Experimental

First of all the polished (100) silicon was cleaned with dilute HF to remove the

native oxide. A layer of 300 nm thick silicon nitride (Si3N4) was then deposited on a

polished (100) silicon wafer by plasma enhanced chemical vapor deposition (PECVD)

technique. A nickel film with a thickness of 5 nm was then evaporated on the silicon

nitride surface using an E-beam evaporating system. The nickel film was then rapid

thermal annealed (RTA) under the forming gas (mixture of H2 and N2) with a flow

rate of 3 sccm at 850°C for 60 s to form nickelclusters, which served as the etch

masks for silicon nitride. Initial nickel thickness of 5 nm has been chosen to get the

higher density and smaller dimension of Ni nano-clusters in our experiment. The

sample was then etched by ICP etcher (ULVAC NE 550) using different gases and

etching parameters which will be discussed in results and discussion section. To

remove the residual nickel mask, the sample was dipped into pure nitric acid (HNO3)

solution for 5 min at room temperature. The morphology of SWS was analyzed by

Scanning electron micrograph (SEM). The reflectances of the SWS were measured

using an N&K 1280 analyzer.

4.2 Results and Discussions

Figure 4.1: Schematic of fabrication process of (a) silicon nitride nanopillars (b) silicon nitride nanocones.

Figure 4.1 shows the schematic illustrations of the fabrication of nanopillars and

nanocones by one- and two-step ICP etching processes, respectively. The one-step

etching process, which is shown in Fig. 1(a), involved non-etching of nickel

nanoclusters, whereas etching of the underlying silicon nitride film. As the size of the

nickel nanoclusters was not changed, the etched area of the underlying silicon nitride

depends on the diameter of the nano clusters. Thus, silicon nitride pillar structures

were readily obtained from the one-step etching process. The two-step etching process

as shown in Fig. 4.1(b) involved minifying of nickel nano clusters and etching of the

underlying silicon nitride film, gradually. As the size of the nano clusters decreased

gradually, the etched area of the underlying silicon nitride increased gradually. Thus,

silicon nitride nanocone structures were readily obtained from the two-step etching

process. One-step etching process was carried out by using a mixture of CF4 and O2

and based on fluorine chemistry. The two-step etching process was carried out by

using a mixture of CF4 and O2 in first step and followed by the etching with Ar. The

mechanism can be confirmed by watching the scanning electron microscope (SEM)

image as shown in Fig. 4.2.

The SEM image of the sample etched by Ar gas for 30 s and 120 s were shown in

Figs. 4.2(a) and 4.2(b), respectively, after the nanoclusters were formed. From these

figures we could see the decrease of nickel nano cluster size after etched in Ar gas for

long time, which confirms the mechanism involved to form nanocones described

previously. Similarly, we could see the effect of ICP etching using the mixture of

CF4/O2 gas to form the nanopillars from the Figs. 4.2(c) and 4.2(d).

Figure 4.2: SEM images of Ar plasma etching of nickel nano cluster on silicon nitride for (a) 30 s (b) 120 s. SEM images of CF4/O2 plasma etching of

nickel nano cluster on silicon nitride for (a) 60 s (b) 120 s.

Using the developed method, experiments were performed to demonstrate

control over nanopillar and nanocone height and structures. In the first series, samples

120 s (b)

60 s 120 s

(c) (d)

30 s

(a)

starting with nickel nano clusters were etched by a gas mixture of CF4/O2 to between

100 to 250 nm heights by different etching time. The resulting nanopillars with

corresponding heights could be seen in Figs. 4.3(a) - 4.3(d).

Figure 4.3: SEM images of fabricated nanopillar structures on silicon nitride film using one-step etching process for etching time: (a) 90 s (b) 120 s (c) 150 s (d)

180 s.

Figure 4.3(a) showed nanopillars with height around 110 nm with CF4/O2

etching time of 90 s. silicon nitride nanopillars with height around 150 nm has been

achieved after etched for 120 s as shown in Fig. 4.3(b). Similarly, Figs. 4.3(c) and

4.3(d) showed the achieved silicon nitride nanopillars with height around 210 and 240

nm for etching time of 150 s and 180 s, respectively.

Figure 4.4 displayed top-view and cross-sectional images of the different

nanocones obtained after varying etching time of the CF4/O2, but keeping the etching

time of Ar constant as 120 s, for the two-step process.

Figure 4.4 : SEM images (top-view) of fabricated nanocone structures on silicon nitride film using two-step etching process for etching time: (a) 90 s (b) 120 s (c) 150 s. Cross-section SEM view of the fabricated nanocone structures

with etching time (d) 90 s and (e) 180 s.

Figures 4.4(a) - 4.4(c) showed the top-view of the nanocones fabricated for the

etching time of 90 s, 120 s, and 150 s respectively. Figures 4.4(d) and 4.4(e) shows

the cross-section view of the structure shown in Fig. 4.4(a) and 4.4(c), which

indicated that etched silicon nitride having a height of around 157 nm for the etching

time of 90 sec and around 215 nm for the etching time of 150 sec, respectively. Note

that in Fig. 4.4(d) and 4.4(e) the base of the nanocones were different from each other,

which were clearly seen.

This behavior of the nanocone base could be explained by our proposed possible

mechanism as shown in Fig. 4.5.

Figure 4.5: Schematic illustration of the base of the fabricated nanocones: (a) shortest etching time (b) medium etching time, and (c) longer etching time.

In the two-step etching process, first step which used the mixture of CF4/O2

forms the nanopillars with different heights for the different etching times as shown in

Fig. 4.5. The longer etching time, the deeper the height of nanopillars. When we used

the Ar gas in the second step of the etching process, Ar starts to etch the nickel nano

clusters and preformed silicon nitride nanopillars isotropically. In the absence of a

covering of nanoclusters, the etching rate of the top region of etched silicon would be

slightly greater than that at the bottom.[49] For this reason, in the deeper nanopillar

structures [i.e., Fig. 4.5(c)] the slower etching rate of the bottom region may form the

longitudinal base of the nanocones.

The relations between etching time with average reflectance and structure height

for nanopillars and nanocones have been shown in Figs. 4.6(a) and 4.6(b),

respectively.

Etching time (s)

100 120 140 160 180

Average reflectance (%)

4 6 8 10 12 14 16 18

Nanocones Nanopillars

(a)

Figure 4.6 : Relations of etching time with (a) Average reflectance and (b) Structure height for the fabricated nanocone and nanopillar structures on silicon

nitride film. (c) The reflectance spectra comparison of nanocone and nanopillar structures with almost same height for the wavelength from 190 to 1000 nm.

From Fig. 4.6(a), it has been observed that nanocone structures shows lesser

average reflectance than the nanopillar structure when we increase the etching time.

Also it has been seen from Fig. 4.6(b) that the height of the nanocone structures

higher than nanopillar structures which is clearly understood effect of the second step

etching process by the Ar gas. The reflectance spectra for both nanocone and

nanopillar structures with almost similar height were compared in Fig. 4.6(c). It has

been observed that though the average reflectance for both nanocone and nanopillar

structures were less than 5 %, the average reflectance of nanopillar structures were

better than the nanocone structure with similar height. Also it was observed that for

the nanocone structures, the reflectance would be less than 6 % for the shorter

wavelength (i.e., from 190 to 300 nm) and longer wavelength (i.e., from 700 to 1000

nm) as well. Whereas, for nanopillar structures, the reflectance would be less than 6 %

for the wavelength range from 400 to 800 nm. Since both the structures could produce

an average reflectance of less than 5 %, which is very good for an ARC to be applied

in solar cell, it is believed that silicon nitride nano structures would be very useful for

replacing the DLAR coatings and can give a better or comparable performance. But,

nanocone structures have been observed to provide lower average reflectance of

below 6 % for a longer range of heights as compared to nanopillar structures. So

nanocone would be better structures to be used as silicon nitride sub-wavelength

structures for solar cell applications due to the tolerance of wider heights variation of

the structure.

4.4 Summary

An easy and scalable non-lithographic approach for creating nanocone and

nanopillar structured antireflection coatings directly on silicon nitride using nickel

nano cluster and ICP etching method for the solar cell application have been

developed. The one step ICP etching process using a gas mixture of CF₄/O₂ can

produce nanopillar structures and two step etching process using gas mixture of

CF4/O2 in first step and Ar in second step can produce nanocone structures on silicon

nitride film using nickel self assembled nickel nanocluster mask. The measured

reflectance for both the structures shows a great promise to be used in solar cell to

improve the efficiency because of its lower average reflectance of less than 5 %.

Nanocone structures have been observed to provide lower average reflectance of

below 6 % for a longer range of heights as compared to nanopillar structures. So it

can be concluded that nanocones are better structures to be used as silicon nitride

sub-wavelength structures for solar cell applications due to the tolerance of wider

heights variation of the structure.

Chapter 5

Solar Cell Fabrication

In this chapter, we will present our preliminary result of the fabricated silicon solar

cell with silicon nitride SWS and will compare with the silicon solar cell with silicon

nitride SLAR. A brief description of the fabrication method will be presented which

was carried out in collaboration with MOTECH Inc., Tainan.

5.1 Fabrication Process

Figure 5.1: Fabrication process steps for Solar Cell with SWS (a) Emitter process ( done at MOTECH Lab) (b) SWS Process (done at CSDLAB) (c) Post

process (done at MOTECH Lab)

.

The flow chart of the fabrication process for a silicon solar cell with silicon nitride

SWS is shown in Fig. 5.1. This fabrication was done by CSDLab in collaboration

with MOTECH Inc., Tainan. As shown in Fig. 5.1, the part shown in Fig. 5.1(a) and (c)

were done at laboratory of MOTECH Inc., Tainan and part shown in Fig. 5.1(b) was

done in our compound semiconductor laboratory (CSDLab).

First, p-type silicon wafer was cleaned with H2SO4/H2O2 and followed by

chemical polishing to remove surface damage. Then emitter was formed by

phosphorous diffusion by supplying phosphorous trichloride oxide (POCl3) to the

silicon wafer. In this process, p-type silicon wafer was loaded into quartz boat which

was slowly moved into the middle of a fused quartz tube in a resistance-heated

horizontal furnace. Furnace temperature for the diffusion was held at about 900 °C.

Nitrogen was used as a carrier gas. During the diffusion process the following

reaction takes place to form phosphosilicate glass (PSG):

2

reported here because of some confidential issue.

Then, in our CSDLab, the silicon nitride SWSs were formed on the front surface

of the solar cell using the fabrication process described in chapter 3, which was

briefly described in Fig. 5.1(b).

After the SWS process was done, the post processing of the cell was done at

laboratory of MOTECH Inc., Tainan again. The electrodes were formed by standard

screen printing method and the cell characterization was performed to get the open

circuit voltage (VOC), short-circuit current density (JSC), efficiency (η), and fill factor

(FF) under AM1.5G conditions.

5.2 Results and Discussion

(a) (b)

Figure 5.2: Fabricated Silicon Solar Cell with (a) Si3N4 SLAR and (b) Si3N4

SWS.

The measured results tabulated in Table 4. The measured results for the fabricated

silicon solar cell with silicon nitride SWS and silicon solar cell with silicon nitride

single layer antireflection (SLAR) coating has been compared in Table 4. It has been

observed that the VOC value of the SWS solar cell is decreased by 0.001 V as

compared to SLAR solar cell. Also, the fill factor of the SWS solar cell is decreased

by 0.04% compared to SLAR solar cell. But, the JSC and efficiency of SWS has been

improved by 2.9 V and 1.09% as compared to SLAR solar cell. The decrease of FF

suggests that there must be some insufficient electrical contact, which is also

confirmed by the high reverse current of the SWS solar cell compared to SLAR solar

cell. So there is a need to improve the fabrication process specifically the post process

steps of Fig. 5.1(c).

Table 5: Measured Solar Cell parameters for Silicon Nitride SLAR and Silicon Nitride SWS. The difference is also tabulated.

VOC (V) JSC (mA/cm²) FF Efficiency (%) Irev (A)

Si

3

N

4

SLAR 0.596 30.8 64.24 11.77 0.22

Si

3

N

4

SWS 0.595 33.7 64.2 12.86 0.63

Difference -0.001 +2.9 -0.04 +1.09 +0.41

The increase of efficiency by 1.09% for a SWS solar cell confirms our simulated

results as described in chapter 2. Since the results obtained so far is preliminary result,

we expect that the efficiency can be improved more if the post process steps of the

Silicon solar cell with silicon nitride SWS will be improved significantly.

5.3 Summary

A silicon solar cell with silicon nitride SWS has been fabricated. The preliminary

results shows an improvement of 1.09% in cell efficiency compared to a solar cell

with 80 nm silicon nitride single layer antireflection coating. This result is agreed with

our simulated results discussed in chapter 2. It is believed that by improving the post

processing in solar cell fabrication, the efficiency can further be improved.

Chapter 6

Conclusions and Future Works

6.1 Conclusion

In this thesis, we have presented simulation and fabrication results of designed silicon

nitride sub-wavelength structures for solar cell applications.

Using the results of rigorous coupled wave analysis simulation for the pyramid,

cone, parabola and cylinder shaped silicon nitride sub-wavelength structures, the ratio

of silicon nitride sub-wavelength structures height and non-textured part of silicon

nitride has been optimized. The reflectance results for the optimized sub-wavelength

structures have been compared in terms of effective reflectivity. From the study of

different shaped Si3N4 SWS, the cone shaped SWS shows the lowest effective

reflectance for same volume as compared to pyramid, cylinder and parabola shaped

Si3N4 SWSs.

Then lowest effective reflectivity sub-wavelength structures were compared with

previously optimized 80nm Si3N4 SLAR and 80 nm / 100 nm Si3N4 / MgF2 DLAR. A

low effective reflectivity of 3.43% can be obtained for a silicon nitride SWS height

and non-etched layer of 150 nm and 70 nm respectively, which is less than 80 nm

Si3N4 SLAR of 5.41% and comparable with Si3N4 / MgF2 DLAR of 5.39%.

Based on our simulation result, we found a great potential of Si3N4 SWS to

increase the efficiency of silicon solar cell as compared to single layer Si3N4 ARC by

reducing the reflection. So, an easy and scalable non-lithographic approach for

creating nanocone and nanopillar structured antireflection coatings directly on silicon

nitride using nickel nano cluster and inductively coupled plasma etching method for

the solar cell application have been developed. The one step ICP etching process

using a gas mixture of CF4/O2 can produce nanopillar structures and two step etching

process using gas mixture of CF4/O2 in first step and Ar in second step can produce

nanocone structures on silicon nitride film using nickel self assembled nickel

nanocluster mask.

We observed the lowest average reflectance for the wavelength from 190 nm to

1000 nm of 4.66 % for the Si3N4 nanopillar using ICP with CF4/O2 etching gas and

etching time of 120 sec for average height of 155 nm. Similarly, for the wavelength

from 190 nm to 1000 nm, lowest average reflectance of 4.82 % has been achieved for

the Si3N4 nanocone structure using ICP with CF4/O2 etching gas and etching time of

100 sec in first step followed by Ar etching for 90 sec in second step.

The measured reflectance for both the structures shows a great promise to be

used in solar cell to improve the efficiency because of its lower average reflectance of

less than 5 %. Nanocone structures have been observed to provide lower average

reflectance of below 6 % for a longer range of heights as compared to nanopillar

structures. So it can be concluded that nanocones are better structures to be used as

silicon nitride sub-wavelength structures for solar cell applications due to the

tolerance of wider heights variation of the structure, which is also seen before from

our calculations.

A silicon solar cell with silicon nitride SWS has been fabricated. The preliminary

results shows an improvement of 1.09% in cell efficiency compared to a solar cell

with 80 nm silicon nitride single layer antireflection coating. This result is agreed with

our simulated results. It is believed that by improving the post processing in solar cell

fabrication, the efficiency can further be improved. Therefore, For the consideration

of working wavelength region of silicon-based solar cells, the SWS with height of 140

- 160 nm is suitable as the antireflective structure and believed to increase the solar

cell performance as compared to solar cell w/ SLAR structure.

6.2 Future Work

1. There is a need of study of silicon nitride sub-wavelength structures with

non-periodic nature with different shape effects to get the exact calculation of

reflectance.

2. More theoretical study is needed with different TE and TM wave in to

consideration for more accurate results. In our calculation we have considered

only TE wave.

3. Genetic algorithm or swarm optimization process can be used with PC1D

calculation to accurate prediction of the electrical characteristics of solar cell.

4. Since the process developed for silicon nitride sub-wavelength structure works

at high temperature (i.e. 850 ^οC), the process is not suitable for fabrication of

silicon nitride SWS on GaAs surface. So there is need to develop a new

fabrication method which can use low temperature rapid thermal annealing to

form nanomasks.

5.

Although, the preliminary results obtained from our silicon solar cell fabrication using silicon nitride sub-wavelength structures show a great

promise in efficiency improvement compared to silicon nitride single layer

antireflection coated silicon solar cell, the efficiency of silicon solar cell is

lower as compared a commercial silicon solar cell available in market. This is

believed to the low electrical contact formed between electrode and SWS

surface. So, more work is needed in electrode formation in the process of solar

cell fabrication to improve the electrical contact of the solar cell so that the

efficiency of the solar cell will be improved.

Appendices

Main Entrance for SWS Reflection Calculation

% Initialization

[last_token,ext] = strtok(line);

inputA(i,1)=str2num(last_token);

[last_token,ext] = strtok(ext);

inputA(i,2)=str2num(last_token);

nc = 2.03;

hc = (6.63*10^34)*(3*10^8);

lamda1 = 1.1071*10^(-6);

eps = 1.16858 * 10^1;

A = 9.39816*10^(-13);

B = 8.10461*10^(-3);

theta = 0;

nc = 2.1;

theta = 0;

% Calculate the reflectance with given height and wavelength

lambda = [400:5:1000];

diameter = 50 * nm;

height_array = [10:5:250];

pyrlen = diameter;

s_array = [10:5:250];

inFP = fopen('test.txt','w');

for height = height_array

for s = s_array

for k = 1: size(lambda, 2) ns(k) =

sqrt(eps+(A/(lambda(k)*nm)^2)+((B*(lamda1^2))/(((lambda(k)*nm)^2)-lam da1^2)));

% ref(k) = (shape, refractive index of Si₃N₄, refractive index of Si, non-etched Si3N4, Base Diameter, Angle of incidence, height of SWS, wavelength)

ref1(k) = rcwa2(2, nc, ns(k), s*nm, diameter, theta, (height+s)*nm, lambda(k)*nm, pyrlen);

rlambdanlambda(k)=(ref1(k)*inputA(k,2))/(hc/lambda(k));

nlambda(k)= inputA(k,2)/(hc/lambda(k));

end

sum_nlambda=sum(nlambda*5)/(1000-400);

eff_ref1=sum(rlambdanlambda*5)/(1000-400);

eff_ref=eff_ref1/sum_nlambda;

savefile =

strcat('_height=',num2str(height),'_s=',num2str(s),'_ref=',num2str(ef f_ref),'.txt');

fprintf(inFP,'%d\t %d\t %f\n',height,s,eff_ref);

datamat = zeros(size(lambda,2), 2);

datamat(:, 1) = lambda';

datamat(:, 2) = ref1';

save(savefile, 'datamat', '-ASCII');

end end

fclose(inFP);

Function for Calculation of Reflectance

% shape: 0 = pyramid 1 = nipple cylinder 2 = parabola 3 = Cone

function ref = rcwa(shape, nc, ns, s, diameter, theta, height, lambda, leng)

nm = 10^(-9); % nanometer value

nair = 1.0; % refractive index of air

layers = 100; % number of layers in the nipple

if (shape == 1)

layers = 2;

end

layer_width = (height-s) / (layers-1); % average layer height

d = ones(layers,1); % height of each individual layer

n = ones(layers, 1);

r = ones(layers, 1);

% The normalized electric field(in the y-direction) for the input and

% output regions and in each of layers may be written as

% E0=[exp(-jk1z)+R*exp(jk1z)]exp(-jkz), z<=0

% El={Pl*exp[-k0*gammal(z-D(l-1))]+Ql*exp{-j[kx*x+k2z(z-Dl)]}, z>=DL

% Et=T*exp{-j[kx*x+k2z(z-DL)]}

k0 = 2 * pi / lambda;

kx = k0 * nair * sin(theta);

k1z = k0 * nair * cos(theta);

k2z = k0 * sqrt(ns^2 - nair^2 * sin(theta)^2);

% Determine the refractive index for each layer

outputdata=[];

for k = 1 : layers

% Dl= summation over p=1 to l dp h = (1-k/(layers-1)) * (height-s);

if (shape==0) % pyramid fill factor

f = 4 * inv_profile(shape, leng, height, h)^2 ...

/(sqrt(3)*diameter^2);

elseif (shape==1) % nipple cylinder

f = pi * inv_profile(shape, leng, height, height-h)^2 ...

/(sqrt(3)*diameter^2);

elseif (shape==2) % parabola

f = pi * inv_profile(shape, leng, height, h)^2 ...

/(sqrt(3)*diameter^2);

r(layers)=i * sqrt(n(layers)^2 - nair^2 * sin(theta)^2);

% loop over all layers from bottom to top for k = layers : -1 : 1

% Constrcut the k-th layer matrices

matt1(1,1) = 1;

matt1(1,2) = exp(-k0*r(k)*d(k));

matt1(2,1) = r(k);

matt1(2,2) = -r(k)*exp(-k0*r(k)*d(k));

matt2(1,1) = exp(-k0*r(k)*d(k));

matt2(1,2) = 1;

matt2(2,1) = r(k)*exp(-k0*r(k)*d(k));

matt2(2,2) = -r(k);

% Multiply the vector vect with the matrix for the new vector vect = matt1 * inv(matt2) * vect;

% This is the inverse of the profile function

function r = inv_profile(shape, diameter, height, z)

if (shape == 0) % pyramid

r = (diameter/2)*(1-(z/(height)));

elseif (shape == 1) % cylinder

r = diameter/2;

elseif (shape == 2) % paraboloid

r = (diameter/2)*sqrt(1-z/(height));

elseif (shape == 3) % cone

r = (diameter/2)*(1-z/(height));

end

return

return

在文檔中 用於太陽電池之氮化矽次波長抗反射結構設計與樣品試製 (頁 85-0)