formation of silicon and silicides by new electrochemical ... · ca-si alloy films using aqueous...

Formation of silicon and silicides

by new electrochemical processes

Yoshihide Sakanaka

i

Contents

Chapter 1 General introduction 1

1.1 Silicon 1

1.2 Overview of electrochemical process 2

1.3 A new process of producing Si from SiO2 3

1.4 Calcium silicide 5

1.4.1 Properties of calcium silicide 5

1.4.2 Electrochemical formation of calcium silicide 6

1.5 Aims of this study 8

References 10

Chapter 2 Experimental procedures 17

2.1 Reagents and their handlings 17

2.1.1 Reagents 17

2.2 Measurements and analyses 18

2.2.1 Electrochemical measurements 18

2.2.2 Surface analysis 18

ii

2.2.2.1 X-ray diffraction (XRD) 18

2.2.2.2 Scanning electron microscope (SEM) and energy

dispersive X-ray spectroscopy (EDS) 18

2.2.2.3 Micro-Raman spectroscopy 19

2.2.2.4 UV-VIS-NIR reflectance spectroscopy 19

2.2.2.5 Photoluminescence spectroscopy 19

2.2.3 Gas chromatography 19

2.2.4 Thermal analysis 20

Chapter 3 Electrodeposition of Si film on Ag substrate in molten LiF-NaF-KF

directly dissolving SiO2 21

3.1 Introduction 21

3.2 Experimental 22

3.3 Results and discussion 24

3.3.1 Cyclic voltammetry 24

3.3.2 Characterization of Si 25

3.3.3 Anode reaction 28

3.4 Conclusion 29

iii

References 30

Chapter 4 Electrodeposition of porous Si film from SiO2 in molten BaCl2-CaCl2-

NaCl 38

4.1 Introduction 38

4.2 Experimental 40


4.3.1 Electrochemical window of BaCl2-CaCl2-NaCl 41


4.3.3 Si deposition 44

4.3.4 Si film growth 45

4.3.5 Optical properties 46

4.4 Conclusion 47

References 48

Chapter 5 Electrodeposition of Si in molten KF-SiO2 58

5.1 Introduction 58

5.2 Experimental 59

5.2.1 Thermal analysis 59

5.2.2 Electrochemical measurements 59

iv


5.3.1 Thermal analysis of binary KF-SiO2 system 60


5.3.3 Deposition of Si 61

5.4 Conclusion 62

References 63

Chapter 6 Electrochemical formation of Ca-Si in molten CaCl2-KCl 68

6.1 Introduction 68

6.2 Experimental 69


6.3.1 Electrochemical window of CaCl2-KCl 70


6.3.3 Formation of Ca-Si alloy 73

6.3.4 Phase control of Ca-Si 73

6.3.5 Optical properties of Ca-Si film 76

6.4 Conclusion 77

References 78

v

Chapter 7 Co-deposition of Ca-Si in molten CaCl2-KCl-K2SiF6 93

7.1 Introduction 93

7.2 Experimental 94


7.3.1 Electrochemical behavior of K2SiF6 95

7.3.2 Formation of Ca-Si alloy 96

7.3.3 Growth of Ca-Si alloy films by co-deposition 97

7.4 Conclusion 98

Reference 100

Chapter 8 General conclusion 104

List of Publications 106

Acknowledgement 108

1

Chapter 1

General introduction

1.1 Silicon

Silicon has been widely used as a material for fabricating solar cells and

electronic devices since it is an intrinsic semiconductor. Silicon is the second most

abundant element on Earth, is environmentally friendly and is physically and chemically

stable. Photovoltaic power generation is a promising alternative energy technology

because it is renewable and environmentally safe. The most common substrate material

used for the fabrication of photovoltaic cells is solar-grade silicon (SoG-Si). Owing to the

growing demand for SoG-Si, production of a lower-cost SoG-Si is an important task for

the solar energy industry. Nowadays, silicon production method has been almost occupied

by Siemens process for several decades. However, as history has illustrated, technology

always has been developing into more sophisticated one than previous one. The

innovation in industry is thus essential to provide against the uncertain future difficulties,

to name a few, energy shortage, climate changes and global warming.

Conventional processes for producing SoG-Si have some challenges. SoG-Si

2

requires contaminant levels to be lower than 10 ppm. In order to achieve such high purity

levels, SoG-Si is produced by carbothermal reduction of SiO2 combined with the Siemens

process. SiO2 extracted from silicate rocks or quartz sand is commonly used as the starting

material for the production of SoG-Si. The carbothermal reduction of SiO2 is conducted

at ~2000 K to obtain metal grade silicon (MG-Si), which has a purity of only 95-98%,

and has carbon, phosphorus and boron contaminants. In order to purify MG-Si further,

the Siemens process is carried out. The Siemens process, however, requires complex

equipment, has high energy needs, and emits carbon dioxide, which is a major

environmental issue. Replacing this technology with a simple, one-step, energy saving

and carbon-free process would be of great benefit. From this background information, an

electrochemical process was selected to evaluate an alternative production process for

SoG-Si.

1.2 Overview of electrochemical process

In the conventional electrochemical reduction of metal oxides, pure metals are

formed on cathodes as represented in Eq. 1-1, where oxide ions generated in the

electrolyte are simultaneously oxidized on a carbon anode according to Eqs. 1-2 and/or

1-3.

3

MOx + 2xe- → M + xO2- (1-1)

C + O2- → CO + 2e- (1-2)

C + 2O2- → CO2 + 4e- (1-3)

Other processes for the production of metal from metal oxide have been reported.

In the FFC Cambridge process [1], direct electrolysis of TiO2 occurred by applying a

potential between the anode and cathode, producing Ti at the cathode. Nohira et al. [2]

reported the direct bulk electro-reduction of SiO2 in CaCl2 at 1123 K. Figure 1-1 shows

the principle of Nohira’s direct electrochemical reduction process. The cathode was made

from a SiO2 plate attached to an electrical conducting material. The reduction of SiO2 at

the cathode in this process is expressed as in Eq. 1-4.

SiO2 (solid) + 4e- → Si + 2O2- (1-4)

In this process, molten CaCl2 at 1123 K was chosen as the electrolyte because the O2- ion

is very soluble in molten CaCl2 [3]. Since SiO2 is an insulator, no electrons can readily

pass through, but electrons from an electron conducting material can be injected into SiO2.

The resulting silicon is brittle, however, because of the solid-phase reaction.

1.3 A new process of producing Si from SiO2

In this dissertation, the author proposed a new process involving the dissolution

4

of SiO2 in a molten salt as represented in Eq. 1-5.

SiO2 → Si(IV) + 2O2- (1-5)

The next step is the electrodeposition of silicon by electrochemically reducing Si(IV) to

Si as shown in Eq. 1-6.

Si(IV) + 4e- → Si (1-6)

The oxide ions are electrochemically oxidized on a graphite anode to form carbon

monoxide or carbon dioxide gas. However, by replacing the graphite anode with an inert

anode, it is possible to generate oxygen gas, as shown in Eq. 1-7.

2O2- → O2 + 4e- (1-7)

The overall reaction in the case of an inert anode is represented by Eq. 1-8.

SiO2 → Si + O2 (1-8)

A schematic representation of this method is shown in Figure 1-2. This process is different

from the direct reduction process, in which silicon ions are electrochemically reduced to

form silicon at a cathode. An interesting potential advantage of this electrochemical

process is that we can possibly fabricate Si film directly on the cathode by adding SiO2

to the electrolyte.

At relatively low temperatures, the low solubility of oxides makes the reduction

of SiO2 difficult. However, low temperature has many advantages from an engineering

5

perspective, such as reduced corrosive damage to the cell and substrate material, and a

wide selection of substrate and electrode materials. The author thus has high expectations

for the development of molten salts that have high oxide solubility at low temperatures.

Additionally, since molten salts with high oxide solubility at lower temperatures must be

developed for this process, it is also important to study the reduction process of oxides at

lower temperatures.

1.4 Calcium silicide

1.4.1 Properties of calcium silicide

Compound semiconductor, such as GaAs, CdTe and GaP, has attention as one of

the most promising materials for optical devices of high power electronic devices because

of their advantages [4]. However, conventional process is inapplicable for producing

compound semiconductor without harmful materials and high energy consumption. From

the background, a new process of compound semiconductor, which does not require

harmful materials and high energy, has been proposed.

Recently, silicides consisting of non-toxic, readily available materials have

attracted much interest as semiconductors and metallic alloys because of their potentially

excellent optical properties and their environmental friendliness. In addition, the bandgap

6

energy of metal silicides can be tuned from the infrared to the visible region by proper

choice of the elemental metals and phase of the compounds [5-8].

In particular, Ca-Si metal alloys are well known as semiconductor materials and

exhibit superconducting properties. For example, it has been reported that Ca2Si is a

semiconductor with an energy gap of 1.9 eV [9]. In addition, CaSi is expected to have a

high hydrogen storage capacity [10]. A new polymorph of the CaSi2 having

superconductive behavior at 14 K was reported [11-12]. Recently, there has been interest

in the use of intermetallic silicides as replacement anodes for graphite in Li-ion batteries

[13-14]. If these alloys can be formed as films with a variety of morphologies, shapes and

sizes, applications to energy conversion devices might be possible.

1.4.2 Electrochemical formation of calcium silicide

Silicide films are usually manufactured by chemical vapor deposition (CVD) or

physical vapor deposition (PVD). However, it is difficult to achieve the growth of

continuous films by deposition from the gas phase [15-20]. Hence, a new method to

produce Ca-Si alloy films is needed.

In searching for a new method for Ca-Si alloy synthesis, electrochemical

methods seemed attractive, but these methods are difficult to apply to the formation of

7

Ca-Si alloy films using aqueous electrolytes, because Ca and Si cannot be

electrodeposited due to the cathodic limit.

In such cases, molten salts are promising alternative electrolytes. Molten salts,

especially alkali metal halides, have large electrochemical windows and are chemically

and physically stable. Electrochemical processes utilizing a molten salt as an electrolyte

have been used to form a variety of functional materials [21-23].

In this study, the author uses a molten salt electrochemical process to form Ca-

Si alloy films by two methods: (1) solid-phase diffusion and (2) co-deposition. Both of

these processes have the following advantages: alloy film deposition can be controlled by

electrochemical parameters and alloy film can be formed on various substrate shapes.

Conceptual representations of the two methods are shown in Figure 1-3 and

Figure 1-4. In both cases, molten CaCl2-KCl is used as an electrolyte. In the case of the

solid-phase diffusion method, when the cathodic electrolysis is conducted on a silicon

electrode, Ca(II) is reduced to a Ca atom to form Ca-Si alloy films at the electrode surface

as shown in Figure 1-5 (a). In the case of co-deposition, K2SiF6 is added to the melt as a

Si ion source. During cathodic electrolysis, Ca and Si ions are reduced to Ca and Si atoms

simultaneously to form a Ca-Si alloy film on the electrode as shown in Figure 1-5 (b).

8

1.5 Aims of this study

In chapter 3, the electrochemical reduction of SiO2 in the molten LiF-NaF-KF at

873 K is discussed. Si films produced by reduction of SiO2 on an Ag electrode are

characterized by XRD, SEM and Raman spectroscopy, to confirm the morphology and

crystallinity control by the proposed method.

In chapter 4, the electrochemical behavior of SiO2 as well as the electrochemical

window of the BaCl2-CaCl2-NaCl eutectic melt (32:48:20 mol%) is discussed. In addition,

the electrodeposition of Si from SiO2 using an Ag electrode is described. To obtain a

porous Si film, electrochemical deposition is demonstrated on an Ag electrode using the

BaCl2-CaCl2-NaCl eutectic melt as the electrolyte.

In chapter 5, the melting point of KF-SiO2 and the electrochemical behavior of

the melt are investigated by measuring thermal properties and by cyclic voltammetry,

respectively. Electrochemical deposition of Si is demonstrated to obtain Si film in the

melt.

In chapter 6, the electrochemical formation and phase control of Ca-Si alloy film

in a CaCl2-KCl melt at 923 K by solid-phase diffusion is presented. A Ca-Si film prepared

by potentiostatic electrolysis is characterized by XRD, SEM and UV-VIS-NIR reflectance

spectroscopy, to confirm the possibility of phase control of the Ca-Si alloy by the

9

proposed method.

In chapter 7, the electrochemical formation of Ca-Si alloy film by the co-

deposition process in CaCl2-KCl-K2SiF6 is discussed. To investigate the growth

mechanism of the obtained film, the thickness of the Ca-Si film is measured by cross-

sectional SEM.

10

References

[1] G. Z. Chen, D. J. Fray and T. W. Farthing, Nature, 407, 361 (2000).

[2] T. Nohira, K. Yasuda and Y. Ito, Nat. Mater., 2, 397 (2003).

[3] D. A. Wenz, I. Johonson and R. D. Wolson, J. Chem. Eng. Data, 14, 252 (1969).

[4] H. Lange, Phys. Stat. Sol., B 201, 3 (1997).

[5] M. Eizenberg and K. N. Tu, J. Appl. Phys., 53, 6885 (1982).

[6] K. Lefki, P. Muret, N. Cherief and C. Cinti, J. Appl. Phys., 69, 352 (1991).

[7] J. F. Morar and M. Wittmer, Phys. Rev., B 37, 2618 (1998).

[8] Y. Imai, A. Watanabe and M. Mukaida, J. Alloy Comp., 358, 257 (2003).

[9] O. Madelung, Semiconductors basic data, 2nd ed, Springer, Berlin (1996).

[10] M. Aoki, N. Ohba, T. Noritake and S. Towata, Appl. Phys. Lett., 85, 387 (2004).

[11] S. Sanfilippo, H. Elsinger, M. Nunez-Reguerio and O. Laborde, Phys. Rev., B 61,

R3800 (2000).

[12] G. Satta, G. Profeta, F. Bernadini, A. Continenza and S. Massidda, Phys. Rev., B 64,

104507 (2001).

[13] A. Netz, R. A. Huggins and W. Weppner, J. Power Sources, 119, 95 (2003).

[14] J. Wolfenstine, J. Power Sources, 124, 241 (2003).

[15] H. N. Acharya, Swapan K. Dutta and H. D. Banerjee, Sol. Ene. Mat., 3, 441 (1980).

11

[16] T. Koga, A. Bright, T. Suzuki, K. Shimada, H. Tatsuoka and H. Kuwabara, Thin Sol.

Films, 369, 248 (2000).

[17] M. Sugiyama and Y. Maeda, Thin Sol. Films, 381, 225 (2001).

[18] T. Nakamura, T. Suematsu, K. Takakura, F. Hasegawa, A. Wakahara and M. Imai,

Appl. Phys. Lett., 81, 1032 (2002).

[19] T. Hosono, Y. Matsuzawa, M. Kuramoto, Y. Momose, H. Tatsuoka and H. Kuwabara,

Solid State Phenom., 93, 447 (2003).

[20] I. Kogut and M. C. Record, Intermetallics, 32, 184 (2013).

[21] Y. Ito and T. Nohira, Electrochim. Acta, 45, 2611 (2000).

[22] T. Iida, T. Nohira and Y. Ito, J. Electrochem. Soc. 46, 2537 (2001).

[23] H. Konishi, T. Nohira and Y. Ito, Electrochim. Acta, 47, 3533 (2002).

12

Figure 1-1 Production of Si by direct electrolytic reduction of solid SiO2

in molten CaCl2 [2].

13

Figure 1-2 Principle of the production of Si from SiO2 in a molten salt.

SiO2 is added into the melt.

14

Figure 1-3 Conceptual representation of the formation of Ca-Si alloy films by solid-

phase diffusion. A CaCl2-KCl melt is used as the electrolyte.

15

Figure 1-4 Conceptual representation of the formation of Ca-Si alloy films by co-

deposition. A CaCl2-KCl melt is used as the electrolyte. K2SiF6 is added into the melt as

a Si ion source.

16

Figure 1-5 Schematic drawings of (a) solid-phase diffusion and (b) co-deposition.

17

Chapter 2

Experimental procedures

2.1 Reagents and their handlings

2.1.1 Reagents

Lithium fluoride (LiF: Wako Pure Chemical Co. Ltd., 98.0%), sodium fluoride

(NaF: Wako Pure Chemical Co. Ltd., 99.0%), potassium fluoride (KF: Wako Pure

Chemical Co. Ltd., 99.0%), barium chloride (BaCl2: Wako Pure Chemical Co. Ltd.,

95.0%), calcium chloride (CaCl2: Wako Pure Chemical Co. Ltd., 95.0%), sodium chloride

(NaCl: Wako Pure Chemical Co. Ltd., 99.5%) and potassium chloride (KCl: Wako Pure

Chemical Co. Ltd., 99.5%) were individually placed in a high purity alumina crucible

(NIKATTO Co. Ltd., 99.5 wt% Al2O3, SSA-S grade) then vacuum dried at 473 K for 24

h before use. Silicon dioxide (SiO2: Wako Pure Chemical Co. Ltd., 98.0%) or potassium

silicofluoride (K2SiF6: Wako Pure Chemical Co. Ltd., 99.0%) was used as a silicon ion

source, which was added into the melt. Lithium oxide (Li2O: Wako Pure Chemical Co.

Ltd., 99.0%) was used as an oxide ion source, which was added into the melt.

18

2.2 Measurements and analyses

2.2.1 Electrochemical measurements

Electrochemical measurements were conducted using an electrochemical

measurement system (Hokuto Denko Co. Ltd., HZ-3000). All measurements were

performed using the three-electrode method.

2.2.2 Surface analysis

2.2.2.1 X-ray diffraction (XRD)

A diffractometer (Rigaku Co. Ltd., Multi Flex) was used and the diffraction

intensity against 2θ was recorded; here the sample was fixed on the stage and the counter

was rotated to produce a given incident angle. Cu-Kα radiation was used as the X-ray

source. The output power was fixed at 40 kV - 40 mA.

2.2.2.2 Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy

(EDS)

Morphology observation of samples and elemental analysis were performed

using a scanning electron microscope and energy dispersive X-ray spectrometer (JEOL

Co. Ltd., JSM-7001), respectively.

19

2.2.2.3 Micro-Raman spectroscopy

Micro-Raman spectroscopic analysis (Jobin-Yvon Co. Ltd., Labram

Spectrometer) was carried out using a He-Ne laser (632.8 nm).

2.2.2.4 UV-VIS-NIR reflectance spectroscopy

For evaluation of the optical properties, UV-VIS-NIR reflectance spectra were

measured with a spectrophotometer (JASCO Co. Ltd., V-670).

2.2.2.5 Photoluminescence spectroscopy

The photoluminescence spectra were measured using a He-Cd laser (KIMMON

Co. Ltd., IK5451R-E) and a spectrometer (Oceanoptics Co. Ltd., USB2000+).

2.2.3 Gas chromatography

Gas chromatography (Yanaco Co. Ltd., G2800) was used in order to analyze

the concentration of oxygen gas.

2.2.4 Thermal analysis

Thermal properties of the KF-SiO2 systems were determined by a differential

20

thermogravimetric analyzer (Shimadzu Co. Ltd., DTG-60). The measurement was

performed under dried Ar gas.

21

Chapter 3

Electrodeposition of Si film on Ag substrate

in molten LiF-NaF-KF directly dissolving SiO2

3.1 Introduction

Electrochemical methods can be used as an alternative for the production of

high purity silicon. Among others, molten salt is employed as an electrolyte for the

electrodeposition of Si. It has been known for many years that Si films can be fabricated

by electrodeposition and most researchers have demonstrated deposition of Si from

K2SiF6 or Na2SiF6 in molten fluoride and chloride electrolytes [1-9]. While there are a

few reports discussing the electrodeposition of Si from SiO2 in BaO-SiO2-BaF2 at above

1688 K [10] and BaF2-CaF2-SiO2 at 1473 K [11]. Recently, direct reduction of SiO2 in

molten chloride was investigated [12]. However, there is no report of electrodeposition

of Si film from SiO2 at low temperature. As mentioned in chapter 1, low temperature is

desirable from an engineering perspective.

From the background, the author selected LiF-NaF-KF eutectic as electrolyte

because molten fluoride has large solubility of oxide and LiF-NaF-KF eutectic has

22

relatively low melting point (m.p.=727 K) [13] among molten fluorides.

The aim of this study was to produce Si from LiF-NaF-KF molten electrolytes

containing SiO2 and characterized the reduction of Si(IV) in the solution, using cyclic

voltammetry and chronoamperometry.

3.2 Experimental

Figure 3-1 shows the schematic drawing of the experimental apparatus. The

electrochemical cell consisted of an alumina crucible placed in a quartz glass holder. The

cell was heated using a programmable furnace and the temperature of the cell was

measured using a chromel-alumel thermocouple.

LiF (Wako Pure Chemical Co. Ltd., 98.0%), NaF (Wako Pure Chemical Co.

Ltd., 99.0%), and KF (Wako Pure Chemical Co. Ltd., 99.0%) were mixed in eutectic

composition (LiF:NaF:KF = 46.5:11.5:42.0 mol%) and placed in a high-purity alumina

crucible (NIKKATO Co. Ltd., 99.5 wt% Al2O3, SSA-S grade). The mixture was placed

under vacuum for more than 24 h at 473 K to expel any water. All the experiments were

performed in the LiF-NaF-KF eutectic melt in a dry Ar atmosphere. SiO2 (Wako Pure

Chemical Co. Ltd., 99.0%) and Li2O (Wako Pure Chemical Co. Ltd., 99.0%) were used

as the source of Si ions and O ions, respectively.

For the electrochemical studies, Ag wires (Nilaco Co. Ltd., 1 mm diameter,

23

99.99%), Ag plates (Nilaco Co. Ltd., 7 mm × 15 mm × 0.5 mm, 99.5%), Fe plates (Nilaco

Co. Ltd., 6 mm × 15 mm × 0.5 mm) and mixed oxide electrode (hercynite and Al2O3)

were used as working electrodes. To prepare the mixed oxide electrode, Fe2O3 and Al2O3

powders were sintered. The surface area of the working electrode was determined after

each experiment by measuring the immersion depth in the bath. Glassy carbon rods (Tokai

Carbon Co. Ltd., 3 mm diameter) were used as counter electrodes. These electrodes are

shown in Figure 3-2. The potentials of the working and counter electrodes were measured

with respect to a Ni wire immersed in the molten electrolyte, which acted as a quasi-

reference electrode. The potential of the quasi-reference electrode was calibrated with

respect to a K+/K electrode, which was prepared by electrodepositing K metal on a Ni

electrode [13]. All potentials in this chapter were reported with respect to the K+/K

electrode. A potentiostat/galvanostat (Hokuto Denko Co. Ltd., HZ-3000) was used for the

cyclic voltammetry and chronopotentiometry experiments.

Si samples were prepared by potentiostatic electrolysis and rinsed with AlCl3

solution. The morphologies and cross-sections of the samples were studied using

scanning electron microscopy (JEOL Co. Ltd., JSM-7001). X-ray diffraction (Rigaku Co.

Ltd., Multi Flex) patterns were obtained using Cu-Kα radiation. Surface chemical states

were measured using Raman spectroscopy, (Jobin-Yvon Co. Ltd., Labram spectrometer)

24

with a He-Ne laser (632.8 nm wavelength). Raman spectra were measured over a

wavenumber range of 200-1000 cm-1. The gases evolved from the electrochemical cell

before and after electrolysis were analyzed using gas chromatography (Yanaco Co. Ltd.,

G2800).

3.3 Results and discussion

3.3.1 Cyclic voltammetry

In order to investigate the dissolution of SiO2 in molten fluoride, cyclic

voltammetry was conducted in a molten LiF-NaF-KF-SiO2 (0.2 mol%) electrolyte at 873

K. Figure 3-3 shows the cyclic voltammograms for an Ag electrode before and after the

addition of 0.2 mol% SiO2 to the electrolyte, acquired at a scan rate of 0.1 V s-1 at 873 K.

The dotted curve in the figure represented the voltammogram obtained before the addition

of SiO2, whereas the solid curve was the voltammogram acquired after the addition of 0.2

mol% SiO2 to the melt. Only one pair of cathodic and anodic current peaks were observed

around 0 V, corresponding to the deposition and dissolution of K metal, respectively [13].

In the case of the voltammogram obtained after the addition of SiO2, the cathodic current

was observed from approximately 1.80 V during the cathodic sweep. Since no alloy

formation is thermodynamically expected [14], the cathodic current observed at potentials

25

more negative than 1.80 V was attributed to Si metal deposition. Furthermore, a sharp

cathodic peak was observed at 0.03 V, attributed to Li-Si alloy formation [15].

A previous study involving Na2SiF6 in LiF-NaF-KF melts [5] reported a clear

cathodic current peak and two-step Si deposition involving Si(II) and Si(IV). In

comparison, the cathodic current peak in Figure 3-3 was rather broad. The broad cathodic

current peak observed at potentials more negative than 1.80 V may be explained as

follows.

SiO2 dissolved in the melt to form silicon fluoride, silicon oxyfluoride and oxide

ion. As a result, various Si ions coordinated by fluoride and oxide coexist in the melt. In

general, the deposition potential varied by the coordination state in the molten salt [16],

and during the cathodic sweep, each ionic species was reduced at the corresponding

potential. Therefore, the various cathodic waves overlapped, as shown in Figure 3-3,

resulting in a broad cathodic current peak.

3.3.2 Characterization of Si

Based on the results of cyclic voltammetry, Si samples were prepared by

potentiostatic electrolysis using an Ag electrode at 0.2 V, 0.4 V, 0.7 V, 1.0 V and 1.4 V for

1 h in LiF-NaF-KF molten electrolytes containing 0.05 mol% SiO2 at 873 K.

26

The obtained Si samples were washed in AlCl3 solution; the appearance of the

samples after washing is shown in Figure 3-4. There was no change in the appearance of

samples prepared at 1.0 V and 1.4 V, but the surfaces of the samples prepared at 0.2 V,

0.4 V and 0.7 V turned gray in color upon washing. Figure 3-5 shows the XRD patterns

of the five samples prepared at different potential values. The diffraction peaks observed

for the samples prepared at 1.0 V and 1.4 V were attributed to Ag and originated from the

substrate, whereas there were no peaks corresponding to Si. These results indicated that

Si deposition did not occur at electrolysis potentials more positive than 1.0 V.

In the case of the Si samples prepared at 0.2 V, 0.4 V and 0.7 V, the observed

XRD peaks were attributed to Ag and Si. Among the samples studied, the intensity of the

peak corresponding to crystalline Si (c-Si) was the strongest for the sample prepared at

0.2 V. This result was consistent with the fact that among the various samples prepared at

different potentials, the maximum amount of charge passed was for the sample prepared

at 0.2 V (45 C cm-2). In comparison, the charges passed for the samples prepared at 0.4 V

and 0.7 V were lower at 35 C cm-2 and 10 C cm-2, respectively.

The lattice parameter for polycrystalline Si prepared at 0.2 V was about 5.38

Å, as determined from the XRD results.

Figure 3-6 shows the SEM image of the surface of the sample prepared by

27

potentiostatic electrolysis at 0.2 V. The SEM image shows that the surface was composed

of smooth grains that were about 1 μm in diameter. Raman spectroscopy was conducted

to examine the crystallinity of the sample. Figure 3-7 shows the Raman spectra obtained

for the sample prepared at 0.2 V and single-crystal Si. In Figure 3-7, the black line

corresponds to the Raman spectrum for single-crystal Si, whereas the red line is the

spectrum for the sample prepared at 0.2 V. A sharp peak centered at 520 cm-1 was

observed in both the spectra and was assigned to the TO phonon of c-Si [17]. The

contribution of the TO phonon in the case of amorphous Si at 480 cm-1 was not observed,

indicating that the sample predominantly consists of the crystalline phase. In addition,

broad peaks at around 300 cm-1 and 950 cm-1 were observed in both the spectra, assigned

to air-formed oxide.

Figure 3-8 shows a cross-sectional SEM image of the Si sample and the

concentration profiles of Si, Ag and O obtained by energy dispersive X-ray spectroscopy

(EDS) line analyses. The results indicated that the Si layer adhered to the Ag substrate,

and the thickness of the Si layer was estimated to be ~1 μm. Based on the film thickness

observed and the amount of electric charge passed during sample preparation, the

coulomb efficiency was estimated to be approximately 10% for the sample prepared at

0.2 V. The deviation from 100% was thought to be caused by alkali metal deposition,

28

which occurs as a side reaction.

An inexpensive electrode material is required for the electrodeposition process

to be economically viable. Therefore, the author conducted similar potentiostatic

experiments using a Fe electrode. Si sample was prepared by potentiostatic electrolysis

using a Fe electrode at 0.2 V for 1 h in LiF-NaF-KF-SiO2 (0.1 mol%) at 773 K. Figure 3-

9 shows the Raman spectrum for the sample. As in the case of the Ag electrode, a sharp

peak centered at about 520 cm-1, attributed to the TO phonon of c-Si, was observed [17].

3.3.3 Anode reaction

From the view point of economic viability, the author fabricated a mixed oxide

electrode, which was composed of Fe, Al and O, as an oxygen gas evolution electrode.

Figure 3-10 shows the XRD pattern of the mixed oxide electrode. The pattern observed

was identified as FeAl2O4 and γ-Al2O3. In order to confirm gas evolution, galvanostatic

electrolysis was performed using the hercynite electrode at 50 mA cm-2 for 1 h in a LiF-

NaF-KF-Li2O (1.0 mol%) molten electrolyte at 773 K. Li2O in the molten electrolyte

acted as the oxide ion source. The gases obtained before and after electrolysis were

analyzed using gas chromatography.

A constant potential of about 5 V was observed in the potential transient curve

29

obtained during electrolysis, and the sampled gas was found to contain oxygen in the gas

chromatography analysis. The concentration of oxygen increased to 1.3% after

electrolysis and the current efficiency for oxygen evolution was estimated to be 100%,

indicating that only oxygen gas was generated on the anode during electrolysis.

3.4 Conclusion

The electrochemical reduction of SiO2 was investigated by cyclic voltammetry

and chronoamperometry in molten LiF-NaF-KF-SiO2 system at 873 K. Potentiostatic

electrolysis at the Ag electrode at 0.2 V, 0.4 V and 0.7 V for 1 h resulted in the formation

of polycrystalline Si. A Si film that was ~1 μm thick was obtained by potentiostatic

electrolysis at 0.2 V for 1 h.

30

References

[1] U. Cohen and R. A. Huggins, J. Electrochem. Soc., 123, 381 (1976).

[2] G. M Rao, D. Elwell and R. S. Feigelson, J. Electrochem. Soc., 127, 1940 (1980).

[3] G. M. Rao, D. Elwell and R.S. Feigelson, J. Electrochem. Soc., 128, 1708 (1981).

[4] D. Elwell and G. M. Rao, Electrochim. Acta, 26, 673 (1982).

[5] R. Boen and J. Bouteillon, J. Appl. Electrochem., 13, 277 (1983).

[6] A. L. Bieber, L. Massot, M. Gibilaro, L. Cassayre, P. Chamelot and P. Taxil,

Electrochim. Acta, 56, 5022 (2011).

[7] A. L. Bieber, L. Massot, M. Gibilaro, L. Cassayre, P. Taxil and P. Chamelot,


[8] S. V. Kuznetsova, V. S. Dolmatov and S. A. Kuznetsov, Russ. Electrochem., 45, 742

(2009).

[9] S. Lee, J. Hur and Ch. Seo, J. Ind. Eng. Chem., 14, 651 (2008).

[10] R. C. D. Mattei, Elwell and R. S. Feigelson, J. Electrochem. Soc., 128, 1712 (1981).

[11] Y. Hu, X. Wang, J. Xiao, J. Shuqiang and H. Zhu, J. Electrochem. Soc., 160, D81

(2013).


[13] H. Quiao, T. Nohira and Y. Ito, Electrochim. Acta, 47, 4543 (2002).

31

[14] T. B. Massalski, Binary Alloy Phase Diagrams, Vol. 1, Editor, ASM International,

93 (1990).

[15] K. Amezawa, N. Yamamoto, Y. Tomii and Y. Ito, J. Electrochem. Soc., 145, 1986

(1998).

[16] T. Goto, T. Nohira, R. Hagiwara and Y. Ito, J. Fluor. Chem., 130, 102 (2009).

[17] C. Smit, R. A. C. M. M. van Swaaij, H. Donker, A. M. H. N. Petit, W. M. M. Kessels

and M. C. M. van de Sanden, J. Appl. Phys., 94, 3582 (2003).

32

A: Quasi-reference electrode (Ni)

B: Counter electrode (Glassy carbon)

C: Thermocouple

D: Working electrode (Ag, Fe or Hercynite)

E: Electrolyte (LiF-NaF-KF)

F: Electric furnace

G: Quartz glass holder

Figure 3-1 Schematic drawing of experimental apparatus.

33

Working electrode

Counter electrode

Quasi-reference electrode

Figure 3-2 Schematic drawings of the electrodes.

34

Figure 3-3 Cyclic voltammograms for an Ag electrode in LiF-NaF-KF eutectic before

and after addition of 0.2 mol% SiO2 at 873 K. Scanning rate: 0.1 V s-1.

Figure 3-4 Photographs of the Ag plate of obtained by potentiostatic electrolysis at 1.4,

1.0, 0.7, 0.4 and 0.2 V for 1 h in the LiF-NaF-KF-SiO2 at 873 K.

35

Figure 3-5 XRD patterns of the samples obtained by potentiostatic electrolysis at 1.4,

1.0, 0.7, 0.4 and 0.2 V for 1 h using Ag electrode.

Figure 3-6 Surface SEM image of the sample obtained by potentiostatic electrolysis

at 0.2 V for 1 h.

36

Figure 3-7 Raman Spectra of the sample obtained by potentiostatic electrolysis at 0.2 V

and single crystal Si.

Figure 3-8 Cross-sectional SEM image and concentration profile of Ag, Si and O for

sample obtained by potentiostatic electrolysis at 0.2 V for 1 h.

37

Figure 3-9 Raman Spectrum of the Fe electrode after potentiostatic electrolysis

at 0.2 V for 1 h.

Figure 3-10 XRD pattern of the mixed oxide electrode.

38

Chapter 4

Electrodeposition of porous Si film from SiO2 in molten

BaCl2-CaCl2-NaCl

4.1 Introduction

Many studies have investigated the use of light-emitting devices based on Si

technology for illumination and display applications. In particular, numerous studies have

focused on the luminescence of porous Si structures since the discovery of visible

luminescence at room temperature [1]. Porous Si can be used in various applications such

as microelectronics, optoelectronics, and chemical and biological sensors.

Electrodeposition is one of the attractive methods available for fabricating metal

and semiconductor films. Studies have already investigated the electrodeposition of Si in

various molten salts [2-4]. Recently, the direct electrochemical reduction of bulk solid [5]

and nanoparticle [6] SiO2 in molten CaCl2 has been reported. As described in chapter 3,

the author has previously reported the electrochemical formation of Si from SiO2

containing LiF-NaF-KF [7]. However, the formation of porous Si by the electrolysis of a

molten salt and a study of its optical properties has not yet been reported. Therefore, in

39

this chapter, the author attempted to produce porous Si from SiO2 by the electrolysis of

molten salt.

In the electrolysis of a molten salt, the choice of electrolyte is an important factor.

Molten fluoride is considered as a suitable electrolyte for this process because it shows

high solubility for oxides. However, fluoride salts are generally insoluble in water, and

the disposal of spent salt is problematic. In contrast, chloride salts are soluble in water,

and the disposal of spent salt is relatively easy. Molten CaCl2 is an attractive solvent that

shows high solubility for oxides [8]. However, the use of a high-temperature molten salt

such as CaCl2 single salt involves corrosive damage to the vessel, thermal damage to the

electrode materials, and higher energy consumption. Therefore, low-temperature molten

salts that nonetheless show high solubility for oxides are strongly desired. In one study, a

BaCl2-CaCl2-NaCl eutectic melt was selected as a candidate electrolyte because it was

expected to show considerable oxide dissolution; this is attributable to the fact that BaCl2

and CaCl2 show large solubility for oxides [8].

In this chapter, the author investigated the fundamental electrochemical

properties of BaCl2-CaCl2-NaCl as well as the electrochemical reduction of SiO2 from

this system. The electrochemical behavior was investigated by cyclic voltammetry. The

samples were produced by potentiostatic electrolysis.

40

4.2 Experimental

Figure 4-1 shows the schematic drawing of the experimental apparatus. An

eutectic composition (32:48:20 mol%) of BaCl2-CaCl2-NaCl was selected based on a

previously reported ternary phase diagram [9]. BaCl2 (Wako Pure Chemical Co. Ltd.,

95.0%), CaCl2 (Wako Pure Chemical Co. Ltd., 95.0%) and NaCl (Wako Pure Chemical

Co. Ltd., 99.0%) were mixed in an eutectic composition (BaCl2-CaCl2-NaCl = 32:48:20

mol%) and placed in a high-purity alumina crucible (NIKKATO Co. Ltd., 99.5 wt% Al2O3,

SSA-S grade). The mixture was kept under vacuum for more than 24 h at 473 K to expel

its water content. All experiments were performed in dry Ar atmosphere.

For the electrochemical studies, Mo wires (Nilaco Co. Ltd., 1 mm diameter,

99.99%), glassy carbon rods (Tokai Carbon Co. Ltd., 3 mm diameter), Ag wires (Nilaco

Co. Ltd., 1 mm diameter, 99.99%) and Ag plates (Nilaco Co. Ltd., 7 mm × 15 mm × 0.5

mm, 99.5%) were used. The glassy carbon rods were used as counter electrodes. The

reference electrode was an Ag wire immersed in the BaCl2-CaCl2-NaCl eutectic melt

containing 1.0 mol% AgCl, placed in an alumina (mullite) tube with a thin bottom. These

electrodes are shown in Figure 4-2.

A potentiostat/galvanostat (Hokuto Denko Co. Ltd., HZ-3000) was used for the

cyclic voltammetry and chronopotentiometry experiments.

41

The morphologies and cross-sections of the samples were studied by scanning

electron microscope (JEOL Co. Ltd., JSM-7001). X-ray diffraction (Rigaku Co. Ltd.,

Multi Flex) patterns were obtained using Cu-Kα radiation. The surface chemical states

were measured using Raman spectroscopy (Jobin-Yvon Co. Ltd., Labram spectrometer)

with a He-Ne laser (632.8 nm wavelength) over the wavenumber range of 400-700 cm-1.

The photoluminescence spectra were measured using a He-Cd laser (KIMMON

Co. Ltd., IK5451R-E) and a spectrometer (Oceanoptics Co. Ltd., USB2000+).


4.3.1 Electrochemical window of BaCl2-CaCl2-NaCl

Cyclic voltammetry of molten BaCl2-CaCl2-NaCl was conducted at 923 K. In

the negative potential region, a Mo wire was used as the working electrode. Figure 4-3

shows the obtained voltammogram. Sharp cathodic current and corresponding anodic

current were observed at around -2.2 V. These currents were attributed to alkali metal or

alkali earth metal deposition and dissolution of the deposits, respectively. The potential

at which the current became zero after the potential sweep reversed to the anodic direction

(-2.2 V) was defined as the cathodic limit potential.

In the positive potential region, a glassy carbon rod was used as the working

42

electrode. Figure 4-4 shows the obtained voltammogram. As only chlorine ions existed

as anions in the melt, the currents were attributed to the oxidation of chlorine ions to form

chlorine gas.

2Cl- → Cl2 + 2e- (4-1)

After the potential sweep was reversed to the cathodic direction, the current constantly

decreased and became zero at 1.1 V. This potential was defined as the anode limit

potential, and the electrochemical window of BaCl2-CaCl2-NaCl at 923 K was determined

to be 3.3 V. The theoretical potentials of the cathode limits, E0, vs. Cl2/Cl- of BaCl2, CaCl2

and NaCl were -3.71, -3.38 and -3.31 V, respectively, at 923 K. They were calculated

from the standard Gibbs free energy of formation, ΔGf0, of each compound [10] according

to the following equation:

𝐸0 =𝛥𝐺𝑓

0+𝑅𝑇 ln (𝑋𝑀𝑛+)

𝑛𝐹 (4-2)

where R is the gas constant; T, the absolute temperature; XMn+, the mole fraction of the

Mn+ cation; n, the number of electrons; and F, the Faraday constant. This calculation

indicated that the potential of the cathode limit as determined in this study was more

positive than that expected theoretically. These results can be explained as follows. The

stability of cations may decrease relative to that in the single melt upon mixing with BaCl2,

CaCl2 and NaCl. The thermodynamic calculations do not necessarily coincide with the

43

actual decomposition voltages in the mixed melt as the stability of the cations in a mixed

melt was rather different from that in a single melt.


Cyclic voltammetry was conducted at an Ag electrode in molten BaCl2-CaCl2-

NaCl-SiO2 (1.0 mol%) at 923 K. Figure 4-5 shows the cyclic voltammograms for an Ag

electrode before and after the addition of 1.0 mol% SiO2 at a scan rate of 0.1 V s-1 at 923

K. The black and red lines in the figure indicated the voltammograms obtained before and

after the addition of 1.0 mol% SiO2 to the melt, respectively. An Ag electrode was selected

because no alloys or intermetallic compounds existed in the Ag-Si binary system at 923

K [11]. Before the addition of SiO2, in the cathodic sweep, a cathodic wave was observed

at -1.80 V. Considering that Ag forms an alloy with Ba [12], Ca [13] and Na [14], the

cathodic currents were considered to be attributable to AgxMy (M = Ba, Ca and/or Na)

compounds. After reversing the scanning direction at -1.85 V, an anodic current peak was

observed at -1.80 V. This current was caused by the anodic dissolution of M (M= Ba, Ca

and/or Na) from AgxMy compounds. On the other hand, after the addition of SiO2, a

cathodic current increasing from -1.10 V was observed. As no alloy formation is expected

thermodynamically [11], the cathodic current observed at -1.10 V was attributed to Si

44

metal deposition. After reversing the scanning direction to positive at -1.85 V, a sharp

anodic current peak was observed at -1.80 V. Furthermore, new anodic peaks were

observed at -1.65 V and -1.55 V compared to before the addition of SiO2. These results

indicated that in the cathodic sweep, SiO2 was reduced and Si metal was deposited on the

electrode. In a region with more negative potential, the deposited Si formed an alloy with

Ba and/or Ca. In the anodic sweep, Ba and/or Ca metal was dissolved from the SixMy

compounds, as in the case of the voltammogram results.

4.3.3 Si deposition

Based on the cyclic voltammetry results, Si samples were prepared by

potentiostatic electrolysis using an Ag electrode at -1.2 V, -1.6 V and -1.7 V for 1 h in

BaCl2-CaCl2-NaCl molten electrolytes containing 1.0 mol% SiO2 at 923 K.

Figure 4-6 (a) shows the XRD pattern of the sample obtained at -1.2 V. All peaks

were identified as Ag. The formation of the Si film could not be confirmed in this result.

Figure 4-6 (b) shows the XRD pattern of the sample obtained at -1.6 V. Ag and Si phases

were identified. Figure 4-6 (c) shows the XRD pattern of the sample obtained at -1.7 V.

Ag, Si and CaSi2 phases were identified. Thus, in a more negative region than -1.7 V, the

deposited Si formed CaSi2. From these results, a single-phase Si film was obtained at -

45

1.6 V.

An additional experiment was conducted to form bulk Si. A sample was prepared

by potentiostatic electrolysis using an Ag electrode at -1.6 V for 7 h in BaCl2-CaCl2-

NaCl-SiO2 (1.0 mol%) at 923 K. Figure 4-7 shows a cross-sectional SEM image of the

sample. The result indicated that the Si layer was 100 μm thick, and its structure was

porous. Base on the film thickness, morphology and the amount of electric charge passed

during sample preparation, the apparent coulomb efficiency was estimated to exceed

100%. As shown in Figure 4-7, Si has a porous structure. Therefore, the apparent film

thickness increased and the current efficiency exceeded 100%.

4.3.4 Si film growth

To investigate the Si film growth, potentiostatic electrolysis was conducted on

an Ag electrode at -1.6 V at 923 K. Electrolysis was performed for durations varying from

1 to 5 h. Figure 4-8 (a)-(c) shows the XRD patterns of the samples prepared at different

electrolysis durations. The observed peaks were attributed to Ag and Si. Figure 4-9 (a)-

(c) shows cross-sectional SEM images of the samples obtained at -1.6 V for 1, 3 and 5 h,

respectively. Si films with thickness of ~1, 3 and 10 μm were observed. From the observed

film thickness and quantity of electricity, the coulomb efficiency was estimated to be

46

~30%. The deviation from 100% was attributed to the decrease in deposition from the

substrate during washing. Figure 4-10 shows the relationship between the Si thickness

and the amount of electricity. The Si thickness was proportional to the amount of

electricity.

4.3.5 Optical properties

Figure 4-11 shows the appearance of the sample prepared at -1.6 V for 3 h. The

sample surface became brown in color.

Figure 4-12 shows the Raman spectrum of the sample obtained at -1.6 V for 3 h.

The obtained sample showed the characteristic asymmetric broadenings of the Raman

lines. The Raman spectrum showed that the main peak was composed of three separate

vibration modes. The first was a broad band in the range of 480-530 cm-1 caused by

amorphous Si (a-Si). The second was a peak at 515 cm-1 attributed to μc-Si [15]. The third

was a Raman band at 519 cm-1 that represented TO phonon modes that were degenerated

at the Brillouin zone center. The frequency shift of porous Si with respect to c-Si increases

in absolute value with increasing porosity, as predicted by the theoretical calculations [16].

In the case, the sample structure showed a mix of c-Si and a-Si.

Figure 4-13 shows the photoluminescence emission spectrum of the sample

47

obtained after potentiostatic electrolysis at -1.6 V for 3 h. This spectrum was recorded at

excitation of 442 nm light selected from the He-Cd radiation. This figure showed weak

red emissions with broad bandwidths, which were centered at 650 and 750 nm in the

range of 530-820 nm. Transition of the plasma states in the laser tube was observed as

sharp emission lines, and the monotonic decreasing background of the He-Cd laser was

spread from the 442 nm laser line throughout the measurement range. As crystalline Si

emits only extremely weak infrared photoluminescence owing to its relatively small and

indirect bandgap, visible light emission was attributable to the porosity of Si.

4.4 Conclusion

The study showed that the electrochemical window of the BaCl2-CaCl2-NaCl

system was 3.3 V at 923 K. Furthermore, the reaction at the anode limit was proposed as

chlorine gas evolution. The electrochemical reduction of SiO2 was investigated in the

molten BaCl2-CaCl2-NaCl-SiO2 system. SiO2 dissolved in the melt to form Si ions; these,

in turn, were electrochemically reduced to form Si metal. Finally, porous Si was formed

by potentiostatic electrolysis at the Ag electrode at -1.6 V for 3 h.

48

References

[1] L. T. Canham, Appl. Phys. Lett., 57, 1046 (1990).

[2] D. Elwell and G. M. Rao, Electrochim. Acta, 26, 673 (1982).

[3] S. V. Kuznetsova, V. S. Dolmatov and S. A. Kuznetsov, Russ. Electrochem., 45, 742

(2009).

[4] A. L. Bieber, L. Massot, M. Gibilaro, L. Cassayre, P. Taxil and P. Chamelot,



[6] S. K. Cho, F. R. F. Fan and A. J. Bard, Electrochim. Acta, 65, 57 (2012).

[7] Y. Sakanaka and T. Goto, Electrochim. Acta, 164, 139 (2015).

[8] D. A. Wenz, I. Johnson and R. D. Wolson, J. Chem. Eng. Data., 14, 252 (1969).

[9] D. V. Sementsova and G. A. Bukhalova, Zh. Neorgan. Khim., 11 [1], 168 (1966).

[10] M. W. Chase Jr, NIST-JANAF Thermochemical Tables, 4th ed., Editor, ACS, New

York (1998).


615 (1990)


14 (1990).

49


21 (1990).


62 (1990).

[15] C. Smit, R. A. C. M. M. Van Swaaji, H. Donker, A. M. H. N. Petit, W. M. M. Kessels


[16] I. H. Campbell and P. M. Fauchet, Solid State Commun., 58, 739 (1986).

50

A: Reference electrode (Ag+/Ag)


C: Thermocouple

D: Working electrode (Ag, Mo or Glassy carbon)

E: Electrolyte (BaCl2-CaCl2-NaCl)

F: Electric furnace



51

Working electrode

Counter electrode

Reference electrode


52

Figure 4-3 Cyclic voltammogram for a Mo electrode in BaCl2-CaCl2-NaCl at 923 K.

Scanning rate: 0.1 V s-1.

Figure 4-4 Cyclic voltammogram for a glassy carbon electrode in BaCl2-CaCl2-NaCl at

923 K. Scanning rate: 0.1 V s-1.

53

Figure 4-5 Cyclic voltammograms for an Ag electrode in BaCl2-CaCl2-NaCl before and

after the addition of 1.0 mol% SiO2 at 923 K. Scanning rate: 0.1 V s-1.

Figure 4-6 XRD patterns of the samples obtained by potentiostatic electrolysis at

(a) -1.2, (b) -1.6 and (c) -1.7 V for 1 h using Ag electrode.

54

Figure 4-7 Cross-sectional SEM image of the sample obtained by potentiostatic

electrolysis at -1.6 V for 7 h.

Figure 4-8 XRD patterns of the samples obtained by potentiostatic electrolysis

at -1.6 V: (a) 1, (b) 3 and (c) 5 h.

55

Figure 4-9 Cross-sectional SEM images of the samples obtained by potentiostatic

electrolysis at -1.6 V: (a) 1, (b) 3 and (c) 5 h.

Figure 4-10 The relationship between the Si thickness and the amount of electricity.

56

Figure 4-11 Photograph of the Ag plate of obtained by potentiostatic electrolysis

at -1.6 V for 3 h.

Figure 4-12 Raman spectrum of the sample obtained by potentiostatic electrolysis

at -1.6 V for 3 h.

57

Figure 4-13 PL spectrum of the sample obtained by potentiostatic electrolysis

at -1.6 V for 3 h.

58

Chapter 5

Electrodeposition of Si in molten KF-SiO2

5.1 Introduction

In chapters 3 and 4, the electrochemical formation of Si in LiF-NaF-KF and

BaCl2-CaCl2-NaCl containing small amounts of SiO2 was described. However, this

process led to some problems concerning the rate of film growth and the current efficiency.

Thus, this chapter described the attempts to increase the efficiency of the production of

Si from SiO2.

Molten SiO2 is the ideal electrolyte, because, in principle, only pure Si is

deposited at the cathode without any side reaction.

However, SiO2 has a high electrical resistivity (106 Ω cm at 1273 K [1]) and

melting point (1923 K). The use of a high-temperature molten salt, such as SiO2 single

salt, involves corrosive damage to the vessel and thermal damage to the electrode

materials. Molten fluoride is a suitable supporting electrolyte because it has high electrical

conductivity. For example, the electrical conductivity of KF is 4.5 Ω-1 cm-1 at 1273 K [2].

Moreover, K forms no alloy with Si [3]. For these reasons, a KF-SiO2 binary melt was

59

selected.

The fundamental thermal properties of the KF-SiO2 system were measured to

confirm the eutectic composition and melting point. Moreover, the electrochemical

behavior was investigated by cyclic voltammetry. Si was prepared by potentiostatic

electrolysis and the obtained sample was analyzed by XRD and Raman spectroscopy.

5.2 Experimental

5.2.1 Thermal analysis

The melting point of KF-SiO2 was determined by differential thermal analysis

(Shimadzu Co. Ltd., DTG-60H). High purity Al2O3 was used as the reference material.

Samples with different compositions of the mixture (30 mg per sample) were

placed in Pt vessels and the measurement was carried out from room temperature to 1273

K at a heating rate of 10 K min-1 under Ar gas flow.

5.2.2 Electrochemical measurements

Figure 5-1 shows a schematic drawing of the experimental apparatus. A 50:50

mol% KF-SiO2 mixture was selected as the electrolyte. KF (Wako Pure Chemical Co.

Ltd., 99.0%) and SiO2 (Wako Pure Chemical Co. Ltd., 99.0%) were mixed (KF-SiO2 =

50:50 mol%), and placed in a high-purity alumina crucible (NIKKATO Co. Ltd., 99.5

60

wt% Al2O3, SSAS-grade). The mixture was kept under vacuum for more than 24 h at 473

K to expel water. All experiments were performed in a dry Ar atmosphere. For the

electrochemical studies, Ag wires (Nilaco Co. Ltd., 1 mm diameter, 99.99%) and Ag

plates (Nilaco Co. Ltd., 7 mm × 15 mm × 0.5 mm; 99.5%) were used. Glassy carbon rods

(Tokai Carbon Co. Ltd., 3 mm diameter) were used as counter electrodes. The potentials

of the working and counter electrodes were measured with respect to a Ni wire immersed

in the molten electrolyte, which acted as the quasi-reference electrode. These electrodes

are shown in Figure 5-2. A potentiostat/galvanostat (Hokuto Denko Co. Ltd., HZ-3000)

was used for cyclic voltammetry and chronopotentiometry experiments. X-ray diffraction

(Rigaku Co. Ltd., Multi Flex) patterns were obtained using Cu-Kα radiation. Surface

chemical states were measured via Raman spectroscopy (Jobin-Yvon Co. Ltd., Labram

spectrometer) using a He-Ne laser (632.8 nm wavelength) over the 300 to 800 cm-1

wavenumber range.


5.3.1 Thermal analysis of binary KF-SiO2 system

Based on the thermal analysis results, a schematic phase diagram was proposed

for the KF-SiO2 system, as shown in Figure 5-3. KF and SiO2 formed an eutectic phase

61

at about 50 mol% KF with a melting temperature of 993 K. No phases other than KF and

SiO2 were detected in the investigated temperature range.


Cyclic voltammetry was conducted with an Ag electrode in molten KF-SiO2

(50:50 mol%) at 1073 K. Figure 5-4 shows the cyclic voltammogram for an Ag electrode

at a scan rate of 0.1 V s-1 at 1073 K. In the negative potential region, a sharp cathodic

current and the corresponding anodic current were observed at about -0.9 V, which were

attributable to the deposition of Si metal and dissolution of the deposits, respectively.

The standard formal potentials of KF and SiF4, calculated from the standard

Gibbs energy of formation, were 4.88 V and 3.94 V, respectively [4]. This calculation

suggested that the reaction at the cathode limit was the formation of Si.

5.3.3 Deposition of Si

Based on the result of cyclic voltammetry, a sample was prepared by

potentiostatic electrolysis on an Ag electrode at -1.0 V for 10 min. Figure 5-5 shows the

XRD pattern of the sample. The peaks were assigned to Ag and Si. Figure 5-6 shows the

Raman spectrum obtained for the sample. The peak centered at 520 cm-1 represented the

62

TO phonon modes of c-Si [5].

The amount of Si that was deposited was measured by comparing the weight of

the electrode before and after electrolysis. As a result, the weight of obtained Si was 3.0

× 10-4 g cm-2 after 10 min of electrolysis and the calculated thickness of the Si layer was

1.3 μm. These values were comparable to 1.8 × 10-3 g cm-2 h-1 and 8 μm h-1. From the

measured weight and quantity of electricity, the coulomb efficiency was estimated to be

40%. The deviation from 100% was attributed to the exfoliation in deposition from the

substrate during washing.

These results indicated that the growth rate of Si and current efficiency were

higher than the results described in chapters 3 and 4.

5.4 Conclusion

The eutectic composition of KF-SiO2 was confirmed to be 50:50 mol% and the

corresponding eutectic temperature was 993 K. The electrochemical formation of Si was

studied in KF-SiO2 (50:50 mol%). Si was deposited by potentiostatic electrolysis at -1.0

V for 10 min and the growth rate of Si was about 8 μm h-1.

63

References

[1] O. V. Mazurin, M. V. Strel’tsina and T. P. Shaviko-Shavikovskaya, Properties of

Glasses and Glass-Forming Melts: A Handbook, Vol. 1, Nauka, Leningrad (1973).

[2] E. W. Yim and M. Feinleib, J. Electrochem. Soc., 104, 626 (1957).


2338 (1990)

[4] M. W. Chase Jr, NIST-JANAF Thermochemical Tables, 4th ed., Editor, ACS, New

York (1998).

[5] C. Smit, R. A. C. M. M. van Swaaji, H. Donker, A. M. H. N. Petit, W. M. M. Kessels


64

A: Quasi-reference electrode (Ni)


C: Thermocouple

D: Working electrode (Ag)

E: Electrolyte (KF-SiO2)

F: Electric furnace



65

Working electrode

Counter electrode

Quasi-reference electrode


66

Figure 5-3 Phase diagram of the KF-SiO2 system obtained by differential thermal

analysis.

Figure 5-4 Cyclic voltammogram for an Ag electrode in KF-SiO2 at 1073 K.


67

Figure 5-5 XRD pattern of the sample obtained by potentiostatic electrolysis

at -1.0 V at 10 min.

Figure 5-6 Raman Spectrum of the sample obtained by potentiostatic electrolysis

at -1.0 V for 10 min.

68

Chapter 6

Electrochemical formation of Ca-Si in molten CaCl2-KCl

6.1 Introduction

As described in section 1.4, Ca-Si alloy films attract great interest of many

researchers for functional properties. Generally, these alloy films are manufactured CVD

and PVD. Nevertheless, the high vapor pressure of Ca makes the growth of a continuous

film layer by deposition from the gas phase [1-6] difficult. Ca atoms are easily evaporated

from the Si substrate, which prevents the formation of Ca-silicide by interdiffusion with

the Si substrate. Therefore, novel methods to produce Ca-Si alloy films will be required.

With this background, the author proposed a novel electrochemical process for

the formation of alloy films. The molten salt electrochemical process has been studied by

Ito and his co-workers [7-10]. This process is advantageous because (1) the phases of the

alloy films can be controlled by adjusting the electrochemical parameters and (2) the films

can be grown on substrates of various shapes. As a typical example, formation and control

of a definite Y-Ni alloy film on a Ni electrode was achieved [7]. Furthermore, they

reported that the anodic polarization of the formed Ni2Y resulted in the dissolution of Y

69

to form other Y-Ni phases.

In this chapter, the author chose a Ca-Si as a model to confirm the feasibility of

this process for producing silicide films, and investigated the electrodeposition of Ca to

form Ca-Si alloys in a molten CaCl2-KCl system at 923 K.

6.2 Experimental

CaCl2 (Wako Pure Chemical Co. Ltd., 95.0%) and KCl (Wako Pure Chemical

Co. Ltd., 99.5%) were mixed in eutectic composition (CaCl2:KCl = 38.4:61.6 mol%), and

introduced in a high-purity alumina crucible (NIKKATO Co. Ltd., 99.5 wt% Al2O3, SSA-

S grade), which was then kept under vacuum for more than 24 h at 473 K to ensure

complete removal of water. All experiments were performed in the CaCl2-KCl eutectic

melt in a dry Ar atmosphere. Temperature measurements were performed using a

chromel-alumel thermocouple, with an accuracy of ±1 K. Figure 6-1 shows the schematic

drawing of the experimental apparatus. To investigate the electrochemical behavior, Mo

wires (Nilaco Co. Ltd., 1 mm diameter, 99.99%), glassy carbon rods (Tokai Carbon Co.

Ltd., 3 mm diameter) and n-type Si plates (Nilaco Co. Ltd., 5 mm × 20 mm × 0.5 mm,

99.5%) were used as the working electrodes. The reference electrode was an Ag wire

immersed in CaCl2-KCl containing 1.0 mol% of AgCl, placed in an alumina tube with a

70

thin bottom to maintain electrical contact with the melt. These electrodes are shown in

Figure 6-2. The potential of the reference electrode was calibrated with reference to that

of an Agn+/Ag electrode. All potentials referred to in this chapter were expressed with

reference to the Agn+/Ag potential. The counter electrode was a glassy carbon rod (Tokai

Carbon Co. Ltd., 3 mm diameter). A potentiostat/galvanostat (Hokuto Denko Co. Ltd.,

HZ-3000) was used for electrochemical measurements. The samples were prepared by

potentiostatic electrolysis and rinsed with ethylene glycol. The obtained samples were

analyzed by X-ray diffractometry (Rigaku Co. Ltd., Multi Flex) with a Cu-Kα line. The

cross-sections of the samples were observed by scanning electron microscopy (JEOL Co.

Ltd., JSM-7001). For evaluation of the optical properties of the films, ultraviolet-visible-

near infrared (UV-VIS-NIR) reflectance spectra were measured using a

spectrophotometer (JASCO Co. Ltd., V-670).


6.3.1 Electrochemical window of CaCl2-KCl

Figure 6-3 shows the cyclic voltammogram obtained in CaCl2-KCl at 923 K. A

Mo wire was used as the working electrode. In the negative potential region, a sharp

cathodic current and the corresponding anodic current were observed at about -2.50 V,

71

which were attributable to the deposition of Ca or K metal and dissolution of the deposits,

respectively. The current observed at a potential more negative than -1.90 V was due to

the formation of Ca metal (activity smaller than 1) and its dissolution in CaCl2. The

potential at the cathodic limit was defined as the potential of the Mn+/M electrode prepared

by the following procedure. Galvanostatic electrolysis was conducted at -50 mA cm-2

using a Mo electrode for 20 s, and then, the open-circuit potential was measured, as shown

in Figure 6-4. The potential measured immediately after the electrolysis was the Mn+/M

potential.

In the positive potential region, a glassy carbon rod was used as the working

electrode. The anodic current increased from about 1.10 V, as shown in Figure 6-5.

Because chloride ions were the only anions in this melt, the anodic current was considered

to be due to the oxidation of chloride ions to chlorine gas.

2Cl-→Cl2+2e- (6-1)

After the potential sweep direction was reversed (to negative), the current

constantly decreased, eventually falling to zero at 3.46 V. This potential was defined as

the potential of the anodic limit, since it was regarded as the Cl2/Cl- potential. From the

CV results, the electrochemical window was determined to be 3.46 V.

72


In order to investigate the electrochemical behavior of the Ca ions, cyclic

voltammetry was conducted in molten CaCl2-KCl at 923 K. Figure 6-6 shows the cyclic

voltammograms for Si and Mo electrode at a scanning rate of 0.1 V s-1 at 923 K. The

electrochemical behavior of a Mo electrode was also investigated for comparison,

because Mo forms no alloy with Ca [11]. For the Mo electrode, a sharp increase in the

cathodic current was observed at -2.50 V. Since Mo forms no alloy with Ca and K, the

cathodic current was considered to be due to Ca or K metal deposition. After reversing

the scanning direction at -2.60 V, a large anodic current peak was observed, because of

the anodic dissolution of Ca or K metal.

For the Si electrode, cathodic currents were observed from -1.80 V. Since this

potential was more positive than the potential for Ca metal deposition, the cathodic

currents were attributed to the formation of Ca-Si alloys. The standard formal potentials

of Ca2Si, CaSi and CaSi2 were 0.35, 0.46 and 0.49 V, respectively, as calculated from the

corresponding standard Gibbs energies of formation [12]. These results agreed with those

deduced from the cyclic voltammograms in Figure 6-6.

When the potential scan direction was reversed at -2.60 V, several anodic peaks

were observed at -2.15, -1.85 and -1.50 V, respectively, indicating Ca dissolution from

73

the different Ca-Si alloy phases.

6.3.3 Formation of Ca-Si alloy

Based on the results of cyclic voltammetry, an alloy sample was prepared by

potentiostatic electrolysis on a Si electrode at -2.60 V for 1 h. Figure 6-7 shows the XRD

pattern of the sample. The peaks were assigned to Ca2Si, CaSi, CaSi2 and the Si substrate,

along with several unknown peaks.

Figure 6-8 shows a cross-sectional SEM image of the sample. The observed alloy

layer with a thickness of approximately 30 μm was considered to comprise Ca-Si layer,

in accordance with the XRD result.

6.3.4 Phase control of Ca-Si

The phase diagram of the Ca-Si system is shown in Figure 6-9 [13]. According

to this diagram, three Ca-Si intermetallic compounds having Ca concentrations lower than

those of CaSi2, CaSi and Ca2Si should exist at 923 K.

Chronopotentiometry measurement was conducted to confirm the possibility of

formation of Ca-Si alloy phases.

Open-circuit potentiometry was carried out to further investigate the formation

74

of Ca-Si alloys. Figure 6-10 shows the open-circuit potential transient curve for a Si

electrode after the deposition of Ca metal by galvanostatic electrolysis at -50 mA cm-2 for

60 s, in molten CaCl2-KCl at 923 K.

As can be seen in the chronopotentiogram, the potential remained at 0 V for the

initial 10 s, probably because of the presence of the deposited Ca metal on the electrode.

Subsequently, plateaus were observed at -2.32, -2.03 and -1.93 V, which were possibly

due to different coexisting Ca-Si phases. Based on this result, samples were prepared by

potentiostatic electrolysis.

First, potentiostatic electrolysis was conducted at -2.60 V for 30 min. Then,

anodic dissolutions of Ca were conducted for 30 min by potentiostatic electrolysis at -

2.42 (sample 1), -2.15 (sample 2), -2.02 (sample 3) and -1.74 V (sample 4). These

potential values were determined by taking into account the potential plateaus before and

after the potential jump indicating complete phase change.

The phases of the samples were analyzed by XRD, and the cross-sections of the

samples were observed by SEM.

Figure 6-11 (a) shows the XRD pattern of sample 1 (obtained at -2.42 V). The

alloy phases were identified as CaSi and CaSi2. Figure 6-11 (b) shows the cross-sectional

SEM of sample 1, indicating that the thickness of the Ca-Si film was about 10 μm. Figure

75

6-12 (a) shows the XRD pattern of sample 2 (obtained at -2.15 V). The alloy phase was

identified as CaSi2. From the cross-sectional SEM image (Figure 6-12 (b)), the thickness

of the CaSi2 film was determined to be about 20 μm.

Therefore, the potential plateau at -2.32 V was considered to correspond to the

following reaction:

CaSi2 + Ca(II) + 2e- ⇔ 2CaSi (6-2)

Figure 6-13 (a) shows the XRD pattern of sample 3 (obtained at -2.02 V),

indicating Si and CaSi2 phases. The cross-sectional SEM image in Figure 6-13 (b) shows

the film thickness to be about 20 μm. Figure 6-14 (a) shows the XRD pattern of sample

4 (obtained at -1.74 V), revealing only Si peaks. Figure 6-14 (b) shows the cross-sectional

SEM image of sample 4, indicating that the transformed film was about 20 μm thick with

a porous structure. Thus, the potential plateau at -1.93 V was considered to correspond to

the following reaction:

2Si + Ca(II)+2e- ⇔ CaSi2 (6-3)

These results suggested that phase control of the Ca-Si alloy was possible by

appropriately adjusting the applied potential.

76

6.3.5 Optical properties of Ca-Si film

The optical properties of the Ca-Si film, such as reflectance and absorption

coefficient, were investigated as a function of wavelength by UV-VIS-NIR reflectance

measurements. The reflection spectrum of the Ca-Si product in the wavelength range 300-

600 nm is shown in Figure 6-15, which revealed reflectance changes in the UV region. In

the UV region, the Kubelka-Munk function [14-15] was used to convert the diffuse

reflectance:

α/S = (1-R)2/2R (6-4)

where α is the absorption coefficient, S is the scattering coefficient which is assumed to

be constant and equal to 1, and R is the diffuse reflectance.

For determining the absorption edge of the semiconductors, the following

equation applicable to a direct bandgap semiconductor was used.

(hνα)2=C(hν-Eg) (6-5)

The bandgap energy was determined from the plot of (hνα/S)2 versus hν, which was linear,

as illustrated in Figure 6-16. The bandgap value was obtained by extrapolating the straight

portion of the graph on the hν axis to (hνα/S)2=0, as indicated by the dotted line in the

figure. Thus, the direct absorption edge was found to be located at 3.1 eV. From the

present measurements, the bandgap of CaSi2 was confirmed to be in the ultraviolet region.

77

6.4 Conclusion

The electrochemical formation of a Ca-Si alloy was investigated in CaCl2-KCl

at 923 K. The electrochemical window of CaCl2-KCl was found to be 3.46 V at 923 K.

Potentiostatic electrolysis of a Si electrode at -2.60 V resulted in the formation of a

multiphase Ca-Si film having a thickness of about 30 μm. The multiphase Ca-Si film was

converted into a CaSi, CaSi2 or Si phase by anodic potentiostatic electrolysis depending

on the potential. The various transformation reactions and the corresponding equilibrium

potentials were clarified. UV-VIS-NIR reflectance measurements confirmed that the

bandgap of CaSi2 was 3.1 eV with a direct absorption edge.

78

References

[1] H. N. Acharya, S. K. Dutta and H. D. Banerjee, Sol. Ene. Mat., 3, 441 (1980).

[2] T. Koga, A. Bright, T. Suzuki, K. Shimada, H. Tatsuoka and H. Kuwabara, Thin Sol.

Films, 369, 248 (2000).

[3] M. Sugiyama and Y. Maeda, Thin Sol. Films, 381, 225 (2001).

[4] T. Nakamura, T. Suematsu, K. Takakura, F. Hasegawa, A. Wakahara and M. Imai,

Appl. Phys. Lett., 81, 1032 (2002).

[5] T. Hosono, Y. Matsuzawa, M. Kuramoto, Y. Momose, H. Tatsuoka and H. Kuwabara,

Solid State Phenom., 93, 447 (2003).

[6] I. Kogut and M. C. Record, Intermetallics, 32, 184 (2013).

[7] G. Xie, K. Ema and Y. Ito, J. Appl. Electrochem., 23, 753 (1993).

[8] G. Xie, K. Ema and Y. Ito, J. Appl. Electrochem., 24, 321 (1994).

[9] K. Tateiwa, M. Tada and Y. Ito, Hyomen Gijyutsu (The Journal of the Surface

Finishing Society of Japan), 46, 673 (1995) (in Japanese).

[10] T. Iida, T. Nohira and Y. Ito, Electrochim. Acta, 46, 2537 (2001).


930 (1990).

[12] F*A*C*T, http://www.crct.polymtl.ca/reacweb.htm.

79


953 (1990).

[14] P. Kubelka and F. Munk, Z. Tech. Phys., 12, 593 (1931).

[15] P. Kubelka, J. Opt. Soc. Am., 38, 448 (1948).

80

A: Reference electrode (Ag+/Ag)


C: Thermocouple

D: Working electrode (Si or Mo)

E: Electrolyte (CaCl2-KCl)

F: Electric furnace



81

Working electrode

Counter electrode

Reference electrode


82

Figure 6-3 Cyclic voltammogram for a Mo electrode in CaCl2-KCl eutectic at 923 K.


Figure 6-4 Open-circuit potential transient curve for a Mo electrode in CaCl2-KCl

system at 923 K. Firstly, Ca metal was electrodeposited at the electrode by galvanostatic

electrolysis at -50 mA cm-2 for 20 s.

83

Figure 6-5 Cyclic voltammogram for a glassy carbon electrode in CaCl2-KCl eutectic

at 923 K. Scanning rate: 0.1 V s-1.

84

Figure 6-6 Cyclic voltammograms for Mo and Si electrode in CaCl2-KCl eutectic

at 923 K. Scanning rate: 0.1 V s-1.

85

Figure 6-7 XRD pattern of the sample obtained by potentiostatic electrolysis with a Si

electrode at -2.60 V for 1 h in CaCl2-KCl eutectic at 923 K.

Figure 6-8 Cross-sectional SEM image for the sample at -2.60 V for 1 h.

86

Figure 6-9 Phase diagram of the Ca-Si system [13].

87

Figure 6-10 Open-circuit potential transient curve for a Si electrode in CaCl2-KCl at 923

K. Before the measurement, Ca metal was electrodeposited at the electrode by

galvanostatic electrolysis at -50 mA cm-2 for 30 s.

Figure 6-11 (a) XRD pattern of the sample 1 obtained by electrolysis

at -2.60 V for 30 min and -2.42 V for 30 min.

88

Figure 6-11 (b) Cross-sectional SEM image of the sample 1.



89




90




91


Figure 6-15 UV-VIS-NIR reflectance spectrum of the sample 2.

92

Figure 6-16 (hνα/S)2 versus hν of the sample 2.

93

Chapter 7

Co-deposition of Ca-Si in molten CaCl2-KCl-K2SiF6

7.1 Introduction

In chapter 6, the author described the successful preparation of several Ca-Si

alloys in a molten CaCl2-KCl system. To be precise, the surface of a Si electrode was

transformed to a Ca-Si alloy by the Ca deposition method. The resultant Ca-Si phase

could be converted to other Ca-Si phases by anodic potentiostatic electrolysis, depending

on the potential. However, by this method, the Ca-Si alloys could be formed only on a Si

substrate. To overcome this problem, the author considered the possibility of the co-

deposition of Ca and Si. By this approach, Ca-Si alloys can be formed on various

substrates, including those that are electrically conductive.

With this background, then, the author described herein the electrodeposition of

Ca and Si to form Ca-Si alloys in a molten CaCl2-KCl-K2SiF6 system at 923 K. In

particular, the relationship between the electrolysis time and thickness of the obtained Ca-

Si film was investigated.

94

7.2 Experimental

The experimental apparatus and the electrodes used in this chapter were very

similar to those described in chapter 6. CaCl2 (Wako Pure Chemical Co. Ltd., 95.0%) and

KCl (Wako Pure Chemical Co. Ltd., 99.5%) were mixed to form an eutectic composition

(CaCl2:KCl = 38.4:61.6 mol%), and introduced into a high purity alumina crucible

(NIKKATO Co. Ltd., 99.5 wt% Al2O3, SSA-S grade). All experiments were performed

in the CaCl2-KCl eutectic melt under dry Ar atmosphere. K2SiF6 (Wako Pure Chemical

Co. Ltd., 95.0%) was added directly to the melts as the Si ion source. For the

investigations of electrochemical behavior, n-type Si plates (Nilaco Co. Ltd., 5 mm×20

mm×0.5 mm, 99.5%) and Mo wires (Nilaco Co. Ltd., 1 mm diameter, 99.99%) were used

as the working electrode. The reference electrode was an Ag wire immersed in CaCl2-

KCl containing 1.0 mol% AgCl, placed in an alumina tube with a thin bottom to maintain

electrical contact with the melt. The potential of this reference electrode was calibrated

with reference to that of a Mn+/M electrode, which was prepared by electrodepositing an

alkali metal on Mo wire. All potentials given in this chapter refered to this Mn+/M

potential. The counter electrode was a glassy carbon rod (Tokai Carbon Co. Ltd., 3 mm

diameter).

A potentiostat/galvanostat (Hokuto Denko Co. Ltd., HZ-3000) was used for

95

electrochemical measurements. The samples were prepared by potentiostatic electrolysis,

and rinsed with ethylene glycol. The obtained samples were analyzed by X-ray diffraction

(Rigaku Co. Ltd., Multi Flex) using a Cu-Kα radiation source. The cross-sections of the

samples were observed by scanning electron microscopy (JEOL Co. Ltd., JSM-7001).


7.3.1 Electrochemical behavior of K2SiF6

To investigate Ca-Si alloy formation, cyclic voltammetry was conducted in

molten CaCl2-KCl at 923 K. The dotted line in Figure 7-1 shows the cyclic

voltammogram of a Si electrode at a scan rate of 0.01 V s-1. On the cathodic sweep,

cathodic currents were observed from approximately -1.50 V, which were attributed to

the following reaction:

Si + xCa(II) + 2xe- → CaxSi (7-1)

Thereafter, large cathodic currents were observed from -2.20 V, which were considered

to correspond to the deposition of Ca metal. After reversal of the sweep direction at -2.40

V, an anodic current peak corresponding to the dissolution of Ca metal was observed. At

potentials in the -1.80 to -1.50 V range, several anodic peaks were observed

corresponding to the dissolution of Ca metal from the Ca-Si alloy. In contrast, the solid

96

line in Figure 7-1 shows the cyclic voltammogram of the after addition of K2SiF6 (1.0

mol%) of a Si electrode at a scan rate of 0.01 V s-1. On the cathodic sweep, a cathodic

wave was evident at approximately -0.40 V, which was considered to be attributed to the

following reaction:

Si(IV) + 4e- → Si (7-2)

Then, new cathodic waves were observed at more negative values, which were

considered to be associated with the co-deposition of the Ca-Si alloy.

xCa(II) + ySi(IV) + (2x+4y)e- → CaxSiy (7-3)

On the anodic sweep, anodic currents were also observed, as was seen for the

blank Si electrode before the addition of K2SiF6. These anodic currents were thought to

be associated with the dissolution of Ca from the Ca-Si alloy.

7.3.2 Formation of Ca-Si alloy

Based on the cyclic voltammetry results, a co-deposited sample was prepared by

potentiostatic electrolysis for 1.0 h using a Si electrode in a molten CaCl2-KCl-K2SiF6

system. The electrolysis potential value was selected as -2.30 V since a steep cathodic

current was observed at this potential. Figure 7-2 shows the XRD pattern of the sample,

which was identified as Ca2Si, CaSi2 and the Si substrate.

97

It was reported that Ca5Si3 and Ca3Si4 thermodynamically existed [1]. However,

these phases were not identified in the sample under the experimental condition, perhaps

because the formation of these phases were kinetically impossible.

The cross-sectional SEM image of the sample in Figure 7-3 revealed an alloy

layer with an approximate thickness of 75 μm, which was considered to be a Ca-Si layer

on the basis of the XRD result. Potentiostatic electrolysis at -2.30 V was also conducted

for 0.5, 1.5 and 2.5 h. Figure 7-4 (a) shows the cross-sectional SEM image of the sample

for obtained after 0.5 h. The thickness of the alloy layer was approximately 45 μm. The

cross-sectional SEM image shown in Figure 7-4 (b) for the sample after 1.5 h electrolysis

corresponds to the thickness of alloy layer was approximately 90 μm. Figure 7-4 (c)

shows the cross-sectional SEM image for the sample after 2.5 h electrolysis, indicating a

thickness of approximately 120 μm for the alloy layer.

7.3.3 Growth of Ca-Si alloy films by co-deposition

The cross-sectional SEM images of the samples obtained at -2.30 V for 0.5, 1.0,

1.5 and 2.5 h revealed adhesive Ca-Si films with thicknesses of about 45, 75, 90 and 120

μm, respectively. In the case of the solid-phase diffusion as shown in chapter 6, the

thickness of the Ca-Si layer was about 30 μm for 1 h. On the other hand, in the case of

98

the co-deposition for 1 h, the obtained layer was 75 μm, indicating that the co-deposition

would be superior to the solid-phase diffusion.

The relationship between the thickness of the alloy and the electrolysis time is

shown in Figure 7-5. Provided that the growth of the alloy was controlled by the supply

of Ca and Si at the surface, the growth was linear with time. However, the growth rate

observed in Figure 7-5 seemed to be not fulfilled with the linear, but parabolic law.

Concerning the observed growth mechanism, the following explanation is

proposed. The initial concentration of Ca ion was about 38 times higher than that of Si

ion and the amount of Si ion was insufficient to maintain the linear growth of the alloy

phase. Therefore, the deposition of Ca metal and its diffusion into the electrode became

the dominant reaction processes over time.

7.4 Conclusion

The electro-codeposition of Ca and Si on a Si substrate was investigated in a

molten CaCl2-KCl-K2SiF6 system at 923 K. The formation of a Ca-Si alloy was confirmed

by XRD analysis and cross-sectional SEM image of a sample prepared at -2.30 V for 1.0

h. The thickness of the film was approximately 75 μm, which was identified as Ca2Si,

CaSi2 and Si.

99

Much thicker alloy films were obtained by prolonged potentiostatic electrolysis.

The samples obtained at -2.30 V for 0.5, 1.5 and 2.5 h revealed adhesive Ca-Si films with

thicknesses of about 45, 90 and 120 μm, respectively.

The growth rate of the film obtained by the co-deposition was superior to that of

the solid-phase diffusion.

100

Reference

[1] T. B. Massalski, Binary Alloy Phase Diagrams, Vol. 1, Editor, ASM International, 953

(1990).

101

Figure 7-1 Cyclic voltammograms for a Si electrode in CaCl2-KCl before and after

addition of K2SiF6 at 923 K. Scanning rate: 0.01 V s-1.

Figure 7-2 XRD pattern of the sample obtained by potentiostatic electrolysis with a Si

electrode at -2.30 V for 1.0 h in CaCl2-KCl-K2SiF6 at 923 K.

102

Figure 7-3 Cross-sectional SEM image of the sample obtained by potentiostatic

electrolysis with a Si electrode at -2.30 V for 1.0 h in CaCl2-KCl-K2SiF6 at 923 K.

Figure 7-4 Cross-sectional SEM images of the samples obtained by electrolysis at

-2.30 V for (a) 0.5, (b) 1.5 and (c) 2.5 h.

103

Figure 7-5 Relationship between the thickness of the alloy films and the electrolysis

time.

104

Chapter 8

General conclusion

In this study, electrochemical reduction of SiO2 in molten fluoride and molten

chloride were investigated. In addition, electrochemical formation of Ca-Si was

investigated in molten chloride by two ways such as solid-phase diffusion and co-

deposition. The results obtained in this study are summarized as follows.

In chapter 3, the electrochemical reduction of SiO2 was investigated by cyclic

voltammetry and chronoamperometry in molten LiF-NaF-KF-SiO2 system at 873 K.

Potentiostatic electrolysis at the Ag electrode at 0.2 V, 0.4 V and 0.7 V for 1 h resulted in

the formation of polycrystalline Si. A Si film that was ~1 μm thick was obtained by

potentiostatic electrolysis at 0.2 V for 1 h. Furthermore, electrodeposition of Si was

achieved using a new inert anode.

In chapter 4, the study showed that the electrochemical window of the BaCl2-

CaCl2-NaCl system was 3.3 V at 923 K. Furthermore, the reaction at the anode limit was

proposed as chlorine gas evolution. The electrochemical reduction of SiO2 was

investigated in the molten BaCl2-CaCl2-NaCl-SiO2 system. SiO2 dissolved in the melt to

105

form Si ions; these, in turn, were electrochemically reduced to form Si metal. Finally,

porous Si was formed by potentiostatic electrolysis at the Ag electrode at -1.6 V for 3 h.

In chapter 5, the eutectic composition of KF-SiO2 was confirmed to be 50:50

mol% and the corresponding eutectic temperature to be 993 K. The electrochemical

formation of Si was studied in KF-SiO2 (50:50 mol%). Si was deposited by potentiostatic

electrolysis at -1.0 V for 10 min and the growth rate of Si was calculated at about 8 μm

h-1.

In chapter 6, the electrochemical formation of a Ca-Si alloy was investigated in

CaCl2-KCl at 923 K. The electrochemical window of CaCl2-KCl was found to be 3.46 V

at 923 K. Potentiostatic electrolysis of a Si electrode at -2.60 V resulted in the formation

of a multiphase Ca-Si film having a thickness of about 30 μm. The multiphase Ca-Si film

was converted into a CaSi, CaSi2 or Si phase by anodic potentiostatic electrolysis

depending on the potential. The various transformation reactions and the corresponding

equilibrium potentials were clarified. UV-VIS-NIR reflectance measurements confirmed

that the bandgap of CaSi2 was 3.1 eV with a direct absorption edge.

In chapter 7, the electro-codeposition of Ca and Si on a Si substrate was

investigated in a molten CaCl2-KCl-K2SiF6 system at 923 K. The formation of a Ca-Si

alloy was confirmed by XRD analysis and cross-sectional SEM image of a sample

106

prepared at -2.30 V for 1.0 h. The thickness of the film was approximately 75 μm. Much

thicker alloy films were obtained by prolonged potentiostatic electrolysis.

107

List of Publications

The main parts of this thesis are constructed from the following papers.

Chapter 3

1. Y. Sakanaka and T. Goto, Electrochim. Acta. 164, 139 (2015).

“Electrodeposition of Si film on Ag substrate in molten LiF-NaF-KF directly dissolving

SiO2”

Chapter 4

2. Y. Sakanaka, A. Murata, T. Goto and K. Hachiya, to be submitted to J Solid State

Electrochem.

“Electrodeposition of porous Si film from SiO2 in molten BaCl2-CaCl2-NaCl”

Chapter 6

4. Y. Sakanaka, T. Goto and K. Hachiya, J. Electrochem. Soc. 162, D186 (2015).

“Electrochemical Formation of Ca-Si in Molten CaCl2-KCl”

Chapter 7

5. Y. Sakanaka, S. Takashima, T. Goto and K. Hachiya, prepare to submit to Thin Sol.

Films.

“Co-deposition of Ca-Si in molten CaCl2-KCl-K2SiF6”

108

Acknowledgement

The author strongly wishes to express his special gratitude to Prof. Masatsugu

Morimitsu for giving valuable suggestions and advice on this study. The author would

like to my sincere thanks to Prof. Takuya Goto for his outstanding supervision including

valuable discussions and prominent suggestions.

The author would like to extend my sincere appreciation to Prof. Toshiyuki

Nohira (Institute of Advanced Energy, Kyoto University) for his invaluable advices. The

author would like to express my sincere gratitude to Prof. Takehiko Ishikawa (Japan

Aerospace Exploration Agency) for his fruitful discussions and precious advices.

Special thanks are given to Ass. Prof. Kan Hachiya (Graduate School of Energy

Science, Kyoto University) for his kind and helpful instructions on the experimental

techniques and discussions on the results. I would like to thank Prof. Ken Hirota, for his

help on the experiments.

I would like to thank to all the members of Prof. Goto’s laboratory for their help,

lively discussions and warm encouragements throughout the study. Among them, I

express my thanks to Mr. Akira Murata, Mr. Akihiro Tabushi and Mr. Shu Takashima for

conducting their research together with me.

109

September, 2015

Yoshihide Sakanaka

formation of silicon and silicides by new electrochemical ... · ca-si alloy films using aqueous...

Documents