New Developments in Solid Electrolytes for Thin-Film Lithium Batteries

Research on lithium-ion secondary batteries began in the 1980s because of the growing demand for power sources for portable electronic devices. After the early 1990s, the demands for higher capacities and even smaller sizes energy systems significantly increased. Further, the explosive growth in the use of limited fossil fuels and their associated environmental issues and economical aspects are major concerns. Hence, the enormous growth in the demand for low-cost, environmentally friendly energy sources over the past decade has generated a significant need for high energy density portable energy sources.


Introduction
Research on lithium-ion secondary batteries began in the 1980s because of the growing demand for power sources for portable electronic devices.After the early 1990s, the demands for higher capacities and even smaller sizes energy systems significantly increased.Further, the explosive growth in the use of limited fossil fuels and their associated environmental issues and economical aspects are major concerns.Hence, the enormous growth in the demand for low-cost, environmentally friendly energy sources over the past decade has generated a significant need for high energy density portable energy sources.
The enormous growth in portable consumer electronic devices such as mobile phones, laptop computers, digital cameras, and personal digital assistants over the past decade has generated a large interest in compact, high-energy density and lightweight batteries.As power requirements become more demanding, batteries are also expected to provide higher energy densities.In recent decades, LiCoO 2 and LiMn 2 O 4 cathode materials and graphite anodes have been developed and are common in most lithium batteries (Kenji et al.;Peled et al., 1996;Peng et al., 1998;Wang et al., 2002;Xu et al., 2002).However, graphite-based materials are less attractive in terms of capacity when compared to lithium metal, 372 vs. 3800 mAh/g, respectively, in spite of graphite's higher cyclability and safer operation than lithium metal anodes (Tarascon and Armand, 2001).Even still, lithium-based batteries have become enormously important batteries due to their relatively high capacity and low weight.A comparison of many different anode (bottom) and cathode (top) combinations are shown in Figure (1) (Tarascon and Armand, 2001).Further, lithium is very lightweight and has a high electrochemical equivalency and these properties make lithium an attractive battery anode.Therefore, rechargeable lithium batteries are attractive for numerous reasons: high voltages, high energy densities, wide operating temperature ranges, good power density, flat discharge characteristics, and excellent shelf life.However, as shown in Figure (1) the 10-fold increase in capacity of Li metal over graphite has prompted continued effort to develop rechargeable lithium batteries based upon lithium metal anodes for use in a wide variety of applications.
Although the implementation of lithium metal as the anode material in lithium batteries is attractive, electrolytes with high ionic conductivity that are stable in contact with metallic lithium are still lacking.Around ten years ago, lithium batteries with lithium metal anodes using liquid electrolytes, which show the highest ionic conductivities, failed because of serious safety issues (Crowther and West, 2008).Lithium metal anodes tend to form dendrites during charging and discharging processes due to plating-out reactions between lithium metal and liquid electrolytes (Crowther and West, 2008).For these reasons, lithium-based solid-state electrolytes instead of liquid electrolytes are now of great interest and many researchers have been examining them in solid-state batteries because solid electrolytes do not have the aforementioned safety issues and show a smaller temperature dependence to the ionic conductivity compared to some liquid electrolytes.In addition, with the recent surge in interest of various kinds of portable electronic devices and electric and hybrid-electric vehicles, the importance of portable energy sources like secondary batteries has increased.It is widely recognized that allsolid-state energy devices show promise towards improving the safety and reliability of lithium batteries (Hayashi et al., 2009).Fig. 1.Voltage versus capacity for positive-and negative-electrode materials presently used or under serious consideration for the next generation of rechargeable Li-based cells adopted from (Tarascon and Armand, 2001).
There is an additional interest in specialized lithium batteries for use in the semiconductor industry and for printed circuit-board applications.These types of batteries are of interest for applications such as non-volatile computer memory chips, smart cards, integrated circuits, and some medical applications (Albano et al., 2008;Souquet and Duclot, 2002).In a d d i t i o n , a s t h e i n c r e a s i n g t e n d e n c y o f m a n y a d v a n c e d t e c h n o l o g i e s i s t o w a r d s miniaturization, the future development of batteries is aiming at smaller dimensions with higher power densities.The development of new technologies and miniaturization in the microelectronics industry has reduced the power and current requirements of small power electronic devices such as smart cards and other CMOS circuit applications (Albano et al., 2008;Souquet and Duclot, 2002).Therefore, developing improved solid-state thinfilm batteries will allow better compatibility with microelectronic processing and components.
Therefore, solid-state lithium secondary batteries have attracted much attention because the replacement of conventional liquid electrolytes with an inorganic solid electrolyte may improve the safety and reliability of lithium batteries utilizing high capacity lithium metal anodes (Jones and Akridge, 1992).
Although solid-state batteries have many potential advantages over competitive batteries, solid electrolytes must have higher Li + ionic conductivity for them to succeed in commercial applications.Solid electrolytes are a key material of all-solid-state energy storage devices and have been extensively studied in the fields of materials science (Scholz and Meyer, 1994), polymer science (Croce et al., 2001;Fauteux et al., 1995;Song et al., 1999), and electrochemistry (Scholz and Meyer, 1994).Much research has been devoted to the preparation of solid electrolytes made of various materials including ceramics (Abe et al., 2005;Jak et al., 1999a;Jak et al., 1999b), glasses (Iriyama et al., 2005;Lee et al., 2002;Lee et al., 2007;Takada et al., 1995) and glass ceramics (Hayashi et al., 2010;Minami et al., 2011;Ohtomo et al., 2005).
Among these materials for electrolytes, amorphous or glassy materials often have superior ionic conductivities over corresponding crystalline materials because they can form over a wide range of compositions, have isotropic properties, do not have grain boundaries, and can form thin-films easily (Angell, 1983;Martin, 1991).Because of their more open disordered structure, amorphous materials typically have higher ionic conductivities than the corresponding crystalline material (Angell, 1983;Martin, 1991).In addition, single ion conduction can be realized because glassy materials belong to decoupled systems in which the mode of ion conduction relaxation is decoupled from the mode of structural relaxation (Kanert et al., 1994;Patel and Martin, 1992).For these reasons, amorphous or glassy materials are thus among the more promising candidates of solid electrolytes because of their properties of single ion conduction and high ionic conductivities.
Oxide-based electrolytes are currently widely studied because of their stability in air, easy preparation, and their long shelf life (Cho et al., 2007;Jamal et al., 1999).However, they show a critical disadvantage which is their low ionic conductivity.Even still, so-called LiPON films formed from sputtering Li 3 PO 4 in N 2 atmospheres are currently one of the primary solid-state thin-film electrolytes in use because of these above mentioned advantages (Bates et al., 1993;Bates et al., 2000a;Bates et al., 2000b;Dudney, 2000;Neudecker et al., 2000;West et al., 2004;Yu et al., 1997).However, this easily prepared material has a relatively low Li + ion conductivity of ~10 -6 S/cm at 25 o C as compared to sulfide-based materials whose Li + conductivities are in the range of 10 -3 S/cm at 25 o C (Hayashi et al., 2003;Komiya et al., 2001;Minami et al., 2006;Mizuno et al., 2005).
Because lithium containing thio-materials show higher ionic conductivities than corresponding oxide materials, much research has been conducted on the use of the thiomaterials as solid electrolytes.Recently, sulfide materials have been investigated such as SiS 2 (Aotani et al., 1994;Hayashi et al., 2002;Hirai et al., 1995;Kennedy, 1989), GeS 2 (Haizheng et al., 2004;Kawamoto and Nishida, 1976;Pradel et al., 1985), P 2 S 5 (Hayashi et al., 2005;Mercier et al., 1981a;Mizuno et al., 2005;Murayama et al., 2004), and B 2 S 3 (Hintenlang and Bray, 1985;Wada et al., 1983).Among these sulfide materials, GeS 2 is particularly attractive as a base material because it is less hygroscopic (Yamashita and Yamanaka, 2003), more oxidatively stable and enables a more electrochemically stable matrix for lithium-ion conduction to be prepared (Xia et al., 2009).While much research has been done on ion-conducting bulk sulfide glasses prepared by melt-quenching, only a few studies of thin-film ion conducting sulfides have been reported because of the difficulty in preparing them.For example, while Kim (Kim et al., 2005) et al. andItoh et al.(Itoh et al., 2006) reported on the Li 2 S + GeS 2 bulk glass system, detailed characterizations of thin-films in this system have not been reported so far.Several thin-film techniques such as pulsed laser deposition (PLD) (Jin et al., 2000;Tabata et al., 1994), radio frequency (RF) sputtering (Bates et al., 1993;Bates et al., 2000a;Bates et al., 2000b;Nakayama et al., 2003;Neudecker et al., 2000;Yu et al., 1997), e-beam evaporation (Bobeico et al., 2003;Wu et al., 2000), physical vapor deposition (PVD) (Kong et al., 2001;Narayan et al., 1992), and chemical vapor deposition (CVD) (Chhowalla et al., 2001) have been used to produce thin-films.Among these techniques, sputtering techniques have been shown to produce high quality thin-films.Furthermore, of the few reports that do exist on sulfide thin-films in the open literature, most show that the films tended to be strongly oxidized either during sputtering, c a u s e d p o s s i b l y b y l e a k a g e o f t h e R F c h a m b e r , o r b y e x p o s u r e t o a i r a f t e r sputtering (Yamashita et al., 1996a).In addition, these thin-films were found to be Li deficient compared to that of the targets from which they were made (Yamashita et al., 1996a).Therefore, although sulfide films may have good potential in thin-film batteries, sulfide thin-films produced so far appear to be less than optimized and for this reason have found limited applications.
In order to investigate sulfide-based thin-films more extensively, lithium thio-germanate thin-films were carefully sputtered under well-controlled conditions in this study.Since GeS 2 -based materials are typically more stable in air than other sulfide materials, GeS 2 -based thin-film electrolytes for Li-ion thin-film batteries were grown by RF magnetron sputtering in Ar atmospheres.The starting materials, GeS 2 and Li 2 S, and the target materials, Li 2 GeS 3 , Li 4 GeS 4 , and Li 6 GeS 5 , were characterized by X-ray diffraction to verify the phase purity of the targets used to produce thin-films.Further structural characterization of the starting materials, target materials, and their thin-films sputtered by RF sputtering in Ar atmospheres was conducted by Raman and IR spectroscopy to verify purity, contamination, and to examine the structures between targets and their thin-films.The surface morphology and the thickness of the thin-films were characterized by field emission scanning electron microscopy (FE-SEM).
The starting materials, target materials, and thin-films were carefully analyzed by x-ray photoelectron spectroscopy (XPS).To minimize contamination of the films produced in this work, every experimental step was performed carefully and in particular, the RF sputtering conditions were optimized to obtain consistency between target and thin-film compositions and to specifically produce films with near stoichiometric lithium concentrations.The starting materials, Li 2 S and GeS 2 , the target materials, and thin-films were characterized by XPS for compositional and chemical shift analysis.In order to determine if a maximum conductivity in the nLi 2 S + GeS 2 system exists, the Li 2 S content ranged from n = 1 to n = 4, 50 mol % to 80 mol %.Ionic conductivities of the thin-films were characterized by impedance spectroscopy.The ionic conductivities were measured over the temperature range from -25 o C to 100 o C in 25 o C increments and over the frequency range from 0.1 Hz to 1 MHz.Before we turn to a detailed description of this work, we first give an overview of the different electrodes that can be used with thin-film electrolytes and of the research progress to date on the chemistry and composition of thin-film electrolytes.

Anodes
As mentioned above in the introduction section, the development of advanced all-solid-state lithium-ion batteries with high energy densities is strongly desired because current lithium-ion batteries using liquid electrolytes potentially have safety issues (Machida et al., 2002;Machida et al., 2004).Because the battery performance strongly depends on the quality of the electrode materials, the electrode materials are very important in battery system.Although many different electrode materials have been developed for the conventional lithium-ion battery which used liquid electrolytes, many of the anode materials developed so far are not suitable for the solid-state lithium-ion batteries.Therefore, in this section, anode and cathode materials which are suitable for the solid-state lithium-ion battery are reviewed.

Graphite/carbon
Graphite/carbon materials have been commonly used as anode materials for the commercial lithium-ion battery using liquid electrolytes because graphite/carbon materials have many advantages including (1) a good cyclability, (2) a relatively large specific capacity of ~370 mAh/g, and (3) a low anode electrode potential of ~0.2 V compared to the Li/Li+ electrode (Buiel and Dahn, 1999;Wu et al., 2003).Although carbon materials have some advantages for conventional Li-ion batteries, not all carbon materials are suitable for all-solid-state lithium-ion batteries with inorganic solid electrolytes.The reason for this is that during charge and discharge processes, the electrochemical lithium insertion into the anode materials, carbon materials, are not completely reversible in solid-state lithium-ion batteries with an inorganic electrolyte.In order to improve the performance of batteries, metallic lithium is very attractive compared to the graphite/carbon materials because metallic lithium has around ten times higher capacity than that of graphite/carbon materials.

Lithium silicide
Lithium silicide (Li 4.4 Si) is a good candidate as an anode material for all-solid-state lithium-ion batteries because Li 4.4 Si has a large theoretical specific capacity of ~4000 mAh/g, has a high negative potential close to that of lithium metal, and Si is very abundant and is a non-toxic material (Armand and Tarascon, 2008;Lee et al., 2001).However, Li 4.4 Si has a severe volume expansion of over 300% for the Li 4.4 Si phase during charge and discharge processes.Thus, in its common form, the material shows poor cyclability compared to graphite and has barriers for commercial application (Kubota et al., 2008).Four different lithium silicides, Li 4.4 S, Li 3.25 Si, Li 2.33 Si, and Li 1.71 Si, as intermetallic phases have been reported in Li-Si system (Sharma and Seefurth, 1976).

Lithium metal
Lithium metal as an anode material has high energy density and it has been recognized as the best candidate for lithium-ion batteries (Tarascon and Armand, 2001).While dendrite formation during cycling is found with liquid electrolytes, lithium metal does not form dendrites in all solid-state lithium-ion batteries.In solid-state batteries, the major challenges are interface resistances and electrochemical stability at the contact area between the anode and electrolyte.Further, extensive effort is demanded before lithium metal is applied to commercial all solid state lithium ion batteries.

Cathode
While anode materials play an important role in supporting the lithium source, the cathode materials also plays an important role in supporting the reducible/oxidizable ion for secondary lithium-ion batteries.There are key requirements for good cathode materials to be used successfully in rechargeable lithium-ion batteries (Whittingham, 2004).The cathode materials should react with lithium metal in a reversible manner, with a high free-energy of formation and react very rapidly both on insertion and removal.In addition, these materials need to be a good electrical conductors, be electrochemically stable, have a low cost, and need to be environmentally safe (Whittingham, 2004).Until now, many cathode materials have been studied.In this section, some of the more representative cathode materials are reviewed.

Vanadium pentoxide (V 2 O 5 )
Vanadium pentoxide, V 2 O 5 , has been studied for three decades (Dickens et al., 1979;Whittingham, 1976).V 2 O 5 as a cathode material is an alternative because of its low cost, plentiful resources, and greater safety compared to commercial cathodes such as LiCoO 2 and LiNiO 2 (Wang and Cao, 2008).The main disadvantages of the V 2 O 5 material are its low capacity, low conductivity, and poor structural stability (Li et al., 2007).Recently, the V 2 O 5 -based material, polyaniline (PAN)-V 2 O 5 composites have been extensively studied to improve conductivity, cyclabilty, and coulombic efficiency of the electrode materials used in lithium batteries (Malta and Torresi, 2005;Pang et al., 2005).

Lithium cobalt oxide (LiCoO 2 )
Lithium cobalt oxide, LiCoO 2 , cathode material was discovered by John Goodenough in 1980 when he worked at Oxford University (Mizushima et al., 1980).Research on LiCoO 2 material has been widely done because of its high energy density and good cyclability (Wang et al., 1999) and relatively high theoretical capacity of 272 mAh/g.LiCoO 2 cathode material is attractive because of its high energy density and reversible lithium-ion intercalation (Chiang et al., 1998;Kumta et al., 1998).Furthermore, LiCoO 2 material has a layer structure which can be suitable for the accommodation of the large changes of the lithium contents.Therefore, it can be cycled more than 500 times with 80-90 % capacity retention (Patil et al., 2008).Thin-film LiCoO 2 cathode materials also show good power density when discharged between 3.0V and 4.2V (Kim et al., 2000) because of the layered LiCoO 2 structure.Amatucci et al. (Amatucci et al., 1996) reported that LiCoO 2 can be reversibly form the Li-ion and CoO 2 .In addition, the preparation of LiCoO 2 is very facile over other comparable materials.For these reasons, it became the most common cathode material in lithium batteries.However, cobalt is relatively expensive compared to other elements such as Ni, Mn, and V.In order to make it cheaper and improve the reversible capacity, Yonezawa et al. (Yonezawa et al., 1998) and Huang et al.(Huang et al., 1999) applied doping materials such as fluorine, magnesium, aluminum, nickel, copper or tin.If LiF is doped, the reversible capacity improved compared to pure LiCoO 2 (Yonezawa et al., 1998).If Al is incorporated partially to substitute for cobalt, the working and open voltages increased.Huang et al. reported that the reversible capacity of LiAl 0.15 Co 0.85 O 2 reached up to 160 mAh/g without volume change after 10 cycles (Huang et al., 1999).Especially, self-discharge effects of the thin-film batteries using LiCoO 2 cathodes are negligible (Dudney, 2005).Thus, thin-film batteries using LiCoO 2 cathode can hold full charge for three years (Dudney, 2005).

Lithium manganese oxide (LiMn 2 O 4 )
Lithium manganese oxide, LiMn 2 O 4 , is an attractive cathode material and has been widely studied because the material has advantages from ecological and economical perspectives as well as easy preparation (Kang and Goodenough, 2000;Lee et al., 2004;Liu and Shen, 2003;Ohzuku et al., 1991).However, LiMn 2 O 4 has a serious drawback.Before cycling, the structure of the LiMn 2 O 4 is cubic.Then, on cycling, the spinel structure is destroyed due to a cubic-tetragonal phase transition induced by Jahn-Teller distortion (David et al., 1987;Gummow et al., 1993;Gummow and Thackeray, 1994).For this reason, batteries with LiMn 2 O 4 cathodes show capacity loss and poor cyclability (Gummow et al., 1994;Myung et al., 2000).Pure LiMn 2 O 4 has been improved by doping.If chromium is doped into LiMn 2 O 4 , it can form Li 1+x Mn 0.5 Cr 0.5 O 2 and the doped Li 1+x Mn 0.5 Cr 0.5 O 2 reveals improved capacity and cyclability (Sigala et al., 1995).It can be assumed that Mn plays an important role to stabilize the structure of the chromium oxide.However, chromium materials are toxic and expensive.Therefore, in order to fabricate successfully stabilized layer structural framework, the doping of other elements into LiMn 2 O 4 has been studied.

Lithium nickel oxide (LiNiO 2 )
Because LiNiO 2 is cheaper than LiCoO 2 and the redox potential is higher than that of LiCoO 2 , The LiNiO 2 material has become an attractive as a cathode material for Li-ion batteries (Campbell et al., 1990).The structure of the LiNiO 2 is layered similar to LiCoO 2 (Zhecheva and Stoyanova, 1993).The layered LiNiO 2 structure has a wide homogeneity range, Li x Ni 2-x O 2 (0.6 x1) (Bronger et al., 1964).Upon cycling, the capacity of the materials fades because Ni 2+ ions migrate to Li + sites.The appearance of Ni 2+ in the Li + sites obstructs Li + diffusion and the lithium-ion transfer during cycling (Li et al., 1992).For this reason, the LiNiO 2 battery shows poor cycle performance compared to LiCoO 2 (Dahn et al., 1991).LiNiO 2 has some drawbacks such as being unstable in the overcharge state as well as easy decomposition at high temperature.Furthermore, lithium oxide contents in the LiNiO 2 decrease when heat treatment is performed due to the volatility of Li 2 O.The Li deficient defect structure results in gradual collapse of oxide structure during cycling and the specific charge decreases during cycling of the LiNiO 2 electrode (Hirano et al., 1995).In order to improve the performance of the LiNiO 2 structure, many researchers have studied this material using doping elements such as Co, Ti, Mn, Al, Mg, Fe, Zn, Ga, Sb, and S (Chang et al., 2000;Chowdari et al., 2001;Cui et al., 2011;Gao et al., 1998;Nishida et al., 1997;Park et al., 2005;Park and Sun, 2003;Pouillerie et al., 2000;Reimers et al., 1993).The reversible capacity of LiNi 0.75 Ti 0.125 O 2 reached up to 190 mAh/g (Gao et al., 1998).Therefore, from this point of view, the doped LiNiO 2 can be a good candidate as a cathode material for secondary Li-ion batteries; however, safety is a serious concern.

Lithium iron phosphate (LiFePO 4 )
Lithium iron phosphate, LiFePO 4 , was reported by John Goodenough's research group in 1996, as a cathode material for rechargeable lithium batteries (Padhi et al., 1997).Conventional cathode materials, LiCoO 2 and LiNiO 2 , have drawbacks such as the high cost, toxicity, safety issues, and electrochemically instability.LiFePO 4 is a promising candidate for secondary lithium-ion batteries because of its relatively high energy density, low cost, good safety, and high thermal stability compared to conventional cathode materials (Padhi et al., 1997).However, there is a key barrier in that LiFePO 4 has an intrinsically low electrical conductivity of 10 -9 to 10 -10 S/cm (Andersson et al., 2000;Barker et al., 2003;Padhi et al., 1997).Therefore, early studies on LiFePO 4 showed that it was not the best cathode material over other conventional cathode materials.These problems were resolved later by reducing the particle size, doping using cations of materials such as Al, Nb, and Zr, and coating the LiFePO 4 particles using a conductive carbon material (Huang et al., 2001;Prosini et al., 2001;Shi et al., 2003;Yamada et al., 2001).By using these methods, greatly improved electrochemical response and full capacity of LiFePO 4 was obtained with prolonged cycle life.Recently, LiFePO 4 can be used up to 90 % of its theoretical capacity, 165 mAh/g, and at high rate capabilities (Yamada et al., 2001).Thus, optimized LiFePO 4 is a good candidate as a cathode material for the solid-state thin-film batteries.

Electrolytes
Among the three components, anode, cathode, and electrolyte, in a battery system, battery performance strongly depends on the performance of the electrolyte.The basic requirements of an appropriate electrolyte for lithium-ion batteries are high ionic conductivity, electrochemical and thermal stability, and good performance at low and high temperatures.Because liquid electrolytes have a higher ionic conductivity compared to polymer and solid electrolytes, they have been widely used.However, liquid electrolytes have strong ionic conductivity temperature dependence and can have safety issues related to the flammability of the organic liquid.Recently, solid-state lithium secondary batteries have attracted much attention because the replacement of conventional liquid electrolyte with an inorganic solid electrolyte may improve the safety and reliability of lithium batteries utilizing high capacity lithium metal anodes (Jones and Akridge, 1992).There are two types of solid state electrolytes; one is thin-film electrolytes grown by RF sputtering (Bates et al., 1993;Bates et al., 2000a;Bates et al., 2000b;Dudney et al., 1999;Neudecker et al., 2000;Seo and Martin, 2011a, b, c;Yu et al., 1997) or PLD (Jin et al., 2000;Tabata et al., 1994) etc. and the other is bulk electrolytes fabricated by typically using melting processes.All-solid-state thin-film batteries using inorganic amorphous electrolytes such as LiPON have been reported and LiPON shows excellent cyclability, over 50,000 cycles, at room temperature (Bates et al., 1993;Bates et al., 2000a;Bates et al., 2000b;Dudney et al., 1999;Neudecker et al., 2000;Yu et al., 1997).However, bulk type batteries using bulk electrolytes have an advantage of improving cell capacity by the addition of large amounts of active materials to the cell.For both the thin-film and bulk batteries, Li 2 S based electrolytes are promising because of their high ionic conductivities compared to oxide electrolytes.In the next section, we report recent data of the thin-film electrolytes for the solid-state batteries.

Polymer electrolytes
Polymer electrolytes for use in lithium batteries were rapidly developed in the 1970s (Fenton et al., 1973).It was found that these materials could offer a safer battery than liquid electrolytes which are corrosive, flammable, or toxic.In recent decades, polymer electrolytes have been widely studied including polyethyleneoxide (PEO) (Appetecchi et al., 2003), polyacrylonitrile (PAN) (Yu et al., 2001), polymethylmethacrylate (PMMA) (Rajendran and Uma, 2000), and polyvinylidenefluoride (PVDF) (Saunier et al., 2004).However, the ionic conductivity of these polymer electrolytes is still too low at room temperature for commercial batteries.In addition, the primary concerns with these electrolytes involve their reactivity with a lithium metal as an anode.Their reactivity with lithium poses safety concerns because lithium dendrites can grow towards the cathode and ultimately shortcircuit the cell.For this reason, gel-type polymer electrolytes were developed and the electrolytes show improved ionic conductivity compared to conventional polymer electrolytes, but they still have lower ionic conductivities than those of liquid electrolytes.The gel-type polymer electrolytes have ionic conductivity of ~10 -4 S / c m a t r o o m temperature.In the gel-type polymer electrolytes, the liquid element can be trapped in polymer matrix so the leakage problems associated with liquid electrolytes can be resolved.The low conductivity of the polymer/gel polymer electrolytes can be overcome by introducing inorganic ceramic particles to form a composite material that is more "solid".These materials have conductivities two or three orders of magnitude lower than aqueous electrolytes.However, thin polymer films on the order of 100 μm thick can compensate for their diminished conductivities (Birke et al., 1999).There is also the opportunity of increasing the operating temperature of the cell to around 90 °C.Therefore, optimized thin polymer electrolytes can be promising electrolytes for the thin-film solid-state batteries.

Solid-state electrolytes
While commercial cells will continue to be fabricated using organic polymeric electrolytes due to their ease of fabrication and low cost, solid-state electrolytes will also attract attention for their possible use in special applications.Solid-state electrolytes are attractive because they provide a hard surface that is capable of suppressing side reactions and inhibiting dendritic growth of lithium that is capable of short-circuiting a cell (Schalkwijk and Scrosati, 2002).However, one disadvantage of these electrolytes is their potential to form cracks or voids if there is poor adhesion to the electrode materials.In order to successfully fabricate all-solid-state lithium batteries with good performance, the design of the electrodes and electrolytes are important.
A number of candidate materials have been investigated for use as solid electrolytes in batteries.The most attractive candidates to date are glassy materials.These electrolytes have many advantages over their crystalline counterparts such as physical isotropy, absence of grain boundaries, good compositional flexibility, and good workability.The anisotropy and grain boundaries present in crystalline materials lead to resistive loss, decreasing cell efficiency, as well as chemical attack, raising safety concerns.A number of different systems have been explored and are discussed specifically below.

Oxide glasses
Oxide glass electrolytes for solid-state batteries have been widely studied because they have the primary advantage of being relatively stable in air allowing for ease of fabrication.However, oxide glasses have received less attention for their use as electrolyte materials because they exhibit very low ionic conductivities and high activation energies.The best of the oxide materials appears to be those glasses with mixed formers such as SiO 2 and B 2 O 3 (Lee et al., 2002;Nogami and Moriya, 1982;Zhang et al., 2004).These glasses have a conductivity at room temperature on the order of ~10 -7 S/cm.These materials might prove promising if produced into thin-films.However, chemistries with a higher ionic conductivity are more desirable.Sulfide materials, discussed in more detail below, are of interest for this reason.In terms of conductivity, it is clear that oxide glasses have significantly lower conductivities than those of sulfide materials (Boukamp and Huggins, 1978;Elmoudane et al., 2000;Ito et al., 1983;Mercier et al., 1981b;Murayama et al., 2004).

Oxinitride glasses
The most commercially viable material in this category is the lithium phosphorus oxynitride (LiPON) glass.This material was first discovered and reported in the 1980s by Marchand(Marchand et al., 1988) and Larson (Larson and Day, 1986).These materials were not thin-films but bulk glass materials.In addition, their properties were not fully characterized until Oak Ridge National Laboratory (ORNL) reported LiPON thinfilms (Bates et al., 1993;Bates et al., 2000a;Bates et al., 2000b;Dudney, 2000;Dudney, 2005;Dudney et al., 1999;Neudecker et al., 2000;Yu et al., 1997).The LiPON thin-films were deposited using a high purity lithium phosphate, Li 3 PO 4 , target by RF magnetron sputtering technique in nitrogen atmosphere.It was found that the resulting thin-film with a typical composition of Li 2.9 PO 3.3 N 0.36 contained 6 at % nitrogen.This additional nitrogen was found to enhance the ionic conductivity at room temperature from ~10 -8 S/cm in the starting Li 3 PO 4 target to a value of ~10 -6 S/cm.Furthermore, these films were found to be highly stable in contact with metallic lithium.It is believed that a thin passivating layer of Li 3 N is formed between the lithium and electrolyte which prevents lithium dendrite growth, but allows ion conduction.The nitrogen was found to substitute for oxygen and form 2 and 3-coordinated nitrogen groups, effectively crosslinking the structure.The schematic of the structural units of LiPON are shown Figure ( 2).This crosslinking is believed to decrease the electrostatic energy of the overall network, allowing for faster ion conduction.Thin-film batteries comprised of Li-LiCoO 2 cells and Li-LiMn 2 O 4 cells have been fabricated using the LiPON electrolyte at ORNL (Bates et al., 1993;Bates et al., 2000a;Bates et al., 2000b;Dudney, 2000;Dudney, 2005;Dudney et al., 1999;Neudecker et al., 2000;Park et al., 2007;Yu et al., 1997).Nam et al. reported LiPON using V 2 O 5 cathode materials (Jeon et al., 2001).These types of batteries are being commercialized and target for applications in implantable medical devices, CMOS-based integrated circuits, and RF identification (RFID) tags for inventory control and anti-theft protection.

Sulfide glasses
Sulfide glasses were first reported on in the 1980s (Kennedy, 1989;Mercier et al., 1981a;Zhang and Kennedy, 1990).These glasses were based on SiS 2 , P 2 S 5 , and B 2 S 3 and are doped with an alkali sulfide such as lithium sulfide.It was found that these materials have exceptional conductivities, ~ 10 -3 S/cm, at room temperature.This is attributed primarily to the larger ionic radius of sulfur and its high atomic polarizability (Kim et al., 2006).This is believed to create weaker covalent bonds between the sulfur and the lithium ions.As a result, the potential energy barrier that must be overcome is decreased, and lithium-ion conduction is facilitated.Unfortunately, these glasses have not been widely used because they are highly reactive in air/moisture and corrosive with silica containers.A high quality, low O 2 and H 2 O, glovebox is absolutely necessary to fabricate such glasses without contamination.
The thin-films related to the Li 2 S-GeS 2 -Ga 2 S 3 glass system have recently been prepared using RF sputtering (Yamashita and Yamanaka, 2003;Yamashita et al., 1996b).Successfully deposited films were produced using this method, however, the authors reported that the ionic conductivities of the thin-films were diminished compared to that of the target material.This was a result of the films being deficient in lithium and enriched in germanium from the XPS composition analysis of the films.In addition, the thin-films were contaminated by O 2 during sputtering due to the leakage of the RF chamber.Although the ionic conductivities of the sulfide materials are higher than oxide materials, if the sulfide thin-films exhibit lithium deficiency and contamination, there are few benefits of sulfide materials.

Oxy-sulfide glasses
Efforts to combine the advantages of oxide and sulfide glasses have resulted in the research of a class of oxy-sulfide materials (Hayashi et al., 1996;Kondo et al., 1992;Takada et al., 1996).It was found that adding a small amount, approximately 5 mole %, of different lithium metal oxides to a base sulfide glass, improved the conductivity over 10 -3 S/cm (Minami et al., 2008;Ohtomo et al., 2005).Furthermore, the stability of the structure was observed to improve from thermal analysis results.Structural analysis of these materials has demonstrated that the oxygen typically occupies a bridging anion site, leaving sulfur at the non-bridging sites for lithium mobility.Some solid-state batteries have been fabricated using these oxy-sulfide compositions, and initial results appear to indicate good electrochemical stability (Hayashi et al., 2010;Minami et al., 2011;Ohtomo et al., 2005).

Thio-nitride glasses
Lithium nitride (Li 3 N) materials are somewhat attractive as a solid electrolyte because of their high ion conductivity, 2-4 × 10 -4 S/cm at 25 °C (Lapp et al., 1983).The single crystal Li 3 N has an impressive high ionic conductivity of 1.2  10 -3 S / c m a t r o o m temperature (Rabenau, 1982).However, it is impossible to use Li 3 N itself as an electrolyte in secondary batteries because Li 3 N decomposes at low voltage (Yonco et al., 1975).For this reason, thio-nitride glasses have been studied with high ionic conduction (Iio et al., 2002;Sakamoto et al., 1999).The motivation behind these materials comes from the fact that doping nitrogen into oxide systems improved the ionic conductivity (Unuma and Sakka, 1987;Wang et al., 1995c).Furthermore, doping of nitrogen into oxide glasses has been found to improve the hardness and chemical durability (Sakka, 1986).

All-solid-state thin-film batteries 4.1 History of thin-film batteries
All-solid-state thin-film batteries were reported first by Hitachi Co. Ltd in Japan in 1982.The TiS 2 cathode material was prepared by chemical vapor deposition (CVD), a Li 12 Si 3 P 2 O 20 electrolyte was grown by radio frequency (RF) sputtering, and a lithium metal anode material was deposited by a vacuum evaporation (Kanehori et al., 1986;Miyauchi et al., 1983).NTT Co. also reported thin-film batteries using a Li 12 Si 3 P 2 O 20 electrolyte with LiCoO 2 or LiMnO 2 cathode materials grown by RF sputtering (Ohtsuka et al., 1990;Ohtsuka and Yamaki, 1989;Yamaki et al., 1996).The performance of the thin-film batteries was not as good as current thin-film batteries.
In 1980s, Union Carbide Corporation and Eveready Battery Co., Ltd. in USA developed thin-film batteries using sulfide glass electrolytes, Li 4 P 2 S 7 or Li 3 PO 4 -P 2 S 5 , and Li metal anode or LiI anode (Akridge andVourlis, 1986, 1988).They improved the battery performance in 1990s to reach over 1000 cycle performance between 1.5V and 2.8V and 10 to 135 A/cm 2 (Jones and Akridge, 1996).Bellcore Co., Ltd. also developed thin-film batteries using a LiMnO 2 cathode, lithium metal as an anode, and lithium borophosphate (LiBP) or lithium phosphorus oxynitride (LiPON) glass as an electrolyte (Shokoohi et al., 1991).The cell showed over 150 cycles with 3.5~4.3V and 70 A/cm 2 .
Recently, Bates and Dudney et al. at Oak Ridge National Laboratory (ORNL) reported significant progress on LiPON-based thin-film batteries which were produced by an RF sputtering technique (Bates et al., 1993;Bates et al., 2000a;Bates et al., 2000b;Dudney et al., 1999;Wang et al., 1995a;Wang et al., 1995b;Wang et al., 1995c;Yu et al., 1997).In order to fabricate LiPON thin-film batteries, the metallic anode was produced by vacuum evaporation and anode and cathodes were produced by RF sputtering.The LiPON thin-film batteries are very stable in air compared to lithium oxide or sulfide based batteries in spite of LiPON's low ionic conductivity of ~10 -6 (S/cm).The LiPON thin-films reported by ORNL showed very good performance between 2-5 V and over 10,000 cycles.Furthermore, ORNL reported also a Li-free thin-film battery with an in-situ plated Li anode on copper electrode (Neudecker et al., 2000).
LiPON is now known as a standard electrolyte for the thin-film batteries and it has been widely studied by a number of research groups.Park et al. in Korea reported "mesa-type" all-solid-state LiPON thin-film battery using a LiMn 2 O 4 cathode (Park et al., 1999).Baba et al. in Japan reported also LiPON thin-film batteries using a Li x V 2 O 5 anode material and V 2 O 5 or LiMn 2 O 4 cathode materials produced by RF sputtering (Baba et al., 2001;Baba et al., 1999;Komaba et al., 2000).et al., 1988).

Thin-film techniques
There are many vapor deposition techniques that can be employed in order to produce thinfilm materials.These include simple heating of a source material, laser-induced vaporization, or bombarding the material with energetic ions.All of these techniques are performed under vacuum and rely on the kinetic theory of gases in order to understand their behavior.

Pulsed laser deposition (PLD)
Pulsed laser deposition (PLD) involves using a laser beam to vaporize the surface of a target material (Chrisey and Hubler, 1994).One of the most common lasers used is the KrF excimer laser, operating at 248 nm with the following parameters: a pulse on the order of 25 ns, a power density of 2.4  10 8 W/cm 2 , and a repetition rate of 50 Hz.In general, the PLD process can be divided into four stages (Chrisey and Hubler, 1994).First, the laser beam is focused onto the target material.The elements in the target are rapidly heated to their evaporation temperature where there are sufficiently high flux densities over a short pulse duration.This ablation process involves many complex physical phenomena such as collisional, thermal and electronic excitation, exfoliation and hydrodynamics.Second, the ablated target elements move towards the substrate according to the laws of gas-dynamics.
In the third stage, the high energy atoms bombard the substrate surface where a collision region is formed between the incident flow and the sputtered atoms.A film begins to grow after a thermalized region develops and when the condensation rate is higher than the rate of sputtered atoms.Finally, nucleation and growth of a thin-film occurs on the substrate.This step depends on many factors such as the density, energy, ionization degree, and the temperature of the substrate.PLD has some advantage over other techniques in that the stoichiometry of the target can be retained in the deposition film and many different materials can be deposited, and can be easily handled compared to other techniques such as CVD and ion implantation techniques (Bao et al., 2005;Kaczmarek, 1996).On the other hand, it has some disadvantages such as the deposition of droplets (Yoshitake et al., 2001), the splashing or the particulates deposition on the thin-film, and lower energy density and lower deposition rate compared to other techniques (Willmott and Huber, 2000).

Radio frequency (RF) sputtering
Sputtering is a technique whereby energetic ions from a plasma are used to bombard a target (which is the cathode of the discharge), and ejecting atoms into the plasma.These atoms then impinge upon the substrate (the anode) and form a coating.Additionally, a magnet can be added to these two setups in order to enhance the deposition rates.RF magnetron sputtering is a reliable technique used to deposit many different types of films, including electrically insulating samples.A high-voltage RF source at a frequency of typically 13.56 MHz is used to ionize a sputtering gas which produces the plasma (Yamashita et al., 1999).The ionized gas then bombards the target where multiple collisions take place, releasing atoms of the target material into the plasma.These atoms condense upon the substrate which is placed in front of the target (Nalwa, 2002).A permanent magnet is added to the sputtering gun in order to enhance the deposition rate.This is done by the trapping of electrons from a Hall effect near the target surface (Nalwa, 2002).This magnet creates lines of magnetic flux that are perpendicular to the electric field or parallel to the target surface.This static magnetic field retains secondary electrons in that region which drift in a cycloidal path on the target and increase the number of collisions that occur.
While many different thin-film deposition techniques could be used in this research, RF magnetron sputtering (RFMS) has been chosen as the technique of choice.The most important reasons for selecting RFMS as the technique of choice are given here (Dudney et al., 1999;Souquet and Duclot, 2002).First, there is no need in the project to produce thick films.To produce a protective barrier for lithium metal anodes, a layer is needed and only needs to be thick enough so that it does not have large numbers of pin holes that will lead to failure of the anode.A layer 50 to 5000 Å is thought to be thick enough.Such layers can easily be produced by RFMS.Secondly, in the thin-film lithium battery research, there is no need for thick films and films 500 to 10,000 Å are thick enough, which are again attainable with sputtering techniques.In addition, sputtering can be done within the confines of a sealed glovebox, can be used with multiple targets and film chemistries, can be used to produce very uniform films of high compositional integrity, and produces films with excellent adherence to the substrate.Finally, it is possible to deposit insulator films through RF reaction sputtering at rates higher than those of DC methods (Davidse, 1967).

Chemical Vapor Deposition (CVD)
Chemical Vapor Deposition (CVD) process is related to transform gaseous molecules, precursor, by chemical reactions in the thin-film or power on the substrate (Mount, 2003).
CVD processing is usually used to apply various fields such as integrated circuits, optoelectronic devices and sensors, micro-machines, and fine metal and ceramic powders.CVD has many advantages compared to physical vapor deposition (PVD) techniques such as sputtering and molecular beam evaporation.While PVD processes may not give complete coverage due to a shadowing effect, CVD can be allowed to coat thin-films of three dimensional structures with large aspect ratios.The deposition rates of the CVD are several times higher than that of PVD.In addition, ultra high vacuum is not needed and high purity film can be produced by CVD process.However, there are some disadvantages of the CVD process.High temperatures of the deposition temperature, over 600 o C, are not suitable for already grown thin-films on substrates.CVD precursors are sometimes dangerous and toxic and many precursors for CVD, for example metal organic chemical vapor deposition (MOCVD) precursors, are very expensive.

4.3
Recent results for the lithium thio-germanate thin-film electrolytes

X-ray diffraction of the starting materials and targets
In this study, GeS 2 and Li 2 S as starting materials were used to synthesize the target material.
To verify the phase purity, XRD pattern of GeS 2 glass powder, Li 2 S crystalline powder and three target materials are shown in Figure (3).While GeS 2 glass powder is verified to be amorphous, Li 2 S powder shows several sharp peaks.The XRD pattern of the Li 2 S powder closely matches the JCPDS data (Cunningham et al., 1972).From the JCPDS data, it is verified that the system and space group of Li 2 S powder are face-centered cubic and Fm-3m, respectively (Cunningham et al., 1972).
The Li 2 GeS 3 target shows an XRD amorphous pattern without dominant peaks because the melt-quenching technique combined with its 50% GeS 2 glass former composition are sufficient to make this phase amorphous on cooling during preparation.The XRD patterns of the Li 4 GeS 4 and Li 6 GeS 5 targets, on the other hand, are polycrystalline and show sharp peaks because the Li 4 GeS 4 and Li 6 GeS 5 target contain only 33 % and 25% of the GeS 2 glass former, respectively, which are not sufficient to vitrify these melts on quenching.To verify the XRD pattern of the Li 4 GeS 4 target material which was quenched on a brass plate in the glovebox, and the reference data (Komiya et al., 2001) of Li 4 GeS 4 is also shown in Figure (3).The XRD pattern of our experimental Li 4 GeS 4 target material shows slightly broader peaks than those of the reference data (Komiya et al., 2001).A possible reason is that the Li 4 GeS 4 target was quenched more quickly on a brass plate.This rapid quenching presumably produces a more defective crystal structure than typical slow cooled or solid-state reaction prepared samples.However, the XRD pattern of the Li 4 GeS 4 target material still appears to closely match the reported reference pattern.Murayama et al. (Murayama et al., 2002) reported that the structure of Li 4 GeS 4 is related to that of -Li 3 PO 4 and is comprised of hexagonal close-packed sulfide ions with germanium ions distributed over the tetrahedral sites.In this structure, the Li + ions are located in both octahedral and tetrahedral sites.Murayama et al.(Murayama et al., 2002) suggested that the distribution of Li + ions in the LiS 4 tetrahedra, the interstitial tetrahedral sites, and the LiS 6 octahedra sites forms conduction pathways in the crystal.For this reason, the Li 4 GeS 4 material shows higher ionic conductivity than oxide materials.
While XRD data of Li 4 GeS 4 do not show peaks related to those of Li 2 S, XRD data of Li 6 GeS 5 show peaks related to those of Li 2 S.This suggests that the XRD pattern for Li 6 GeS 5 agrees well with the expectation that it is composed of equi-molar mixture of Li 4 GeS 4 and Li 2 S. The Li 4 GeS 4 and Li 6 GeS 5 targets are crystalline as shown in Figure (3).The fact can also be seen from the Raman spectra in Figure ( 4).The Li 2 GeS 3 , Li 4 GeS 4 , and Li 6 GeS 5 thin-films were not characterized by XRD because our standard XRD system is not sensitive enough to examine such thin-films as are reported here.

Raman spectroscopy
Starting materials, GeS 2 and Li 2 S, targets, and thin films were characterized by Raman spectroscopy in order to analyze their purity and to determine their chemical structure and are shown in Figure ( 4).In the Raman spectrum of GeS 2 , a strong main peak appears at ~340 cm -1 that agrees well with that of literature (Cernosek et al., 1997) and is assigned to the Fig. 4. Raman spectra of Li 2 S, GeS 2 , targets, and thin films.
www.intechopen.comsymmetric stretching of bridging sulfur, S (BS), (Ge-S-Ge) in the GeS 4/2 tetrahedra.The Raman spectrum of Li 2 S shows a single strong peak at ~375 cm -1 which is assigned to Li + -S  stretching modes.The Raman spectrum of the Li 2 S is sharper than that of GeS 2 glass because Li 2 S is crystalline while the GeS 2 is glassy.
In the spectrum of the Li 2 GeS 3 target, there are three dominant peaks at 340, 375 and 415 cm -1 .The peak at 340 cm -1 is found in GeS 2 and is assigned to bridging sulfur (Ge-S-Ge) bonding.The peak at 375 cm -1 is found in the Raman spectrum of Li 2 S and for this reason is assigned to Li + S -ionic bonding.The peak at 415 cm -1 is assigned to non-bridging sulfur (NBS) Ge-S -ionic bonding.While there are three peaks in the Raman spectrum of Li 2 GeS 3 target, the Raman spectra of the Li 4 GeS 4 and Li 6 GeS 5 targets show only one dominant peak at 375 cm -1 .The strong main Raman peak in the both Li 4 GeS 4 and Li 6 GeS 5 target materials appears at 375 cm -1 which is at the same peak position of Li 2 S.This indicates that the 375 cm -1 peak in both of the target materials was related to that of the Li 2 S component.The narrowing of the Raman peaks in spectra of Li 6 GeS 5 and Li 4 GeS 4 compounds compared to that of Li 2 GeS 3 arises from the polycrystalline structure of the former compound and the glassy structure of the latter.
The Raman spectrum of the Li 2 GeS 3 thin-film shows three dominant peaks at 340, 375, and 415 cm -1 .The peak at 340 cm -1 coincindes with GeS 2 main peak position and is assigned to the BS (Ge-S-Ge) mode.The 375 cm -1 peak is assigned to Li + -S -modes and the 415 cm -1 peak is assigned to NBS (Ge-S -) modes.
Among the three peaks, the peak at 340 cm -1 has the highest intensity.This is due to the high fractions (50%) of GeS 2 glass former in Li 2 GeS 3 .The Raman spectrum of the Li 4 GeS 4 thinfilm also shows three peaks at 340, 375 and 415 cm -1 , like the spectrum of the Li 2 GeS 3 thinfilm, and another broader peak of lower intensity at 460 cm -1 .The intensities of the peaks at 375 and 415 cm -1 are higher than those in the spectrum of the Li 2 GeS 3 thin-film.This is consistent with the increased Li 2 S content in the Li 4 GeS 4 compared to Li 2 GeS 3 which would increase the concentration of both Li + S -and Ge-S -NBS modes.The Raman spectrum of the Li 6 GeS 5 thin-film which has an even higher Li 2 S content compared to the other thin-films only has one dominant peak at 375 cm -1 which is assigned to the Li + S -vibrational mode.This indicates that the Li 6 GeS 5 thin-film contains the highest Li 2 S content compared to the other two thin-films.There are three low intensity peaks at 340, 415 and 460 cm -1 in the spectrum of Li 6 GeS 5 .As described above, the peak at 340 cm -1 is assigned to the bridging sulfur (Ge-S-Ge bonding) and the peaks at 415 and 460 cm -1 are assigned to modes of the NBS (Ge-S -).The peak at 460 cm -1 is assigned to the 1 NBS bonding and the peak is not present significantly in thin-films.The peak at 415 cm -1 is assigned to 2 NBS and the peak is present in thin-films.On the other hand, the peak at 340 cm -1 is assigned to 0 NBS and the peak is present in thin-films.The Raman spectra of all other thin-films do not show sharp peaks, but rather show broad peaks compared to those of crystalline targets (Li 4 GeS 4 and Li 6 GeS 5 ) and are consistent with the films being amorphous.As the Li 2 S content increases in the targets (Li 2 , Li 4 , and Li 6 ), the Li 2 S content in the thin-film increases.It can be concluded that although the previous reported literature showed Li 2 S deficiency in GeS 2 -based thin-films after sputtering compared to that of target, (Yamashita et al., 1996a) the amount of Li 2 S in the thin-films in this study increases with the increase of Li 2 S in the target and are consistent with the Li 2 S content in the targets.

Infrared (IR) spectroscopy
To further characterize the starting materials, Li 2 S crystalline powder and GeS 2 glass powder, targets, and their thin-films were characterized by infrared spectroscopy.Attention is focused on the far-IR region (900 to 100 cm -1 ) in order to evaluate the nature of the chemical bonding in the materials, as well as the mid-IR region (4000 to 400 cm -1 ) in order to determine how these materials might be contaminated by oxygen and/or moisture before and/or after processing.However, due to the lack of any significant O or OH contamination in the films and the very thin dimension observed, the mid-IR spectra are not shown here.However, in the far-IR region, strong absorptions were observed and arise from the framework structure species Li, Ge, and S.  5).The IR peak in the far-IR spectrum of Li 2 S at ~345 cm -1 is assigned to the ionic bonding of Li + S  and the strong peak at ~370 cm -1 in the spectrum of glassy GeS 2 is assigned to the BS, (Ge-S-Ge, BS) mode of the GeS 4/2 tetrahedra.(Zhou et al., 1999) It is possible that the broad peak in the IR spectrum of GeS 2 can be deconvoluted into two additional peaks, one centered at ~325 cm -1 and the other centered at ~435 cm -1 .These two additional peaks also arise from vibrational modes of the GeS 4/2 tetrahedra.(Frumarova et al., 1996) The shift in wavenumbers can be due to the presence of compressive stress in the film which is expected for films deposited by RF sputtering.In the IR spectra of the GeS 2 , there is one broad and low intensity peak at ~800 cm -1 .(Zhou et al., 1999) This peak is assigned to the preparation and handling giving rise to a Ge-O bonding mode.It can be assumed that GeS 2 might be slightly contaminated by oxygen during IR sample measurement.In the IR spectra of both the starting materials, there is no peak at ~ 1500 cm -1 (O-H vibration mode) or at ~3500 cm -1 (O-H stretching mode) so this suggests that these two starting materials are not significantly contaminated by oxygen or moisture.
The IR spectra of Li 2 GeS 3 , Li 4 GeS 4 and Li 6 GeS 4 targets show dominant peaks at ~360 cm -1 and 415 cm -1 and a low intensity peak at ~750 cm -1 .The IR peak at ~360 cm -1 is assigned to the BS, (Ge-S-Ge, BS) mode and the 415 cm -1 peak corresponds to the vibration stretch of Ge with two non-bridging sulfur atoms.The broad IR peak at ~360 cm -1 can be deconvoluted into two peaks one centered at 345 cm -1 corresponding to Li + S -mode and the other centered at ~360 cm -1 corresponding to Ge-S-Ge mode.In addition, one low intensity peak which is assigned to oxide impurities, (Ge-O-Ge) appears at ~750 cm -1 .It is possible that contamination occurs when the target materials are melted in the glovebox because a background level of several ppm O 2 exists in the glovebox.Another possibility is that the oxygen comes from the GeS 2 , its spectrum in Figure ( 6) shows that there is a low intensity peak at ~ 750 cm -1 assigned to GeO 2 .
To the best of our knowledge, the IR spectra of the thio-germanate based thin-film materials have not been reported in the open literature.In this research, in order to characterize the thin-films by IR spectroscopy, the Li 2 GeS 3 , Li 4 GeS 4 and Li 6 GeS 5 thin-films were deposited directly on the top side of pressed CsI pellets that provided a mid-and far-IR transparent support for the films.The Li 2 GeS 3 , Li 4 GeS 4 and Li 6 GeS 5 thin-films were deposited directly on the pressed CsI pellets and the IR spectra were then collected in transmission.The intense peak at ~360 cm -1 can be deconvoluted into two peaks one centered at 345 cm -1 corresponding to the Li + S -mode and the other centered at ~360 cm -1 corresponding to the Ge-S-Ge mode as described above and the intensity of this peak decreases with added Li 2 S. In addition, one low intensity peak which is assigned to oxide impurities, (Ge-O-Ge) appears at ~750 cm -1 .A new band appears at 445 cm -1 as a result of the formation of nonbridging sulfurs -Ge-S --Li + (NBS).This NBS band was reported at ~450 cm -1 in the IR spectra of binary xNa 2 S + (1-x)GeS 2 glasses.(Barrau et al., 1980) The NBS band at 445 cm -1 diminishes as another NBS band at 415 cm -1 grows stronger with further additions of Li 2 S and this suggests that the number of NBS per Ge increases with the addition of Li 2 S. Indeed, it is expected from the compositions that these would be two NBS in Li 2 GeS 3 and four NBS in Li 4 GeS 4 and Li 6 GeS 5 .

Surface morphology and thickness of the thin-film
In order to determine the sputtering rate, the thickness of the thin-films were measured in the cross-section direction by FE-SEM as shown in Figure (6-a).A Ni adhesion layer (~120 nm) is used to improve the adhesion between the Si wafer and the thin-film.The Ni adhesion layer is also very useful for Raman spectroscopy.In particular, when one characterizes the films using micro-Raman spectroscopy, the dominant silicon peak, ~520 cm -1 , appears in Raman spectra unless a barrier layer is used.Therefore, the Ni adhesion layer also acted to prevent the appearance of the peak from the silicon substrate.Furthermore, it has been found that Ni is chemically stable in contact with the lithium thio-germanate thin-film electrolytes.(Bourderau et al., 1999) The sputtering power and pressure of 50 W and 25 mtorr (~3.3 Pa) were used, respectively, and the total thickness of the thin-film after two hours of sputtering was ~1.3 m which gives a sputtering rate of ~5 nm/minute.The thin-film surface is mirror-like without any defects or cracks.This suggests that the thin-film electrolytes are homogeneous and have a flat surface morphology.The smooth surface enables the thin-films to decrease the contact resistance between thin-film and the electrodes.

Impedance analysis
In order to measure the ionic conductivity of the thin-films, they were deposited on a single crystal Al 2 O 3 substrate.The sapphire substrates were loaded into a d.c.sputtering chamber in the glovebox and covered by the stainless steel mask with two 2 mm  10 mm slits at 2 mm apart parallel to each other to produce two 2 mm  10 mm parallel electrodes 2 mm apart on the sapphire substrate.Au electrodes of ~100 nm thickness were sputtered for 20 min.at a sputtering rate of 5 nm/min.through the mask.Lastly, the substrate then was loaded into the RF magnetron sputtering chamber to grow the thin-film electrolytes.
The conductivities of the thin-films were determined from the resulting complex impedance spectra.The semicircle at high frequency represents the response of the thin-film materials to an applied electric field.Thus, the d.c.resistance can be calculated from the semicircle plot.The ionic conductivities of the thin-films can be calculated from the measured d.c.resistance, R, the thickness of the electrolyte t, its area A, and t/A is the cell constant.
The ionic conductivities of the Li 6 GeS 5 thin-film grown in Ar atmosphere at various temperatures from -25 o C to 100 o C with 25 o C increments are shown in Figure (7).The ionic conductivities of the Li 6 GeS 5 thin-film in Ar atmosphere at 25 o C and at 100 o C are 1.7  10 -3 S/cm and 3.0  10 -2 S/cm, respectively.As the temperature increases from -25 o C to 100 o C, the ionic conductivity continually increased and was found to be stable over this temperature range.This thin-film appears to be stable wider temperature range compared to liquid electrolytes (Guyomard and Tarascon, 1995).The ionic conductivities were calculated from the resistance and cell constant relations and are listed in Table (1).The d.c.ionic conductivities of the Li 2 GeS 3 , Li 4 GeS 4 , Li 6 GeS 5 and Li 8 GeS 6 thin-films are also shown in Table (1).The ionic conductivities of all four thin-films were characterized with the same temperature ranges, from -25 o C to 100 o C with 25 o C increments, and same frequency ranges, from 0.1Hz to 1 MHz.As shown in Table 1, the ionic conductivities of the Li 4 GeS 4 thin-film are higher than those of Li 2 GeS 3 thin-film at each temperature.The reason for the Li 4 GeS 4 thin-film having a higher ionic conductivity than the Li 2 GeS 3 thin-film is that the Li 4 GeS 4 thin-film contains a higher Li 2 S content.The ionic conductivities at room temperature and 100 o C of the Li 4 GeS 4 thin-film are 7.5  10 -4 S/cm and 1.3  10 -2 S/cm, respectively.As the temperature increased, the ionic conductivities increase without decreasing and hence the thin-films are stable over wide temperature ranges. Temp.
To determine if a maximum Li + ionic conductivity occurs for this series of materials, a Li 8 GeS 6 thin-film, n = 4 in nLi 2 S + GeS 2 , was prepared and the ionic conductivities were analyzed over the same temperature and frequency ranges.The ionic conductivities of the Li 8 GeS 6 thin-film at 25 o C and 100 o C are 7.3  10 -5 S/cm and 1.4  10 -3 S/cm, respectively.While the d.c.ionic conductivities of the thin-films from n = 1 to 3 in nLi 2 S + GeS 2 increased with n, the d.c.ionic conductivity of the thin-film for n = 4, Li 8 GeS 6 , decreased and this is caused by the activation energy increasing, see discussion below.
In this series of films, the n = 3 composition, Li 6 GeS 5 , is the optimized composition with the highest ionic conductivity in the nLi 2 S + GeS 2 system, n = 1, 2, 3, and 4.Although the n = 4 composition, Li 8 GeS 6 , thin-film showed lower ionic conductivities than those of the n = 1(Li 2 GeS 3 ), 2(Li 4 GeS 4 ), and 3(Li 6 GeS 5 ) compositions, the ionic conductivities of all four of these thin-films are significantly higher than that of LiPON.In addition, all compositions n = 1, 2, 3, and 4 of the sulfide thin-film electrolytes are very stable over wide temperature ranges compared to liquid or polymer electrolytes.Therefore, Li-ion batteries using these sulfide thin-film electrolytes are promising for use in solid-state lithium-ion batteries.
For all thin-films, the ionic conductivities were found to follow an Arrhenius law,   9) and are compared to that of LiPON.The activation energies of conduction, E a , were calculated from the slope of the Arrhenius plots.The ionic conductivities of the thin-films at room temperature, the activation energies, and pre-exponential factors are listed in Table (2).
The ionic conductivities of the all-sulfide thin-films higher than that of LiPON (Yu et al., 1997).In the case of the Li 6 GeS 5 thin-film, the ionic conductivity at 25 o C is approximately three orders of magnitude higher than that of LiPON (Yu et al., 1997).The composition dependence of the ionic conductivities of all thin-films, n = 1, 2, 3, and 4 in nLi 2 S + GeS 2 system, at 25 o C and their activation energies are shown in Figure ( 10) to show how they depend upon Li 2 S content.The thin-films showed that as Li 2 S content increases, the ionic conductivities increase up to n = 3, 75 % Li 2 S.
In addition, while the conductivity of the bulk sulfide glasses are less than that of the thinfilms, the ionic conductivities of the sulfide bulk glasses (Kim et al., 2006) over the range from 35% to 50 % the ionic conductivities also increased.It is significant to note that the ionic conductivity decreased and the activation energies increased for the thin-films at n = 4. LiPON [Yu et al., 1997] Fig. 9. Arrhenius plot of the ionic conductivities at various temperatures for Li 2 GeS 3 , Li 4 GeS 4 , Li 6 GeS 5 , and Li 8 GeS 6 thin-films in Ar atmosphere and comparison of ionic conductivities between thin-films and LiPON (Yu et al., 1997).Further, the effective basicity of the counter and charge compensating anion in the structure of these materials is also expected to change significantly with n.In the n = 1, 2, 3, and 4 films, the structure is expected to consist of increasing numbers of sulfurs possessing a single negative charge, and recent XPS studies of these same films show that these films are comprised of the nominal structures shown in Figure (11).In these structures, the average charge on the sulfur is expected to change from -2/3, -4/4, -6/5 to -8/6.At Li 2 S, the formal charge of the sulfur is expected to -2/1.Hence, while increasing the number of Li + is important to increasing the ionic conductivity, the negative charge density on the sulfur increases by a factor of 2 in this series and as a result the columbic binding energy of these increasingly basic sulfurs will increase as well.It appears that for the n = 3 composition, the larger number of Li is still important because the appearance of the full -2/1 negatively charged Li 2 S unit does increase the conductivity activation energy, 35 kJ/mol for n = 3 versus 34.5 kJ/mol for n = 2, but the conductivity is still higher, presumably because of the composition (n) dependence of the pre-exponential factor.We have shown in our other studies of these thin-films that the fraction of NBS Ge-S - increases with n in these series.For n = 1, the fraction of bridging sulfurs S-Ge-S and nonbridging sulfurs Ge-S -are 1/3 and 2/3, respectively.At n = 2, these fractions are 0 and 1, respectively.For n = 3 and n = 4, these units are expected to be connected to non-bridging sulfur units and new S = units.Hence, the fraction of sulfur in non-bridging Ge-S -units and S = units are expected to be 4/5 and 1/5, respectively.In Li 8 GeS 6 these fractions change to 4/6 and 2/6, respectively.The increase in the fraction of Li + ions bound to S = units increases from 0 for Li 2 GeS 3 and Li 4 GeS 4 to 2/6 (33 %) and 4/8 (50 %) for Li 6 GeS 5 and Li 8 GeS 6 , respectively.Due to the high binding energy expected for Li + ions about these S = ions it is therefore not surprising to see that the activation energy passes through a minimum at the n = 2 composition and increases for n = 3 and 4. Such a maximum in conductivity and minimum activation energy have been observed in other high alkali glass forming system where the anionic basicity of the host network increases significantly in the high alkali modifier range (Martin and Angell, 1984).

Analysis of the Li 2 S and GeS 2 starting materials
In order to verify the purity of the starting materials, the Li 2 S and GeS 2 were examined by XPS.In the case of the commercially purchased Li 2 S material, One possibility is that the Li 2 S was slightly contaminated on the surface in the glovebox because the glovebox contained several ppm level of oxygen.Another possible reason for this can also include the "see-through" effect due to the double-sided tape used to adhere the powder to the XPS sample holder.The ratio of Li to S, however, 1.95 : 1.00 is very close to the expected value of 2 : 1.
From Table (3), while the Li 2 S shows relatively high O content, the GeS 2 material was not contaminated by oxygen due in part to the fact that GeS 2 material is less hygroscopic than other sulfide materials, but also due to the fact that this material was prepared from high purity starting materials, Ge and S (99.9999%), in the very controlled conditions of our laboratory.GeS 2 contains a small percent of C, presumably surface C, see Table (3 160.7 eV and 163.2 eV ±0.2 eV, respectively.The reason for the difference in the S2p binding energies between Li 2 S and GeS 2 is that S in Li 2 S is the fully ionic S = sulfide anion and the S in the GeS 2 is the fully covalent BS, Ge-S-Ge, and hence the binding energy of the covalent BS is higher than that of the ionic sulfide.For the S2p spectra of sulfur species, there is a doublet consisting of S2p 3/2 and S2p 1/2 spin-orbit coupled electrons in the intensity ratio of 2:1.The S2p core peaks of Li 2 S show one doublet.This doublet arises from the Li 2 S bonding and means that only the Li 2 S bonding exists.This result agrees well with the literature data (Foix et al., 2001).
If Li 2 S was significantly contaminated by oxygen, the deconvoluted S2p spectra would be expected to show additional peaks related to sulfite SO 3 2-(166 eV) and/or sulfate SO 4 2- (172 eV) contamination (Volynsky et al., 2001).Both the Li 2 S and GeS 2 materials do not show significant peaks at 166 eV and 172 eV and suggests that these materials are of high purity.
The deconvoluted S2p core peaks of the GeS 2 also show as expected only one doublet arising from the single bridging sulfur structure, Ge-S-Ge.

Target material analysis
After the target materials for RF sputtering were made using nLi 2 S + GeS 2 , n = 1, 2 and 3, their compositions were determined by XPS.The compositional data of the three target materials are shown in Table ( 4).The target materials show C and O contents and therefore the Li, Ge and S contents are lower than the expected values.If C and O elements are ignored, the compositional data of Li, Ge and S for all three target materials nearly match with the expected values.Although the XPS compositional data are different between the collected and expected data, the differences are within the confidence error limit, ± 3%.Considering the ± 3% error of the XPS data, the small Li deficiency can be ignored.From the spectra in Figure ( 13), the binding energies of the NBS, Ge-S -Li + and BS, Ge-S-Ge can be obtained.By comparing these XPS spectra to that of the standard materials, it can be determined that the sulfur spectra do not contain peaks related to sulfite SO 3 2-or sulfate SO 4 2-anions which would be shifted to significantly higher binding energies due to their S +4 and S +6 oxidation states, respectively.This suggests that although the target materials show some oxygen content as shown in Table ( 5), the contamination may only be on the surface.The low oxygen content is associated with the high purity of the starting materials as well as the fact that the target materials were made in a N 2 filled high quality glovebox.As shown in Figure ( 13), while two doublets appear in the deconvoluted S2p core peaks of the Li 2 GeS 3 target material indicating that there are two chemically distinct surface species, the The binding energies of sulfur in the target compositions and the NBS and BS ratios of the three target materials are shown in Table (5).In order to compare the chemical shifts, the binding energies of the GeS 2 and Li 2 S are also listed in Table (5).
While the binding energy of S in GeS 2 shows the highest value due to its BS structure, the binding energy of S in Li 2 S shows the lowest value.The binding energies of the target materials are similar to one another and, as expected, are between that of GeS 2 and Li 2 S. In the S XPS spectrum of the Li 2 GeS 3 target, the low energy doublet is assigned to the NBS and the other higher energy doublet is assigned to the BS.For the Li 2 GeS 3 target material, the ratio of the NBS to BS is 65.3 % to 34.7 %.The expected ratio of NBS to BS in the Li 2 GeS 3 target composition agrees well with that calculated from the composition of 67 % to 33 %.(Foix et al., 2002) Theoretically, the ratio of the NBS to BS in the Li 4 GeS 4 target should be 100 % and 0 %, respectively.
As shown in Table ( 5), the Li 4 GeS 4 target material shows 100 % NBS to 0 % BS ratio.Additionally and as expected, the Li 6 GeS 5 target material shows 100 % NBS and 0 % BS.As described above, Li 2 S consists of only the S = anion whereas Li 4 GeS 4 consists of 100 % NBS.
From the composition, it is expected that the Li 6 GeS 5 target should be composed of an equimolar mixture of Li 2 S and Li 4 GeS 4 .However, the XPS spectra data shown in Table (5) shows that Li 6 GeS 5 consists of only 100 % NBS and 0 % BS.Strictly speaking, Li 6 GeS 5 should consist of Li 4 GeS 4 which has 100 % NBS and Li 2 S which has 100 % ionic sulfur, S = .While the binding energies of the S = anion and the NBS are very close, the resolution of our XPS instrument appears to be insufficient to differentiate the chemical shift of S = anion and the NBS unit, Ge-S -Li + .

Lithium thio-germanate thin-film analysis
After sputtering thin-films on Ni-coated Si substrates in Ar atmospheres, they were characterized by XPS to determine their compositions and chemical shifts.It was found that a Ni protective layer on the Si was necessary to prevent reaction of the Si with the Li which produces highly Li deficient films.This is described below in the experimental section.The compositional data of all of the thin-films sputtered in Ar atmospheres using the three different conditions are shown in For the as-prepared thin-films, the C and O contents are slightly higher than those of the targets and the Li, Ge and S contents are slightly lower than their expected values.It is assumed that this arises due to the intrinsically higher chemical reactivity of the surface of thin-films compared to bulk materials.In order to obtain more accurate compositional data of the thin-films, Ar etching was performed on the thin-film surfaces for 1 min.and 5 min.at a rate of ~1 nm/min.As shown in Table ( 6), after Ar etching for 1 min.the C content in the thin-films reduced to 0 % and the O content decreased significantly.Although some O content still exists in the thin-films, the Li, Ge, and S contents in the thin-films are very close to their expected values.In order to examine deeper profiles of the thin-film, Ar etching for 5 min.was performed at ~1 nm/ min.etching rate.After Ar etching for 5 min.was performed, the compositional data are almost the same compared to the data obtained after Ar etching for 1 minute.This suggests that the thin-films show high uniformity and quality except for the top 1 nm of the surface.Previous literature reported (Yamashita, M., et al., 1996a) that thio-germanate thin-films produced from Li 2 S + Ga 2 S 3 + GeS 2 by sputtering showed severe Li deficiency.While these ternary thin-films showed as high as ~30 to 40 % Li deficiency compared to the Li in target composition, the thin-films produced in this study only show ~3-5 % Li deficiency.The compositions of the thin-films in this study are consistent with those of the target and it is therefore assumed that the sputtering conditions reported here are optimized and the thin-film compositions are reliable.In order to determine the fractions of NBS and BS in the thin-films, the deconvoluted S2p core peaks for the Li 2 GeS 3 , Li 4 GeS 4 , and Li 6 GeS 5 as-prepared thin-films (without Ar etching treatment) are shown in Figure ( 14).While the XPS spectra of the Li, Ge, and S species are unchanged in binding energy with and without Ar etching, Ar etching could reform the chemistry of the Ar sputtered surface.For this reason, it is believed that a better representation of the bonding chemistries, the chemical speciation, of these thin-films are therefore found in the as-prepared surfaces of the thin-films.For example, Foix et al. reported the fractions of NBS and BS in lithium thio-germanate and thio-arsenate bulk glasses and to do so they broke the glasses in the glove box and they characterized the newly exposed broken surface of the glasses without Ar etching (Foix et al., 2001;Volynsky et al., 2001).
In addition, Atashbar et al. reported the XPS deconvoluted data of TiO 2 thin-films without Ar etching (Atashbar et al., 1998).These approaches suggest that although accurate compositional data could be obtained from the Ar etched surface, the data could also be to use the XPS deconvoluted structural analysis is also obtained from the as-prepared surface without Ar etching.The fractions of the NBS and BS in the thin-films were calculated from Figure ( 14) and are shown in Table (7).In Figure ( 14) as described above, the Li 2 GeS 3 thin-film shows two doublets.The doublet on the low energy side (lower binding energy) is attributed to the NBS and the other doublet on the high energy side (higher binding energy) is associated to BS.The ratios of the NBS and BS in the Li 2 GeS 3 thin-film are 64.4 % NBS and 35.6 % BS, respectively.Although the ratios are not exactly the same as the expected values, 67 % NBS and 33 % BS, the differences between those of the Li 2 GeS 3 thin-film and expected values are within the error range of ± 3%.
In addition, the ratios of the NBS and BS in the Li 2 GeS 3 target and thin-film are very close.This suggests that the target compositions and thin-film compositions are quite consistent.While the deconvoluted S2p core spectra of the Li 2 GeS 3 thin-film show two doublets, the deconvoluted S2p core spectra of the Li 4 GeS 4 and Li 6 GeS 5 thin-films show only one doublet.
As described above, the Li 4 GeS 4 and Li 6 GeS 5 targets also show only one doublet from the NBS.In agreement with these Li 4 GeS 4 and Li 6 GeS 5 targets, the two thin-films show only one doublet arising from only NBS structures.
Recently, a few XPS studies of Ge-S thin-films have been reported in the literature but the analyzes were very brief.(Gonbeau et al., 2005;Mitkova et al., 2004) However, in this study, the compositions and chemical shifts of the Li-Ge-S thin-films have been thoroughly investigated (Gonbeau et al., 2005;Mitkova et al., 2004).As shown in Figure ( 14), the spectrum for GeS 2 shows a higher binding energy than those of Li 2 S and the thinfilms because the GeS 2 is assigned to the BS as described above.As the Li 2 S content increases, the binding energy of the thin-films shifts to lower values than that of GeS 2 .
While the binding energies of the thin-films are similar to one another, with the S peak for Li 2 GeS 3 being broader than that for Li 4 GeS 4 and Li 6 GeS 5 due to the presence of both BS and NBS, the binding energies of the thin-films slightly shifted to lower values.As expected, the binding energy of the Li 2 S shows the lowest binding energy of the materials studied here.
Thin-films E b S2p

Conclusion
For the first time, lithium thio-germanate thin-film electrolytes for the solid-state lithium-ion batteries grown by RF sputtering were characterized thoroughly by XRD, FE-SEM, Raman, IR, impedance spectroscopy, and XPS.From the XRD pattern, the Li 2 GeS 3 (n = 1) target was amorphous and the Li 4 GeS 4 (n = 2) and Li 6 GeS 5 (n = 3) targets were crystalline as expected from compositions.The Li 6 GeS 5 target appears to be consistent with an equi-molar mixture of Li 2 S and Li 4 GeS 4 .FE-SEM of the thin films deposited on Ni-coated Si substrates shows a mirror-like surface without cracks or pits.The Raman spectra of all of the thin-films do not show sharp peaks, rather they show much broader peaks compared to those of crystalline targets (Li 4 GeS 4 and Li 6 GeS 5 ) and are consistent with the thin-films being amorphous.This shows that RF sputtering can be used to extend the formation range of amorphous materials from ~50 to ~75 mole % Li 2 S.
The Raman and IR spectra also showed the structural and compositional consistency between targets and the thin-films and that the Li 2 S content of thin-films increased as expected with Li 2 S addition in the targets.These results suggest that the thin-films did not show significant Li deficiency as seen in previous reports after sputtering.
The ionic conductivities of the thin-films at 25 o C obtained are the highest reported for Li + ion in a glassy materials and are at least two orders of magnitude higher than those of commercial LiPON thin-film electrolytes.The thin-films materials are stable over wide temperature ranges, so that it can be said that the lithium-ion batteries based on these sulfides materials are very stable over wide temperature ranges and are very promising to apply to commercial products.The purpose of the XPS work was to provide information on the compositional data and the structures of lithium thio-germanate thin-films by means of XPS studies.High purity starting materials were used and targets were produced under well-calibrated and optimized conditions.
For the first time, highly reproducible compositions and chemical shifts of the starting materials, targets, and thin-films of nLi 2 S + GeS 2 materials were determined by XPS.
Although the as-prepared thin-films contained C and O on the surface, the thin-films showed that the C was completely removed and O content decreased significantly after Ar etching for 1 min.This suggests that the thin-films were contaminated by C only at the top 1 nm of the surface and the thin-films contained low oxygen contents in the interior of the film.After Ar etching, the compositions of the thin-films were very close to those expected.Therefore, the thin-films produced by sputtering are very close to their corresponding target materials.Thio-germanate thin-film materials have not been as widely studied as their oxide materials because of the difficulties in preparation.However, in this study, the lithium thiogermanate thin-films were successfully prepared and the compositional data and the chemical shifts were carefully characterized by XPS.
By successfully making thin-films of high quality and high conductivity, they can be applied as thin-film electrolytes for solid-state thin-film batteries.Further extensive effort for solid-state full battery fabrication, however, is needed before this thin-film electrolyte is put to practical use.

Acknowledgement
This research was supported by NSF under grant number DMR-0312081 and this research support is gratefully acknowledged.The authors thank Mr. James Anderegg who helped collect all of the XPS spectra and assisted with the experimental details to load and examine the samples without contamination.

Fig. 5 .
Fig. 5. Infrared spectra of Li 2 S, GeS 2 , targets, and thin-films The IR spectra of polycrystalline Li 2 S, glassy GeS 2 , targets and thin films are shown in Figure (5).The IR peak in the far-IR spectrum of Li 2 S at ~345 cm -1 is assigned to the ionic bonding of Li + S  and the strong peak at ~370 cm -1 in the spectrum of glassy GeS 2 is assigned to the BS, (Ge-S-Ge, BS) mode of the GeS 4/2 tetrahedra.(Zhou et al., 1999) It is possible that the broad peak in the IR spectrum of GeS 2 can be deconvoluted into two additional peaks, one centered at ~325 cm -1 and the other centered at ~435 cm -1 .These two additional peaks also arise from vibrational modes of the GeS 4/2 tetrahedra.(Frumarova et al., 1996) The shift in wavenumbers can be due to the presence of compressive stress in the film which is expected for films deposited by RF sputtering.In the IR spectra of the GeS 2 , there is one broad and low intensity peak at ~800 cm -1 .(Zhou et al., 1999) This peak is assigned to the preparation and handling giving rise to a Ge-O bonding mode.It can be assumed that GeS 2 might be slightly contaminated by oxygen during IR sample measurement.In the IR spectra

Fig. 6 .
Fig. 6.FE-SEM images of cross-sectional view (a) and top view (b) of Li 4 GeS 4 thin film grown on a Ni/Si substrate in an Ar atmosphere.

Figure ( 6
shows the surface morphology of the thin-films produced in an Ar atmosphere.The thin-film surface is mirror-like without any defects or cracks.This suggests that the thin-film electrolytes are homogeneous and have a flat surface morphology.The smooth surface enables the thin-films to decrease the contact resistance between thin-film and the electrodes.

Figure ( 8
Figure (8) shows a Nyquist plot of the complex impedance for the Li 6 GeS 5 thin-film grown in an Ar atmosphere over the temperature ranges from 25 o C to 100 o C with 25 o C increments.

Fig. 7 .Fig. 8 .
Fig. 7.The ionic conductivity of Li 6 GeS 5 thin-film grown in Ar atmospheres over the temperatures from -25 o C to 100 o C with 25 o C increments.
d.c.(T) =  o exp(-E a /RT), over the measured temperature ranges.The Arrhenius plots of the d.c.ionic conductivities of the thin-films over the temperature range from -25 o C to 100 o C with 25 o C increments are shown in Figure (
), and after ignoring C, the compositional data of GeS 2 agrees well with expected values.Deconvoluted S2p core XPS spectra for crystalline Li 2 S (a) and glassy GeS 2 (b) starting materials are shown in Figure (12).The binding energies for sulfur in Li 2 S and GeS 2 are at www.intechopen.com
Jourdaine et al. in France reported thin-film batteries produced by RF sputtering.They successfully fabricated the cell using metallic lithium as an anode, Li 2 O-B 2 O 3 -P 2 O 5 or Li 2 O-B 2 O 3 glasses as electrolytes, and V 2 O 5 -TeO 2 or V 2 O 5 -P 2 O 5 as cathodes, respectively(Jourdaine

Table 3 .
XPS compositional analysis of the Li 2 S and GeS 2 starting materials.

Table 4 .
Ar etching treatments were not performed to remove surface C and O because the target materials were in the form of powders and Ar etching does not work well for powders that do not have large flat smooth surfaces.The deconvoluted S2p spectra of the three target materials are shown in Figure(13).XPS compositional analysis of the Li 2 GeS 3 , Li 4 GeS 4 and Li 6 GeS 5 target materials.

Table 5 .
Li 4 GeS 4 and Li 6 GeS 5 targets show only one doublet indicating a single chemical species for sulfur.The XPS binding energies and the ratio of NBS to BS for the starting and target materials.

Table 6 .
Table (6).XPS compositional analysis of the Li 2 GeS 3 , Li 4 GeS 4 and Li 6 GeS 5 thin-film grown on Ni-coated Si substrates in an Ar atmosphere.

Table 7 .
The XPS binding energies (E b ) and fractions of NBS to BS of the Li 2 GeS 3 , Li 4 GeS 4 and Li 6 GeS 5 thin-films.
, Lithium-Ion Transfer at the Interface Between Lithium-Ion Conductive Ceramic Electrolyte and Liquid Electrolyte-A Key to Enhancing the Rate Capability of Lithium-Ion Batteries: Journal of the Electrochemical Society, Vol.152, No. 11, pp.A2151-A2154, 00134651.