Composition Control of Pd-Cu-Si Metallic Glassy Alloys for Thin Film Hydrogen Sensor

Transcription

1 Materials Transactions, Vol. 51, No. 12 ( pp to 2138 # The Japan Institute of Metals Composition Control of Pd-Cu-Si Metallic Glassy Alloys for Thin Film Hydrogen Sensor Susumu Kajita 1, Shin-ichi Yamaura 2, Hisamichi Kimura 2 and Akihisa Inoue 2 1 Advanced Materials Development Department, Panasonic Electric Works Co., Ltd., Kadoma , Japan 2 Institute for Materials Research, Tohoku University, Sendai , Japan Pd-Cu-Si metallic glassy alloys were investigated as materials for a hydrogen sensor. We prepared thin films of Pd-Cu-Si metallic glassy alloys with varying compositions by RF magnetron sputtering method. And effects of their composition on thermal properties ( : glass transition temperature, : crystallization temperature and response were examined. The response was examined by measuring changes of electric resistance of the thin films exposed in and. Thermal stability depending on thermal properties and large response to hydrogen are key requirements of the hydrogen sensor. and and response were significantly affected by Si content. and became higher and response decreased when Si content was increased. As for metals, Pd and Cu, increased and response decreased with decreasing Pd/Cu atomic ratio among the samples having almost the same Si contents. These results can be explained by a trigonal prism cluster that is reported as a structural unit of Pd-based amorphous alloy. We discussed the effects of a trigonal prism cluster on thermal properties and response, as well as the effects of Si content and Cu content on constructing trigonal prism clusters. [doi:.23/matertrans.m254] (Received July 28, ; Accepted September 6, ; Published October, Keywords: palladium-copper-silicon alloy, metallic glassy alloy, sputtering, thin film, thermal properties, glass transition temperature, crystallization temperature, hydrogen sensor 1. Introduction In recent years, research and development relating to hydrogen energy and fuel cells have been conducted all over the world in order to solve the environmental problems such as a drying up of the fossil fuel and a global warming caused by excess amount of CO 2 emission. Above all, development of fuel cell vehicles is now in the limelight, and two types of hydrogen sensors are required for the fuel cell vehicles. 1 One is a fuel sensor to detect the hydrogen gas in fuel processor output and the other is a safety sensor to detect a hydrogen gas leak. As for the fuel sensor, rapid response and good stability are required in wide concentration range (e.g. 25 % of hydrogen gas containing impurity gases such as O, CO, CH 4 and CO 2. In addition, to detect hydrogen under oxygen-free atmosphere is essential specification, because the sensor is used in hydrogen gas pipes. In order to develop the hydrogen sensor, we adopted a Pd- Cu-Si metallic glassy alloy 2 that is a typical composition of Pd-based metallic glassy alloys for the sensor material. Palladium has excellent properties as a hydrogen sensor material. Pd absorbs about 9 times larger amount of hydrogen than the volume of itself at room temperature. The electric conductivity of Pd decreases as the amount of hydrogen absorbed in Pd increases. 3 Moreover, Pd possesses catalytic activity to dissociate hydrogen molecules to hydrogen atoms, 4 and this facilitates diffusion and dissolution of hydrogen atoms into Pd. A metallic glassy alloy is a kind of amorphous alloys, and has ( : glass transition temperature and wide composition range for forming amorphous phase. A metallic glassy alloy has not grain-boundaries and crystalline defects, resulting in good corrosion resistance and excellent mechanical properties. 5 It also shows no plateau pressure in pressure composition isotherms using hydrogen gas, 6 therefore it does not show hysteresis in the change of resistivity relating to concentration. Thus, the Pd-Cu-Si metallic glassy alloy is expected to show excellent properties for the hydrogen sensor. As we mentioned above, hydrogen sensing ability and stability are key requirements for the hydrogen sensor of fuel cell vehicles. As for hydrogen sensing ability, large response to hydrogen is naturally required for accuracy of the hydrogen sensor. As for stability, we focused on the thermal stability in this study. A metallic glassy alloy is a kind of amorphous alloy and it will crystallize gradually by receiving heat. If amorphous phase is transformed into crystalline phase, its resistivity also changes causing inaccuracy of the hydrogen sensor. Therefore, higher and ( : crystallization temperature are needed to prevent the crystallization of Pd-Cu-Si metallic glassy alloy, because the fuel sensor will be used in temperature of around 373 K. And it is obvious that the composition of Pd-Cu-Si alloys affects these two key requirements. For applying a Pd-Cu-Si metallic glassy alloy to the fuel sensor, it must be fabricated in the form of thin films. Since rapid response to hydrogen is required for the fuel sensor, such as within 1 s for 9% response. 1 Therefore, it is important to shorten the time for hydrogen to diffuse into the alloy and the time for the alloy to be saturated with hydrogen by fabricating the alloy in the form of thin films. In this study, we fabricated Pd-Cu-Si thin films by sputtering method and examined the effect of their composition on thermal properties ( and and response. The results were discussed considering a formation of trigonal prism cluster that is reported as a structural unit of Pd-based amorphous alloys.

2 2134 S. Kajita, S. Yamaura, H. Kimura and A. Inoue Cu plates Pd target Si plates Cu (at % Fig Experimental Photograph of a composite target for Pd-Cu-Si thin films Pd (at % Crystalline Crystalline/Amorphous Amorphous Amorphous w/o with with w/o 2.1 Alloy film preparation The Pd-Cu-Si thin films of varying compositions were deposited on glass substrates using a RF magnetron sputtering equipment (SH-35, ULVAC, Inc. In order to fabricate Pd-Cu-Si alloy thin films, composite targets consisting of a disk-shaped Pd target and square Cu and Si plates were prepared. Cu and Si plates ( 1 mm, mm were placed on the Pd target of 8 mm in diameter. Composition ratios of thin films were controlled by changing the number and the size of Cu and Si plates. An example of this composite target is shown in Fig. 1. The experimental conditions of the film deposition are as follows. Ar-pressure during sputtering was.3 Pa. RF power was W. Deposition times were 2.5 min and 1 min. Film thickness were approximately nm and 48 nm by deposition for 2.5 min and for 1 min, respectively. The films of nm in thickness were used for the examination of response and the films of 48 nm were used for the characterization of the alloys. 2.2 Alloy film characterization The compositions of thin films were analyzed with an electron probe X-ray microanalyzer (EPMA, JXA-8621MX, JEOL Ltd.. The amorphous nature was examined with an X-ray diffractometer (XRD, D8ADVANCE, Bruker AXS using CuK radiation. The thermal properties ( and were measured with a differential scanning calorimeter (DSC, DSC6, SII NanoTechnology Inc.. The samples that indicated were identified as metallic glassy alloys. Ar gas was used for the atmosphere gas and the heating rate was.67 K/s. 2.3 Measurement of the response The response was examined by measuring changes of electric resistance of the thin films exposed in and. A thin film sample with a glass substrate was placed in the Fig. 2 Compositional distribution and phase mapping of Pd-Cu-Si thin films. : glass transition temperature stainless chamber set up in an electric oven. And its electric resistance was measured by 4-probes method using Au plated brass electrodes after exposing the thin film sample in % and % under atmospheric pressure. The hydrogen pressure of % at atmospheric pressure was kpa. The experiment temperature was controlled at 3 K. The response is described as normalized value (R=R where electric resistance (R of the sample in each concentration is divided by electric resistance (R in %. The schematics of the experiment to measure the response were described in a previous report Results and Discussion 3.1 Alloy film preparation and characterization Figure 2 shows the summarized composition map of the thin films indicating amorphous/crystalline phases and the presence of. In this study, the Pd-Cu-Si thin films having different microstructures were fabricated, that is, 1 crystalline phase, 2 composite microstructure with crystalline phase in amorphous matrix with, 3 amorphous phase with and 4 amorphous phase without. 3.2 Effect of composition on thermal properties Figure 3(A and Fig. 4(A show the compositional effect on (K and (K, respectively for the samples indicated in Fig. 2. From these figures, it can be seen that Si content has some effects on and. Figure 3(B and Fig. 4(B show the correlations between and Si content, and Si content, respectively. Both figures show strong positive correlation between thermal properties ( and and Si content. From these results, it can be stated that and

3 Composition Control of Pd-Cu-Si Metallic Glassy Alloys for Thin Film Hydrogen Sensor 2135 (A (A Cu (at % Cu (at % Pd (at % Pd (at % 9 6 < < 6 6 < < < < < < < 7 (B, T / K Fig. 3 (A compositional dependence of and (B correlations between and Si content for the samples indicated in Fig. 2. were significantly affected by Si content: they became higher when Si content was increased. Pd content indicated weak negative correlations with and in Fig. 3(A and Fig. 4(A, respectively. In contrast, the clear correlations of Cu content with and were not recognized in these figures. In order to explain the result concerning the effect of Si content on and, we considered a trigonal prism cluster that is reported as a structural unit of Pd-based amorphous alloys. Figure 5 illustrates a trigonal prism capped with three half-octahedra consisting of 9 transition metals and a metalloid atom. 5 The metalloid atom (Si in the present case is a essential element for constructing a trigonal prism. Therefore, the number density of trigonal prisms in Pd-Cu-Si alloys probably increases with Si content. The trigonal prisms construct a cluster structure and the clusters form random cluster networks. 8 The formation of trigonal prism clusters increases and. Saida has suggested that a formation of trigonal prism clusters provides 25 (B, T / K Fig. 4 (A compositional dependence of and (B correlations between and Si content for the samples indicated in Fig. 2. transition metal metalloid Fig. 5 Schematic illustration of a trigonal prism capped with three halfoctahedra (9 transition metals around a metalloid atom. 5

4 2136 S. Kajita, S. Yamaura, H. Kimura and A. Inoue DSC Heating rate.67 K/s Pd atom Cu atom Si atom Pd 76.5 Cu 8.9 Si 14.6 (Pd/Cu = K Exothermic (arb. unit Pd 76.1 Cu 9.2 Si 14.7 (Pd/Cu = 8.27 Pd 72.8 Cu 12.3 Si 14.9 (Pd/Cu = K 626 K 676 K 631 K Fig. 7 Schematic illustration of trigonal prism cluster capped by a Cu atom. 11 Pd 7.7 Cu 14.3 Si 15. (Pd/Cu = K 634 K 672 K Temperature, T / K Fig. 6 DSC charts of the samples with Si = ca. 15 at% and different Pd/ Cu atomic ratio. good stability of glass structure of metallic glassy alloys, because additional thermal energy is required to deconstruct trigonal prism clusters for glass transition and to rearrange atoms for crystallization. 9 Therefore, an increase in Si content increases the number density of trigonal prisms to increase and. It was found that Si content of Pd-Cu-Si metallic glassy alloys has strong effects on and. And then, we examined the effects of the compositions of two metals, Pd and Cu, on thermal properties. In order to eliminate the effect of Si content, the sample group having almost the same Si content (Si = ca. 15 at% and different Pd/Cu atomic ratios were selected. Their DSC charts are shown in Fig. 6. This figure shows that tends to increase when Pd/Cu ratio is decreased, that is Cu content is increased. In contrast, the clear correlation between and Pd/Cu ratio can not be observed. From these results, it was found that Cu content shows the effect on, but it is not as strong as that of Si content, because it can not be recognized clearly in Fig. 3(A and Fig. 4(A. Chen has measured of Pd-Cu-Si alloys with Si content fixed at 16.5 at% and Cu contents ranging from 2 to 12 at%. In his study, increased with Cu content, and which shows the same tendency as this present result. In order to explain this result, we considered a trigonal prism cluster in the same way as a case of the correlation between the thermal properties ( and and Si content. Nishi et al. have compared viscosity and activation energy of Pd 84 Si 16 with those of Pd 78 Cu 6 Si 16, and obtained the result that Pd 78 Cu 6 Si 16 showed higher viscosity and activation energy than Pd 84 Si And they supposed that the addition of Cu makes the construction of trigonal prism clusters easy, because Cu atoms coordinate as capping-atoms of a Pd-Si trigonal prism, and two Pd atoms and one Cu atom make a new basic plane of a new trigonal prism. The schematic illustration of trigonal prism cluster capped by a Cu atom is shown in Fig Therefore, an increase in Cu content increases the number density of trigonal prisms to increases. Nishi et al. explained the reason why Cu atoms can coordinate at the capping-atom position easier than Pd atoms using the interaction parameters of Pd-Si, Cu-Si and Pd- Cu. 11 The interaction parameters of them take negative values. And this leads to that Pd-Si, Cu-Si and Pd-Cu have stronger interatomic forces of attraction than those of Pd-Pd and Cu-Cu. Therefore, Cu atoms coordinate at the cappingatom position easier than Pd atoms. On the result that the clear correlation between and Pd/Cu ratio was not recognized, we can not explain it at the present time. The further investigation is needed, which is determining the initial crystal of Pd-Cu-Si alloys after heating them to of each composition, because changes depending on the initial crystal. 3.3 Effect of composition on response The % response transients were measured at 3 K, and the results for (A Pd 72:1 Cu 6: Si 21:9 with most high Si content in this study and (B Pd 78:8 Cu 7:2 Si 14: with most high Pd content in this study are shown in Figs. 8(A and (B, respectively. These figures indicate significant difference in the % responses depending on the composition of Pd-Cu-Si alloys. Figure 9 shows the compositional effect on the % response at 3 K for the samples indicated in Fig. 2. The % responses at 3 K are plotted against the

5 Composition Control of Pd-Cu-Si Metallic Glassy Alloys for Thin Film Hydrogen Sensor 2137 (A (B % % % 3 K 5 Time, t / s % % % 3 K 5 Time, t / s Fig. 8 Response transients of (A Pd 72:1 Cu 6: Si 21:9 and (B Pd 78:8 - Cu 7:2 Si 14: to % at 3 K. R: electric resistance of sample in each concentration, R : electric resistance of sample in % (A (B Pd (at % 3 K K Pd (at % R/R < R/R < R/R < R/R < R/R < R/R < R/R < 1.35 Cu (at % Fig. 9 Compositional dependence of the % response (R=R at 3 K for the samples indicated in Fig. 2. Fig. Correlations between % response at 3 K and contents of (A Pd and (B Si. contents of Pd and Si in Figs. (A and (B, respectively. Figure (A shows weak positive correlation between the response and Pd content, and Fig. (B shows negative correlation between the response and Si content. The possible explanation of the effect of Pd content on response is that Pd is a main element which absorbs hydrogen in a Pd-Cu-Si alloy, therefore response increases with Pd content. The effect of Si content on response can be explained by the number density of trigonal prisms. An increase in Si content enhances the growth of the formation of trigonal prism clusters in the Pd-Cu-Si metallic glassy alloy, and also leads to a decrease in the response. The response depends on the amount of hydrogen dissolved in a Pd-Cu-Si metallic glassy alloy. Consequently, the amount of dissolved hydrogen probably decreases with an increase in the number density of trigonal prisms. Additionally, we examined the effect of Pd/Cu ratio on response. The responses of two sample groups are plotted against Pd/Cu atomic ratio in Fig. 11. The unfilled and filled circles indicate the samples with Si = ca. 15 at% and Si = ca. 19 at%, respectively. The positive correlations between the response and Pd/Cu ratio can be seen in both sample groups.

6 2138 S. Kajita, S. Yamaura, H. Kimura and A. Inoue This result admits of two different explanations, and the first one is as follows. The addition of Cu makes the construction of trigonal prism clusters easy, and the amount of hydrogen dissolved in a Pd-Cu-Si metallic glassy alloy decreases with an increase in the number density of trigonal prisms. Therefore, the Pd/Cu ratio showed positive correlation with the response. This means that Cu content shows negative effect on response. The second one is that a decrease in Pd/Cu ratio leads to a decrease in Pd content. As we mentioned above, Pd is a main element that absorb hydrogen in a Pd-Cu-Si alloy. Therefore the response decreased with decreasing in Pd/Cu ratio. Thus, both explanations will be appropriate for the result. 4. Summary Pd / Cu Si content ca.15 at% ca.19 at% Fig. 11 Correlations between % response at 3 K and Pd/Cu atomic ratio. Unfilled and filled circles indicate samples with Si = ca. 15 at% and Si = ca. 19 at%, respectively. In this study, thin films of Pd-Cu-Si metallic glassy alloys with varying compositions were fabricated by RF magnetron sputtering method. The effects of their compositions on thermal properties and response were examined, and they were clearly observed. The obtained results can be explained by the possible correlation between the composition and the number density of trigonal prisms. And it was found that, and response of Pd-Cu-Si metallic glassy alloys can be affected by the number density of trigonal prisms in the alloys. The results obtained in this study are summarized as follows. (1 and were significantly affected by Si content. They became higher when Si content was increased. This result can be explained by the formation of trigonal prism clusters. The formation of trigonal prism clusters in metallic glassy alloys increase and, because additional thermal energy is required to deconstruct trigonal prism clusters for glass transition and to rearrange atoms for crystallization. And the number density of trigonal prisms in a Pd-Cu-Si alloy naturally depends on Si content of it (2 increased with decreasing Pd/Cu ratio, that is increasing Cu content. This result can be explained by the mechanism that Cu makes the construction of trigonal prism clusters easy by coordinating as cappingatoms of a trigonal prism. Cu atoms can coordinate at the capping-atom position easier than Pd atoms. (3 response was affected by Pd content and Si content. Pd content showed positive correlation with the response, and Si content showed negative correlation with it. As for Cu, response decreased with decreasing Pd/Cu ratio, that is increasing Cu content, among the samples having almost the same Si contents. (4 The positive effect of Pd content on response can be explained by that Pd is a main element which absorbs hydrogen in a Pd-Cu-Si alloy. The negative effects of Si content and Cu content on response can be explained by the formation of trigonal prism clusters. Increases in Si content and Cu content enhance the growth of the formation of trigonal prism clusters in the Pd-Cu-Si metallic glassy alloy. And the amount of hydrogen dissolved in a Pd-Cu-Si metallic glassy alloy probably decreases with an increase in the number density of trigonal prisms. The response depends on the amount of hydrogen dissolved in a Pd-Cu-Si metallic glassy alloy. The higher thermal stability, that is higher and, are required for applying the Pd-Cu-Si metallic glassy alloys to the hydrogen sensor of fuel cell vehicles. From this point of view, the composition having high Si content and low Pd/Cu ratio meets the requirement. On the other hand, the composition having high Pd content and low Si content meets the requirement of large response to hydrogen. These conflicting effects of the composition on key requirements for the hydrogen sensor suggest that we can decide the composition of Pd-Cu-Si metallic glassy alloys considering the balance between thermal stability and response. REFERENCES 1 U.S. Department of Energy: Fuel Cells, Technical Plan, Multi-Year Research, Development and Demonstration Plan, (7. 2 A. Inoue, T. Aoki and H. Kimura: Mater. Trans., JIM 38 ( F. A. Lewis: The Palladium Hydrogen System, (Academic Press, London 1967 p M. Okada: Kinzokuzairyo no Kagaku, (The Nikkan Kogyo Shimbun Ltd p A. Inoue: Acta Mater. 48 ( K. Aoki, M. Kamachi and T. Masumoto: J. Non-Cryst. Solids 62 ( S. Kajita, S. Yamaura, H. Kimura, K. Yubuta and A. Inoue: IEE Trans. SM 128 ( T. Takeuchi, D. Fukamaki, H. Miyazaki, K. Soda, M. Hasegawa, H. Sato, U. Mizutani, T. Ito and S. Kimura: Mater. Trans. 48 ( S. Saida: New Functional Materials, Fundamentals of Metallic Glasses and their Applications to Industry, (TECHNOSYSTEM, Japan 9 p. 33. H. S. Chen: Acta Metall. 22 ( Y. Nishi, N. Kayama, S. Kiuchi, K. Suzuki and T. Masumoto: J. Japan Inst. Metals 44 (