Si heterojunction solar cells

Si heterojunction solar cells

Accepted Manuscript Hole-selective NiO:Cu contact for NiO/Si heterojunction solar cells Xueliang Yang, Jianxin Guo, Yi Zhang, Wei Liu, Yun Sun PII: S...

926KB Sizes 0 Downloads 77 Views

Accepted Manuscript Hole-selective NiO:Cu contact for NiO/Si heterojunction solar cells Xueliang Yang, Jianxin Guo, Yi Zhang, Wei Liu, Yun Sun PII:

S0925-8388(18)30928-9

DOI:

10.1016/j.jallcom.2018.03.067

Reference:

JALCOM 45297

To appear in:

Journal of Alloys and Compounds

Received Date: 1 December 2017 Revised Date:

4 March 2018

Accepted Date: 6 March 2018

Please cite this article as: X. Yang, J. Guo, Y. Zhang, W. Liu, Y. Sun, Hole-selective NiO:Cu contact for NiO/Si heterojunction solar cells, Journal of Alloys and Compounds (2018), doi: 10.1016/ j.jallcom.2018.03.067. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Hole-selective NiO:Cu Contact for NiO/Si Heterojunction solar

2

cells

RI PT

1

3 4

Xueliang Yanga,b, Jianxin Guoc, Yi Zhanga,b, Wei Liu*, a,b, Yun Suna,b

SC

5 6

a

7

Tianjin 300071, P. R. China

8

b

9

Nankai University, Tianjin 300071, P. R. China

M AN U

Institute of Photoelectronic Thin Film Devices and Technology, Nankai University,

Key Laboratory of Photoelectronic Thin Film Devices and Technology of Tianjin,

10

c

11

China

12

TE D

College of Physics Science and Technology, Hebei University, Baoding 071002, P. R.

Abstract:

14

Heterojunction architectures become a major trend in the development of high power

15

conversion efficiency (PCE) c-Si solar cell in recent. However, NiO as a prevailing

16

material in other application is not favored in c-Si solar cells. In this study, a novel solar

17

cell architecture using nonstoichiometric p-type NiO thin films as a hole-selective,

18

dopant-free contact to n-type crystalline silicon (n-Si) is successfully fabricated. We

19

achieved a power conversion efficiency (PCE) of 4.3% in the pure NiO/n-Si

20

heterojunction solar cell, which is the first case to be reported. Further Cu-incorporation

AC C

EP

13

1

ACCEPTED MANUSCRIPT

enhanced the work function and conductivity of NiO films (NiO:Cu, for short),

2

increasing built-in potential and favoring hole transport in the NiO:Cu/n-Si

3

heterojunction, which results in a record PCE of 9.1% among all bulk-like NiO/Si

4

heterojunction solar cells. Analysis of the spectra of Cu 2p and Cu LMM Auger lines,

5

measured by X-ray photoelectron spectroscopy (XPS), shows that the fabricated NiO:Cu

6

film is a mixture of metallic Cu, Cu2O and CuO multi-phase coexistence. Among these

7

species, the proper Cu incorporation contributes the improvement of acceptor

8

concentration due to the dominant phase of Cu+, but the excessive Cu incorporation

9

creates defect states within the band gap, increasing recombination and deteriorating

SC

M AN U

10

RI PT

1

interface quality, which leads to S-shape current density-voltage (J-V) characteristics.

12

TE D

11

Key words: hole-selective contact, NiO, transition metal oxide, heterojunction solar cells

15 16 17 18

AC C

14

EP

13

19

*Corresponding author

20

E-mail address: [email protected] 2

ACCEPTED MANUSCRIPT

1

1. Introduction Currently, high power conversion efficiencies (PCE) of crystalline silicon (c-Si) solar

3

cells, such as passivated emitter and rear contacts (PERC) [1], tunnel oxide passivated

4

contacts (TOPCon) [2], interdigitated back contact (IBC) [3] and heterojunction with

5

intrinsic thin-layer (HIT) solar cells [4], are still dependent on phosphorus/boron doped

6

Si bulk or thin film materials as the emitter or the back surface field (BSF) to separate

7

photogenerated electrons and holes within both industry and research laboratories. There

8

are a host of issues, most notably parasitic absorption, auger recombination and other

9

heavy doping effects. Another technological complexities associated with doping, such as

10

high processing temperatures, dopant glass removal and junction isolation, have to be

11

also

12

hetero-contacts (DASH cells) based on transition metal oxides (TMOs), such as MoO3,

13

WO3, and V2O5 as hole contact, have attracted much attention because this kind of solar

14

cell can alleviate issues associated with doping as stated above, using their advantages of

15

wide band gap and high work function [8-11]. In the family of TMOs, another attractive

16

member, nonstoichiometric NiO thin film, as a p-type transparent conductive oxide

17

material, has been introduced into organic solar cells [12], perovskite solar cells [13] and

18

other type of devices, such as lithium ion batteries [14], gas sensors [15], electrochromic

19

display devices [16], and antiferromagnetic layers [17] owing to its excellent optical and

20

electrical properties. However, for this plausibly promising material, the progress in

[5-7].

Recently,

phosphorus/boron

dopant-free

asymmetric

AC C

EP

TE D

challenged

M AN U

SC

RI PT

2

3

ACCEPTED MANUSCRIPT

conjunction with Si solar cells is not smooth, and there are almost no reports about pure

2

NiO/Si heterojunction (HJ) solar cells resulting from several intractable problems, such

3

as an intricate defect mechanism, carrier transport associated with Ni vacancies (VNi) in

4

the films, work function tuning, NiO/Si interface states and material conductivity. Hsu et

5

al. reported a Li doped NiO (NiO:Li)/Si heterojunction solar cell with an optimized PCE

6

of up to 6.3% using RF magnetron sputtering [18-20], offering an additive-optimized

7

roadmap for the development of these devices. In fact, a great number of studies only

8

focus on the improvement of NiO material properties using metal additive, for example,

9

monovalent atoms such as Li, Na and K, thanks to their similar ionic radii with Ni2+

10

[21-23]. In addition, several studies have been conducted to investigate the changes in the

11

optical and electrical properties of NiO that occur when incorporating Cu (NiO:Cu)

12

[24-26]. To date, however, some issues are still open in this direction: (i) not much work

13

has been done on application in Si-based solar cells; (ii) the exact mechanism of Cu

14

incorporation is not well understood due to Cu being a variant valence metal with usual

15

valences of +1 and +2; (iii) the effect of Cu incorporation on the performance of c-Si

16

solar cell has not been investigated; and (iv) the corresponding physical insights in this

17

field are lacking.

AC C

EP

TE D

M AN U

SC

RI PT

1

18

In this work, we introduce a solar cell architecture using NiO films with different Cu

19

contents as a hole-selective, dopant-free contact to n-type Si. We first achieved a PCE of

20

4.3% using the pure NiO as a hole-selective contact, which is the first case to be reported, 4

ACCEPTED MANUSCRIPT

and then obtained a highest PCE of 9.1% after incorporating the proper amounts of Cu

2

into the NiO contact layer. The evolution from microscopic mechanisms in NiO:Cu thin

3

film materials, such as Ni vacancies, Cu valances variations, defect states and

4

microstructure phase to macroscopic photoelectric properties of NiO:Cu and performance

5

of solar cell devices, have been investigated in detail.

2. Experimental

M AN U

7

SC

6

RI PT

1

n-type CZ (100) Si wafers (1-3 Ω·cm, 180 µm) after texture, phosphorus-diffusion and

9

glass remove were selected. An ultra-thin nitric acid oxidation of Si (NAOS) layer was

10

formed by immersion of the Si wafer in 68 wt% HNO3 aqueous solution for 15 min for

11

the passivation of interface quality. Then, NiO and NiO:Cu thin films were deposited on

12

the selected wafers by RF magnetron sputtering with a substrate temperature of 300 ˚C,

13

pressure of 0.8 Pa, power of 50 W and Ar flow of 50 sccm from a NiO ceramic target

14

without and with numbers of bonded Cu square chips (see supplementary Fig. S1).

15

Several atoms’ contents in the range of 0-95.8 at. % were adjusted by placing Cu chips

16

with different areas, which were measured by X-ray photoelectron spectroscopy (XPS)

17

(Table 1). At the end of plasma, the samples were cooled under an O2 atmosphere of 103

18

Pa to avoid an oxygen deficiency.

AC C

EP

TE D

8

19 20

5

ACCEPTED MANUSCRIPT

1 Table 1 Chemical composition in NiO:Cu thin films Thin film for devices

Cu area (mm2)

C 1s (at. %)

O 1s (at. %)

Ni 2p (at. %)

0 60

52.4 57.0

27.9 26.2

19.7 13.9

Dev. 3 Dev. 4

110 160

54.2 56.4

26.8 23.0

13.9 8.2

Dev. 5

260

53.1

27.1

0.8

Cu/(Cu+Ni) (at.%)

0.0 2.9

0.0 17.1

5.2 12.5

27.1 60.2

19.0

95.8

SC

Dev. 1 Dev. 2

Cu 2p (at. %)

RI PT

2

3

The n-type double-sides polished wafers (Topsil) and glass (Schott AF 32TMeco)

5

substrates always accompany the growth of NiO thin films for the characteristics of

6

material properties. ITO was deposited at room temperature as an antireflection layer. A

7

magnesium (Mg) thin film with a low work function (3.6 eV) was introduced to obtain a

8

downward energy band at the rear surface of the device to strengthen the function of a

9

back surface field (BSF) [27]. The Ag metal grid was thermally evaporated onto the front

10

side of the samples (1.5×1.5 cm2) using a shadow mask. Lastly, these 1.5×1.5 cm2 cells

11

with the top electrode were cut into 1×1 cm2 cells to remove the conductive edge by the

12

metal thin film deposited by sputtering. All samples were annealed at 200 ˚C to improve

13

the conductivity of ITO and partially recover sputtering damage [28, 29]. The geometry

14

of the device is shown in the supplementary Fig. S2. To facilitate discussion later in this

15

article, the solar cell device with a Cu/(Cu+Ni) ratio (Cu content) of 0-95.8 at. % (Table 1)

16

is denoted by dev. 1-5.

AC C

EP

TE D

M AN U

4

6

ACCEPTED MANUSCRIPT

Photovoltaic (PV) performances of the devices were characterized by current

2

density-voltage (J-V) at the standard test condition (AM1.5, 100 mW/cm2 and 25 °C) and

3

the external quantum efficiency (EQE) (R3011, Enlitech). The microstructure, element

4

valance, and optical and electrical properties of the thin films were measured by X-ray

5

diffraction (XRD) (D8-advance, Bruck), X-ray photoelectron (XPS) spectroscopy

6

(Thermo escalab 250XI, Al-Ka 1486.6 eV), double-beam uv-vis spectrophotometer

7

(Hitachi U4100) and Hall measurement (ET9000). The capacitance vs. voltage (C–V)

8

measurement was carried out with a semiconductor device analyzer (Agilent B1500A).

9 10

3. Results and discussion

M AN U

SC

RI PT

1

Fig. 1 plots the photovoltaic (PV) parameters (open circuit voltage (Voc), short current

12

density (Jsc), fill factor (FF) and PCE) of all devices, and Table 2 provides data and

13

statistics based on three cells for each condition. The devices with NiO hole contact

14

achieve a Voc of 260 mV, Jsc of 36.6 mA/cm2, FF of 45.4%, and PCE of 4.3%, which is a

15

record PCE in pure NiO/Si heterojunction solar cells to the best of our knowledge. When

16

the NiO:Cu with the proper Cu content (27.1%) is developed as hole contact, the PCE of

17

the device increases significantly from 4.3% to 9.1%, corresponding to an improved Voc

18

from 260 mV to 378 mV and a FF from 45.4% to 67.7%. Fig. 1 also presents Suns-Voc

19

and pseudo FF (pFF) measured by a “Suns-Voc” procedure, which was developed by Dr.

20

Sinton based on the measurement of the Voc

AC C

EP

TE D

11

7

ACCEPTED MANUSCRIPT

(a)450

(b) 80 FF&pFF (%)

350 300 250 17.1

27.1

60.2

30

0

17.1

27.1

(d) 10 8

Jsc (mA/cm )

36

PCE (%)

2

40

95.8

34 32 16

6 4 2

0

17.1

27.1

0

60.2 95.8

0

60.2 95.8

17.1

27.1

60.2 95.8

Cu/(Cu+Ni) (at. %)

M AN U

Cu/(Cu+Ni) (at. %)

1

Fig. 1. PV parameters of the devices.

2 3

Table 2 Photovoltaic parameters of NiO:Cu/Si heterojunction solar cells Solar cells

Voc (mV)

Dev. 1

260

Dev. 3

Dev. 4

PCE (%)

36.6

45.4

4.3

36.8 (±0.3)

47 (±1.5)

4.2 (±0.1)

36.5

51.5

6.3

331(±13.4)

36.0(±0.4)

49(±1.9)

5.9(±0.5)

378

35.6

67.7

9.1

377 (±1.3)

35.5(±0.2)

64(±3.4)

8.6 (±0.4)

367

35.9

20.9

2.8

369(±2.2)

35.9(±0.1)

19(±2.2)

2.6(±0.3)

382

30.2

11.5

1.3

367(±21.4)

28.8(±1.9)

12(±0.1)

1.2(±0.1)

AC C

Dev. 5

FF (%)

336

EP

Dev. 2

Jsc (mA/cm2)

TE D

244 (±14.2)

7

Diamond: pFF Rectangle: FF

50

RI PT

0

(c) 38

5 6

60

20

200

4

70

SC

Voc (mV)

Diamond: Suns-Voc 400 Rectangle: Voc

Note: Data and statistics based on three cells of each condition; Numbers in bold are the maximum record values.

8

as a function of the irradiance (e.g., 0-5 suns) without external applied bias voltage or

9

current [30]. It is observed that the Suns-Voc is equal to the standard Voc (std-Voc) for the

10

good devices (dev. 1-3) but are less than std-Voc for the bad ones (dev. 4-5). Generally, for 8

ACCEPTED MANUSCRIPT

the solar cells with metal-Si contact, Suns-Voc is greater than std-Voc; for the solar cells

2

without metal-Si contact, e.g., HIT solar cells using ITO contact to Si, Suns-Voc is similar

3

to std-Voc because the metal-Si interface is recombination free [31, 32]. However, std-Voc

4

is greater than Suns-Voc for the dev. 4 and 5, which is overestimated because of the

5

non-exponential (S-shape) dependence of the J–V curves (see Fig. 2(a)). FF is improved

6

as the proper Cu incorporation into NiO contact. However, FF dramatically decreases

7

while Cu content increase to 60.2%. It is well-known that pFF is not the dependence of

8

series resistance (Rs), but pFF-FF is related to Rs [30]. The large pFF-FF is consistent

9

with their large Rs for dev. 4 and 5 (Fig. 2 (b)). However, the origin of the S-shape J–V

10

curves of dev. 4 and 5 is not Rs, which will be unveiled later. Jsc decreased with the

11

increase of Cu content, which can be explained by the light filtering effect of NiO:Cu thin

12

films (see the EQE measurement of the device in Fig. 2 (c) and absorption spectra of thin

13

films in Fig. 2 (d)) with the exception of the large decrease of Jsc in dev. 5. The

14

absorption coefficient increases with the increase of Cu content but changes exceptionally

15

when the Cu content is 95.8 at. % since that the dominant phase in NiO with the Cu

16

content of 95.8 at. % becomes Cu oxides rather than NiO species (see XRD data in Fig.

17

4). Correspondingly, dev. 5 is a CuxO/Si HJ solar cell with very low Jsc and FF, as

18

reported by the previous work [33, 34].

AC C

EP

TE D

M AN U

SC

RI PT

1

9

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

1

Fig. 2. (a) J-V curves, (b) series resistances, (c) EQE of the devices and (d) absorption of

3

NiO:Cu films with the different Cu contents.

4

TE D

2

Fig. 3 (a) shows the electrical properties of NiO and NiO:Cu thin films, respectively.

6

The conductivity (σ) of the NiO films increases first and then slightly decreases with the

7

increase of the incorporated Cu concentration. The increase of σ mainly results from the

8

very high carrier concentration (n) in NiO:Cu thin films despite the carrier mobility (µ)

9

decreasing with the increase of Cu content. The increase of σ can explain the

10

AC C

EP

5

improvement of FF from dev. 1 to dev. 3.

11

10

ACCEPTED MANUSCRIPT

(a)

(b)

4

50

10

-5

10

-6

10

-7

30 20

15

10

10

5m 60

80

100

(d)

Cu/(Cu+Ni) (at. %) 467.3

4

1

17.

27

502.7 478.9

.1%

341.2

1%

95.8%

0 334.5

-1.0

-0.5

0.0

0.5

1.0

Voltage (V)

1

2

0

20

40

60

10

17

10

16

10

15

400

80

100

Cu/(Cu+Ni) (at. %)

0 17.1% 27.1% 60.2% 95.8%

-3

17 .1%

2 6 0 .2 %

16

40

2

600

SC

20

-2

-2

×10 C (cm F )

3

0

Doping density (cm )

(c)

3

1

0

J0 (A/cm )

17

10

RI PT

-3

n (cm )

σ (S/cm)

n

µ

40 -3

10 σ

30

µ (cm )

19

Diode factor n

60

800

1000

1200

SCR width (nm)

Fig. 3. (a) Conductivity, carrier concentration and mobility of NiO:Cu films, (b) the

3

calculated ideal factor and J0, (c) C-2–V plots, and (d) acceptor concentration as a function

4

of Cu incorporation concentration.

M AN U

2

5

The ideal factor and the saturation current density (J0) (Fig. 3 (b)) of the devices are

7

calculated by fitting the linear region II of the dark J-V curves (supplementary Fig. S3).

8

The proper Cu incorporation (17.1% and 27.1%) can make J0 of NiO/Si HJ solar cells

9

decrease clearly, which explains the improvement of Voc in Fig. 1 (a). The ideal factor

10

increases slightly but always remains less than 1.5 with the increase of Cu content from 0

11

to 60.2 at. %, indicating that the recombination of these solar cells is mainly originated

12

from the depletion region, but the recombination in the electrically neutral region also

13

slightly increases with the increase of Cu content. All devices adopted the same geometry

14

except for the hole-selective contact layer, and thus all the variations in photovoltaic

15

parameters should be derived from the modification of the hole-selective contact layer.

AC C

EP

TE D

6

11

ACCEPTED MANUSCRIPT

The increase of the ideal factor suggests that the recombination centers exist in NiO:Cu

2

thin films. For dev. 5, J0 dramatically increases and results in the decrease of Voc. In

3

particular, the ideal factor is close to 3, indicating that the p-n junction property of the

4

solar cell has changed and is not a Schottky behavior. The transport behavior of carriers

5

in this device is studied to be a bulk limited Poole–Frenkel conduction mechanism (see

6

the supplementary material, specifically Fig. S4 and Note 1), resulting from

7

defect/impurity-generated electron traps in the material [35, 36]. It can be deduced that

8

defect states may occur in NiO:Cu material with Cu content of 96.5 at. %.

M AN U

SC

RI PT

1

The capacitance C-2–V plots for a 1-kHz signal frequency of the devices are shown in

10

Fig. 3 (c). Here, C-2 was linear with the applied voltage, and the intercept corresponds to

11

the built-in potential [37]. With the increase of Cu contents, the built-in potential

12

increases from dev. 1 to dev. 3 and then decreases from dev. 3 to dev. 5. In addition, the

13

acceptor concentration profiles of the samples are shown in Fig. 3 (d). Dev. 3 possesses

14

the highest acceptor concentration, corresponding to the best device performance in Fig.

15

1. The previous studies showed that the origin of the S-shape lies in a reduced built-in

16

potential in the c-Si depletion region, which, in combination with a hole accumulation at

17

the interface, causes an increase in recombination losses [38]. Lower acceptor

18

concentration and built-in potential can explain the occurrence of the S-shape J-V curve

19

in dev. 5. The acceptor concentration in dev. 4 is low, but its built-in potential is qualified,

20

so the origin of the S-shape J-V curve is different from dev. 5. Its problem may be from

AC C

EP

TE D

9

12

ACCEPTED MANUSCRIPT

2

that the enhanced interface recombination also causes that S-shape J-V curve [39]. The

3

clarification of this aspect is linked to the phase species of NiO:Cu films. ∇Cu2O ♦NiO ◊CuO ∀Cu

(200)

(111)



∇ ♦

Intensity (a.u.)

◊ ∇♦

4 5

◊ ∇ ♦ 95.8%

SC

60.2% 27.1%

17.1%

M AN U

30

RI PT

the NiO:Cu/Si interface due to the excessive incorporation of Cu. It has been reported

(220)

1

40

50

2θ (°)

60

0%

70

Fig. 4. XRD patterns of NiO:Cu films with the different Cu contents.

TE D

6

Fig. 4 shows XRD patterns of those thin films. Three obvious diffraction peaks are

8

observed for the pure NiO thin film, which are ascribed to the diffraction from the NiO

9

(111), NiO (200) and NiO (220) planes, respectively. Cu addition leads to the diffraction

10

angle shift of 0.3˚ toward a higher angle, meaning that the lattice slightly shrinks.

11

According to the view that Ni2+ ions are arranged in (100) planes and the high VNi density

12

in (111) ones [18, 40], the depression of the (111) peak in XRD patterns of NiO:Cu film

13

indicates that VNi (Ni3+) may be occupied by the additive Cu. When the additive Cu

14

content reaches 95.6%, the main peaks of NiO depress, and those of Cu oxides occur. It is

15

not easy to distinguish which one of the phase constitutions in the material is Cu2O, CuO

AC C

EP

7

13

ACCEPTED MANUSCRIPT

and Cu because their diffraction peaks are posited at the close angle range of 42-43˚ from

2

XRD. However, it can confirm the phase transformation from NiO to Cu oxides in the

3

hole-selective contact layer of dev. 5. This point corresponds to a series of abnormal

4

changes of PV performance in dev. 5 and material properties in the 95.6 at. % NiO:Cu

5

above, such as Suns-Voc
6

abnormal absorption and ideal factor of >2. CuxO/n-Si HJ solar cells were previously

7

investigated to obtain high Voc but very low Jsc and FF [33, 34], which is consistent with

8

the performance of dev. 5 in this work. Thus, we speculate that the hole-selective contact

9

layer in dev. 5 is not NiO but instead mainly CuxO.

M AN U

SC

RI PT

1

Regarding CuxO/n-Si HJ solar cells, their deteriorative PV performances have been

11

explained in the previous study by the presence of an amorphous interfacial layer,

12

SiaCubOc complex oxide, between p-CuxO and the Si substrate [33]. However, this cannot

13

explain the poor performance of dev. 4. To explore this physical origin, first-principles

14

total-energy calculations were performed and shown in Fig. 5 (the details of the

15

calculated method can be seen in the supplementary materials, specifically Note 2 and

16

Fig. S5). Fig. 5 (a) shows the total density of states (DOS) of the stoichiometric NiO,

17

demonstrating its insulator property. In the presence of 2 VNi, as shown in Fig. 5 (b), the

18

electronic structure shows a half-metallic property, suggesting that nickel vacancies are

19

the dominant point defects responsible for the enhanced p-type conducting electrical

20

properties of NiO, which is consistent with reference [41]. When one of 2 VNi was

AC C

EP

TE D

10

14

ACCEPTED MANUSCRIPT

replaced by one Cu, the states near Ef were reduced, and a small number of defect states

2

emerge within the NiO band gap (Fig. 5 (c)). When two Cu were added to replace 2 VNi,

3

the states near Ef were reduced to zero below, and defect states were increased within the

4

NiO band gap (Fig. 5 (d)). Dev. 4 may belong to this case: the excessive Cu incorporation

5

into NiO leads to the formation of the defect states, increasing interface DOS to favor

6

hole transport limitation, eventually resulting in S-shape J-V characteristics.

-35

-5

0

Energy (eV)

5

1 VNi+1 Cu

40 20 0 -20 -40

EP

-60

7

Spin up Spin down

-5

0

0

-35

-5

(d) 60

5

Energy (eV)

2 VNi

35

-70

TE D

DOS (states/eV)

(c) 60

DOS (states/eV)

0

M AN U

35

-70

(b) 70

Stoichiometric NiO

DOS (states/eV)

DOS (states/eV)

(a) 70

SC

RI PT

1

0

5

Energy (eV) 0 VNi+2 Cu

40 20 0 -20 -40 -60

-5

0

Energy (eV)

5

Fig. 5. The calculated total DOS for (a) a perfect NiO lattice, (b) NiO with two VNi, (c)

9

NiO with two VNi but one VNi was replaced by one Cu, and (d) NiO with two VNi but two

10

AC C

8

VNi were replaced by two Cu.

11 12

In the calculations above, the additive Cu is located at the substitutional VNi since these

13

configurations are more stable, which was proved by the corresponding binding energy of 15

ACCEPTED MANUSCRIPT

-3.27 eV (see the supplementary material, specifically Note 3 and Fig. S6). Due to the

2

replacement of nickel vacancy at lattice points by Cu, the Cu valence may be +2, and we

3

find that this replacement mechanism reduces DOS near Ef and weakens the p-type

4

property of NiO. This is inconsistent with the optimized devices’ (dev. 2 and dev. 3)

5

performance by Cu addition. Li et al reported that Cu+ incorporation into NiO can

6

increase the p-type conductivity of NiO; in their first-principles calculation results, DOS

7

showed an obvious impurity level near Ef, which is contributed by the Cu 3d states with

8

some hybridization with O 2p states [42]. The Cu+-induced DOS improvement is

9

beneficial to solar cell device performance thank to the high acceptor concentration in the

SC

hole-selective contact layer.

(a)

(b) 2+

60.2%

(c)

1+

0+

Cu /Cu

TE D

95.8%

Cu

17.1%

AC C

11

864

Binding energy (eV)

868 930

2+

0+

*

95.8%

Cu /Cu



2+

Cu 60.2%

60.2%

27.1%

860

Cu

* ♦Cu1+

raw data fiting *

Satellite

856

0+

95.8%

raw data fiting

0%

852

2+

raw data fitting

EP

Intensity (a.u.)

Ni Ni3+

M AN U

10

RI PT

1

*

27.1%

17.1%

17.1% 932

934

936

938

27.1%

564

Binding energy (eV)

568

572

576

Binding energy (eV)

12

Fig. 6. XPS patterns of (a) Ni 2p and (b) Cu 2p and (c) Cu LMM Auger lines NiO:Cu

13

films with the different Cu contents.

14

16

ACCEPTED MANUSCRIPT

To identify phase species in the material, the XPS photoemission and Cu LMM Auger

2

spectra were studied. Fig. 6 (a) shows Ni 2p XPS spectra of NiO and NiO:Cu thin films,

3

and the Ni 2p3/2 main peak and its satellite are observed. The main peak consists of two

4

peaks at 854.4 eV and 855.9 eV, which are ascribed to Ni2+ and Ni3+ (VNi) electronic

5

states [43-45], respectively. Each valance state can be quantified from the integrated peak

6

area after Gaussian fitting of a multi-peak deconvolution of the spectra. When the

7

additive Cu content reaches 95.8 at. %, no Ni signal is observed. Fig. 6 (b) shows Cu 2p

8

XPS spectra, and it can be seen that the Cu2+ species increases with the increase of the

9

additive Cu, especially for 95.8 at. % NiO:Cu, and Cu2+ becomes its dominant species,

10

proving that CuO is the main phase in the hole-selective contact layer of dev. 5. Cu 2p

11

spectra did not distinguish Cu1+ and Cu metal state (Cu0) because of the very close

12

binding energies. This distinction can be made, however, by examining the Cu LMM

13

Auger spectra, as shown in Fig. 6 (c). Cu1+ species increases as the additive Cu increases

14

from 17.1 to 60.2 at. % and then decreases as the Cu increases to 95.8%. The addition of

15

monovalent cations, e.g., Li+, can increase the hole concentration of NiO films thanks to

16

their lower valence than Ni [18, 25]. Due to the difference in valance between Cu+ and

17

Ni2+, the work function of NiO:Cu films is improved (see the supplementary Fig. S7),

18

which favors the performance improvement of the NiO/n-Si solar cell device. In addition,

19

from the Cu LMM spectra in Fig. 6 (c), we can still observe: (i) Cu metal state exists in

20

NiO:Cu thin films, and its content increases with the increase of the additive Cu. This is

AC C

EP

TE D

M AN U

SC

RI PT

1

17

ACCEPTED MANUSCRIPT

likely related to the thin film deposition way of a NiO ceramic target in combination with

2

Cu metal chips. (ii) The LMM spectra of 60.2 at. % NiO:Cu can be decomposed into

3

several species, for example, the peak at the binding energy of 571.2 eV contains a

4

mixture of CuO and metallic Cu, which is consistent with reference [46]. In this case,

5

NiO:Cu may be a mixture of material with a transition phase from Cu2O to CuO.

AC C

6

EP

TE D

M AN U

SC

RI PT

1

7

Fig. 7. (a) VB XPS spectra of NiO:Cu films with different Cu contents. (b) and (c) are the

8

close-ups for defect states near Ef and maxima observation, respectively, for (a).

9

Fig. 7 shows the valence band spectra for the NiO and NiO:Cu thin films. It can be

10

observed that an obvious defect state is within the band gap of 96.5 at. % Cu-doped NiO

11

thin film and a small defect state 60.2 at. % Cu-doped NiO film, corresponding to the 18

ACCEPTED MANUSCRIPT

poor performance of these two devices (dev. 4 and 5) while supporting the first-principles

2

calculation results above (Fig. 5 (d)). The existence of defect states would increase the

3

recombination of carriers in the electrically neutral region and deteriorate the device

4

performance, which explains the S-shape J–V curves in Fig. 2 (a). The region near the

5

Fermi level (Ef) for each spectrum is re-plotted for a clearer view, as shown in Fig. 7 (b).

6

Of particular interest is the reduced state density near Ef after Cu incorporation,

7

originating from the effect of Cu2+ doping, which is consistent with the calculated data

8

(Fig. 5 (c) and (d)). The valance band (Ev) positions are revealed, as shown in Fig. 7 (c).

9

Ev-Ef is 0.77 eV for pure NiO thin film but increases to 1.02 eV after Cu incorporation

10

and then decreases with the addition of more Cu. This result is consistent with work

11

function variations (Supplementary Fig. S7).

TE D

M AN U

SC

RI PT

1

According to the chemical modulation of valence band (CMVB) theory [47],

13

monovalent cations into metal oxide would extend the valence band structure, leading to

14

an increase of the hole concentration, but the carrier concentration is as high as 1020 cm-3

15

in the previous and present works [18], plausibly not only resulting from the CMVB

16

contribution but also likely thanks to the existence of metallic Cu in these NiO:Cu films,

17

which was found in this study. The metallic Cu improves the conductivity of material

18

while introducing deep level recombination (see the ideal factor increase in Fig. 3 (b)).

AC C

EP

12

19

We also find the importance of interface passivation, e.g., the NAOS layer was

20

introduced to the devices in this work to ensure good interface quality; if we remove this 19

ACCEPTED MANUSCRIPT

layer, the device performance becomes poor (see supplementary Fig. S8). In addition, we

2

do not discuss the effect of oxygen in NiO:Cu films on material and device performance

3

because a method is used in which all samples are cooled after NiO deposition under an

4

O2 atmosphere of 103 Pa to ensure sufficient oxygen. In fact, if it is lack of oxygen, e.g.,

5

the sample was cooled under a vacuum atmosphere of 10-4 Pa, the device performance is

6

also poor (supplementary Fig. S9a) since the metallic Cu increases (supplementary Fig.

7

S9b). More detailed studies regarding the influence of oxygen will be discussed in

8

another paper.

M AN U

SC

RI PT

1

NiO:Cu films, fabricated by way of an NiO ceramic target in combination with Cu

10

metal chips, were demonstrated to be a mixture of metallic Cu, Cu2O and CuO

11

multi-phase coexistence. Cu2O is beneficial to an NiO:Cu/Si HJ solar cell device because

12

the difference of valance states between Cu+ and Ni2+ increases the hole concentration

13

and improves the work function. CuO is adverse due to the creation of defect states with

14

band gap. The metallic Cu may also be negative for these devices in that the

15

recombination in the material is increased, which is likely the main reason why Voc is still

16

low (<400 mV) in this work. The next improved scheme should adopt the composite

17

(NiO)x:(Cu2O)y ceramic target including the NiO and Cu2O phase.

19

EP

AC C

18

TE D

9

4. Conclusions

20

ACCEPTED MANUSCRIPT

In summary, we successfully constructed a novel NiO:Cu/c-Si heterojunction solar cell

2

by Cu-incorporation, and obtained a highest PCE of 9.1% among all bulk-like NiO/Si

3

heterojunction solar cells. The hole-selective contact layers of the solar cell devices in

4

this work are divided into four cases: (1) pure NiO (dev. 1); (2) NiO with proper Cu

5

incorporation (dev. 2 and 3); (3) NiO with excessive Cu incorporation (dev. 4); and (4)

6

Cu incorporation induced phase transition of NiO material to Cu oxides (dev. 5). From

7

case 1 to case 2, the carrier concentration and work function of the contact material,

8

built-in potential and acceptor concentration of HJ, leading to a decrease in Rs and J0, and

9

then an increase in the PCE of solar cells from 4.3% to 9.1%. However, Jsc always

10

decreased with the increase of Cu content because Cu incorporation results in high

11

absorption. XRD, XPS and Cu LMM spectra show a gradual phase transformation from

12

NiO:Cu to CuO and a multi-species coexistence phase including metallic Cu, Cu2O and

13

CuO. For case 3, the excessive Cu incorporation introduces more Cu2O, CuO and even

14

Cu, with the latter two creating defect states with band gap, which increases

15

recombination, as proved by the increased ideal factor, J0 and decreased built-in potential

16

and acceptor concentration. These result in the presence of S-shape J-V characteristics

17

and deteriorating device performance. In case 4, NiO:Cu/n-Si HJ has been changed to a

18

CuO/n-Si HJ device, and the very low FF and Jsc are consistent with the previous work.

19

The huge Rs and J0, the least little built-in potential and acceptor concentration, owing to

20

the sever defect states within band gap, explains why these devices possess poor

AC C

EP

TE D

M AN U

SC

RI PT

1

21

ACCEPTED MANUSCRIPT

1

performance. Further improvement in Voc has been deduced by suppressing the metal Cu

2

and CuO phase for the next higher device performance.

4

RI PT

3

Acknowledgements

This work is supported by the National Natural Science Foundation of China (No.

6

61774089, 61274053, 51572132), Yang Fan Innovative & Entrepreneurial Research Team

7

Project (No. 2014YT02N037), and NSF of Hebei Province (No. E2015201203).

M AN U

SC

5

8 9

References

[1] M.A. Green, Solar Energy Materials and Solar Cells 143 (2015)190-197.

11

[2] C. –W. Chen, M. Hermle, J. Benick, Y. Tao, Y. –W. Ok, A. Upadhyaya, A. M. Tam, A.

12

Rohatgi, Progress in Photovoltaics Research & Applications 25 (2017).

13

[3] P. Procel, A. Ingenito, R. D. Rose, S. Pierro, F. Crupi, M. Lanuzzaa, G. Cocorullo, O.

14

Isabella, M. Zeman, Prog. Photovolt: Res. Appl. 25 (2017) 452-469.

15

[4] K. Yoshikawa, H. Kawasaki, W. Yoshida, T. Irie, K. Konishi, K. Nakano, T. Uto, D.

16

Adachi, M. Kanematsu, H. Uzu, K. Yamamoto, Nature Energy 20 (2017) 1-8.

17

[5] B. Paviet-Salomon, A. Tomasi, A. Descoeudres, L. Barraud, S. Nicolay, M. Despeisse,

18

S. D. Wolf, C. Ballif, IEEE Journal of Photovoltaics 5 (2015) 1293-1303.

19

[6] M. Balucani, K. Kholostov, P. Nenzi, R. Crescenzi, D. Ciarniello, D. Bernardi, L.

20

Serenelli, Z. Izzi, M. Tucci, Energy Procedia 43 (2013) 54-65.

AC C

EP

TE D

10

22

ACCEPTED MANUSCRIPT

[7] Q. Li, K. Tao, Y. Sun, R. Jia, S. –M. Wang, Z. Jin, X. –Y. Liu, solar energy 135 (2016)

2

487-492.

3

[8] J. Bullock, M. Hettick, J. Geissbühler, A. J. Ong, T. Allen, C. M. Sutter-Fella, T. Chen,

4

H. Ota, E. W. Schaler, S. D. Wolf, C. Ballif, A. Cuevas and A. Javey, E-cient Nature

5

Energy 31 (2015) 1-14.

6

[9] J. Geissbühler, J. Werner, S. Martin de Nicolas, L. Barraud, A. Hessler-Wyser, M.

7

Despeisse, S. Nicolay, A. Tomasi, B. Niesen, S. De Wolf, C. Ballif, Applied Physics

8

Letters 107 (2015) 081601.

9

[10] L.G. Gerling, S. Mahato, A. Morales-Vilches, G. Masmitja, P. Ortega, C. Voz, R.

M AN U

SC

RI PT

1

Alcubilla, J. Puigdollers, Solar Energy Materials and Solar Cells 145 (2016) 109-115.

11

[11] C. Battaglia, X. Yin, M. Zheng, I.D. Sharp, T. Chen, S. McDonnell, A. Azcatl, C.

12

Carraro, B. Ma, R. Maboudian, R.M. Wallace, A. Javey, Nano letters 14 (2014) 967-971.

13

[12] M. D. Irwin, J. D. Servaites, D. B. Buchholz, B. J. Leever, J. Liu, J. D. Emery, M.

14

Zhang, J. –H. Song, M. F. Durstock, A. J. Freeman, M. J. Bedzyk, M. C. Hersam, R. P. H.

15

Chang, M. A. Ratner, T. J. Marks, Chem. Mater. 23 (2011) 2218-2226.

16

[13] W. Chen, F. –Z. Liu, X. –Y. Feng, A. B. Djurišić, W. K. Chan, Z. –B. He, Adv.

17

Energy Mater. 1700722 (2017) 1-8.

18

[14] B. Varghese, M. V. Reddy, Z. Yanwu, C. S. Lit, T. C. Hoong, G. V. S. Rao, B. V. R.

19

Chowdari, A. T. S. Wee, C. T. Lim, C. –H. Sow, Chem. Mater. 20 (2008) 3360-3367.

20

[15] X. Deng, L. Zhang, J. Guo, Q. Chen, J. Ma, Materials Research Bulletin 90 (2017)

AC C

EP

TE D

10

23

ACCEPTED MANUSCRIPT

170-174.

2

[16] N. Chopra, L. Claypoole, L. G. Bachsa, J Nanopart Res 12 (2010) 2883-2893.

3

[17] A. F. Khapikov, J. W. Harrell, H. Fujiwara, C. Hou, J. Appl. Phys. 87 (2000) 4954.

4

[18] F.H. Hsu, N.F. Wang, Y.Z. Tsai, Y.S. Cheng, M.P. Houng, Journal of Physics D:

5

Applied Physics 46 (2013) 275104.

6

[19] F. H. Hsu, N. F. Wang, Y. Z. Tsai, Y. S. Cheng, C. F. Yang, M. P. Houng,

7

International Scholarly and Scientific Research & Innovation 8 (2014) 683-686.

8

[20] F.-H. Hsu, N.-F. Wang, Y.-Z. Tsai, C.-Y. Wu, Y.-S. Cheng, M.-H. Chien, M.-P.

9

Houng, Thin Solid Films 573 (2014) 159-163.

M AN U

SC

RI PT

1

[21] T. Dutta, P. Gupta, A. Gupta, J. Narayan, J. Appl. Phys. 108 (2010) 083715.

11

[22] Y. R. Denny, K. Lee, C. Park, S. K. Oh, H. J. Kang, D. –S. Yang, S. Seo, Thin Solid

12

Films 591 (2015) 255-260.

13

[23] N. Wang, C. Q. Liu, B. Wen, H. L. Wang, S. M. Liu, W. W. Jiang, W. Y. Ding, W. P.

14

Chai, Thin Solid Films 616 (2016) 587-593.

15

[24] S.C. Chen, T.Y. Kuo, Y.C. Lin, H.C. Lin, Thin Solid Films 519 (2011) 4944-4947.

16

[25] W.-Y. Chen, J.-S. Jeng, K.-L. Huang, J.-S. Chen, Journal of Vacuum Science &

17

Technology A: Vacuum, Surfaces, and Films 31 (2013) 021501.

18

[26] L. Zhao, G. Su, W. Liu, L. Cao, J. Wang, Z. Dong, M. Song, Applied Surface

19

Science 257 (2011) 3974-3979.

20

[27] B. Chen, J. H. Chen, Y. Shen, K. Ge, J. Guo, F. Li, H. Liu, Y. Xu, Y. Mai, Appl. Phys.

AC C

EP

TE D

10

24

ACCEPTED MANUSCRIPT

Lett. 110 (2017) 133504.

2

[28] B. Demaurex, S. D. Wolf, A. Descoeudres, Z. C. Holman, C. Ballif, Appl. Phys. Lett.

3

101 (2012) 171604.

4

[29] D. Zhang, A. Tavakoliyaraki, Y. Wu, R. A. C. M. M. Swaaij, M. Zeman, Energy

5

Procedia 8 (2011) 207-213.

6

[30] R. A. Sinton and A. Cuevas, 16th European Photovoltaic Solar Energy Conference,

7

1-5 May Glasgow, UK.

8

[31] A. Descoeudres, Z. C. Holman, L. Barraud, S. Morel, S. D. Wolf, C. Ballif, IEEE

9

journal of photovoltaics 3 (2013) 83-89.

M AN U

SC

RI PT

1

[32] M. R. Page, E. Iwaniczko, Y. Xu, L. Roybal, R. E. Bauer, H. C. Yuan, Q. Wang, D. L.

11

Meier, IEEE Photovoltaic Specialists Conference 1-4 (2008).

12

[33] S. Masudy-Panah, G.K. Dalapati, K. Radhakrishnan, A. Kumar, H.R. Tan, E. Naveen

13

Kumar, C. Vijila, C.C. Tan, D. Chi, Progress in Photovoltaics: Research and Applications,

14

23 (2005) 637-645.

15

[34] F. Gao, X. J. Liu, J. S. Zhang, M. Z. Song, N. Li, Journal of Applied Physics 111

16

(2012) 084507-4.

17

[35] G.W. Pabst, L.W. Martin, Y.-H. Chu, R. Ramesh, Applied Physics Letters 90 (2007)

18

072902.

19

[36] E. Arslan, S. Bütün, E. Ozbay, Applied Physics Letters 94 (2009) 142106.

20

[37] Y. F. Zhang, W. Cui, Y. W. Zhu, F. S. Zu, L. S. Liao, S. T. Lee, B. Q. Sun, Energy

AC C

EP

TE D

10

25

ACCEPTED MANUSCRIPT

&Environmental Science 8 (2015) 297–302.

2

[38] M. W. M. van Cleef, F. A. Rubinelli, J. K. Rath, R. E. I. Schropp, W. F. van der

3

Weg, R. Rizzoli, C. Summonte, R. Pinghini, E. Centurioni, R. Galloni, Journal of

4

Non-Crystalline Solids 227–230 (1998) 1291–1294.

5

[39] A. Froitzheim, K. Brendel, L. Elstner, W. Fuhs, K. Kliefoth, M. Schmidt, Journal of

6

Non-Crystalline Solids 299–302 (2002) 663–667.

7

[40] T.F. Chen, A.J. Wang, B.Y. Shang, Z.L. Wu, Y.L. Li, Y.S. Wang, Journal of Alloys

8

and Compounds 643 (2015) 167-173.

9

[41] W.-L. Jang, Y.-M. Lu, W.-S. Hwang, T.-L. Hsiung, H.P. Wang, Applied Physics

M AN U

SC

RI PT

1

Letters 94 (2009) 062103.

11

[42] J.-C. Li, X.-Y. Hou, Q. Cao, Journal of Applied Physics 115 (2014) 164507.

12

[43] S. Uhlenbrock, C. Scharfschwerdt, M. Neumann, G. IIIing, H. Freund, Journal of

13

Physics Condensed Matter 4 (1999) 7973.

14

[44] S. Oswald, W. Brückner, Surface & Interface Analysis 36 (2004) 17-22.

15

[45] A. N. Mansour, Surface Science Spectra 3 (1994) 231.

16

[46] J. P. Espinós, J. Morales, A. Barranco, A. Caballero, J. P. Holgado, and A. R.

17

González-Elipe, J. Phys. Chem. B 106 (2002) 6921-6969.

18

[47] H. Kawazoe, H. Yanagi, K. Ueda, H. Hosono, H. Kawazoe, Mrs Bulletin 25 (2000)

19

28-36.

AC C

EP

TE D

10

20

26

ACCEPTED MANUSCRIPT 

The nonstoichiometric p-type NiO thin film is used as hole-selective contact for a novel solar cell architecture.



An efficiency of 4.3%, for the first time, is achieved in the pure NiO/n-Si heterojunction solar cell. A record PCE of 9.1% is obtained after incorporating moderate Cu into NiO contact layer.



The fabricated NiO:Cu film is a mixture of metallic Cu, Cu2O and CuO multi-phase

RI PT



coexistence.

Cu2O contributes the improvement of acceptor concentration and CuO creates defect states

EP

TE D

M AN U

SC

within the band gap.

AC C