The effect of surface impurities on the permeation of hydrogen through palladium

THE EFFECT OF SURFACE IMPURITIES ON THE PERMEATION

OF HYDROGEN THROUGH PALLADIUM

by .

DELFT

Armando Bruno Antoniazzi

THE EFFECT OF SURFACE IMPURITIES ON THE PERMEATION

OF HYDROGEN THROUGH PALLADIUM

by

Armando Bruno Antoniazzi

Submitted July 1988

.,J

ACKHOWLEDGEKENTS

I wou1d 1ike to express my gratitude to my supervisors, Professors A.A. Haasz and P.C. Stangeby, for their suggestion of the topic of research as we11 for their guidance and contributions throughout the work. I a1so wish to thank Professor J.B. French and Dr. W. Shmayda for serving on my Thesis Committee and for he1pful discussions.

Many thanks to our engineering techno10gist, Mr. C. Perez, for the continual aid and friendship he has given me and to our 1ibrary technician, Ms. P. Cooke, for help through the years in obtaining reference material as weil as partaking in interesting conversations.

The members of our group, past and present, are grateful1y acknowledged for their help, friendship and many hours of fascinating discussions.

The support staff at UTIAS is gratefu11y acknowledged for all the aid given to me over the years as weil as for the kindness displayed.

1 would like to extend my gratitude to Mr.

W.

Park and

Mr.

G. Fitzgera1d for the development and modifications, respective1y, to the computer fitting program FLUX.

This research was supported by the Canadian Fusion Fue1s Technology Program, the Natura1 Sciences and Engineering Research Council of Canada and the Ontario Ministry of Energy. Appreciation goes to the Institute for Aerospace Studies for scho1astic aid and financial support.

Last1y, but most important, I wish to express my deepest gratitude and love to my wife, paula; as we11 as, my parents and family for the

understanding, patience and aid given to me throughout this and preceding university education.

..

ABSTRACT

The effect of surface impurities (in particular carbon and sulphur) on the permeation of atomie or molecular hydrogen through a palladium membrane was examined. The elemental coverages on one of the membrane surfaces was measured by in-situ Auger spectroscopy. The upstream, driving, pressure for

-5 2

molecular permeation ranged from ~10 to 3x10 Pa. For atomie permeation

15 20 0 2

the incident sub-eV atomie flux varied from -1x10 to 3x10 H /m .s. The palladium temperature ranged from 355 to 630K. Atomie hydrogen was used to measure the membrane asymmetry in the surface limited permeation regime. The phenomenological recombination rate coefficient, K , and consequently

r

.

-4

the permeation rates were observed to decrease by a factor of ~10 for an increase in the tmpurity coverage from 0.13 to 0.45. The change observed is compatible with a mechanism which views the reduction in the permeation rate as resulting from a blockage of surface adsorption sites. Molecular

-6 -2

sticking coefficients measured at 625K ranged from 3.2x10 to 7.0x10 depending on tmpurity coverage. Observed permeabilities in the diffusion ltmited regime were found to be in agreement with literature values •

.

,.

TABU OF CONTBNTS Acknowledgements Abstract List of Appendices List of Tables List of Figures List of Symbols

I.

INTRODUCTION 11. REVIEW 2.1 Historical Introduction 2.2 Hydrogen-Metal Interaction 2.2.1 Terminology

2.2.2 Mechanisms of Hydrogen Sorption 2.2.3 Desorption

2.3 Palladium

2.3.1 Hydrogen in the Bulk 2.3.2 Hydrogen Adsorption

2.3.2a Energies of Adsorption 2.3.2b Sticking Ooefficients 2.3.2c Nature of Adsorption 2.3.3 Hydrogen Permeation 2.4 Impuri ties 2.5 Fusion I I I. THEORY

3.1 Diffusion Limited Permeation 3.2 Mechanisms of Hydrogen Transport

3.3 Impurity Effects on the H-M Interaction: stickingcoefficient

3.3.1 Form of the molecular sticking coefficient

3.4 General Equation for permeation

3.4.1 Model of Waelbroeck and co-workers IV. INSTRUKENTATION AND BQUIPMENT

4.1 The Vacuum System

4.1.1 High Vacuum Chambers 4.1.2 Pumps

4.2 System Bakeout

4.3 Membrane Support and Heating Stage 4.4 Electronic Instrumentation

4.4.1 Ionization Gauges 4.4.2 Auger System

4.4.3 Quadrupole Mass Spectrometers 4.4.4 Pirani Gauges

4.4.5 Spinning Rotor Gauge

Page ii

iii

v~

vi

vii

~x 1 5 6 6 7 9 9 9 11 11 14 15 22 24 28 30 32 36

37

38 43 50 50 50 51 51 52 52 52 52 53 53

4.4.6 Computers and Data Acquisition 4.4.7 Thermocouples

4.5 Palladium Samples 4.6 Gas Handling System

4.6.1 Calibrated Gas Leak 4.7 production of Atomie Hydrogen

V.

PROCEDURES AND TECHNIQUES

5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 Sample Preparation General UHV Techniques Permeation Rate

Calibration of Equipment 5.4.1 Pumping Speeds 5.4.2 Ionization Gauges

5.4.3 Quadrupole Mass Spectrometers Auger Spectra

Atomie Hydrogen Production FLUX Fitting Program

Corrections to the Incident Flux Estimates of Errors

5.9.1 HO produetion

5.9.2 Permeation Rate (~p) and Recombination 5.9.3 Permeability (DS)

VI. DISCUSSION OF DESIGN AND EXPERIKENTAL PROBLEMS

6.1 Heating of the Membrane

6.2 Produetion of HO

6.3 Translational Movement of OMA and HO-filaments 6.4 Techniques for Measuring ~

VI I.

RESULTS

53 54 54 54 55 55 56 56 57 57 57 .59 60 61 63 64 64 65 65 Coefficient (K ) 67 r 67 68 69 70 71 7.1 Reading of Appendix E 74 7.2 Surfaee Contaminants 77

7.2.1 Exposure to H&: effect on carbon levels 78

7 .2.2 Exposure to H 80

7.2.2a Removal of Carbon by Atomie Hydrogen 80 7.2.2b HO: probe for surface changes 81 7.2.3 Exposure to 02: sulphur removal 83 7.2.4 Effect of Heating Pd to 1100K 84

7.3 H₂ permea tion 85

7.3.1 Surfaee Limited permeation 85 7.3.2 Diffusion Limited Permeation 87

VIII. DISCUSSION

8.1 Bulk Effects 89

8.1.1 Grain Boundary Diffusion 89

8.1.2 Hydride Formation 89

8.1.3 Carbon Solution in Pd 90

8.2 Permeation Rates 91

8.2.2 Comparison to the Results of Davis 94 8.2.3 Comparison to the Results of Auer and Grabke 95

8.3 Effect of Impurities 96

8.4 Enhanced Segregation of Carbon in Palladium 102

8.5 Removal of Surface Carbon 103

8.6 Scatter in Ky Values 104

8.7 Diffusion Limited Permeation 104

8.8 Relevanee to Fusion 105

IX. CONCLUSIONS AND UCOMMENDATIONS 107

9.1 Contributions 109

REFERENCES 111

List of Appendices

Appendix A: Correction due to Non-thermal Equilibrium Appendix B: Listing of Atomie Flux Program

Appendix C: Listing of FLUX Program

Appendix D: Connection of Model of Waelbroeck et al to Aller and Grabke' s model

Appendix E: Membrane History and Surface Composi tion Appendix F: States of Surface One

Appendix G: Variations of the Permeation M:>del

List Of Tables

Table 1. Proper ties of Palladium

Table 2. Hydrogen Solubility and Diffusivity in Palladium Table 3. Limits of a-phase

Table 4. Bulk Metallic Impurities Table 5. Sensitivity Factors

List Of Figures

Figure 1. The Tokamak configuration

Figure 2. Pressure-composition isotherms for Pd(H) Figure 3a. Crystalline structure of Pd

Figure 3b. Pd(lll) surface Figure 4. H

2 adsorbed on Pt(lll) Figure 5. H

2 desorption from a Pd(100) surface Figure 6. Potential energy diagram for palladium

Figure 7. Variation of initia! sticking coefficeint with impurity coverage

Figure 8a. Fluxes at gas-surface, surface-bulk interfaces Figure 8b. Rates for model of Waelbroeck et al

Figure 9. Plot of log (u2) vs log (W2)

Figure 10. Schema tic of the Hydrogen Permeation Facility Figure 11. Photograph of the Hydrogen permeation Facility Figure 12a. Photograph of present heating stage

Figure 12b. Schema tic of original heating stage Figure 13. Photograph of the membrane

Figure 14. Pumping speed as a function of pressure Figure 15. Calibration curves of ig₁, ig₂ vs. SRG

Figure 16. Quadrupole output (QM2) versus lonization gauge (IG2) Figure 17a. Quadrupole (QM1) output versus IGl

Multiplier Mode Figure 17b. Faraday Plate Mode

Figure 18a. Auger spectra af ter initial bakeout Figure 18b. Auger spectra af ter carbon removed Figure 19. Series I.a permeation curves

Figure 22. Series I.c permeation curves

Figure 23. Permeation curves I.d, I.e, and I.f Figure 24. Series I.h permeation curves

Figure 25. Series I.i permeation curves Figure 26. Permeation curves I.j and I.k

Figure 27. Permeation curves 1.1, 1.1' and I.m Figure 28. Permeation curves

Figure 29. Permeation curve I.u

Figure 30. Permeation curves I.a (600K) compared to Davis Figure 31. The variation in the recombination coefficient (K

r2) with temperature

Figure 32. The variation in the permeation probability with temperature Figure 33. Molecular sticking coefficient results of Dean et al

Figure 34a. Relative change in Kr

1 versus total impurity coverage Figure 34b. Relative change in K

r1 versus impurity coverage (carbon, sulphur) Figure 35. Figure 36. Relative change in Kr 1 versus (1-01) o

List of Syabols

Subscripts

u,d - upstream or downstream variables respectively

1,2 - chamber one or chamber two variables respectively i,j - chamber i or j variables respectively

Superscripts

A - variables for surface state A on surface one

r - variables for reference, initial, state of surface two

Symbols

- tungsten filament area

- membrane area (m2)

- orifice area (m2)

c

NbkBDI J.

- bulk hydrogen concentration (H/m3)

c

e - equilibrium bulk concentration (H/m

3₎

<c ) - mean gas particle speed, upstream chamber

(mis)

u c

i c

f

- initial bulk concentration - final bulk concentration

C

=

2(rlO)uKAB Cl - chamber one C2 - chamber two d - Auger scale i D - diffusivity Do - prefactor of E = 2(rlO)dKAB

factor for element i

(m2j s)

2

D (m Is)

(all energies are in kJ/mol H unless otherwise stated in the text) E

a - activation energy for dissociative adsorption

E - enthalpy of adsorption

A

E - activation energy for diffusion

D

E - activation energy

k for recombinative release

E - activation energy for diffusion limited permeation (= E + E )

P D s

.

..,

E - enthalphy of solution s

I - peak to peak Auger signal for element i i

k - rate constant for hydrogenation or - Boltzmann's constant 1.38x10-23 JK-1 k' - rate constant for dehydrogenation

kA - equilibriun constant for the reaction H(A) + H(bulk)

..

kB - equilibriwn constant for the reaction H(B) + H(bulk)

..

k 4S

=

r4KAB

+

rS

K - equilibriwn constant for adsorption (Chap. 11, Appendix D) - Kr

2: recombination coefficient for surface two in initial, reference,

r 4 2

state (H

2.m IH .s)

- equilibrium constant for the reaction H(bulk) +

..

HA (Cha p • II) : H

B (Chap. II)

- equilibriwn constant for the reaction H(bulk) - kAk

B (Chap. 11)

4 2 Kr - recombination coefficient for desorption (H

2.m IH .s) <K

> -

effective recombination coefficient

r r 3 r 4 rS r 6 2K K ru rd K

+

K ru rd - membrane thickness -3 - gas density in upstream chamber (m )

28 .

I

3 - density of bulk palladiun (6.8x10 atoms m ) - pressure (Pascals)

1/2

- permeability (H/m.s.Pa ) (~ DS) prefactor for P

- flow rate through calibration orifice

- fraction of molecules adsorbing without overcoming an activation energy barrier

(the rate constants r

3+r8 to followare in mIs, while r9,10 are in 2

H/m .s), adjacent to the particular rate constant is listed the reaction to which it applies)

- H(C) + H(bulk) - H(A) + H(bulk) - H(B) + H(bulk) - H(bulk) + H(C)

r 7 r 8 - H(bulk) + H(A) r 9

riO

r 78

<riO>

R S S o Sp Si T T g T w t t ss u

v

w

X - H(bulk) + H( B) - 2H(C) + H (gas) 2 - H(A)

+

H(B) + H (gas) 2 r + r 7 8 = 2(r lO)u(r10)d (rlO)u + (rlO)d - effective ra te constant - gas constant (8.3143 J/K.mol)

- solubility of hydrogen in the metal ( Hlm3.pa l/2) - prefactor of S

pumping speed (m3/s)

- Auger sensitivity for element i - membrane temperature

- mean gas temperature - mean wall temperature - time

- time to achieve steady state conditions for diffusion limited permeation

(~12/D)

reduced downstream concentration - chamber volume (m3)

- permeation number

- fractional absorbed hydrogen concentration (= e/N b) Greek Alphabet

a - bulk phase of palladium (low concentration)

- weak adsorption states for hydrogen (which applies elear from the

'context)

a - atomic sticking coefficient a

a - moleeular sticking coefficient m

. a - prefactor for a

mo m

a

m1 - moleeular stieking eoeffieient for dissoeiative adsorption H/gas) + 2H( C)

a

m2 - moleeular stieking eoeffieient for reaetion H2(gas) + H(A) + H(B) aai - atomie stieking eoefficient for surfaee in state i

. '"

aaA - atomie sticking coefficient for surface in state A ~ I' ~ ,~ - strong adsorption states for hydrogen

2 3 2

r

-

incident molecular flux (H₂/ m .s)

~i - asymmetry ratio (upstream recombination coefficient in state i r

divided by K (z K ri

»

~A - asymmetry ratio (K IK) (upstream recombination coefficient in ru

state A divided by K.

9

H - fractiona1-hydrogen-surface coverage (superscripts (A, B, C, etc.)

refer to the coverage for adsorption sites A, B, C, etc. 9

1 - total impurity fractiona1-surface coverage 9 - impurity coverage by su1phur

s

9

0 - impurity coverage by oxygen 9

c - impurity coverage by carbon

~

=

I.S6xI024/IT (H 2/m 2 .s.pa) g 2 VI v 2 v 3 v 4 V

s

v 6 V 7 va V 9 vlO v~ l;

=

~~ ~ _a ~D (all rates vl+v

lO are in ~/m .s, the reactions to which theyapp1y are given to the right of the particu1ar rate)

- H (gas) + 2H(C) 2 - H (gas) + H(A) + H(B) 2 - H(C) + H(bu1k) - H(A) + H(bu1k) - H( B) + H(bu1k) - H(bu1k) + H(C) - H(bu1k) + H(A) - H(bulk) + H(B) - 2H(C) + H 2(gas) - H(A) + H(B) + H 2(gas) o - rate of H production

<rIO>/(rIO)d

=

<Kr>/Krd - asymmetry factor - work function change

_ atomie incident flux (Ho

Im

2 .s) (subscrits i, A refer to an incident

o

H flux onto a surface in a general state i or a particu1ar state A) - diffusing flux (H/m2.s)

2

- molecular incident flux (H

2/m .s)

2

_ permeating flux (H/m .s) (superscripts T, R, and m refer to total, atomie or molecular permeating flux, (i+j) refers to direction of

permeation chamber i to chamber j)

2

- recycled flux (H/m .s)

2

_ flux penetrating the surface (H/m .s) for an incident Ri flux _ flux penetrating the surface (R/m2.s) for an incident H flux - permeation probability (= ~ /~ )

p ~

_ permeation probability (= ~ /~ ), subscripts i, A indicate whether p a

the upstream surface is in the general state i or in the particular state A)

. ."

I.

lRTllODUCTIOH

Gas transport fram a high pressure region to a low pressure region via asolid membrane represents the permeation process. The permeation of hydrogen through metals has, for the last hundred years, attracted numerous investigators, both theoretical and experimental. Interest in the hydrogen metal (H-M) systems continues to increase. The reasons range fram a basic desire to understand the fundamentals of the interaction of the simplest atom (hydrogen) with the metal lattice to more practical uses of the H-M

1

systems • Hydrogen in metals can also lead to many structural problems as it supports the development of cracks, formation of gas bubbles and

blisters, the reduction of plasticity and is a cause of embrittlement. Hence, any parts operating in a hydrogen media can expect to encounter difficulties.

One area of application which bas an intense interest in the H-M interaction is that of fusion. !Wo nuclei with sufficient momentum to overcome their mutually repulsive potential will 'fuse' into a heavier nucleus; this represents the fusion reaction. Man has over the last three decades attempted to barness the fusion process as source of inexhaustible

2

energy. The fusion reaction with the highest power density and the lowest 3

igni tion temperature requirement. is ,

2 3 4 1

D + T ~ He(3.5 Mev)

+

n(14.1 Mev) [1.1)

Most of the energy is carried away by a highly energetic neutron (n). 0 and T represent the hydrogen isotopes deuterium and tritium respectively.

Tritium is radioactive with a half-life of 12.3 yrs. Achievement of energy production from the fusion process requires plasma temperatures of ~100

million K, and a plasma density and confinement time product of greater than -3

cm .s. This constraint is referred to as the Lawson criterion. Several concepts using magnetic fields for the confinement of the

3,4

fusion plasma exist • In the tokamak concept (see figure 1) the magnetic fields twist around the torus' axis. The confinement of the plasma however is not ideal and leakage to the vacuum vessel wall (the first wall) occurs. The first wall is exposed to a variety of species and conditions:

mechanical and electromagnetic loading; alternating thermal stresses; irradiation by 14 MeV neutrons (altering material bulk properties); bombardment by ions, energetic neutrals, photons, runaway electrons and arcing. It has been recognized that the interaction of hydrogen and its

5 6 isotopes with the first wall is an important technological problem ' • Knowledge of the factors affecting hydrogen (or D,T ) permeation, recycling

and inventory by the first wall is required. In werking devices no surface is completely clean and therefore an understanding as to how the hydrogen metal interaction behaves in the low vacuum region (present in the fusion vacuum vessel) in the presence of surface wall impurities is essential.

It is well known that molecular hydrogen in the high pressure regime,

"

atmospheric or greater, permeates through metals as a function of the square

"

1

root of the driving pressure • Experimentalists have observed deviations 8 9

from this square root dependance in the low vacuum reg ion ' These variations in power dependances have led to the realization that the metallic surface can play a significant role in the achievement of

steady-state permeation. Many models have, over the years, been proposed to 6-17

explain the transport of hydrogen across the surf ace interface •

Experimental attempts at measuring the recombination coefficient for hydrogen on the material surface have not been very successful. Orders of

'

..

'

5

steels • The main reason given for the severe discrepancies is that of different sample treatments, hence different surface conditions.

Surface conditioning is also suspected to be the reason for varying results in permeation studies of

~

through palladium (Pd)18. Up to the present, few researchers have conducted permeation studies within ultra-high vacuum systems and fewer still have used modern methods of surf ace analysis (eg. AES, LEED, etc.) 13 • There have been a handful of studies examining, systematically, the effects of surface impurities on hydrogen permeation.

Mbtivated by the situation described above, the following set of objectives were defined for the thesis:

a) to design and build a hydrogen permeation facility wi th in-situ surface analysis of impurities,

b) to observe, experimentally, the effects of surface ~purities on the permeation of hydrogen,

c) to provide data for fusion technology,

d) to correlate the experimental results with available theoretical modeis.

The metal selection for the permeation study was based on the following criteria:

a) a pure element was desirabie in order to keep the hydrogen-impurity-metal interaction as simple as possible,

b) a high throughput for hydrogen was desirabie to provide a clear signal above the background,

c) it had to be reasonably weil understood and investigated with respect to its bulk properties of diffusivity and solubility. d) it had to be under consideration for use or be in present use in

Palladium (Pd), a transition metal, met the criteria above. The H-Pd system 1

is one of the most extensively studied , although the hydrogen-palladium

18

surf ace interaction is not weIl understood • Palladium, Pd-alloys or Pd coated metals are being considered as pumps in vacuum technology and fusion research (eg. to separate H-isotopes from the fusion exhaust 19-23 , to extract tritium from the tritium producing lithium or lithium lead alloy blanket24).

Chapter 11 will present a review of the hydrogen-metal interaction with emphasis on the hydrogen-palladium system. Chapter 111 follows with a

presentation of theoretical models used to describe the permeation process. The instrumentation and equipment used toobtain the results are presented and described in chaper IV. Chapter V provides a description of the sample preparation and experimental techniques used. Chapter VI briefly presents and discusses design and experimental problems encountered and solved. The results obtained during this investigation and the analysis of these results are presented in chapter VII. A discussion of the results follows in

chapter VIII. Chapter IX culminates with the presentation of the conclusions, recommendations and the author's contributions.

" .

.-..

I l . REVIEW

2.1 Historical Introduction

The first reported studies of hydrogen permeation through metals was by Deville and Troost (1863)25 and Deville (1864)26. They examined the metals iron and platinum. Palladium was investigated soon af ter by Thomas Graham (1866)27. Graham not only observed the high throughput of hydrogen by

palladium but that it, as he termed it, occluded (absorbed) large quantities of the gas. From this time on, the hydrogen-palladium system has been the

1 18 28 most extensively investigated' , •

The relationship describing the steady-state permeation of hydrogen in the diffusion limited regime was first derived by O.W. Richardson

(1904)29,30 and has since become known as Richardson's equation (see Theory chapter). In essence, the permeability is proportional to the square root of the driving pressure and inversely proportional to the sample thickness.

The metal surface was suspected quite early on as being capable of 31

affecting the permeation when Holt et al. (1913) attempted to explain anomolies in their data as the effect of adsorption. Holt later abandoned

32 this argument

It was not until the advent of modern surface analysis techniques in the early 1960's that the surfaces of the metals could be examined and studies could be performed on well defined surfaces (crystallographic orientation as well as the nature of surface impurities). Even though the techniques have now been available for two decades most of the work

conducted on adsorption-desorption behaviour on the H-Pd system bas been 18

performed on metals with poor surface characterization To distinguish the surface behaviour from the b~lk, most studies are conducted at very low

temperatures reducing the diffusion of H into the bulk. Interpretation becomes difficult if absorption into the bulk is not negligible,

particularly for a material like palladium which is an exothermic occluder of hydrogen (ie. has a large absorption capacity).

Most permeation studies have still not employed ultra-high vacuum

techniques and a significantly small number have attempted to use an in-situ surface analysis technique to complement their permeation experiments. As a result of a lack of proper (characterized) surface treatment, there is still speculation as to how surface impurities (type, quantity) affect the H-M interaction. Surface impurities are, however, recognized as affecting the H-M interaction, usually hindering uptake. This has resulted in various

6-17

models which attempt to provide a clear understanding of the hydrogen metal interaction. The section to follow will briefly review the

terminology used in studies of gas-metal interaction af ter which the mechanisms of hydrogen uptake by the metal will be presented.

2.2 Hydrogen-Metal Interaction

2.2.1 Terminology

Atoms in metals are surrounded by neighbouring atoms. The surface layer of atoms represents a sudden, abrupt, end to the crystal structure. The surface atoms with no neighbours in one direction have incomplete bonds, bonds which dangle into space. These atoms provide an attractive force normal to the plane of the surface. Gas atoms which adhere to the surface are referred to as adsorbates and the solid matrix as the adsorbent.

Adsorption is distinguished by being either physical or chemical in nature. Physical adsorption represents the weak bonding of the gas

Waals forces. Chemical adsorption involves the breaking and forming of new bonds and are characterized by higher enthalpies (energies of adsorption 50-400 kJ/mol) than physisorption. Chemical adsorption is divided into two types: associative and dissociative. In associative adsorption the nature of the adsorbed species is similar to that of the molecule in the gas phase. A molecule which has to dissociate to adsorb as atoms on the solid surf ace is said to be dissociatively adsorbed. The insertion of atoms into the bulk is the absorption process.

Absorption entails the dissolution of the adsorbed species into the solid bulk. Once in the metal, the absorbed atoms may affect the crystal structure,such as, hydrogen forming a hydride by the expansion of the

original crystal lattice. The above processes are generally referred to as sorption.

The reversal of gases adsorbed on the solid surface is the act of molecules leaving the surface known as desorption.

The various mechanisms proposed over theyears in the literature for the sorption of hydrogen in metals will be reviewed in the following section.

2.2.2 Mechanisms of Hydrogen Sorption

Different mechanisms for the hydrogen adsorption and subsequent absorption have been proposed since studies in the hydrogen metal

interaction began, roughly, a hundred years ago. How and where hydrogen bonds onto (into) tbe surface is still an area of discussion and

investigation.

35 36

Taylor (1931) and Lennard-Jones (1932) proposed that H

2 adsorbed onto the metal surface in two states: physisorbed and dissociatively

adsorbed. In 1932, Wagner 33 proposed that hydrogen transfer from the gas phase into the lattice involved two steps: the dissociative chemisorption of H

2 on the surface and the transfer of H atoms from the surface sites into the bulk sites, viz.,

+

m(d)

+ [2.1]

H(ad) + H(Me)

+ [2.2]

where H

2(g) is the gas phase molecule, H(ad) the adsorbed hydrogen atom and H(Me) the absorbed hydrogen atom. Attempting to examine these rate limiting

steps, Wagner observed that the particulars of the rate law depended significantlyon the history of the sample surface.

34

In 1938, Wagner proposed an intermediate step between the adsorbed and absorbed state. Adsorbed atoms, in this step, migrate on the surface until a passage is found for entering the bulk. The activation energy for surface diffusion is believed to be typically one fifth to one seventh of

41

the desorption energy • However, the lifetime of the adatom at a site is

4

~10 times longer than its time of movement so that the adsorbed layer is considered localized.

29 30 Diffusion limited permeation has been described by Richardson ' (1904); it proceeds at a rate proportional to the square root of the

upstream pressure. If transition of the gas to the absorbed state involves the dsorbed state, it would be expected that when the surface is saturated by adsorbed atoms that permeation would saturate. This was, however, not the case for H

2 permeation through nickel (Ni). The Ni surface was expected to saturate at sub-atmospheric pressures, but experiments37 showed that the

·

.

greater than one hundred atmospheres. To deal with this, wang38 (1936) proposed a new mechanism. The mechanism involved the dissociation of the gas molecule into an adatom and an atom dissolved (adsorbed 1) into the subsurface layer, viz.,

+ H(ad) + H(ab)

+ [2.3]

These represent the basic mechanisms proposed for hydrogen sorption by the metal matrix.

2.2.3 Desorption

The desorption process occurs when the adsorbates acquire sufficient energy to leave the chemisorption potential well. The energy in this situation is provided by lattice vibrations of the adsorbent. When the thermal energies of the adsorbed species is larger than the activation energy for desorption, the species returns to the gas phase. Desorption then involves the reverse mechanisms to adsorption as given by equations 2.1 and 2.3.

Having presented the terminology and the mechanisms by which hydrogen enters the bulk, an extensive examination of the hydrogen-palladiun system follows below.

2.3 Palladiua

2.3.1 Hydrogen in the Bulk

Palladium belongs to the class of materials called hydrides. The pressure-composition isotherms were first obtained by Troost and

39 1

Hautefeuille (1874) • Figure 2 presents the pressure-composition isotherm for the hydrogen-palladium system. This figure indicates what pressure is required to give the fractional bulk concentration (no. of absorbed H/no. of Pd atoms) indicated on the abscissa. The solution of H

2 in Pd is

<

non-stoichiometric. For temperatures ~ 573K palladium has two phases, commonly referred to as the a and ~ phase (see the areas marked off in the

figure). The a-phase occurs for low hydrogen pressures (small H-concentrations in the bulk ). At some higher pressure for a given

temperature the nucleation of the ~-phase commences. Palladium in the

~-phase differs from the a-phase structurally by an isotropic expansion of the face centered cubic lattice (see figure 3). From the a state to the

max

1

~. state the lattice constant, a, changes from 3.894A to 4.025A. The m~n

absorbed hydrogen occupies the octrahedral sites in the bulk57,58.

<

Pressure-composition isotherms for temperatures ~ 393K are generally difficult to obtain due to the excessively long time required to attain

equilibrium. Impurity layers on the Pd surface hindering the adsorption of Hare viewed as the reason. Pretreatment or addition of a dissociating

2

catalyst (such as Pd black (PdO, Pd0

2

»

to the surface reduces the time 28

taken to reach equilibrium •

Measurements of the solubility of hydrogen in the a-phase of palladium shows a decrease of the solubility with an increase in the temperature (expected for an exothermic occluder of hydrogen). Data extracted from

109 110

Clewley et al and Holleck indicates that the enthalpy of solution decreases with an increase in temperature. Table 2 presents the values taken from the above references. In the a-phase (at low bulk

H-concentration) the solubility of hydrogen is described by Sievert's Law; that is, the equilibrium concentration, ce' follows the following

relationship,

c

=

Sp1/2 e

The solubility, S, follows an Arrhenius type behaviour.

[2.4)

In the table is listed the enthalpy of solution in J/mole and the temperature of applicability. Also listed are the pre-exponential factors for each range as well as the diffusivity of H in Pd. The limits of the a-phase as a function of the temperature are given in table 3. The

fractional bulk concentration have been converted to pressures Which would be needed to attain a _max •

The section to follow will deal with hydrogen on the surface.

2.3.2 Hydrogen Adsorption

The experimentally determined heats of adsorption for H

2 on various Pd faces are extracted from the literature and presented below. The heats of adsorption will be seen to have little correlation to the type of surface

face exposed.

2.3.2a Energies of Adsorption

Aldag and Schmidt (1971)46 using the flash-filament desorption 45

technique on a polycrystalline Pd wire observed four different desorption states. Three of the desorption states (~1' ~2' ~3) occurred above 350 K with enthalpies of 92, 105 and 147 kj/mol H₂ respectively. [Unfortunately the notation a, ~ ••• used in the chemisorption field is also used to designate different solid hydride phases. Which applies is usually clear

order desorption kinetics. The fourth state, a, desorbed according to first order kinetics and was seen if adsorption occurred at

lOOK.

Desorption was seen at ~250K with an activation energy of 55-59 kj/mol H 2• This a state was interpreted by the authors as hydrogen released from the bulk. No analysis was made by the authors with respect to the surface composition of their samples.

48 (.

Lynch and Flanagan 1973) conducted their studies in a greaseless and mercury-free system capable of ultra-high vacuum (UHV). The authors

investigated the sorption of hydrogen by samples in the form of a wire, a thin sheet and a black. They studied the weak, reversible adsorption of hydrogen on Pd black af ter the strongly chemisorbed states became saturated. The bulk contribution was substracted (10-20%) from the total amount taken up. Measurements conducted between 273 and 311K yielded isosteric heats of desorption ranging from 46 to 36 kj/mol H

2 as the coverage decreased. They observed this weakly bound state to be in "rapid equilibrium" with absorbed hydrogene No surface composition of the samples were determined.

Conrad et al (1974)53 appear to be the first to publish a study of the absorption-desorption behaviour on clean, well characterized, single crystal faces of Pd. The techniques employed were LEED, contact potential

measurements and thermal desorption measurements. The electron werk function (~~) increased with hydrogen coverage; this indicated a net

electron transfer from the metal to the chemisorbed H-atoms. Chemisorption -7 -3 isotherms were measured for T

=

308-398K and pressures from 10 to 10 Pa.

1/2

The isotherms at low coverages followed ~~ ~ (PH) which confirmed 2

dissociative chemisorption. Thermal desorption spectroscopy (TDS), af ter hydrogen adsorption at room temperature, showed a single desorption peak at

be surface adsorbed

H.

The area under the broader peak and the temperature of its maximum increased with exposure. This was explained as H released from the bulk. Using the temperature dependance of the isotherms at Pd(llO) and (111) surfaces, isosteric enthalpies of (E

A in figured 6) desorption of 103 ± 4 and 88 ± 4 kj/mol H

2 respectively were obtained. Results of TDS from the Pd(llO) surface gave an activation energy of 97 kj/mol H

2 (EA + Ea in figure

6).

This near agreement between the activation energy and the enthalpy indicated that dissociative chemisorption took place with almost

zero activation energy. The enthalpy values of the ~l' ~2 states given by Aldag and Schmidt, appear to correspond to the above. Conrad et al assume the Pd(lll) faces (most densely packed) predominate the surface of

polycrystalline Pd. LEED analysis of H chemisorbed on Pd(lll) plane yielded a lxI structure, (figure 3b). Auger electron spectroscopy (AES) indicated that the surfaces studied were known to be free from contaminants. Work function changes were found to be a more sensitive detector of surface cleanliness than AES.

Behm et al 80 (1980) observed the adsorption of hydrogen on a Pd(lOO) surface using TDS, work function and LEED measurements. An isosteric heat of adsorption equal to 103 kj/mol H

2 was measured and observed to remain constant up to a fractional surface-hydrogen coverage (eH) of ~0.9.

In summary these represent the most confident values for the adsorption

energies for H . 53 53

2: on Pd(lll) , 88±4 kj/mol H2; on Pd(llO) ,

103±4 kj/mol H 80

2 and on Pd(lOO) , 103 kj/mol H2• The ~ l' ~ 2 states of Aldag

and Schmidt46 appear to correspond to these values while their

~3

value is too large and has not been observed by other workers • Many studies have observed a fourth, weakly, chemisorbed state

(a,

using Aldag and Schmidt's

whether the a-state is desorption of bulk hydrogen or a true chemisorbed state varies with authors and is still in question. Aldag and Schmidt observed an activation energy of 55-59 kj/mol

Hz

for this state. Lynch and

48

Flanagan observed the weak chemisorbed (their conclusion) state as having isosteric heats of desorption ranging from 46+36 kj/mol H

2 as eH increased. How effectively hydrogen, molecular or atomic, adsorbs onto the Pd surface is reviewed in the next se.ction.

2.3.2b Sticking Coefficients

Dissociative chemisorption of hydrogen on clean Pd occurs with almost 53

no activation energy ; this is also the conclusion of Livshits and 76

Samartsev •

46

Aldag and Schmidt measured a molecular sticking coefficient of 0.13 for their

f3

states and "'10-3 for their a sta te. Conrad et a153 estimated a molecular sticking coefficient of 0.1-0.2. Couper and John 64 , while

investigating the sorption properties of Pd wires, observèd initial sticking coefficients of the order of 0.2. Initial sticking coefficients of 0.16 at

56 lowe

H coverages to 0.006 for eH

=

0.78 were recorded by Eley and Pearson • 76

In contrast to these values Livshits and Samartsev , investigating the permeation of hydrogen through palladium, observed a sticking coefficient as high as 0.65. Only Conrad et al have analyzed the sample surface and

observed it to be free of impurities.

Results, then, indicate that for clean Pd the molecular sticking coefficient is in the order of 0.1-0.2 with negligible activation energy.

For an incident atomic flux it is expected that the sticking coefficient, a , should be )0.2 since there" is no need to dissociate.

a 89

Dean et al measured a = 0.5±0.2 on polycrystalline palladium. In the a

temperature range of investigation (360-500K) no temperature dependance was observed. The high potential energy of the atom (220 kj/mol H) is the

contributing factor to the lack of a temperature dependance. The impurities 89

on the Pd sample of Dean et al were ~19% sulphur and 2% oxygen. Livshits et al77 observed a

=

0.2-0.4 for contaminated Pd surfaces.

a No

analysis of the surface impurities was made. They also observed aa to be practically independent of the membrane temperature in the range of

293-793K. A value of 0.2_+0.24 0 Pd79.

0•08 was observed for D adsorption on

Atomie sticking coefficients, then, are in the range ~0.2-0.7 and temperature independent. There are indications that a may be rather

a

insensitive to the level of impurities on the surface; more work in this area is required.

The nature of molecular hydrogen adsorption onto the Pd surface is the subject of review in the next section.

2.3.2c Nature of Hydrogen Adsorption on pd

The strong chemisorbed states are characterized by enthalpies of adsorption in the order of 100 kj/mol H

2 (the enthalpy of solution is

~9.7 kj/mol H). The enthalpy of molecular adsorption is less than half of

that of the strongly chemisorbed H. It is expected for low pressures that the sites for strong chemisorption will be filled first; once these sites are close to being filled, the weaker chemisorption sites proceed to be occupied.

46

Recall the conclusion by Aldag and Schmidt that their a-state

represented dissolved hydrogen in the bulk. Wicke and Brodowski argue that 1 this could not be the case. If it were absorbed hydrogen then the

as increasing amounts of hyd rogen were sorbed. 'nle a:-peaks of Aldag and Schmidt remain sharp and maintain their position at 250K irrespective of the

amount of hydrogen sorbed. Wicke and Brodowski state that this probably

represents molecular chemisorbed hydrogene

Wicke and Brodowskil point out that the observations of Conrad et alS3

of a lxI H-chemisorbed structure on Pd(lll) (figure 3b) observed by LEED agrees with the result of model calculations for hydrogen chemisorbed on

SS

platinum(lll) planes done by Weinberg and Merrill There occurs a

potential weIl for H

2 molecules on Pt(lll) in position I (see figure 4).

The dissociation of the chemisorbed molecule then moves along a minimum

energy path to position 2, the transition state. Finally the movement ends

at position 3. Position 2 is a saddle point in the energy surface.

Weinberg and Merrill considered this chemisorbed molecule as a precursor to

atomic adsorption and reported observlng the proper, transient, 2x13 LEED

pattern for molecular adsorption of D

2 on Pt(lll). Conrad et al were unable

to demonstrate the existance of such a 2x/3 structure inthe case of H

2 on

Pd(lll). Wicke and Brodowsky, however, present arguments which support the

application of Weinberg and Merrill's results to H₂ on Pd(lll), at least '

qualitatively. They suggest that af ter the strongly chemisorbed atoms on

the octahedral surface sites are occupied the still empty tetrahedral holes

act as passages for hydrogen atoms to enter the bulk. Wicke and Brodowsky

connect the surface tetrahedral sites with the weak chemisorption state se en

48

by Lynch and Flanagan • Hydrogenation and dehydrogenation proceed via

rather flat chemisorption wells on a Pd surface, where the potential wells

have been filled up by strong chemisorption. 'nle rate determining step of

hydrogenation then is the molecular dissociation of H

2 chemisorbed in

[2.5]

with the HA(ad) atoms remaining in position 1 or migrating along the 1

surface. Wicke and Brodowsky work out that the enthalpy of adsorption for H

2(g)

t

2Had(B) is -27.3 kj/mol H2•

The rates of hydrogen absorption and desorption were investigated by

Auer and Grabke (1974)54. They used, in order to measure surface processes

only, very thin pd foils (2.5 and5~m thick). The change in electrical

resistivity of the Pd foils with hydrogenation and dehydrogenation was used

to monitor hydrogen content. Results obtained indicated that the rate

limiting step was dissociative chemisorption on the surface and not

diffusion. The samples were pretreated by heating in a pure helium flow for

2 hrs. at 773K. No analysis of the surface was performed. The temperature

range of the experiment was 293 to 423K. They observed the following rate

law:

dt 1 + IQ{

k'. KX2

1 + KX [2.6]

dX kp. 1

where X is the bulk atomie ratio, p the hydrogen pressure, k and k'

represent rate constants for hydrogenation and dehydrogenation respectively.

The constant K is the equilibrium constant for adsorption. This type of

rate law, however, required a second type of H atom chemisorption (weak).

The two types are then labelled A and B. An activation energy of 28.6

kj/mol H

2 was measured for rate constant of hydrogenation, k. According to

Auer and Grabke the type B adsorption sites bloek, reversibly, the passages

between the surface and the bulk, hence the retardation terms (1

+

KX).

Auer and Grabke have postulated an activated complex, but unlike the

proposed position of the chemisorbed H argued by Wicke and Brodowsky they

. 2

suggested that the solution precursor is an H

2 molecule adsorbed with its axis perpendicular to the surface. One atom (or proton) of the molecule adsorbs in an interstice directly beneath the metal surface (the

sub-surface), referred to as the B state; the other adsorbed atom is on the surface, A state. The surface concentrations of HA'

Ha

(BA' BB) are

determined by the dissolved hydrogene For the equilibrium H(ab)

t

HA the following was assumed,

o

I-BA'" 1 [2.7]

BA

=

KAX [2.8]

For the equilibrium H(bulk)

t

Ha

reaction a Langmuir isotherm was assumed,

B

B [2.9]

KA B represent equilibrium constants for the above reactions.

_,

The authors conclude that the ra te limiting step occurs at one site only, since the reaction site can be blocked by one H atom [(1 + KBX) term].

The change in X then was given as,

dX = kp(1-0_B) - k' B A BB dt

[2.10]

Substituting relations 2.8 and 2.9 yielded au equation similar to 2.6 relating the equilibrium constant K to KA and KB'

considers that the elastic energy of dilatation is less for interstitial sites at the surface than in the bulk. If the lattice parameter were

increased, for example by the addition of silver, the forward reaction rate should increase; such was observed by tbe authors.

51

In 1973, R. Dus used a rapid-recording statie capacitor technique to measure dynamie work function (6~) changes of evaporated Pd films at 78K when exposed to hydrogene The change, 6~, was initially positive but

gradually became negative as the sorbed dose increased and Pd hydride began to form. A positive change in the work function indicated a net electron transfer fran the metal to the chemisorbed H atoms. The author interpreted this as successive filling of two surface states (~1' ~2' respectively). The ~1 population first increasing then decreasing as that of ~2 increased. It was hypothesized that ~1-H was located on tbe surface; whereas, ~2 was positioned in the surface (ie. sub-surface) and was a precursor to the bulk phase. These correspond to the two states observed by Toya 43 • Increasing the pressure further significantly lowered the work function; this was recognized as a third state labelled a. The a state could hold many monolayers of H

2 and was reversible. The author attributed this to a molecularly adsorbed state. Again, no analysis of the surface composition was performed.

Results in contrast to the above ware presented by Lynch and

52

Flanagan (1974) • They used an adiabatic calorimeter to determine differential heats of sorption of hydrogen by various forms of Pd.

Palladium black showed rapid sorption at 195 and 78K in quantities in excess of those expected for chemisorption. This hydrogen could not be removed by pumping. They assumed that a strongly chemisorbed monolayer of hydrogen covered their sample; therefore, all the observed sorption was attributed to bulk solution. This is a similar conclusion to that of Aldag and Schmidt 46 •

Even though studies o~ chemisorption of H on Pd have been conducted, there is still a lack of clarity on the nature of the bond to the surface. Adsorption of H on Pd(lll) at 80K produced a peak in the difference curve at

60

6.4eV below the Fermi energy for a photon energy of 21eV ,but room

temperature adsorption did not produce a bonding state, split from the bulk bands61• Eberhardt et a159, using angle-resolved photoelectron

spectroscopy, similarly observed the above and found that it is independent of the coverage. Eberhardt et al studied the bonding of H to the (111) surface of Ni, Pd and Pt • For a temperature of -lOOK, the adsorption of H produced a new energy level split off from the d-bands causing drama tic changes in the d-band region of the spectra. The split-off state had a periodicity corresponding to a lxl overlayer. The authors state that the substrate d-electrons were involved in the surface bond. For pd the split off state vanished for T ~ 270K and was irreversible for all H exposures. The conclusion was that Hadsorbed on a low-temperature substrate as atoms bound in the threefold surface site, which is not the lowest energy site. When the metal was warmed there existed a phonon-assisted conversion to a lower energy site. The substrate atoms moved to accomodate the H atom. The most likely site, as concluded by Eberhardt et al, was the octahedral site

under the surf ace plane. This is the site occupied by H in the bulk. The multiple peaks observed in TOS spectra by other investigators, they claim, were a result of competition between conversion and desorption. When the coverage is smal 1 , all H converted. At higher coverages, part of the adsorbed H desorbed.

Theoretical analysis of sub-surface bonding of hydrogen to metallic 62

surfaces has been carried out by Lagos (1982) • Lagos in his paper 59

53

adsorption energy of H on a Pd(lll) surface is -0.45eV ; whereas, the ca1cu1ated sub-surface energy was -0.50eV, which suggested that it may be a sub-surface bond. lts physica1 origin is exp1ained by Lagos as

hydrogen-induced se1f-trapping distortion. 63

Kubiak and Stu1en used e1ectron-stimu1ated desorption as a probe of hydrogen adsorption on and diffusion into Pd(lll). The authors stated that theoretica1 studies indicate that for 1

<

eH

<

~ that H energetica11y

3

3

favoured .the occupation of sub-surface octahedral bonding sites. Kubiak and Stu1en claim that their experimenta1 resu1ts i11ustrated this sub-surface bonding site. Also, that for T

>

200K hydrogen diffusion into deep 1ayers of the bulk became significant, and hence the disappearance of the distinct peak separated from the d-bands observed in the photoemission spectra of Eberhardt et a159•

Further discussions into the bonding of H2' H on the Pd surface was

56 64

presented by E1ey and Pearson and Couper and John Couper and John have investigated the sorption proper ties of Pd wires at 104K. The pretreatment of the wire entai1ed repeated outgassing of the wires at high temperatures and exposures to 02. Thermal desorption spectroscopy for adsorption at 273, 210 and 90K (eH

<

0.7 for T = 90K) showed on1y a single peak, ~, with a temperature of -330K at maximum. Adsorption at 90K for e

>

0.7 and

subsequent TOS had two peaks with the lower temperature peak labelled as a; it was not possib1e to saturate the a-peak. The authors assigned the a-peak to desorption from the bulk. Adsorption at 400K with coo1ing to 90K before the TDS spectra showed no a-peak for eH

<

1.0. The adsorbed precursor states were not popu1ated until the sites for strong chemisorption were occupied. They state that the strong1y chemisorbed atoms on the Pd(lll) are a1ternate octahedral holes on the surface. The precursor can then be seen

as occupying these octahedral holes that are left following the occupation by the sorrounding four holes.

66

Rye and Ricco (1987) concluded from their studies of hydrogen

sensitive Pd metal/insulator/semiconductor diodes that the observed results were inconsistent with a mechanism Which involved a direct transfer of

hydrogen from the strongly chemisorbed state to the bulk. They suggested

that adsorption occurred into a weakly bound precursor state Which populated both the bulk states and the strongly chemisorbed surface states.

Concluding this section, it is apparent that the nature of the

precursor to absorption remains a subject of investigation and debate. What may be concluded with some certainty is that the strongly chemisorbed

hydrogen (adsorption enthalpy ~100 kj/mol H

2) is not in equilibrium with the

bulk absorbed hydrogene The strong chemisorption sites will fill before the

weaker chemisorption sites. The precursor to absorption is weakly adsorbed

and is in equilibrium with the absorbed hydrogene The precursor is adsorbed

in a subsurface (possibly octahedral) site or has one of its atoms adsorbed

at this site. The mechanism which continues to allow hydrogen permeation at

very high pressures would involve this precursor.

A brief review of hydrogen permeation studies in the literature

.

follows.

2.3.3 Hydrogen Permeation

There have been numerous investigations on the hydrogen permeability of

Pd; however, most have been in the diffusion limited regime or in the

surface limited regime with no surface characterization attempted. The

measurements for the diffusion limited permeation followed the expected relation given by equation 3.4 (next chapter) even into the vacuum regime

12 75 76 80

•

membrane.

12

Davis (1954) observed the permeation of hydrogen through a palladium cylinder for various surface treatments. He observed that when the low vacuum (in which the palladium was present) was maintained by a mechanical pump the permeation rate of hydrogen was reduced and deviated from the square root dependance in the low pressure regime. The conclusion was that hydrocarbonspresent in such a vacuum system were the cause of the decrease. Exposure to atmosphere 'reactivated' the palladium. No surface analysis

techniques were used. 75

Koffier et al (1969~ investigated the permeation of hydrogen through a-palladium. Their samples were 99.95% Pd and were conditioned differently in order to vary the grain size. All samples underwent additional exposure to 800 Pa of 02 for 10 min. at 673K to remove impurities. They obtained the

21 1/2

following permeability: P

=

1.3±0.07x10 p exp(-15.7±0.2 (kj/mol H)/RT)

2

H

2/m .s. This relationship was obtained for a temperature range of 300 to

-2

709K and a pressure range of 3.9x10 to 6.7 Pa. The varying grain sizes (0.01 to 1mm), over the temperature range investigated, had no effect on the permeation rate. The authors state that there was no previous permeation data for T<443K or p(1.3Pa; thus, their results extended the range.

A significant contribution to H

2 (D2) permeation through palladium for

various surface conditions (showing both the surface and diffusion limited 76-79

regimes) has been performed by Livshits and co-workers Unfortunately no analysis was conducted to classify the impurities nor how the various treatments altered the permeation levels. The results, however, do

illustrate that changes in the permeation rates of orders of magnitudes are 76

possible for different treatments. Livshits and Samartsev observed

-5

what appeared to be a clean Pd membrane. No previous work had ever gone to this low a pressure. The results of these authors will be reviewed in more detail in the discussion (see 8.5.1) when they are compared to those of this study.

Works examining the effects of surface impurities on the hydrogen adsorption by metals are reviewed in the next section.

2.4

Impurities

Impurities on the surface have been known to affect studies measuring hydrogen sorption and permeation. This section will present the current understanding of impurity effects on the hydrogen-metal interaction. The impurities of interest to this study were carbon, sulphur and to a minor degree oxygen.

Sulphur bas been shown to play a significant role in the desorption of hydrogen from the palladium surf ace. Comsa et al 69 , using time of flight techniques, observed the velocity distributions of D_{2 and H2 desorbing from} a Pd(100) surface. Two distinct groups of molecules were observed, fast and slow. The slow molecules were characterized by a cos0 (0 is measured with respect to the surface normal) spatial distribution; whereas, the fast

n

molecules had a spatial distribution which was described by cos 0 (n)l). For a sulphur fractional coverage of ~.5 the desorbing fast molecules had a

10

spatial distribution described by cos 0. There continued to be an increase of the number of fast molecules with sulphur to 0 = 0.65, at which point

s

the authors have no further data. The activation energy for desorption associated with the fast molecules was 15±2 kJ/mole in the temperature range of 360-800K. No temperature dependance was observed.

A copper sample resulted in an activation energy of ~55 kJ/mole while

For Ni fast molecules desorbed also from a clean surface; whereas, for Pd the presence of sulphur was essential for the appearance of fast

molecules. For 0

s ~ 0.01 a cos0 spatial distribution and a Maxwellian speed distribution characterized by the sample temperature was evident. As the

sulphur coverage increased the flux of fast molecules increased (figure 5); however, the total flux (fast + slow) remained constant. The authorsnote

that the ratio of integrated fluxes varied with sulphur coverage, but was independent of the sample temperature in the range of observation 400-800K. This last observation provided, for the authors, proof that there was no activation energy for adsorption. If there was an activation energy, Ea' for adsorption then the fast molecules would have an activation for

desorption of 2(E

a + EA), where EA is the enthalpy of adsorption (see figure 6). The slow molecules would have an activation energy for desorption of 2E

A• The difference in energies of ~30 kJ/mole would show a temperature dependance for the ratio.

Comsa et al then presented a mechanism where the fast desorbed H₂ arose from the direct recombination of two absorbed atoms, viz.,

[2.11]

Sulphur acting to 'open' channels for desorption from the bulk would have explained such a mechanism, though it was pointed out by the authors that a similar result was possible if the sulphur actually blocked the channels for adsorption.

68

Comsa et al observed that for Ni the fast molecules were also present on the clean surf ace, contrary to the Pd results. This is in apparent

69

contradiction to the Ni results of Bradley et al where a cos0 spatial distribution was observed for clean Ni. Bradley et al observed the most

4

peaked spatial distribution (cos e) occurred for es - 0.5 and broadened to cose for es = 1 or eS = O. Same general conclusions were drawn for C on Ni and C or S on Fe, Pt and Nb.

The effect on the initial sticking coefficient of H

2 on polycrystalline nickel surf aces has been studied for various coverages of sulphur or

70

oxygen

A

significant change in the initial sticking coefficient as a function of sulphur or oxygen was observed (figure 7). The observations were conducted only at 283K. At saturation coverage (0.57 for S, 0.37 for

0) the initial sticking coefficient, to within the measurable accuracy, was according to the authors essentially zero. Unfortunately this does not

provide a quantitative result, for even a small sticking coefficient, e.g.,

-4

10 is still of practical importance.

It is known, generally, that electronegative preadsorbates like Cl, S or 0 decrease the sticking probabilities for molecules like H

2, CO or N2• This is attributed to a decreased binding of the molecule to the surface as

71

a result of the coadsorbate • Since H

2 when itadsorbs on Pd shows a net 53

electron transfer from the metal to the hydrogen , it is easy to understand how electronegative adsorbates reduce the adsorption capability of the pure metal.

The effect of coadsorbed sulphur on H

2 adsorption on a Pt(110) 107

surface showed that the hydrogen coverage decreased rapidly with

increasing sulphur coverage. A saturation fractional coverage of 0.8 for S "completely" inhibited the hydrogen uptake. A plot of hydrogen coverage vs. sulphur coverage was non-linear, it showed a rapid decline in H coverage for low sulphur coverage which slowed as the sulphur coverage increased. The initial slope indicated that 12±3 adsorption sites were blocked by one sulphur atom. As the sulphur coverage (e

>

1/3) increased the number of

affected sites decreased; that is, interactions between adsorbed sulphur 107

atoms tended to lower the poisoning effect. Marcus et al observed strong blocking effects but no major effect on the energetics of adsorption on the

free surface sites.

109

Kiskinova and Goodman studied the effects of electronegative adatoms on the

Hz

chemisorption by Ni(100). They observed that the reduction in the initial sticking coefficient of H

2 on Ni(100) followed approximately a 2

(1-40

i) relationship up until es ~ 0.2. This indicated that one sulphur atom blocked 4 neighbouring adsorption sites.

110

In contrast to this Johnson and Madix observed that sulphur adsorbed (es ~ 0.5) on the Ni(100) surface greatly affected the desorption energy (a decrease of more than 42 kJ/mole) and the pre-exponential (a decrease of six

orders of magnitude). Desorption at e ~ 0.5 no longer obeyed a simple

s 111

similar effects second order rate law. Ko and Madix observed on H

2 desorption kinetics for carbon coverage of Ni(100)(e

=

0.25). The

c

desorption energy was reduced by 52.5 kJ/mole and the pre-exponential by 10-7 •

115

Wampler , while studying iron, observed a decrease in the recombination coefficient for H

2 desorption of approximately two orders of magnitude for an oxygen coverage of ~0.5. A linear decrease of the

recombination coefficient occured up to e

O ~.5. The author suggested that the effect could result from a decrease in the number of surface sites available for adsorption.

Experimental results causing differing conclusions are again evident here as they were in the section dealing with adsorption sites and

mechanisms of hydrogen transfer from the gas to the bulk.

A

clear