is built in the physical model, which is located at 3 mm above the silicon wafer (see Fig. 3(e)). The concentration of the reactive group particles at the near silicon wafer can be more accurately known. The ratio of reactive groups can be obtained on the detection line. The radical group concentrations of NH3, NH2, SiH3, SiH2, SiNH3, SiNH4, SiN2H4, SiN2H5 and SiN2H6 at the data monitoring line are also investigate, which can be seen from Fig. 3(f). Among them, SiN2H5 particles have the highest proportion. In the subsequent optimization study, the concentration of SiN2H5 is used to represent the concentration of the reactants in the in-line PECVD chamber. According to the analysis, since the reactant contains a large amount of hydrogen particle groups, a good passivation is generated after sintering, so that the minority carrier lifetime is increased. This also indirectly illustrates the correctness of the simulation model.

and structural parameters. Through this simulation strategy, the dynamic coating process is avoided. As the reaction zone becomes wider and thicker, the silicon wafer will pass through the reaction zone for a longer reaction time, which makes more reactants falling into the silicon wafer substrate and causes more reactive particles to react. Also, as the intensity of the reaction field increases, more reactive particles will react and the reaction rate will be accelerated. The coating efficiency can be further improved by the above method. As for process parameter optimization, parameters such as total gas flow, temperature and pressure is adjusted. Though analysis of the structure of the in-line PECVD reactor, it can be found that the size of microwave shield (X1) and the microwave tube position (X2) may be the main influencing factors (see Fig. 4). This is because X1 and X2 significantly affect the location of the plasma reaction zone. Hence, the optimization of the structure is mainly focused on the structure of microwave system. 4.1. Heat flow field distribution In this study, the in-line PECVD equipment is assumed to keep in a stable state during the coating process. Simulation experiments have been completed many times in this research under different parameter conditions. As is shown in Fig. 3(e), the simulation cloud of flow field is obtained inside the in-line PECVD reactor by the simulation experiments. The flow rate of reaction gas can be visually observed from image of simulate clouds. It is observed from Fig. 5(a) that the rate of reaction gas flow under the microwave tube is relatively slow because of the blockage of microwave tube. Therefore, the position of the microwave tube (X2) should be considered. For the thermal field, the data of the temperature is collected on the monitoring line. The temperature distribution in the in-line PECVD reactor is relatively uniform and difference in temperature within the reactor is small, which is shown in Fig. 5(b). 4.2. Reaction radical distribution The SiNxHy radicals are the main deposition precursors in silicon nitride deposition. The gas phase reactions of the precursors can be seen from Table 2. Reaction field distribution inside the in-line PECVD reactor is shown on Fig. 6(a). The main reaction region inside the in-line PECVD reactor can be shown from the concentration distribution of SiNxHy particles. The concentration of SiNxHy particles, near the microwave tube, is relatively higher, which also indicates that the reaction in this region is more intense and more SiNxHy radical groups are formed. In Fig. 6(b), taking SiN2H5 as an example, the concentration distribution of SiN2H5 is opposite to the flow field distribution, which also indicates that the flow field and the chemical field interact with each other inside the in-line PECVD reactor.

4. Results and discussion After verifying the physical model, this paper conducts optimization research. As is shown in Fig. 4, optimization strategy of this paper aims to lock the moving silicon wafer into a wider and stronger reaction region in the in-line PECVD reactor by changing the process parameters 473

Solar Energy 198 (2020) 469–478

Y. Cao, et al.

Fig. 3. Diagram of experiment and simulation results. (a) Simulation thickness of silicon nitride film on silicon wafer. (b) Experimental and simulated thickness of silicon nitride film on silicon wafer. (c) Silicon nitride film thickness and refractive index on the silicon wafer samples which located on carrier (Y-direction). (d) The lifetime of minority charge carrier and iVoc on different thickness. (e) Velocity distribution of gas flow in in-line PECVD chamber. (f) Radicals percentage in the reaction region.

microwave shield increases, which is shown in Fig. 7(a) and (b). Through analysis, it can be found that when X1 becomes larger, the reaction area under the U-shaped microwave shield becomes larger. As the process parameters are constant, the occupancy rate of the plasma inside the in-line PECVD chamber becomes low, thereby reducing the reaction intensity and slowing down the coating rate. Therefore, the width of the microwave shield is a parameter worth considering. For the position of microwave tube, as is shown in Fig. 7(c) and (d), the influence of the position of moving microwave tube on coating is more obvious. When the distance between the microwave tube and the silicon wafer becomes closer, relative reaction intensity of the reaction field increase and the average molar concentration of SiN2H5 is also increases. However, it can be found that the thickness of the average silicon nitride film is significantly reduced when X2 = 32.5 mm (see Fig. 7(d)). This may be that the microwave tube is too close to the silicon wafer, so that the microwave tube blocks the flow of the reaction

4.3. Adjusted structure parameters For the structural optimization of in-line PECVD reactor, this paper investigates the size of microwave shield (X1) and the distance between the microwave tube and the silicon wafer (X2). The initial value of X1 is 110 mm and the initial value of X2 is 66 mm. The microwave tube emits microwaves outward to excite the reaction gas into a plasma state, which means that the position of the microwave tube should be considered. In addition, the size of the microwave shield also has influence on the reaction area of the coating. In this paper, average SiN2H5 molar concentration at the position of the monitoring line is calculated to indirectly indicate the overall concentration of the reactor. Additionally, relative intensity of the reaction in the reaction field is represented by the variance of the SiN2H5 molar concentration. The intensity of the reaction field, the molar concentration of SiN2H5 and deposited thickness of film decrease when the size of the 474

Solar Energy 198 (2020) 469–478

Y. Cao, et al.

Fig. 4. Illustrated drawing of optimization design scheme.

from the microwave tube. In addition, during the coating process, the in-line PECVD reactor is always in a low pressure state. As for the influence of pressure, it can be seen from Fig. 8(c) and (d) that relative reaction intensity and average molar concentration of SiN2H5 will decrease when internal pressure of the reactor increases. This is due to the fact that higher pressure will suppress the coating reaction. Hence, average molar concentration of SiN2H5 decreases, because reaction intensity is weaker at the same gas flow rate (Soppe et al., 2005). In addition, temperature inside the PECVD reactor also affects the coating reaction. As are shown in Fig. 8(e) and (f), when internal temperature of the in-line PECVD reactor becomes higher, the deposition rate of the silicon nitride film becomes slower. This is because when the temperature rises, the SiN2H5 particle groups itself will gain more energy and then run out from the film. It is also found that the average concentration of SiN2H5 species also decreased as the temperature becomes higher (see Fig. 8(e)). In addition, high temperature also causes thermal damage to the silicon nitride film on the surface of the silicon wafer, and therefore the deposition rate and the quality of the silicon nitride film are affected. Through the optimization of the process parameters above, it can be found from the simulation results that the total flow of reaction gas and temperature of the in-line PECVD reactor will have a more significant effect on the silicon nitride film. Results of simulation optimization research reveal the influence of different parameters on the growth height of silicon nitride film and relative value of reaction intensity. Besides, optimization results are

gas. The reactant is relatively difficult to deposit on the silicon wafer for reaction. The reaction intensity of the central portion of the silicon wafer becomes low. This further illustrates that the location of the microwave tube greatly affects the flow field distribution inside the inline PECVD chamber. For the design and production of plate PECVD equipment, this is worth considering.

4.4. Adjusted process parameters As for process parameters, the effects of temperature, pressure and total reactant gas flow on the silicon nitride film are investigated. As are shown in Fig. 8(a) and (b), the gas concentration and reaction intensity inside the in-line PECVD reactor increase with the increase of total gas flow of the reaction gas (NH3/SiH4 = 2.25). The growth thickness of the silicon nitride film is also thicker at the same time. It can be known from the analysis that when the total flow rate of reaction gas increases, more reaction gas is injected into the reaction zone, which greatly increases the density of the reaction particles in the in-line PECVD chamber. Hence, more reactive particles fall on the surface of the silicon wafer for surface growth reaction. However, it can also be seen from Fig. 8(a) and (b) that the change tendency of the concentration of particles, intensity of reaction field and the thickness of silicon nitride film are gradually reduced. When the reaction gas injected into the chamber is continuously increased, the reaction gas cannot be completely ionized into a plasma state due to the limited energy emitted

Fig. 5. Diagram of simulation results on heat flow field. (a) Flow velocity magnitude on monitoring line. (b) Temperature magnitude on monitoring line. 475

Solar Energy 198 (2020) 469–478

Y. Cao, et al.

Fig. 6. Diagram of simulation results on reaction field. (a) Cloud chart of steady-state SiNxHy concentration distribution. (b) Molar concentration of SiNxHy on the monitoring line.

Fig. 7. Optimized results by adjusting structure parameters. (a) Average molar concentration of SiNxHy and intensity of reaction at different X1. (b) Thickness of silicon nitride film at different X1. (c) Average molar concentration of SiNxHy and intensity of reaction at different X2. (d) Thickness of silicon nitride film at different X2.

5. Conclusion

shown in Table 4, and it is hoped to provide some reference for the optimization of the in-line PECVD equipment system in the future. Through Table 4, it can be more clearly seen that the position of microwave tube in the in-line PECVD equipment has a greater influence on the coating thickness and the intensity of the reaction area, and the size of microwave shield is more inclined to affect the molar concentration of reaction particles in the chamber. In terms of process parameters, total flow rates of the reaction gas and the chamber temperature have a greater influence on the coating effect. And the temperature inside in-line PECVD chamber should be controlled within a reasonable range, otherwise the deposition height of the film will be greatly affected.

(1) Based on COMOSL simulation software, a physical model of in-line PECVD reactor is constructed and verified. The verification of the physical model is made based on the experimental data. (2) In view of the dynamic coating problem, this study proposed a simulation strategy to solve the dynamic coating problems, lowering the difficulty of simulation convergence. At the same time, it successfully obtained internal flow and heat field clouds inside the inline PECVD chamber, the concentration of SiNxHy particles, the proportion of the main reactants and the main reaction region the in-line PECVD chamber. 476

Solar Energy 198 (2020) 469–478

Y. Cao, et al.

Fig. 8. Optimized results by adjusting process parameters. (a) Average molar concentration of SiNxHy and intensity of reaction in different total gas flow. (b) Thickness of silicon nitride film in different total gas flow. (c) Average molar concentration of SiNxHy and intensity of reaction in different pressure. (d) Thickness of silicon nitride film in different pressure. (e) Average molar concentration of SiNxHy and intensity of reaction in different reactor temperature. (f) Thickness of silicon nitride film in different reactor temperature. Table 4 Key dates on the influence of different parameters on coating film. Variable name

Range of optimal parameter

Variable of film thickness

Variable of SiNxHy molar concentration

Size of microwave shield (mm) Position of Microwave tube (mm) Total gas flow (sccm) Pressure (Pa) Temperature (K)

[90,100] [52.5,72.5] [650,1100] [25,30] [498,698]

3.086 nm 9.310 nm 19.270 nm 7.962 nm 27.041 nm

1.78E−7 1.59E−7 1.27E−7 0.46E−7 1.92E−7

477

mol/m3 mol/m3 mol/m3 mol/m3 mol/m3

Variable of relative reaction intensity 1.954 1.963 2.524 0.650 1.736

Solar Energy 198 (2020) 469–478

Y. Cao, et al.

(3) The simulation optimization is carried out by adjusting internal structural parameters of PECVD chamber and the process parameters coating film. And optimal parameter distribution range is obtained. microwave shield optimal size (X1) is in the range of 100–110 mm, and microwave tube optimal position (X2) is in the optimum range of 52.5–72.5 mm. As for process parameters, when total flow of the reaction gas is in the range of 650–1100 sccm, it is always benefit for the film growth and the reaction strength, but economic cost should also be taken into consideration. In addition, as for the pressure inside the chamber, 25–30pa is optimal range, and the temperature should be maintained in the range of 498–698 K. In summary, this paper provides a reference for the optimization of in-line PECVD equipment.

absorbing plasma-enhanced chemical vapor deposited antireflection coatings for silicon photovoltaics. Appl. Opt. 36 (30), 7826–7837. Dullweber, T., Schmidt, J., 2016. Industrial silicon solar cells applying the passivated emitter and rear cell (PERC) concept—a review. IEEE J. Photovoltaics 6 (5), 1366–1381. El Amrani, A., Menous, I., Mahiou, L., Tadjine, R., Touati, A., Lefgoum, A., 2008. Silicon nitride film for solar cells. Renewable Energy 33 (10), 2289–2293. Green, M.A., 2015. The Passivated Emitter and Rear Cell (PERC): from conception to mass production. Sol. Energy Mater. Sol. Cells 143, 190–197. Hezel, R., Schörner, R., 1981. Plasma Si nitride—A promising dielectric to achieve highquality silicon MIS/IL solar cells. J. Appl. Phys. 52 (4), 3076–3079. Howling, A.A., Legradic, B., Chesaux, M., Hollenstein, C., 2012. Plasma deposition in an ideal showerhead reactor: a two-dimensional analytical solution. Plasma Sources Sci. Technol. 21 (1). Kabir, E., Kumar, P., Kumar, S., Adelodun, A.A., Kim, K.-H., 2018. Solar energy: potential and future prospects. Renew. Sustain. Energy Rev. 82, 894–900. Kim, B., Park, K., Lee, D., 2005. Use of neural network to model the deposition rate of PECVD-silicon nitride films. Plasma Sources Sci. Technol. 14 (1), 83–88. Kumar, M., Khanna, S., Gupta, N., Gupta, R., Sharma, S.C., 2019. Numerical simulation and parametric study of carbon deposition during graphene growth in PECVD system. IEEE Trans. Nanotechnol. 18, 401–411. Kushner, M.J., 1988. A model for the discharge kinetics and plasma chemistry during plasma enhanced chemical vapor deposition of amorphous silicon. J. Appl. Phys. 63 (8), 2532–2551. Kushner, M.J., 1992. Simulation of the gas-phase processes in remote-plasma-activated chemical-vapor deposition of silicon dielectrics using rare gas–silane-ammonia mixtures. J. Appl. Phys. 71 (9), 4173–4189. Li, Q., Liu, J., Dai, Y., Xiang, W., Zhang, M., Wang, H., Wen, L., 2016. Fabrication of SiNx thin film of micro dielectric barrier discharge reactor for maskless nanoscale etching. Micromachines (Basel) 7 (12). Liehr, M., Dieguez-Campo, M., 2005. Microwave PECVD for large area coating. Surf. Coat. Technol. 200 (1–4), 21–25. Liu, G., Zhou, B., Liao, S., 2018. Inverting methods for thermal reservoir evaluation of enhanced geothermal system. Renew. Sustain. Energy Rev. 82, 471–476. Moschner, J.D., Henze, J., Schmidt, J., Hezel, R., 2004. High-quality surface passivation of silicon solar cells in an industrial-type inline plasma silicon nitride deposition system. Prog. Photovoltaics: Res. Appl. 12 (1), 21–31. Ruan, T., Qu, M., Wang, J., He, Y., Xu, X., Yu, C., Zhang, Y., Yan, H., 2019. Effect of deposition temperature of a-Si: H layer on the performance of silicon heterojunction solar cell. J. Mater. Sci.: Mater. Electron. 30 (14), 13330–13335. Sansonnens, L., Bondkowski, J., Mousel, S., Schmitt, J.P.M., Cassagne, V., 2003. Development of a numerical simulation tool to study uniformity of large area PECVD film processing. Thin Solid Films 427 (1–2), 21–26. Sobbia, R., Sansonnens, L., Bondkowski, J., 2005. Uniformity study in large-area showerhead reactors. J. Vacuum Sci. Technol. A: Vacuum, Surf., Films 23 (4), 927–932. Soppe, W., Rieffe, H., Weeber, A., 2005. Bulk and surface passivation of silicon solar cells accomplished by silicon nitride deposited on industrial scale by microwave PECVD. Prog. Photovoltaics Res. Appl. 13 (7), 551–569. Strobel, C., Zimmermann, T., Albert, M., Bartha, J.W., Kuske, J., 2009. Productivity potential of an inline deposition system for amorphous and microcrystalline silicon solar cells. Sol. Energy Mater. Sol. Cells 93 (9), 1598–1607. Wei, M.C., Chang, S.J., Tsia, C.Y., Liu, C.H., Chen, S.C., 2006. SiNx deposited by in-line PECVD for multi-crystalline silicon solar cells. Sol. Energy 80 (2), 215–219. Wu, X., Zhang, Z., Yuan, L., Chu, X., Li, Y.J.S.E., 2015. Process parameter selection study on SiNx:H films by PECVD method for silicon solar cells. 111, 277–287. Xia, H., Xiang, D., Yang, W., Mou, P., 2016. Multi-model simulation of 300mm siliconnitride thin-film deposition by PECVD and experimental verification. Surf. Coat. Technol. 297, 1–10. Xiang, D., Xia, H., Yang, W., Mou, P., 2019. Parametric study and residual gas analysis of large-area silicon-nitride thin-film deposition by plasma-enhanced chemical vapor deposition. Vacuum 165, 172–178. Yousfi, M., Benabdessadok, M.D., 1996. Boltzmann equation analysis of electron-molecule collision cross sections in water vapor and ammonia. J. Appl. Phys. 80 (12), 6619–6630. Yuan, J., Na, C., Xu, Y., Zhao, C., 2015. Wind turbine manufacturing in China: a review. Renew. Sustain. Energy Rev. 51, 1235–1244.

Acknowledgements This work was funded by Hunan Province Science and Technology Department ‘Innovative Venture Technology Investment Project’ number 2017GK5002 and by Fundamental Research Funds for the Central Universities of Central South University ‘general project’ 2018zzts492. We thank S. Liu and T.C. Chen for discussions and experimental and technical assistance. Appendix A. Supplementary material Supplementary data to this article can be found online at https:// doi.org/10.1016/j.solener.2020.01.054. References Aberle, A.G., 2001. Overview on SiN surface passivation of crystalline silicon solar cells. Sol. Energy Mater. Sol. Cells. Armaou, A., Christofides, P.D., 1999. Plasma enhanced chemical vapor deposition: modeling and control. Chem. Eng. Sci. 54 (15–16), 3305–3314. Bavafa, M., Ilati, H., Rashidian, B., 2008. Comprehensive simulation of the effects of process conditions on plasma enhanced chemical vapor deposition of silicon nitride. Semicond. Sci. Technol. 23 (9). Bouherine, K., Tibouche, A., Ikhlef, N., Leroy, O., 2016. 3-D numerical characterization of a microwave argon PECVD plasma reactor at low pressure. IEEE Trans. Plasma Sci. 44 (12), 3409–3416. Caquineau, H., Despax, B., 1997. Influence of the reactor design in the case of silicon nitride PECVD. Chem. Eng. Sci. 52 (17), 2901–2914. Caquineau, H., 1996. Reactor modeling for radio frequency plasma deposition of SiNxHy: comparison between two reactor designs. J. Vacuum Sci. Technol. A: Vacuum, Surf., Films 14 (4), 2071–2082. Crose, M., Tran, A., Christofides, P., 2017. Multiscale computational fluid dynamics: methodology and application to PECVD of thin film solar cells. Coatings 7 (2). Crose, M., Zhang, W., Tran, A., Christofides, P.D., 2018. Multiscale three-dimensional CFD modeling for PECVD of amorphous silicon thin films. Comput. Chem. Eng. 113, 184–195. Ding, J., Zhao, Y., Yuan, N., Chen, M., Wang, S., Ye, F., Kan, B., 2011. Effect of electrode architecture and process parameters on distribution of SiH3 in a PECVD system. Vacuum 86 (3), 344–349. Dollet, A., Couderc, J.P., Despax, B., 1995. Technology Analysis and numerical modelling of silicon nitride deposition in a plasma-enhanced chemical vapour deposition reactor. I. Bidimensional modelling. Plasma Sources Sci. Technol. 4 (1), 107. Doshi, P., Jellison, G.E., Rohatgi, A., 1997. Characterization and optimization of

478