2024年4月11日发(作者：)

具有选择性前表面场的高效N型背结太阳能电池

王伟，盛健，袁声召，盛赟，蔡文浩，张淳，冯志强， Pierre Verlinden

光伏科学与技术国家重点实验室，天合光能有限公司，天合路2号，常州新北区光伏产业园，江苏，213031，

@

摘要：高效N型单晶太阳能电池由于其低的衰减，一直受到广泛关注。本文针对N型单晶背结电池，

正面制绒面磷扩散形成前表面场（FSF），并进行选择性刻蚀优化，PECVD沉积SiNx钝化后既得到了

低的正面J

0

（轻掺区36.4fA/cm

2

），又保证了低的正面接触电阻。背面采用硼扩散形成P-N结，钝化后

激光开槽，最后双面采用传统的丝网印刷技术，形成正背面电极，在154.83cm

2

电池片上得到平均20.11%

的电池效率， 659.6mV的开压（V

oc

），39.34mA/cm

2

的电流密度（J

sc

），和77.49%的填充因子（FF），

最高效率达20.17%。

关键词：N型太阳能电池；硼扩散；背结；选择性前表面场；

High-efficiency n-type silicon solar cells

with

Selective front surface

field and rear boron-doped junction

Wei Wang, Jian Sheng, Shengzhao Yuan, Sheng Yun, Wenhao Cai, Chun

Zhang, Zhiqiang Feng, Pierre Verlinden

State Key Laboratory of PV Science and Technology, Trina Solar, No.2 Trina Road, Trina PV

Park, New District, Changzhou, Jiangsu, China, 213031. @

Abstract: High-efficiency n-type silicon solar cells are actively developed in the photovoltaic industry,

because of low Light-Induced Degradation (LID). This paper presents the results of the development of

high-efficiency n-type mono-crystalline silicon solar cells with a rear junction. Large area (154.83cm

2

)

n-type solar cells were fabricated using a Boron-doped rear junction and a selective Phosphorous-doped

Front Surface Field (FSF). Using a PECVD-grown silicon nitride front passivation layer, an excellent

front surface passivation was achieved with a measured J

0

as low as 36.4fA/cm

2

on a textured surface,

and at the same time a low contact resistivity. Laser ablation was used to open local contacts to the rear

junction through the rear passivation layer. Finally, front and back screen-printed contacts were formed as

on a conventional solar cell. Through this technology, 20.11%-average efficiency n

+

-n-p

+

cells were

obtained with a V

oc

of 659.6mV, a J

sc

of 39.34mA/cm

2

, and a fill factor of 77.49%, while the champion

efficiency reached 20.17%.

Keywords: n-type silicon solar cells; boron diffusion; rear junction; selective FSF.

1

Introduction

The structural feature of n-type rear junction solar cells requires the base of the cells to have

excellent bulk lifetime and front surface passivation. The minority carriers, generated by the

illumination coming from the front surface, have to travel through the base of the cells to the rear

side, where they are collected by the Boron emitter. The phosphorous diffusion Front Surface

field (FSF) is a good method to passivate the front surface with a High-Low junction and to

ensure a low contact resistance. The FSF can make the cell more tolerant to changes in surface

recombination velocity than without a FSF [1]. Low surface doping concentration of the FSF

reduces the recombination at the front surface and improves the internal quantum efficiency

(IQE) at short wavelengths. A shallow FSF gives a broad degree freedom in the choice of base

doping concentration [1]. Homogeneous FSF have a low manufacturing cost because of its

simpler procedure. However, the surface doping concentration of FSF should be high enough to

obtain low contact resistivity [2]. Therefore, the best design of the FSF is a trade-off between the

series resistance and the IQE at short wavelengths. Selective FSF, i.e. designing the FSF with a

high surface doping concentration and low sheet resistance under the contacts and with a low

surface doping concentration and high sheet resistance in the passivated regions, allows to

increase the cell V

oc

and IQE without increasing the contact resistance [2] [3]. The improvement

in efficiency due the Selective FSF comes, however, with an increase in manufacturing cost.

There are several methods for forming Selective FSF [4], including using a double diffusion

process, an etch-back process or a laser doping process. The etch back, is considered as simple,

low cost, and easy to implement in industrial production lines.

2 Experimental details

Figure 1 shows the process flow for a Homogeneous and a Selective FSF n-type rear

junction cells. N-type Cz, 154.83cm

2

, 200m-thick, 10~13 wafers were used for the

cells fabrication. A TMAH solution was applied to etch the saw-damaged layer and to polish the

raw wafers. The boron-doped rear emitter was performed in a tube furnace with a BBr

3

diffusion

process. The front surface Boron-Silicate Glass (BSG) was removed by diluted hydrofluoric acid

(HF). A TMAH solution was used to form the low-reflection pyramid structure on front surface,

and a standard POCl

3

tube furnace diffusion was used to form two kinds of Homogeneous FSF,

the heavier doping one being used for the etch back process to form the Selective FSF solar cells.

The contact area in the Selective FSF process was masked by a resist paste, while the no-contact

area was etched by a hydrofluoric and nitric acid (HF/HNO

3

) mix-solution. The sheet resistance

after etching was controlled by the ratio of HF to HNO

3

and the etching time. Subsequently, the

Phosphorous-Silicate Glass (PSG) and BSG were removed by dipping the wafers in a diluted HF

solution. A PECVD-grown silicon nitride was used for the front passivation layer, and laser

ablation was used to open local contacts to the rear junction through the rear passivation layer.

Finally, front and back screen-printed contacts were formed and fired as on a conventional p-type

solar cell.

Surface doping concentration profiles were tested by ECV Measurement. A QSSPC lifetime

tester WCT-120 was used to monitor the passivation effect of the different FSF’s and the

recombination parameter J

0

of the rear emitter. The cells electric performance was measured by

conventional I-V tester with a steady-state solar simulator. A Quantum Efficiency Measurement

System and High-Performance Lambda spectrometers were used to characterize the quality of

the FSF passivation, the bulk lifetime and the rear emitter , between 300~1200nm wavelength.

Homogeneous FSF Selective FSF

Damage etching and polishing

BBr

3

furnace diffusion

Front surface BSG removal and texturing

POCl

3

furnace diffusion

Selective FSF formation

Both side passivation layer depositions

Rear passivation layer laser ablation

Screen printing and co-firing

Figure 1: Process flow for a Homogeneous solar cell and Selective FSF design solar cell using

the etch-back method.

3 Results and discussion

3.1 Doping profile

Figure 2 shows the surface phosphorous doping concentration profile of the Homogeneous

FSF and the etched back area of the Selective FSF. The surface doping concentration of

Homogeneous FSF is 6e20 cm

-3

, the junction depth was 0.28m. For the Selective FSF cells,

no-contact area surface doping concentration was etched from 7.2e20 cm

-3

to 2.6e19 cm

-3

by A

time and to 1.6e19 cm

-3

by B time (B is longer than A). The doping depth was etched from

0.32m to 0.28m and 0.25m respectively. Resulting from to the decrease in surface doping

concentration and doping depth, a significant improvement of front surface J

0

was expected. In

the contact area of the Selective FSF cells, protected by the resist mask, the surface concentration

was 7.2e20 cm

-3

, which is considered as enough for obtaining a good contact resistivity.

3

D

o

p

i

n

g

c

o

n

c

e

n

t

r

a

t

i

o

n

(

a

t

o

m

s

/

c

m

)

1E20

Homogeneous FSF diffusion

Selective FSF diffusion

Selective FSF diffusion etching back A

Selective FSF diffusion etching back B

1E19

1E18

1E17

0.000.050.100.150.200.250.30

Depth

(



m

)

Figure 2: ECV measurement of Phosphorous doping profiles of Homogeneous FSF and

Selective FSF.

3.2 Reducing the J

0

of front surface

To evaluate the contribution of low surface concentration in the no-contact area, obtained by

the etch back process; the J

0

in no-contact area was compared with the Homogeneous FSF

samples. The results are showed in Table 1. The J

0

of Selective FSF is decreased to 36.4fA/cm

2

for the A etch back process, and 31.5fA/cm

2

for the B etch back process. As expected, the

Selective FSF in the no-contact area with the longest etch back process has the lowed J

0

since it

has a lower surface doping concentration and shallower doping depth (Figure 2). During the

QSSPC measurement of the J

0

and effective lifetime, the Minority Carrier Density (MCD) value

was set at 1e16cm

-3

for J

0

testing and 1e15cm

-3

for effective Lifetime or implied V

oc

testing. The

average substrate doping concentration of the wafers is 4.54e14 cm

-3

. The implied V

oc

improved

10mV with the etching back time increasing from A to B.

Table 1: Effective lifetime, J

0

and implied V

oc

testing results on samples with Homogenous FSF

and Selective FSF

Groups

Homogeneous FSF

Selective FSF A

Selective FSF B

Lifetime(s)

466.5

610.6

581.9

Implied V

oc

(mV)

648

656

666

J

0

(fA/cm

2

)

156.3

36.4

31.5

3.3 Cell efficiency

Resulting from the lower J

0

in the passivated no-contact area, the cells with the Selective

FSF demonstrated a better V

oc

as expected. The electric performance of the different cells

presented in Table 2. Selective FSF A cells have a 11.2mV increase in V

oc

compared to the

Homogeneous FSF cells and a 0.26mA/cm

2

gain in J

sc

. The FF of the Selective FSF cells is

almost identical to the FF of the Homogeneous FSF cells despite the increase in FSF sheet

resistance. The sheet resistance of the Selective FSF with the B process is, however, higher than

with the A process, which results in a slightly lower FF. The improvements in V

oc

and J

sc

, for

Selective FSF cells with the B process compared to the A process or compared to the

Homogeneous FSF process, come from the lower recombination and better passivation of the

front surface. Influenced by metallization area, the cell V

oc

of the Selective FSF cells did not

reach the implied V

oc

value showed in Table 1. The V

oc

improvement between the Selective A

and B is small, which is in agreement with the J

0

results shown in Table 1 well. The J

0

for

Selective FSF B was only 5fA/cm

2

lower than for Selective FSF A. The J

sc

of Selective FSF B is

slightly lower than for Selective FSF A. The reason is assumed be because of the pyramid

rounding. The best average cell efficiency is obtained with the Selective FSF A process with a

value of 20.11%. With this FSF A process, the champion cell efficiency reached 20.17%.

Table 2: Performance of Homogeneous and Selective FSF cells

J

sc

[mA/cm

2

]

V

oc

[mV]

FF[%]

Eta[%]

Champion Eta[%]

Homogeneous FSF

39.08

648.4

77.47

19.63

19.90

Selective FSF A

39.34

659.6

77.49

20.11

20.17

Selective FSF B

39.24

660.2

76.75

19.88

20.07

3.4 Quantum efficiency analysis

The IQE and reflectance of typical solar cells with either the Homogeneous FSF or the

Selective FSF A process are presented in Figure 3. Significant reduction in J

0

in the passivated

no-contact area of the Selective FSF A process (shown in Table 1) is responsible for the

improvement in internal quantum efficiency in the short wavelength region compared to

Homogeneous FSF cells. This result illustrates that a Selective FSF process can improve the IQE

of short wavelength light. Figure 3 also shows that the reflectance of Selective FSF cells is

slightly larger than the reflectance of Homogeneous FSF cells, probably due to the pyramid

rounding during the etch back process.

100

80

I

Q

E

&

R

e

f

l

e

c

t

a

n

c

e

(

%

)

60

40

IQE-Homogeneous FSF

Refl.-HomogeneousFSF

IQE-Selective FSF

Refl.-SelectiveFSF

20

0

41200

Wavelength(nm)

Figure 3: IQE and reflectance curve of Homogenous FSF and Selective FSF A cell.

4 Summary

With FSF, Large area (154.83cm

2

) n-type rear junction solar cells with different FSF

processes were fabricated. The QSSPC measurements shows that, compared to Homogeneous

FSF cells, Selective FSF cells fabricated with a simple etch back process can have a significantly

lower J

0

and lower front surface recombination. The optimization of the FSF process results in

improvement in a 11.2mV improvement in V

oc

, a 0.26mA/cm

2

improvement in J

sc

and a 0.48%

absolute improvement in efficiency, in average. The IQE curve also shows a better response in

short wavelength range (300-500nm). With this technology, 20.11%-average efficiency n

+

-n-p

+

cells were obtained with a V

oc

of 659.6mV, a J

sc

of 39.34mA/cm

2

, and a fill factor of 77.49%,

while the champion efficiency reached 20.17%.

References

[1] M. Hermle, F. Granek, O. Schultz and S.W. Glunz: . 103, 054507(2008).

[2] V. E. Lowe and A.C. Day: IEEE Trans. Electr. Dev. 31,626(1984).

[3] D.K. Schroder and D.L Meier: IEEE Trans. Electr. Dev. 31,637(1984).

[4] J. H. Bultman, R. Kinderman, J. Hoomstra and M. Koppes: 16

th

EU PVSEC Glasgow, 2000,

pp. 1424-1426 (VA1/56).