X-ray photoelectron spectroscopy study of lithium thio-germanate thin films

Abstract

In this study, the compositions of and chemical speciation in lithium thio-germanate thin films grown by radio frequency (RF) magnetron sputtering were determined by means of X-ray photoelectron spectroscopy (XPS). Three different targets, nLi₂S + GeS₂ (n = 1, 2, and 3) for RF sputtering were prepared. Thin films of ∼1 μm thickness were grown on single crystal Si [111] substrates using 50 W power and ∼3.3 Pa Ar gas atmospheres. As is commonly observed for RF sputtered non-oxide films, the surfaces of these lithium thio-germanate films were contaminated with both C and O at ∼10 at %. Etching the top surface of the films with Ar removed all of the surface carbon and reduced the oxygen contamination down to an average value of ∼5 at %. After such surface etching, the compositions of the films were found to agree very closely with the nominal composition of the target from which it was made. The Li₂GeS₃ (n = 1) thin film was found as expected to exhibit a 2 : 1 non-bridging sulfur (≡Ge–S⁻Li⁺, NBS) to bridging sulfur (≡Ge–S–Ge≡, BS) ratio. As the Li₂S content increased, the Li₄GeS₄ and Li₆GeS₅ thin films, n = 2 and 3, exhibited as expected, 100% NBS. For the S2p core peaks, the binding energy of S in GeS₂ is the highest due to the nearly covalent bonding in the BS structure, ≡Ge–S–Ge≡. As the Li₂S content increased, the binding energy of S in the thin films progressively decreased due to the progressive increase in the fraction (and number) of NBS (more ionic), ≡Ge–S⁻Li⁺, in the films.

---

Introduction

Renewable energy harvesting, conversion, storage, and distribution are of great interest due to the demand created by portable electronics and transportation.^1–3Lithium-based batteries have begun to meet these demands due to the high energy densities of lithium-carbon anodes. However, the theoretical energy capacity of lithium metal is ∼10 times higher than that of lithium-carbon anodes at ∼3860 mAh g⁻¹vs. ∼372 mAh g⁻¹.⁴ Recently, lithium based thin film batteries have been of interest in miniaturized power electronics such as smart cards.⁵ In addition, thin film batteries are candidates for high power applications such as electric/hybrid vehicles.⁶ Thin film batteries show smaller temperature dependences and are able to address the safety issues of aqueous flammable electrolytes.⁷

So far, lithium phosphorous oxy-nitride (LiPON) thin film electrolytes are the most commonly used because LiPON shows good chemical and electrochemical stability in contact with metallic lithium, high cyclability, and facile fabrication.^8–9 However, this easily prepared and chemically stable solid electrolyte has a relatively low Li⁺ ion conductivity of ∼10⁻⁶ (S cm⁻¹) at 25 °C compared to sulfide materials which have Li⁺ ion conductivities in the range of ∼10⁻³ to ∼10⁻⁴ (S cm⁻¹) at 25 °C.^10–16

For these reasons, sulfide materials have attracted attention for electrochemical applications. While many reports have been made of bulk sulfide glasses,^17–21 there are very few reports of sulfide thin films because of the difficulty in preparing them.^22–23 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, caused possibly by leakage of the RF chamber, or by exposure to air after sputtering.²³ In addition, these thin films were found to be Li deficient compared to that of the targets from which they were made.²³ Therefore, although sulfide films may have good potential in thin film batteries, the 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, in this study lithium thio-germanate thin films were carefully sputtered under well controlled conditions and then carefully analyzed by X-ray photoelectron spectroscopy (XPS). XPS is one of the best methods to determine the composition of thin film materials, even of light elements such as lithium, and it is also effective in investigating the characteristics (valence) of the constituent atoms (ions) in the thin films by monitoring their binding energies. 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, Li₂S and GeS₂, the target materials, and thin films were characterized by XPS for compositional and chemical shift analysis.

Results and discussion

Analysis of the Li₂S and GeS₂ starting materials

In order to verify the purity of the starting materials, the Li₂S and GeS₂ were examined by XPS. In the case of the commercially purchased Li₂S material, Table 1 shows that the concentration of C and O were ∼12% and ∼21% (±3%), respectively, and as such are relatively high. One possibility is that the Li₂S 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.

Table 1 XPS compositional analysis of the Li₂S and GeS₂ as starting materials

At%	Li1s	Ge2p3	S2p	C1s	O1s	Comments
Li2S	44.7	—	22.9	11.7	20.7	As-prepared
	66.1	—	33.9	—	—	Ignore C and O
	66.7	—	33.3	0.0	0.0	Expected values
GeS2	—	34.2	59.2	6.6	0.0	As-prepared
	—	35.7	64.3	—	0.0	Ignore C
	—	33.3	66.7	0.0	0.0	Expected values

---

The ratio of Li to S, however, 1.95 : 1.00 is very close to the expected value of 2 : 1. From Table 1, while the Li₂S shows relatively high O content, the GeS₂ material was not contaminated by oxygen due in part to the fact that GeS₂ 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. GeS₂ contains a small percent of C, presumably surface C, see Table 1, and after ignoring C, the compositional data of GeS₂ agrees well with expected values.

Deconvoluted S2p core XPS spectra for crystalline Li₂S (a) and glassy GeS₂ (b) starting materials are shown in Fig. 1. The binding energies for sulfur in Li₂S and GeS₂ are at 160.7 eV and 163.2 eV ±0.2 eV, respectively. The reason for the difference in the S2p binding energies between Li₂S and GeS₂ is that S in Li₂S is the fully ionic S²⁻sulfide anion and the S in the GeS₂ 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 Li₂S show one doublet. This doublet arises from the Li₂S bonding and means that only the Li₂S bonding exists. This result agrees well with the literature data.²⁴ If Li₂S was significantly contaminated by oxygen, the deconvoluted S2p spectra would be expected to show additional peaks related to sulfite SO₃²⁻ (166 eV) and/or sulfate SO₄²⁻ (172 eV) contamination²⁵. Both the Li₂S and GeS₂ 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 GeS₂ also show, as expected, only one doublet arising from the single bridging sulfur structure, ≡Ge–S–Ge≡.

Fig. 1 Deconvoluted S2p core XPS spectra for Li₂S (a) and GeS₂ (b).

Target material analysis

After the target materials for the RF sputtering were made using nLi₂S + GeS₂, n = 1, 2 and 3, their compositions were determined by XPS. The compositional data of the three target materials are shown in Table 2. 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.

Table 2 XPS compositional analysis of the Li₂GeS₃, Li₄GeS₄ and Li₆GeS₅ target materials

At%	Li1s	Ge2p3	S2p	C1s	O1s	Comments
Li2GeS3 target	26.1	16.4	41.3	10.3	5.9	As-prepared
	30.4	19.1	48.1	—	—	Ignore C and O
	33.3	16.7	50.0	0.0	0.0	Expected values
Li4GeS4 Target	36.5	9.6	40.2	8.6	5.1	As-prepared
	42.3	11.1	46.6	—	—	Ignore C and O
	44.4	11.2	44.4	0.0	0.0	Expected values
Li6GeS5 target	40.4	8.0	37.2	6.5	7.9	As-prepared
	47.2	9.3	43.5	—	—	Ignore C and O
	50.0	8.3	41.7	0.0	0.0	Expected values

---

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 Fig. 2. From the spectra in Fig. 2, 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 SO₃²⁻ or sulfate SO₄²⁻ anions which would be shifted to significantly higher binding energies due to their S⁺⁴ and S⁺⁶oxidation states, respectively. This suggests that although the target materials show some oxygen content as shown in Table 2, 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 N₂ filled high quality glovebox. As shown in Fig. 2, while two doublets appear in the deconvoluted S2p core peaks of the Li₂GeS₃ target material indicating that there are two chemically distinct surface species, the Li₄GeS₄ and Li₆GeS₅ targets show only one doublet indicating a single chemical species for sulfur.

Fig. 2 Deconvoluted S2p core XPS spectra of the Li₂GeS₃ (a), Li₄GeS₄ (b), and Li₆GeS₅ (c).

The binding energies of sulfur in the target compositions and the NBS and BS ratios of the three target materials are shown in Table 3. In order to compare the chemical shifts, the binding energies of the GeS₂ and Li₂S are also listed in Table 3. While the binding energy of S in GeS₂ shows the highest value due to its BS structure, the binding energy of S in Li₂S shows the lowest value. The binding energies of the target materials are similar to one another and, as expected, are between that of GeS₂ and Li₂S. In the S XPS spectrum of the Li₂GeS₃ target, the low energy doublet is assigned to the NBS and the other higher energy doublet is assigned to the BS. For the Li₂GeS₃ target material, the ratio of the NBS and BS is 65.3% to 34.7%. The expected ratio of NBS to BS in the Li₂GeS₃ target composition agrees well with that calculated from the composition of 67% to 33%.²⁶ Theoretically, the ratio of the NBS and BS in the Li₄GeS₄ target should be 100% and 0%, respectively. As shown in Table 3, the Li₄GeS₄ target material shows 100% NBS and 0% BS ratio. Additionally and as expected, the Li₆GeS₅ target material shows 100% NBS and 0% BS.

Table 3 The XPS binding energies and the ratio of NBS to BS for the starting and target materials

Compounds	E b S2p3/2−1/2 (eV)	Ratio
GeS2	162.8–164.0	100% BS
Li2GeS3 target	160.9–162.1	65.9% NBS
	161.7–162.9	34.1% BS
Li4GeS4 target	160.8–162.0	100% NBS
		0% BS
Li6GeS5 target	160.7–161.9	100% NBS
		0% BS
Li2S	160.5–161.7	100% NBS

---

As described above, Li₂S consists of only the S⁼ anion whereas Li₄GeS₄ consists of 100% NBS. From the composition, it is expected that the Li₆GeS₅ target should be composed of an equimolar mixture of Li₂S and Li₄GeS₄. However, the XPS spectra data shown in Table 3 shows that Li₆GeS₅ consists of only 100% NBS and 0% BS. Strictly speaking, Li₆GeS₅ should consist of Li₄GeS₄ which has 100% NBS and Li₂S which has 100% ionic sulfur, S²⁻. While the binding energies of the S²⁻ anion and the NBS are very close, the resolution of our XPS instrument appears to be insufficient to differentiate the chemical shift of S²⁻ anion and the NBS unit, ≡Ge–S⁻Li⁺.

The structure of Li₂GeS₃, Li₄GeS₄ and Li₆GeS₅ targets consist of the progressive addition of Li₂S to GeS₂ to create increasing amounts of NBS which replace the BS. The chemical short range structures of GeS₂, Li₂GeS₃, Li₄GeS₄, and Li₆GeS₅ are shown in Fig. 3 which shows that the fraction of NBS to BS progressively changes from 0 : 2, 2 : 1, 4 : 0, and 4 : 0 across this series. The latter composition, Li₆GeS₅, would be expected to consist of equimolar amounts of Li₂S and Li₄GeS₄.

Fig. 3 The schematic structures of GeS₂, Li₂GeS₃, Li₄GeS₄, and Li₆GeS₅.

Compositional analysis of the lithium thio-germanate thin films

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 targets are shown in Table 4. 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⁻¹. As shown in Table 4, 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⁻¹etching 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 min. This suggests that the thin films show high uniformity and quality except for the top 1 nm of the surface.

Table 4 XPS compositional analysis of the Li₂GeS₃, Li₄GeS₄ and Li₆GeS₅ thin films grown on Ni coated Si substrates in an Ar atmosphere

At (%)	Li1s	Ge2p3	S2p	C1s	O1s	Comments
Li2GeS3 thin film n = 1	27.2	8.5	37.1	18.6	8.6	As-prepared
	32.6	15.9	47.8	0.0	3.7	Etching for 1 min.
	31.7	16.1	48.1	0.0	4.1	Etching for 5 min.
	33.3	16.7	50.0	0.0	0.0	Expected values
Li4GeS4 thin film n = 2	31.0	5.5	32.1	18.3	13.1	As-prepared
	40.6	12.6	41.3	0.0	5.5	Etching for 1 min.
	41.9	12.9	40.5	0.0	4.7	Etching for 5 min.
	44.4	11.2	44.4	0.0	0.0	Expected values
Li6GeS5 thin film n = 3	35.9	4.9	33.2	14.7	11.3	As-prepared
	43.7	8.9	41.8	0.0	5.6	Etching for 1 min.
	44.6	11.1	41.2	0.0	3.1	Etching for 5 min.
	50.0	8.3	41.7	0.0	0.0	Expected values

---

Previous literature reported²³ that thio-germanate thin films produced from Li₂S + Ga₂S₃ + GeS₂ by sputtering showed severe Li deficiency. While these ternary thin films showed Li deficiency as high as ∼30 to 40% 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 compare the compositional data for the thin films, the spectra of the as-prepared Li₂GeS₃ thin film and thin film with Ar etching for 1 and 5 min are shown in Fig. 4. In Fig. 4, the XPS spectra intensity of the Li, Ge and S elements changed slightly after Ar etching treatment, but the binding energies are unchanged. This suggests that the relative compositions of the Li, Ge, and S components changed because the C and O contents changed due to Ar etching, but the chemical speciation, i.e. the chemical bonding of Li, Ge, and S, are unchanged from the surface to the bulk of the thin film. Strictly speaking, the XPS spectra of the sulfur in Fig. 4c are slightly different before and after Ar etching treatment. The reason for this is that the Ar etching modifies in the local chemical structure very slightly. A similar phenomenon can be seen in the literature.²⁷

Fig. 4 XPS spectra of the Li₂GeS₃ thin films in Ar atmosphere without Ar etching and with Ar etching for 1 min and 5 min.

The intensity of the C1s XPS spectra decreased to zero after Ar etching and this suggests that the thin films were contaminated by C only in the top 1 nm of the surface. The spectral intensity of the O element decreased after Ar etching treatment as well as and its binding energy shifted to lower values. This suggests that the Ar etching treatment destroys the surface oxygen bonding structure and removed O from the thin film. The XPS spectra of the Li₄GeS₄ and Li₆GeS₅ thin films, as-prepared, and with Ar etching for 1 and 5 min show very similar patterns to those of the Li₂GeS₃ spectra. The two XPS spectra of the as-prepared and with Ar etching of the Li₄GeS₄ and Li₆GeS₅ thin films are available in the ESI.† From these spectra, as mentioned above, the thin films were contaminated by C only in the top 1 nm of the surface and were contaminated by O with much lower amounts in the interior of the films. In order to correct the peak positions of the XPS spectra, the C1s peak (284.5 eV) was used for the raw materials before Ar etching. However, after Ar etching was performed, the C1s peak disappeared. For the correction of the peak position after Ar etching, the peaks were shifted by the same amount as used before Ar etching. The different samples were installed together on one sample holder and the XPS spectra were collected before and after Ar etching using the exact same conditions. Thus, it can be anticipated that peak position correction after Ar etching using same amount as shifted for the before Ar etching are reasonable.

In order to determine the fractions of NBS and BS in the thin films, the deconvoluted S2p core peaks for the Li₂GeS₃, Li₄GeS₄, and Li₆GeS₅ as-prepared thin films (without Ar etching treatment) are shown in Fig. 5. 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.^24–25 In addition, Atashbar et al. reported the XPS deconvoluted data of TiO₂ thin films without Ar etching.²⁸

Fig. 5 Deconvoluted S2p core peaks of the Li₂GeS₃, Li₄GeS₄ and Li₆GeS₅ thin films grown in Ar atmosphere.

These approaches suggest that although accurate compositional data could be obtained from the Ar etched surface, a better approach is to use the XPS deconvoluted structural analysis obtained from the as-prepared surface without Ar etching. The fractions of the NBS and BS in the thin films were calculated from Fig. 5 and are shown in Table 5. In Fig. 5 as described above, the Li₂GeS₃ 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 Li₂GeS₃ 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 Li₂GeS₃ thin film and expected values are within the error range of ±3%. In addition, the ratios of the NBS and BS in the Li₂GeS₃ 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 Li₂GeS₃ thin film show two doublets, the deconvoluted S2p core spectra of the Li₄GeS₄ and Li₆GeS₅ thin films show only one doublet. As described above, the Li₄GeS₄ and Li₆GeS₅ targets also show only one doublet from the NBS. In agreement with these Li₄GeS₄ and Li₆GeS₅ 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.^29–30 However, in this study, the compositions and chemical shifts of the Li–Ge–S thin films have been thoroughly investigated.^29–30

Table 5 The XPS binding energies (Eb) and fractions of NBS to BS of the Li₂GeS₃, Li₄GeS₄ and Li₆GeS₅ thin films

Thin films	E b S2p3/2−1/2 (eV)	Fraction	Expected values
Li2GeS3 as-prepared	160.9–162.1	65.4% NBS	66.7% NBS
	161.5–162.7	34.6% BS	33.3% BS
Li4GeS4 as-prepared	160.8–162.0	100% NBS	100% NBS
		0% BS	0% BS
Li6GeS5 as-prepared	160.7–161.9	100% NBS	100% NBS
		0% BS	0% BS

---

In order to compare the binding energies and chemical shifts of these films, the deconvoluted S2p core peaks for GeS₂, Li₂S, and three thin films without Ar etching are shown in Fig. 6. The spectrum for GeS₂ shows a higher binding energy than those of Li₂S and the thin films because the GeS₂ is assigned to the BS as described above. As the Li₂S content increases, the binding energy of the thin films shifts to lower values than that of GeS₂. While the binding energies of the thin films are similar to one another, with the S peak for Li₂GeS₃ being broader than that for Li₄GeS₄ and Li₆GeS₅ 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 Li₂S shows the lowest binding energy of the materials studied here.

Fig. 6 Deconvoluted S2p core peaks of the Li₂S, GeS₂, Li₂GeS₃, Li₄GeS₄ and Li₆GeS₅ thin films.

Experimental

GeS₂ starting material preparation

Glassy GeS₂ was prepared by reacting stoichiometric amounts of germanium metal powder (Alfa, 99.999%) and sulfur (Alfa, 99.999%) in an evacuated silica tube at 900 °C. First, a bare silica tube was cleaned with a 2% aqueous ammonium bifluoride, NH₄HF₂, solution. Surface moisture on the inside of the tube was then removed by passing the tube over a natural gas/oxygen flame. Once this moisture was removed, the valve was closed and the tube was transferred to the glovebox where a previously weighed mixture of appropriate amounts of Ge and S were placed into the tube. The sealed tube was removed from the glove box and the sealed with a gas/oxygen torch under vacuum of ∼4 Pa and a liquid nitrogen trap. The sealed tube was then placed into a furnace held at an angle of ∼5° where it was heated at a rate of 1 °C min⁻¹. from room temperature to 900 °C. The tube was rotated at ∼7 rpm to promote mixing and reaction of the components. After the tube was held at 900 °C for 16 h, it was air quenched to room temperature. The air quenched GeS₂ material inside the silica tube was put into the glovebox and silica tube was broken to remove the GeS₂ glassy product. The final product from this process was a homogeneous, transparent yellow colored glass.

Preparation of the target materials

To prepare the Li₂GeS₃, Li₄GeS₄, and Li₆GeS₅ target materials, stoichiometric amounts of Li₂S (Alfa, 99.9%) and GeS₂ were used. These powders were vibratory Spex milled for ∼15 min inside a steel container with one steel grinding ball inside the glovebox to ensure good mixing and to start the reaction process. Batches of 3 to 4 grams were melted in a covered vitreous carbon crucible at 950 °C for ∼15 min inside a mullite tube lined muffle furnace attached hermetically to the side of the glovebox. The melted charge was poured out of the crucible onto a brass plate and allowed to cool to room temperature. Once enough material, ∼15 g, was prepared, the material was vibratory Spex milled again for 15 min to make fine powder. Fifteen grams of the powder was weighed out and transferred into a hardened steel 2′′ diameter die set. The top press was placed into the die set and allowed to settle on top of the powder. The die set assembly was placed on a sheet of 1/4′′ rubber between the press and the die set and loaded and held at ∼60 MPa overnight. The target was attached to a 2′′ diameter and 0.125′′ thick copper backing plate using silver paste so that it could be water cooled during the RF deposition process to prevent overheating.

Substrate preparation

Single crystal silicon substrates, 1 cm × 1 cm × 360 μm, cut and polished along the [111] plane were put into a “piranha” solution (H₂SO₄ : H₂O₂ = 3 : 1) for 1 h, ultrasonically cleaned with acetone for 20 min, and then rinsed with deionized water for 20 min. The samples were dried with high purity N₂ gas. The samples were put in a drying oven at 120 °C overnight to dry. The cleaned silicon substrates were loaded into a d.c. sputtering chamber in the glovebox and a Ni adhesion layer of ∼120 nm thickness was grown over ∼40 min at ∼3 nm min⁻¹ on the surface of the silicon substrates. The Ni/Si substrates were then loaded into the RF magnetron sputtering chamber on which the thin films were grown.

Thin-film depositions

In order to produce the desired thin films, the target assembly was attached to the sputtering gun inside a high quality (<5 ppm O₂ and H₂O) glovebox and the sputtering gun assembly was then attached to the sputtering system by removing it from the glovebox and quickly attaching it to the deposition chamber to minimize oxidation of the target. The deposition chamber was then evacuated to ∼9 × 10⁻⁵ Pa during which the deposition chamber was heated for two days using tape heaters to drive off any remaining surface physisorbed water in the system. Substrates were then introduced into the deposition chamber through the glovebox via a sliding track load-lock system. A “sputtering up” configuration was used where the target and sputtering gun were placed below the substrate. In this way, if particles fractured off the target during sputtering they would remain on the target and not contaminate the substrate. In this configuration, it was found that these ceramics targets would crack during extended sputtering, but they could continue to be used for sputtering.

Fig. 7 shows a schematic side view diagram of the sputtering system which is connected to the N₂ filled glovebox. After pump down and bake out, the sputtering chamber was filled with high purity Ar gas (99.9999%) to ∼13 Pa and the plasma was lit by turning on the RF power to the target. Pre-sputtering was performed for ∼30 min on the target with the substrate shielded in order to clean the surface of the target of any impurities from the brief exposure to air during assembly. After pre-sputtering, all films were produced with a power level of 50 W and a ∼3.3 Pa dynamic pressure of the Argon sputtering gas. Lower pressures and lower powers were found to produce Li deficient films, whereas higher powers were found to quickly and severely degrade the sputtering target.

Fig. 7 A schematic side view diagram of the sputtering system which is connected to the N₂ filled glovebox.

XPS analysis

X-ray photoelectron core level spectra were obtained using a Scienta ESCA-300 spectrometer with monochromatic Al-K_α (1486.6 eV) radiation. Photoelectron spectra were determined at a take-off angle of 45°. The pressure inside the chamber was ∼6.7 × 10⁻¹⁰ Pa at room temperature. The binding energy scale was calibrated by the C1s peak position (284.5 eV). However, after Ar etching treatment was performed, the C1s peak disappeared. Therefore, corrected peaks after Ar etching were obtained by shifting the energies by the same amount as before Ar etching. The XPS intensity scale and sensitivity factors were calibrated by using standard materials. For the sensitivity factors of the Li, Ge, O and S, Li₂O, LiF, Li₂SO₄, GeO₂, and GeS₂ were used. The starting materials, glassy GeS₂ and crystalline Li₂S powder, were characterized by XPS to calibrate Li, Ge, and S and to determine the chemical shift limits for the bridging (purely covalent sulfur), ≡Ge–S–Ge≡, and the purely ionic sulfur, Li⁺S²⁻Li⁺. In addition, XPS was performed on the bulk pressed Li₂GeS₃, Li₄GeS₄, and Li₆GeS₅ target materials, and thin films Ar sputtered from these target materials in order to determine the composition as well as the nature of the chemical bonding (chemical speciation) for each element. The target material, in a compacted powder pellet, was fractured and scraped in the glovebox before loading it into the XPS chamber so as to provide a fresh, nearly uncontaminated surface. After being sputtered, the thin films on Ni/Si substrates were moved directly into the XPS chamber in tightly closed XPS sample containers. In order to get more accurate compositional data, Ar etching for 1 min and 5 min at ∼1 nm min⁻¹etching rate were performed to remove unavoidable residual contaminants on the surface.

Conclusions

The purpose of this 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 nLi₂S + GeS₂ materials were determined by XPS. The experiments were done with care to prevent contamination and the RF sputtering conditions were optimized to produce thin films whose compositions agreed quite well with those of the targets. Although the as-prepared thin films contained C and O on the surface, the thin films after Ar etching for 1 min showed that the C was completely removed and O content decreased significantly. 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 thio-germanate thin films were successfully prepared and the compositional data and the chemical shifts were carefully characterized by XPS.

Acknowledgements

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.

References

1. C. K. Dyer, J. Power Sources, 2002, 106, 31.
2. S. Prakash, W. E. Mustain and P. A. Kohl, J. Power Sources, 2009, 189, 1184.
3. N. J. Dudney, Mater. Sci. Eng., B, 2005, 116, 245.
4. J. M. Tarascon and M. Armand, Nature, 2001, 414, 359.
5. J. L. Souquet and M. Duclot, Solid State Ionics, 2002, 148, 375.
6. J. N. Harb, R. M. LaFollette, R. H. Selfridge and L. L. Howell, J. Power Sources, 2002, 104, 46.
7. M. Dolle, L. Sannier, B. Beaudoin, M. Trentin and J. M. Tarascon, Electrochem. Solid-State Lett., 2002, 5, A286.
8. X. Yu, J. B. Bates, J. G. E. Jellison and F. X. Hart, J. Electrochem. Soc., 1997, 144, 524.
9. J. B. Bates, N. J. Dudney, B. J. Neudecker, F. X. Hart, H. P. Jun and S. A. Hackney, J. Electrochem. Soc., 2000, 147, 59.
10. S. Kondo, K. Takada and Y. Yamamura, Solid State Ionics, 1992, 53–56, 1183.
11. F. Mizuno, A. Hayashi, K. Tadanaga and M. Tatsumisago, Adv. Mater., 2005, 17, 918.
12. H. Morimoto, H. Yamashita, M. Tatsumisago and T. Minami, J. Am. Ceram. Soc., 1999, 82, 1352.
13. E. Robinel, B. Carette and M. Ribes, J. Non-Cryst. Solids, 1983, 57, 49.
14. Y. Kim, J. Saienga and S. W. Martin, J. Phys. Chem. B, 2006, 110, 16318.
15. J. Saienga and S. W. Martin, J. Non-Cryst. Solids, 2008, 354, 1475.
16. W. L. Yao and S. W. Martin, Solid State Ionics, 2008, 178, 1777.
17. N. Aotani, K. Iwamoto, K. Takada and S. Kondo, Solid State Ionics, 1994, 68, 35.
18. J. H. Kennedy and Y. Yang, J. Solid State Chem., 1987, 69, 252.
19. R. Mercier, J.-P. Malugani, B. Fahys and G. Robert, Solid State Ionics, 1981, 5, 663.
20. A. Pradel, T. Pagnier and M. Ribes, Solid State Ionics, 1985, 17, 147.
21. J. L. Souquet, E. Robinel, B. Barrau and M. Ribes, Solid State Ionics, 1981, 3–4, 317.
22. R. Creus, J. Sarradin, R. Astier, A. Pradel and M. Ribes, Mater. Sci. Eng., B, 1989, 3, 109.
23. M. Yamashita, H. Yamanaka and H. Wakabayashi, Solid State Ionics, 1996, 89, 299.
24. D. Foix, D. Gonbeau, G. Taillades, A. Pradel and M. Ribes, Solid State Sci., 2001, 3, 235.
25. A. B. Volynsky, A. Y. Stakheev, N. S. Telegina, V. G. Senin, L. M. Kustov and R. Wennrich, Spectrochim. Acta, Part B, 2001, 56, 1387.
26. D. Foix, D. Gonbeau, D. Granier, A. Pradel and M. Ribes, Solid State Ionics, 2002, 154, 1613.
27. V. Yufit, M. Nathan, D. Golodnitsky and E. Peled, J. Power Sources, 2003, 122, 169.
28. M. Z. Atashbar, H. T. Sun, B. Gong, W. Wlodarski and R. Lamb, Thin Solid Films, 1998, 326, 238.
29. D. Gonbeau, V. Pamukchieva, R. Dedryvere, E. Skordeva and D. Arsova, J. Optoelectron. Adv. Mater., 2005, 7, 341.
30. M. Mitkova, M. N. Kozicki, H. C. Kim and T. L. Alford, Thin Solid Films, 2004, 449, 248.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c1ra00165e

---