Hydrogen diffusion in nanocrystalline PD by means of quasielastic neutron scattering

Hydrogen diffusion in nanocrystalline PD by means of quasielastic neutron scattering

Pergamon NanoStructuredMaterials,Vol.9. pp.579-582.1997 l~lsevierScienceLtd © 1997ActaMetallurgicaInc. PrintedintheUSA.Allrightsreseawed 0965-9773/97...

187KB Sizes 0 Downloads 73 Views

Pergamon

NanoStructuredMaterials,Vol.9. pp.579-582.1997 l~lsevierScienceLtd © 1997ActaMetallurgicaInc. PrintedintheUSA.Allrightsreseawed 0965-9773/97$17.00+ .00 PH S0965-9773(97)00129-3

HYDROGEN DIFFUSION IN NANOCRYSTALLINE PD BY MEANS OF QUASIELASTIC NEUTRON SCATTERING S. Janflen 1,2, H. Natter 1 , R. Hempelmann 1 , T. Striffier 3, U. Stuhr 3, H. Wipf 3, H. Hahn 4, J. C. Cook 5

1physikalische Chemie, Universitat des Saarlandes, Germany 2Lab. for Neutron Scattering, Paul Scherrer Institute, Switzerland 3Institut ftir FestkOrperphysik, TH Darmstadt, Germany 4Institut ffir MaterialwissenschafL TH Darmstadk Germany 5 It.1. Grenoble, France

Abstract--Hydrogen Diffusion in nanocrystalline Palladium has been investigated by means of quasielastic neutron scattering (QENS) in the temperature range 220-300K within the ¢t-phase. The QENS -results reveal broad and narrow line widths with the first ones being significantly broader than those obtained from the coarse grained PdH reference sample indicating the presence of a fast diffusion process within the grain boundaries. The narrow line width seems to emerge from the diffusion within the nanocrystalline grains. A large elastic contribution to the spectra reveals hints on hydrogen traps due to internal surfaces and defects. For comparison of effects emerging from the preparation process samples have been investigated obtained from different preparation techniques, i.e. pulsed electrodeposition (PED) and inert gas condensation (IGC). © 1997 Acta Metallurgica Inc. 1. INTRODUCTION Coarse grained palladium-hydride (PdH) is one of the most extensively studied metalhydrogen systems, see e.g.(l) and references therein. Contrary, the H-diffusion in the respective nanocrystalline metal-hydrid (n-PdH) has been only scarcely investigated so far. Former solubility measurements (2-4) and examinations of the vibrational properties (5) reveal a shift of the solubility limit to higher H-concentrations exhibiting a broader a-region and a ~l-phase with less H-content than the polycrystalline reference material. Furthermore, a strong increase of the macroscopic diffusion coefficient is found. The results are interpreted by the assumption that a large H-fraction is located close to internal surfaces and in grain boundaries, where rapid local diffusion is possible provided that hydogen traps axe sufficiently saturated. Thus H in nano-Pd provides an interesting model system for the investigation of diffusion processes dominated by grain boundary effects.

579

580

S JANSSEN,H NATTER,R HEMPELMANN,T STRIFFLER,U STUHR,H WlPF, H HAHNANDJC COOK

Due to the microscopic properties of the neutron as a probe for condensed matter physics quasielastic neutron scattering (see e.g. (6)) reveals an effective method for the investigations of H-diffusion in metals. The large incoherent cross section of hydrogen allows for the investigation of single particle movements from the Fourier-transform of the self correlation function G(r,t). From the broadening of the elastic line as a function of momentum transfer Q and temperature T microscopic information on jump mechanisms, activation energies, diffusion coefficients, etc. is obtainable. Thus QENS is an important tool to investigate the mechanism of the H-diffusion process also in nanocrystalline materials. 2. EXPERIMENTAL n-Pd samples have been prepared both by pulsed electrodeposition (PED) (7) and inert gas condensation (IGC) (8). Thus effects emerging from the sample preparation could be detected. Due to its high efficiency PED preparation of n-Pd is well suited especially for neutron scattering experiments where 20-30g of material are needed for one sample. After the preparation process the samples have been loaded with 2.9 at-%(IGC) and 3.7 at-%(PED) hydrogen to be still within the a-phase. The samples were then sealed within an Al-container for the scattering experiment. The QENS-experiments have been performed on the backscattering spectrometer IN16 and the multi-chopper Time-of-Flight spectrometer IN5 both at the ILL/Grenoble, France. The respective resolutions (HWHM) have been chosen to be 0.3 geV (IN16) and 30 laeV (IN5), respectively such that a broad dynamic range could be investigated. For both instruments Qvalues between --0.4/~-1 and 1.8 A-1 were accessible. The experiments have been done at temperatures between 220K and 300K. The data have been treated in a standard way, where the instrumental resolutions were measured either by a Vanadium-standard or by the frozen-in sample at low temperatures (-7K). 3. RESULTS AND DISCUSSION Fig. 1 presents a typical QENS result for the PED prepared sample at a temperature T = 300K and a Q-value Q=l.24A "1 obtained from IN5. The included fit (solid line) represents a weighted sum of two Lxwentzians and an additional elastic contribution:

s(e, o )=

(1- ){ Z, )+ O-

h )}

(i)

with the Lorentzian line shapes:

~l,i2 + (hOJ)2

(2)

The 2i denote the respective HWHM's. The elastic part had to be included because hydrogen could not be unloaded from both samples even during a 24h vacuum treatment. As a consequence the scattering from Pd lattice producing an elastic contribution could not be subtracted. As one can see from the statistical behaviour of the residuals below Fig. 1 the fit describes the data well.

HYDROGENDIFFUSIONIN NANOCRYSTALLINEPd BY MEANSOF QUASlELASTICNEUTRONSCATTERING

581

0.8 0.7

PED Q=1.24,~"l T=300K

0.6 ,-~

0.5

2 Lorentzians

0.4 ~

0.3 I!

i

0.2 0.1

. . . . -~. . . . . ?.... ~. . . . . .._.

0

-1

-0.5

0

0.5

1

Fig. 1: QENS-result (IN5) for the PED-prepared sample. Besides the described fit (solid line) the two Lorentzians (both dashed) and the elastic contribution (dotted line) are included. Below the figure the residuals of the fit are visible.

0.12

.

I

T'~V300K

0.1

.

.

,

-

.

.

,

.

.

.

,

.

.

,

-

.

.

.

.

} E. =35meV

0.08

i

.

0.06

i•

0.04

r~

0.02

0

~ .,59meV

Q-1.24 A" '

,

,

,

|

*

0.5

.

.

.

.

,

.

.

I

Q [,~- 1]

.

.

,

1.5

.

.

2

0.G



3.2

.

.

,

3,4

.

.

.

,

3,0

.

.

.

,

.

.

3,0

.

,

4

.

4.2

T'I[IO'3K "I]

Fig. 2: Narrow line width as a function of Q (left part) and Arrhenius representation for both obtained line widths with their respective activation energies (right part).

S JANSSEN,H NAI-rER,R HEMPELMANN,T STRIFFLER,U STUHR,H WIPF, H HAHNAND,JC COOK

582

Contrary, the best fit obtained with only a single Lorentzian showed systematic deviations from the data. The two line widths obtained from the nanocrystalline sample are not measurable for the polycristalline reference sample. They indicate rapid diffusion processes being absent in ordinary PdH systems. In the left part of Fig. 2 the Q-dependence of the narrow line width is presented for Tf300K. Surprisingly, even for small Q the line width shows not the usual Q2-behaviour. Instead a linear increase with Q is found. This result is obtained for both line widths and for all the measured temperatures. Additionally, the line widths are plotted in an Arrhenius representation in the right part of Fig. 2 providing activation energies of 35 and 59 meV, respectively, being 4-7 times smaller than for coarse grained PdH. The results indicate a rapid H-diffusion process within the grain boundaries of the nanocrystalline Pd. The fact that the two activation energies are quite similar gives a hint on the assumption that the local diffusion might not completely be described by two well separated single processes but by a distribution of thermally activated diffusion processes. The relatively large uncertainty in the estimation of the activation energies will be overcome by further experiments in an extended temperature range. On the high resolution back.scattering spectrometer IN16 it was possible to resolve the elastic part observed on IN5 into a quasielastic and another elastic contribution. A fit with a single Lorentzian plus elastic component described the data statistically well. The obtained line width is of the same order as for coarse grained PdH but the T- and Q-dependence is not pronounced enough to state yet that here the H-diffusion within the nanocrystalline grains is visible. The high contribution of the elastic component indicates that a large fraction of hydrogen is trapped and does not diffuse at all. This result might be due to the refined grain size. A lot of hydrogen-traps are formed close to internal surfaces and defects within the grains such that more hydrogen is necessary to saturate these traps than in coarse grained Pd. For both kinds of sample the results are qualitatively the same indicating that the sample history and especially the higher porosity in the case of the IGC prepared sample are not decisive for the H-diffusion in n-PdH. At least for the PED preparation technique it is possible to adjust the nanocrystalline grain size in a sensitive manner. Here QENS-experiments under variation of the grain size are planned. Together with the presented results these experiments should provide the basis for the microscopic understanding of the H-diffusion in a nanocrystalline metal-hydride.

4. REFERENCES

. .

3. 4. 5. . . .

E. V01kl, G. Alefeld in G. Alefeld, E. VOlkl, Hydrogen in Metals I, Springer Tracts in Appl. Phys. 28, Springer, Berlin (1978) T. Miitschele, R. Kirchheim, Script. Metall. 21, 135 (1987). T. Mtttschele, R. Kirchheim, Script. Metall. 21, 1101 (1987). J.A. EaslJnan, L.J. Thompson, B.J. Kestel, Phys. Rev. B 48, 84 (1993). U. Stuhr, H. Wipf, TJ. Udovic, J. WeitmUller, H. Gleiter, J. Phys.: Condensed Matter 2, 219 (1995). M. Bte, Quasielastic Neutron Scattering, Adam Hilger IOP Publ., Bristol and Philadelphia (1988). H. Natter, T. Krajewski, R. Hempelmann, Bet. Bunsenges. Phys. Chem. 100, 55 (1996). H. Gleiter, Progress in Material Science 33, 223 (1989).