Studies of transport in some oxides by gas phase analysis

Studies of Transport in Some Oxides by Gas Phase Analysis

Qian DONG

Licentiate Thesis

Department of Materials Science and Engineering Division of Corrosion Science

Royal Institute of Technology, KTH Stockholm 2004

ISRN KTH/MSE-04/70-SE+CORR/AVH

ISBN 91-7283-923-6

Licentiatavhavhandling

Som med tillstånd av Kungliga Tekniska Högskolan i Sockholm framlägges till offentlig granskning av avläggande av teknisk licentiatexamen, fredagen den 17 december 2004, kl.13.00 i sal Q2, seminarierummet, Osquldas väg 4, KTH

Granskare är Tekn.Dr Rachel Pettersson, Institutet för Metallforskning, SE-11428

Stockholm

The effects of porous Pt on the oxidation of Cr at 800°C have been studied by Gas Phase Analysis (GPA), Secondary Ion Mass Spectrometry (SIMS) and X-ray Photoelectron Spectroscopy (XPS). In the oxide areas with Pt, a substantial oxide growth near the Cr substrate takes place by a pronounced inward oxygen transport. Oxide grown on areas without Pt, the SIMS counts of CrO ions depend on the distance from the area with Pt. The experimental observations are believed to be a consequence of a high dissociation efficiency of O

on areas with Pt in combination with a high diffusivity of O on external and internal oxide surfaces on areas both with and without Pt.

An analysis of literature data of transport of hydrogen in vitreous silica indicates a contribution of atomic hydrogen to the overall hydrogen transport.

Transport of hydrogen through a 0.1 cm thick quartz wall was studied by applying 70-1200 mbar hydrogen at the high-pressure side at 550

C. The steady state flux of permeated hydrogen was found to increase proportionally to the applied hydrogen pressure, while the steady state concentration of hydrogen did not increase proportionally to the applied hydrogen pressure.

An iterative fitting procedure of the experimental data revealed a concentration dependent diffusivity of hydrogen. The observation can be linked to a situation where the transport of atomic hydrogen is retarded in reversible traps. It is found that the fraction of retarded hydrogen decreases when the hydrogen concentration in the material increases.

A novel and relatively straightforward non-destructive method for evaluation of diffusivity, concentration and effective pore size is obtained by quantitative evaluation with a mass spectrometer of gas release from oxides during outgassing. The method is validated in measurements of diffusivity and solubility of He in quartz at 80°C and applied for characterization of Zr- and Fe-oxides.

Keywords:

Gas Phase Analysis (GPA), Secondary Ion Mass Spectrometry (SIMS),

chromium, platinum, quartz, porous oxides, hydrogen, oxidation,

dissociation, trapping, diffusion, outgassing, diffusivity

This thesis is a summary of the following papers:

I. Platinum-induced oxidation of chromium in O

at 800°C G. Hultquist, E. Hörnlund, Q. Dong

Corr. Sci., 45(2003), 2697

II. Concentration-dependent hydrogen diffusion in quartz related to trapping

Q. Dong, G. Hultquist and J. Rundgren Submitted to J Appl. Phy.

III. A method for characterization of gas transport in porous oxides

C. Anghel, Q. Dong, J. Rundgren, G. Hultquist, I. Saeki and M. Limbäck Manuscript

Paper not included in the thesis:

IV. First-principles study of hydrogen diffusion in α-Al

O

and liquid alumina

A. B. Belonoshko, A. Rosengren, Q. Dong, G. Hultquist, C. Leygraf

Phys. Rev. B, 69(2004), 024302-1

1 Introduction………...………1

2 Experimental……….3

2.1 Gas Phase Analysis (GPA)………...…3

• Two-stage oxidation……...………..4

• Flux measurement………4

• Concentration measurement by outgassing………..5

2.2 Secondary Ion Mass Spectrometry (SIMS)………..………5

2.3 X-ray Photoelectron Spectroscopy (XPS)………6

2.4 Scanning Electron Microscopy (SEM)………...…………..7

2.5 Materials studied………...…………7

3 Summary of appended papers………...…………8

3.1 Platinum-induced oxidation of chromium in O

at 800°C…..…….….8

3.2 Concentration-dependent hydrogen diffusion in quartz related to trapping.………....10

3.3 A method for characterization of gas transport in porous oxides…...14

4 Conclusions……….…19

5 Suggestions for future work……….…...20

6 Acknowledgments………...……21

7 References……….……..22

Supporting papers I-III

1 Introduction

Matter and charge transport properties of oxides are of profound importance for a number of oxide characteristics. These include oxide ability to resist degradation in chemically aggressive environments and ability, as oxide membranes, to separate chemical species. Depending on application and desired properties, the transport rates should be either high or low. In both cases it is of scientific as well as technological importance to understand the underlying chemical and physical principles and find means of influencing the transport properties.

Knowledge of conditions for the growth of adherent metal oxides with low densities of pores and other macroscopic faults is essential for optimal protection of high-temperature materials in services. When oxide growth takes place predominantly at the oxide-gas interface, for example on Cr, poor adherence of the scale is observed

. In systems where oxides growth takes place predominantly at the substrate-oxide interface, cracks can often be found in the oxide scale. An effective way to repair an oxide containing pores, voids and cracks is to fill these with new “material”.

This material is comprised basically of metal and oxygen, which implies the need for transport of both oxygen (ions) and metal (ions) to these positions. To realize this, one needs to be able to vary the position of oxide growth in the metal-oxide system of interest.

The beneficial influence on the oxidation behaviour of high temperature

materials by the addition of Rare Earth Metals (REM) such as Y, Hf, Ce

and La is well known phenomenon

. The reactive element effect

manifests itself giving a good resistance to spallation at high temperature

and during cooling down to room temperature

. Several mechanisms

have been proposed to explain these REM effects

. There are similar

positive effects by the addition of certain noble metals such as Pt

.

However, the mechanisms behind the effects of all these additions are not

clearly identified and several possible mechanisms have been discussed

for a long time

. It have been pointed out that these additions catalyze

the dissociation of oxygen molecules and this can be a key factor for the

mechanistic understanding of their effect on improving oxide scale adherence

. In this thesis (Paper I) the influence of porous platinum on the oxidation of chromium at 800°C in O

is addressed with the use of mainly gas phase analysis with the

O-SIMS technique.

Hydrogen has been found to influence the oxide growth mechanism in oxidation of many metals at high temperatures. Hydrogen shows to increase the fraction of oxide growth at the oxide-gas interface during high-temperature oxidation of chromium and a number of alloys

. Studies of the silica-hydrogen system show that several factors such as impurities, thermal history and the amount of trapped hydrogen in the bulk

can influence the transport of hydrogen itself in quartz. Lee et al

22-23

investigated the diffusion of hydrogen in silica and observed two diffusion regimes, called “normal” and “abnormal diffusion”. The features were explained by hydrogen traps in the silica because of a relatively deep associated potential well. The consequence of trapping is a decrease in the rate of transport of hydrogen in metals, alloys and oxides, because there is a certain probability for hydrogen to jump into trap sites where the residence time is longer than in normal bulk/lattice diffusion sites. Therefore the trap sites essentially act as sinks for hydrogen, and obviously a comprehensive treatment is required for the quantitative analyses of traps. Later Shelby

established that permeability and solubility vary little among different silica glasses. It is generally believed that hydrogen is trapped by forming OH-groups, and several mechanisms for the formation of OH are proposed. However, effects of trapping on the diffusivity in oxide systems can be further clarified, which is the aim of Paper II.

Evaluation of diffusion (diffusivity, concentration, permeability) of gas in oxides is of both fundamental and technical interest. Paper III presents a novel and relatively straightforward method to quantify diffusion and amount of gas present in oxide scales.

The overall aim of this thesis work has been to contribute to the knowledge of transport mechanisms of oxygen and hydrogen in oxides.

Especially, a possible fast transport of atomic/ionic oxygen and hydrogen,

relative to molecular oxygen and hydrogen, has been in focus.

2 Experimental

2.1 Gas Phase Analysis (GPA)

Figure 1 shows schematically the equipment used to study gas transport.

It consists of three main parts: a enclosed reaction chamber of approximately 70 cm

, a Mass Spectrometer (MS) placed in an Ultra High Vacuum (UHV) chamber with an ion pump, and a gas handling system containing a rough pump. These three parts are joined with stainless steel coupling. A tube furnace that can be moved back and forth on a rail with rollers supplies temperatures in the range of 25-1000

C.

Figure 1. Gas Phase Analysis (GPA) experimental setup

The GPA technique offers a versatile tool for studying many different processes. Reactions taking place on a sample are studied by analyzing the gas in the closed reaction chamber. The equipment is basically used in two different modes:

• Mode I: The mode is used to monitor the release of gaseous

components from samples. Gases released from a sample are

analyzed by the MS and evacuated by the ion-pump. By calibration

of MS signal and the ion pump rate

, the gaseous components

released can be measured. A typical example of Mode I measurement is outgassing of hydrogen from metallic samples.

• Mode II: The reaction chamber is filled with gas from the gas handling system and the leak valve connected to the MS is either closed or slightly opened to probe the gas composition. The probing enables measurements of isotopic exchange due to different reactions. The consumption of gas due to probing is generally negligible in comparison with the gas consumption due to reaction with the sample. Typical measurement in Mode II is oxide formation, hydrogen uptake and calculation of dissociation rates.

In the attached papers, GPA is used for the following measurements and experiments.

• Two-stage oxidation

Two-stage oxidation is performed by two consecutive exposures of the sample and the oxygen isotope is changed. For example oxidation in

16,16

O

followed by oxidation in

O

. By SIMS analysis of the oxide formed in the two-stage oxidation, information about growth processes is obtained. If the oxide grows by exclusive metal transport, the isotope used in the second stage of oxidation will be found at the oxide-gas interface. If the oxide grows by oxygen transport, the isotope used in the second stage of the oxidation can be found at the oxide-metal interface.

By using a mixture of

O

and

O

in the second stage, additional information about molecular dissociation rates and isotopic exchange rates with the oxide may be retrieved.

• Flux measurement

The experimental procedure for flux measurement is: the specimen is

held at a certain temperature sufficiently long to provide a constant and

low background for gas. At zero time, a known gas pressure (normally

less than 1000 mbar) is introduced into the outer volume. The flux of gas

through the specimen is continuously monitored by the mass

spectrometer. Steady-state flux of gas is judged when virtually no further increase of gas signal in the mass spectrometer can be seen.

• Concentration measurement by outgassing

The outgassing of hydrogen from the specimen is carried out in the following way. After steady state flux has been established, the influx gas is evacuated by a rough pump, and the specimen is heated by the furnace rapidly. The aim is that the concentration profile in the material created at steady state diffusion remains when the outgassing starts. A decay curve is obtained for the release of gas into the mass spectrometer system. The concentration of hydrogen can be obtained by integrating the flux during the decay process to yield the total amount of gas released from the specimen.

2.2 Secondary Ion Mass Spectrometry (SIMS)

SIMS is a mass-sensitive spectroscopic method that the surface of a sample placed in UHV is sputtered with an ion beam

. Fragments in the form of both negative and positive secondary ions and neutral species are knocked off from the sample surface. As the name of the method indicates, the secondary ions are analyzed. The detection limit of SIMS is in the ppm range.

SIMS is widely used for analysis of trace elements in solid materials, especially semiconductors and thin films. The SIMS primary ion beam can be focused to less than 1µm in diameter. Controlling the location where the primary ion beam strikes the sample surface permits microanalysis, the measurement of the lateral distribution of elements on a microscopic scale can be done. Continuous analysis with a MS while sputtering produces information as a function of depth, called a depth profile. When the sputter rate is extremely slow, the analysis can be performed while consuming less than a tenth of an atomic monolayer.

This slow sputtering mode is called static SIMS to separate it from

dynamic SIMS used for depth profiles. In the work presented in this

thesis, only dynamic SIMS has been used. The SIMS technique is very

useful to analyze samples exposed to isotopes. The growth mode of an

oxide (metal or oxygen transport) can be determined by a two-stage oxidation followed by a depth profile in SIMS.

The main advantages of SIMS are the possibility to detect hydrogen, different isotopes and the low detection limit (in the ppm range). The main drawback is the difficulty to quantify SIMS data, due to the complexity of the ionization process.

The samples are coated with gold in vacuum before they are introduced into the SIMS apparatus, to avoid difficulties with charging of the sample. These measurements are performed at Chalmers University of Technology.

2.3 X-ray Photoelectron Spectroscopy (XPS)

XPS is one of the most widely used methods for surface analysis and the reason is that the interpretation is well established

. The measured parameter is the kinetic energy of electrons ejected from the atoms upon X-ray radiation.

E

=hν-E

-Φ (1) where E

is the kinetic energy of the electrons, hν is the energy of the X- ray radiation, E

is the binding energy of the electrons to the atoms, and Φ is the work function of the spectrometer. Different elements have different electron binding energies and therefore element identification is possible by XPS. The binding energies also give information about the chemical state in which an element is.

One advantage with XPS compared to SIMS is that the composition of the surface (both concentration of the elements and their chemical state) can be determined. The disadvantage is that hydrogen is not directly detectable. Hydrogen bonded to other species will give a chemical shift in these species, but the interpretation of this shift is not straightforward.

The intensity of the signal from an oxide overlayer depends on the mean

free path of detected electrons and thickness of oxide. The probing depth

is 1-5 nm.

2.4 Scanning Electron Microscopy (SEM)

In scanning electron microscopy an electron beam is scanned over an area of the sample surface, and electrons emitted from the surface are collected and amplified to form a video signal.

FEG-SEM analysis was performed with a Leo 1530 Field Emission Scanning Electron Microscope equipped with a GEMINI field emission column.

2.5 Materials studied

The materials used in this study together with the methods used for the characterization of the samples are summarized in Table 1.

Table 1. Characterization of the samples used in this study Analysis methods Sample Paper Shape

GPA SIMS XPS SEM

Cr-Pt I Plate x x x x

Quartz II, III Tube x - - -

Oxide on Zircaloy-2 III Plate x - - x

Oxide on Fe III Plate x - - x

3 Summary of appended papers

The main results from Papers I, II and III are briefly described and discussed topic by topic.

3.1 Platinum-induced oxidation of chromium in O

at 800°C(Paper I)

Oxidation of a Cr sample, of which approximately 1/3 has been covered by porous Pt, was studied in Paper I. A two-stage oxidation was performed in a closed reaction chamber at 800°C with near 20 mbar O

: the first stage in oxygen with >99%

O and the second stage in oxygen with 67%

O. SIMS was used for depth profiling of the oxide formed at different positions from the Pt coated area. The sample with indicated analyzed positions in SIMS is shown in Figure 2.

Figure 2. Oxidized Cr sample with indicated area of porous Pt and analyzed positions in SIMS.

By examination of

Cr

O/

Cr

O and

O/

O ratios in SIMS signals at

the gas-oxide interface and oxide-metal interface, interesting results were

found as shown in Figure 3. At the gas-oxide interface, the two ratios

were found to correspond to the gas composition in the second stage of

oxidation (indicated by horizontal dash line) at all positions of the sample

surface, where oxide grows by metal transport from the substrate. At the

oxide-metal interface, the ratios are found to vary along the sample. The

ratios in the Pt area are high, which means that Pt promotes oxide growth near the substrate by oxygen transport through the oxide.

-4 -2 0 2 4 6

0 1 2 3

Ratio

Distance from Pt-interface, mm

52Cr¹⁸O/ ⁵²Cr¹⁶O near substrate

18O/ ¹⁶O near substrate

52Cr¹⁸O/ ⁵²Cr¹⁶O near gas interface

18O/ ¹⁶O

near gas interface

Figure 3. Ratios of

Cr

O/

Cr

O and

O/

O near the gas interface and near the substrate interface of oxides at different positions on the sample.

By integrating the count rate of

Pt and

Cr

O over sputter time, counts of

Pt and

Cr

O are shown versus distance from the Pt-area in Figure 4. A strong enhancement of the

Cr

O counts in the Pt-area is found.

Also some enhancement of the

Cr

O counts is observed adjacent to the Pt-area, but not at distances exceeding 4-5 mm.

-4 -2 0 2 4 6

0

5x10⁶ 1x10⁷

195 Pt counts

0 5x10⁸

1x10⁹

mm-ranged influence of Pt

52 Cr18 O counts

Distance from Pt-interface, mm

195Pt

52Cr¹⁸O

Figure 4. Counts of

Cr

O and

Pt at different positions on the sample.

The counts have been obtained by integration of count rate over sputter time until the count rate of

Cr

O has decreased to 1% of its maximum.

These observations strongly point out a fast surface diffusion of dissociated oxygen, O

, primarily in the external surface of the oxide, but also in the internal oxide surfaces including grain boundaries. This scenario can be explained by a “spill-over” of O

from the Pt-area, which acts as a generator for O

. The spill-over effect is indicated on rapid surface diffusion

.

3.2 Concentration-dependent hydrogen diffusion in quartz related to trapping (Paper II)

The quartz specimen was a vitreous silica tube sealed at one end and enclosed in an outer quartz tube. Steady state flux was measured with applied hydrogen pressure of 70, 460, 840 and 1200 mbar at 550

C. The mean concentration of hydrogen was obtained by integrating the amount of hydrogen released from the specimen after steady state flux was established.

The result is that the steady state flux of hydrogen through the quartz wall increases proportionally to the applied hydrogen pressure, as shown in Figure 5. Concurrently, the mean concentration of hydrogen in the material, C

, increases with a declining gradient as seen in Figure 6.

0 200 400 600 800 1000 1200 1400

0,0 0,5 1,0 1,5 2,0 2,5

expt.

- - - - fitted line

Flux, pmol cm-2 s-1

Pressure of hydrogen, mbar

Figure 5. Steady state flux of hydrogen through quartz at 550

C. The

fitted straight line indicates the proportionality with pressure.

0 200 400 600 800 1000 1200 1400 0,00

0,05 0,10 0,15 0,20 0,25

expt.

- - - - fitted line

Mean concentration, µmol cm-3

Pressure of hydrogen, mbar

Figure 6. Mean concentration of hydrogen in quartz at 550

C obtained from outgassing after finished steady state permeation experiment. The

fitted line indicates a non-proportional dependence on pressure.

Based on experimentally obtained steady state flux and mean concentration of hydrogen with applied hydrogen pressure of 70, 460, 840 and 1200 mbar at 550

C, the diffusivity was obtained by the following two ways.

• A computer routine fits the fluxes and mean-concentrations together agree with experimental value. The self-consistency was achieved with great accuracy after 15 turns.

• Analysis of steady state diffusion based on the assumptions of a constant diffusivity, D

, and of a triangular concentration distribution across the quartz wall, 2C

(1-x/L), gives rise to the approximation D

=FL/(2C

). This analysis is labeled “current” in Figure 7.

In Figure 7, the concentration dependent diffusivity is plotted together

with the current diffusivity as function of the mean hydrogen

concentration in the specimen. It is found that the concentration

dependent diffusivity is 30-40 % higher than the current one. This

difference is significant and the use of concentration dependent

diffusivity for describing the diffusion of hydrogen in quartz is necessary for obtaining the correct diffusivities from experimental data.

0,00 0,05 0,10 0,15 0,20 0,25

0,0 0,2 0,4 0,6 0,8

Diffusivity, x10-6 cm2 s-1

Mean concentration, µmol cm^-3 current

iterated

Figure 7. The diffusivity of hydrogen in a quartz wall at 550

C versus mean concentration of hydrogen.

Figure 8 illustrates the diffusivity D of He, Ne, Ar, oxygen, and hydrogen in vitreous silica at 550°C by using data obtained from Refs. 30 and 31.

For the hydrogen atom H, D value is extrapolated from the noble-gas line

(determined by the noble gases He, Ne and Ar) with a size half that of the

H

-molecule

. It is observed that the measured values of diffusivity of

hydrogen are higher than those expected from an H

-molecule positioned

on the noble-gas line. These differences are labeled ∆D and can be

explained by a certain contribution from diffusion of H. Other

experiments strongly indicate a considerable contribution of H diffusion

to the overall hydrogen transport in vitreous silica

.

1,0 1,5 2,0 2,5 3,0 3,5 4,0 10^-7

10^-5 10^-3 10^-1 10¹ 10³ 10⁵ 10⁷

∆D Hydrogen H

Diffusivity, x10-6 cm2 s-1

Oxygen Ar H₂

Ne He

Molecular collision diameter, Å

Figure 8. Measured diffusivity D of He, Ne, Ar oxygen and hydrogen in vitreous silica at 550

C from Ref. 30 versus molecular collision diameter from Ref. 31. Diffusivity of H

and H according to solid line determined

by the noble gases and diameter of H=½ diameter of H

(Ref.32).

The migration of hydrogen in quartz is described in terms of two migration modes for the H atom, one mode of free motion with D

and one mode of retarded motion with D

. Considering the diffusion lengths associated to the modes of motion, the thickness of the quartz wall is L, and it is assumed that a fraction x (0<x<1) of L carries free motion and a fraction 1-x carries retarded motion. Figure 9 is plotted by using equation D

= D

(1-x)

for three assumptions about D

. The high curve corresponds to pure H

diffusivity =3x10

H2

D

cm

s

. The diffusivities

measured at 70 and 1200 mbar H

pressure (the mean concentration are

0.034 and 0.21 µmol cm

, respectively) are indicated with horizontal

solid lines. The intersections of these lines with the high D

curve define

fractions 0.65<x<0.75 that are consistent with our experiment.

0,0 0,2 0,4 0,6 0,8 1,0 10^-10

10^-8 10^-6 10^-4 10^-2

D_Hret:

D_eff at 70 mbar H₂ D_eff at 1200 mbar H₂ 3x10^-8

5x10^-2

3x10^-9 3x10^-10

D_H=

D eff, cm2 s-1

x

Figure 9. Effective diffusivity, D

, versus normalized diffusion length x for free H migration. Curves depict D

= D

(1-x)

for D

equal to 3x10

, 3x10

, and 3x10

cm

s

. D

is H atom diffusivity from Figure 8.

Horizontal solid lines designate diffusivity measured at 70 and 1200 mbar. Intersection of a horizontal line and a curve gives H atom diffusion

length x corresponding to a measured D

for a given value of D

. It is concluded that to maintain the experimentally determined variation of D

, the diffusion length of free H atom migration exceeds 65% of the wall thickness and increases slightly, when the hydrogen pressure is increased from 70 to 1200 mbar. The increased diffusivity is thus explained by a reduction of unfilled traps (equivalent to an increase of occupied traps) upon increased hydrogen concentration.

3.3 A method for characterization of gas transport in porous oxides (Paper III)

If oxide is considered as a plate with thickness L, the flux from a single surface of plate (as shown in Figure 10) is expressed by the formula

= ∑

_ ^ − − _ ^

= 2

2 2 1

n 1

L 4 ) Dt 1 n 2 ( L exp

DC

F 2 π (2)

where F is flux [µmol cm

s

], D is diffusivity [cm

s

], C

is

concentration [µmol cm

], t is time [s], L is thickness of plate[cm]. Eq.

(2) describes the gas release from a scale attached to an impenetrable substrate.

metal substrate oxide

UHV

Figure 10. Outgassing from a single surface of plate with thickness L.

The magnitude of the terms of the series decreases with n, and in a long time experiment only the first term is significant. This means that in an

e

log[F] versus t diagram (bracket [] signifies dimensionless value) the flux approaches an asymptote in the form of a straight line of slope when t gets very long. The first measurement value of the outgassing experiment is the absolute value of the asymptotic slope of Eq. (2),

2 2

1

4 L

m ₌ π D (3)

When the asymptote is drawn backward in time as a straight line, it intersects the diagram axis t = 0 at the value

log [2DC

/L]. Calculating the exponential of the value of intersection, one gets the second measurement value,

L DC

m

= 2

(4)

For a given size L, C

and D are determined by the values m

and m

. The third measurement value is the integrated flux,

m

=LC

(5)

For completeness, measurement values are given for two-side gas release

from a plate of thickness L (Figure 11) on the assumption of a uniform

initial concentration . On doubling the flux in Eq. (2) one obtains for the gas release,

C

L / D

m

= × π

(6)

L DC

m

= 8

/ (7)

L

UHV oxide UHV

Figure 11. Outgassing from two-side of plate with thickness L.

In Figure 12, the horizontal axis is time t, and, in accordance with current convention, the vertical axis is flux F (in mol cm

s

) is drawn exclusively with such a power-of-ten scale corresponding to the proper logarithmic scale

log [F].

Using the above-described method, the diffusion parameters of He at 80°C in a quartz plate of thickness L = 0.1 cm after equilibration in an exposure to 250 mbar He was investigated.

The parameters m

and m

are obtained from Figure 12. He diffusivity is

thereby calculated using Eqs. (6) and (7) and the results of diffusivity are

presented in Table 2. By integrating the flux data from Figure 12 over

time, Figure 13 is obtained. The concentration of He in quartz at 80°C is

calculated using Eq. (5) and the result is also presented in Table 2.

0 10000 20000 30000 40000 50000 60000 70000 10^-11

10^-10 10^-9 10^-8 10^-7 10^-6 10^-5 10^-4

m₂

m₁=slope of dashed line

Quartz Flux of He , µmol cm-2 s-1

time,s

Figure 12. Flux of He during outgassing at 80

C after background subtraction.

0 10000 20000 30000 40000 50000 60000 70000 0,000

0,003 0,006 0,009 0,012 0,015

Quartz C₁ x 0.1cm( m₃)

intergrated F over time

µmol He cm-2

time, s

Figure 13. Integrated flux of He over time for a quartz plate of thickness 0.1 cm at 80

C.

The diffusivity, concentration and permeability (permeability equals

concentration gradient times diffusivity) of He in quartz at 80°C are

compared with the literature data

in Table 2. It can be seen that the

diffusivity, concentration and permeability of He in quartz evaluated by

this method are in similar with the literature data. Therefore, it indicates that the novel method is valid and can be used to characterize the gas transport in oxides.

Table 2. Transport parameters for He in quartz at 80°C

Parameters Results from this method Literature data

D

1.2 ⋅ 10

Diffusivity

cm

s

D

1.2 ⋅ 10

(1-2) ⋅ 10

Concentration

µmol cm

0.12 -

Permeability

atoms s

cm

atm

3.5 ⋅10

(4-6) ⋅10

4 Conclusions

Oxygen transport in oxidation of Cr

• Pt acts as a generator for dissociated oxygen, and an influence of 20-30 nm sized Pt-particles is identified up to mm distance from the particles. This can be explained by a fast diffusion of atomic and/or ionic oxygen on the external surface of the oxide.

• The main influence of a Pt-addition on the mechanisms of Cr-oxide growth is an increased transport of dissociated oxygen towards the metal-oxide interface, which predominately takes place in internal oxide surfaces. This leads to increased oxide growth near the chromium substrate, which is favorable for the adherence of the oxide.

Gas transport in quartz

• An analysis of literature data, including noble gases, indicates that diffusion of atomic hydrogen is a main contribution to the overall hydrogen transport in vitreous silica.

• It is found that diffusivity of hydrogen is concentration dependent.

The effective diffusivity is found to increase from 3x10

to 7x10

cm

s

in the hydrogen pressure range 70 to 1200 mbar applied over a quartz wall. The obtained effective diffusivity is interpreted in terms of migration of nontrapped H atoms, and migration of retarded H atoms, the latter corresponding to H atoms visiting traps together with H atoms transported as part of H

molecules.

Method for study of molecular gas transport in oxides

• A novel and relatively straightforward method is presented to

quantify diffusion and concentration of gas present in oxide scales

equilibrated in controlled atmosphere.

5 Suggestions for future work

Some suggestions of investigation that may be interesting to carry out in the future are listed below:

In this thesis the effect of Pt on oxidation of only Cr has been studied. To get a wider perspective, other metals (Al, Ni) and alloys (NiAl, GaAs and FeCrSi) should be studied in the similar way. It is also interesting to compare the effect of Pt-coating with Au-coating on the oxygen transport during the oxidation. It may leads to development of new application in the formation of protective metal-oxide.

It is found that the hydrogen uptake rate at steady state is not linearly pressure-dependent

. Together with concentration dependent hydrogen diffusivity in quartz, it should be pointed out that the linear pressure dependence of flux in Paper II is rather a coincidence than a fundamental relation. The finding deserves further examination in other systems so as to verify the knowledge.

The literature data presented in Figure 8 suggest that diffusion of hydrogen, nitrogen and oxygen take places in two modes in quartz:

molecular and atomic. It is proposed that further studies of isotopic hydrogen, nitrogen and oxygen should be performed in order to find out if this kind of diffusion in silica is applicable for nitrogen and oxygen.

Comparison of transport of hydrogen, nitrogen and oxygen in different oxides such as silica, alumina, zirconia and chromia should be done, which is helpful to fundamentally understand the importance in high temperature corrosion research.

It would also be interesting to perform electrochemical measurement of

charge transport to give a more comprehensive picture of transport in

oxides.

6 Acknowledgments

I would like to take this opportunity to express my gratitude and appreciation to the people who have contributed in different ways towards the completion of this thesis:

Prof. Christofer Leygraf for creating the chance to undertake this study.

Doc. Gunnar Hultquist, my supervisor, for your inspiring guidance and help. It is a pleasure to work and discuss with you and to take part of your knowledge and enthusiasm.

Doc. Jinshan Pan and Doc. Inger Odnevall Wallinder for your all kinds of help and consideration.

All my colleagues at the Division of Corrosion Science for providing such a nice working atmosphere. It has been a pleasure working with all of you.

Dr. John Rundgren for fruitful discussions and many valuable ideas and Dr. M. J. Graham for interest shown in our work.

Dr. Peter Szakalos is acknowledged for his help on microscopy, good comments and correcting the language.

To my family in China and my husband, Yanbing Cai, thanks so much for your continuous support and encouragement. Specially, to our coming baby, he/she has shown patience and has not disturbed me much during the licentateexam preparation.

Finally, Financial support from Swedish Foundation for Strategic Research (SFS) and Center of Competence in High-Temperature Corrosion (HTC) is gratefully acknowledged.

Qian Dong (

₎

7 References

1. B. Tveten, G. Hultquist, and T. Norby, Oxid. Met., 51(1999), 221 2. F. A. Golightly, F. H. Stott, and G. C. Wood, Oxid. Met., 10(1976),

163

3. D. P. Whittle and J. Stringer, Phil. Trans. R. Soc. London A, 295(1980), 309

4. J. Stringer, Mater. Sci. Eng. A, 120(1989), 129

5. P. Hou, V. Chia and I. Brown, Surf. Coatings Technol., 51(1992) 73

6. R. J. Hussey and M. J. Graham, Oxid. Met., 45(1996), 349 7. C. H. Xu, W. Gao, and S. Li. Corr. Sci., 43(2001), 671

8. A. Bautista, F. Velasco, and J. Abenojar. Corr. Sci., 45(2003), 1343

9. F. H. Stott, G. C. Wood, and F. A. Golightly, Corr. Sci., 19(1979), 869

10. J. R. Nicholls and P. Hancock, Role of Active Elements in the Oxidation Behaviour of High-Temperature Metals and Alloys (E.

Lang ed. Elsevier, London, 1989)

11. C. M. Cotell, G. J. Yurek, R. J. Hussey, D. F. Mitchell, and M. J.

Graham, Oxid. Met., 34(1990), 173

12. A. Strawbridge and R. A. Rapp, J. Electrochem. Soc., 144(1994), 1905

13. B. A. Pint, Oxid. Met., 45(1996), 1

14. H. Liu, M. M. Stack, and S. B. Lyon, Solid State Ionics, 109(1998), 247

15. E. J. Felton, Oxid. Met., 10 (1976) 23

16. D. L. Douglass, P. Kofstad, A. Rahmel and G. C. Wood, Oxid.

Met., 45(1996) 529

17. G. Hultquist, B. Tveten, E. Hörnlund, M. Limbäck, and R.

Haugsrud, Oxid. Met., 56(2001), 313

18. G. Hultquist, B. Tveten, E. Hörnlund, Oxid. Met., 54(2000), 1

19. W. Beallfower and A. H. Edwards, J. Non-Cryst. Solids,

222(1997), 33

20. P. Campone, M. Magliocco, G. Spinolo, and A. Vedda, Phys. Rev.

B, 52(1995), 15903

21. C. Giradet, J. Plata, J. Breton, and S. Hardisson, Phys. Rev. B, 38(1988), 5648

22. R. E. Lee, J. Chem. Phys., 38(1963), 448

23. R. W. Lee, R. C. Frank, and D. E. Swets, J. Chem. Phys., 36(1962), 1062

24. J. E. Shelby, J. Appl. Phys., 48(1977), 3387

25. T. Åkermark, G. Hultquist and L. Gråsjö, J. Trace Microprobe Techniques, 14(1996), 377

26. J. C. Vickermann, A. Brown, and N. M. Reed, Secondary Ion Mass Spectrometry (Craendon Press-oxford, 1989)

27. J. F. Moulder, W. F. Stickler, P. E. Sobol, and K. D. Bomben, Handbook of X-ray Photoelectron Spectroscopy (J. Chastain ed.

Perkin-Elmer Corporatiopn, Minnesota, 1992)

28. D. Briggs and M. P. Seah, Auger and X-ray Photoelectron Spectroscopy (Practical Surface Analysis, Sec. Ed., volume 1, John Wiley &Sons, Chichester, 1994)

29. G. A. Somorjai, Introduction to Surface Chemistry and Catalysis, John Wiley & Sons, INC, 1994, 345

30. J. E. Shelby, Handbook of Gas Diffusion in Solids and Melts (ASM international, 1996)

31. CRC Handbook of Chemistry and Physics, edited by D. R. Lide, 80

ed., CRC press, Boca Raton, 1999)

32. http://chemlab.pc.maricopa.edu/periodic/H.html

33. E. Hörnlund, G. Hultquist, J. Appl. Phy., 94(2003), 4819

34. J. Crank, The Mathematics of Diffusion (Clarendon, Oxford, 1976) 35. M. Abramowitz and I. A. Stegun, Handbook of Mathematical

Functions (NIST, 1972)

Paper I

Letter

Platinum-induced oxidation of chromium in O

at 800
C

G. Hultquist

, E. H€ o ornlund, Q. Dong

Division of Corrosion Science, Department of Materials Science and Engineering, Royal Institute of Technology, Dr Kristinas V €aag 51, 100 44 Stockholm, Sweden

Received 24 March 2003; accepted 7 May 2003

Abstract

The effects of porous Pt on the oxidation of Cr at 800
C have been studied with the¹⁸O- SIMS technique, gas phase analysis and XPS. In oxide areas with Pt a pronounced inward oxygen transport takes place and a substantial oxide growth near the Cr substrate is observed.

In oxide grown on areas without Pt the counts of CrO ions in SIMS and the binding energy of O (1s) in XPS depend on the distance from the area with Pt. The experimental observations are believed to be a consequence of a high dissociation efficiency of O₂ on areas with Pt in combination with a high diffusivity of O in external and internal oxide surfaces on areas both with and without Pt.

 2003 Elsevier Ltd. All rights reserved.

Keywords: A. Chromium; A. Platinum; B. SIMS; C. Oxidation; Surface diffusion

1. Introduction

The formation of dense and adherent chromium-oxide is essential for many alloys in numerous applications. Oxide grown on ‘‘pure’’ chromium at high temperatures often suffers from spallation but upon removal of hydrogen from the metal substrate the scale adherence can be improved [1]. A more established way to improve the scale adherence is to add oxides of rare earth metals [2,3], but also certain noble metals have a positive effect on oxide scale performance [4]. However, the mechanisms behind the effects of all these additions are not clearly identified and several possible

*Corresponding author. Tel.: +46-8-790-8208; fax: +46-8-20-8284.

E-mail address:gunnarh@kth.se(G. Hultquist).

0010-938X/$ - see front matter  2003 Elsevier Ltd. All rights reserved.

doi:10.1016/S0010-938X(03)00117-3

mechanisms have been discussed for a long time [5]. We have pointed out that these additions catalyse the dissociation of oxygen molecules and that this can be a key factor for the mechanistic understanding of their effect of improved oxide scale adherence [6]. In this work the influence of porous platinum on the oxidation of chromium at 800
C in O2 is addressed with the use of mainly the¹⁸O-SIMS tech- nique.

2. Experimental

A 12 mm· 5 mm · 2 mm piece of Cr (Alfa) was mechanically polished to 2400 mesh with SiC-paper and partly sputter coated with a porous Pt film, with a thickness of approximately 20 nm. The partly Pt-coated sample was then heated stepwise in ultra-high vacuum up to 800
C whereby most of the hydrogen in the sample was outgassed. The remaining hydrogen content was thereby 65 wt. ppm. A two-stage oxidation was performed in a closed reaction chamber at 800
C near 20 mbar O2: the first stage in oxygen with >99%¹⁶O and the second stage in oxygen with 67%¹⁸O. The apparatus used in the oxidation has been described earlier [6,7].

Secondary ion mass spectroscopy, SIMS (Cs^þ source, 10 kV, 200 lm
200 lm sputtered areas with detection from the inner area with £ 70 lm) was used for depth profiling of the oxide formed at different distances from the Pt coated area. The count rates of negative (⁵²Cr¹⁸O) ions were found after subtraction of the contri- bution of (⁵⁴Cr¹⁶O) ions to the emitted species of mass 70. Monoenergetic Al Ka was used in X-ray photoelectron spectroscopy, XPS. The O (1s) photoelectrons were recorded from different areas, 0.4 mm²each, on the sample. The sample with indi- cated analysed positions in SIMS and XPS is shown in Fig. 1.

3. Results and discussion

The oxygen uptake by the sample (based on the pressure decrease in the reaction chamber of known volume) and the isotopic composition of the gas are shown in

Fig. 1. Oxidised Cr sample with indicated area of porous Pt and analysed positions in SIMS and XPS.

Fig. 2. The gas introduced in the second stage of oxidation was^16;16O2 and^18;18O2

with 13%^16;18O2. In Fig. 3, the abundance of ^16;18O2 in the second stage is seen to increase as a result of dissociation–diffusion–association of oxygen. The overall rate for this chain of events can be calculated with good accuracy as long as the abun- dance of the^16;18O2molecules is far from statistical equilibrium [6,8]. Two calculated values are shown with triangles in Fig. 3, which indicate a decreasing dissociation rate. This can be interpreted as a result of a decreasing Pt content in the oxide surface since rates in the order of 1000 and 100 lmol O/(cm², h) represent pure Pt [7] and pure Cr-oxide [9], respectively.

In the second stage of the oxidation the¹⁸O/¹⁶O ratio in the gas was approximately two as seen in Fig. 2. This ratio is present in the outer part of the oxide (near the gas interface) at all distances from the Pt-area as found from the⁵²Cr¹⁸O/⁵²Cr¹⁶O ratios in

Fig. 2. Oxygen uptake in sample (N_{) and}¹⁸_O/¹⁶O ratio in the O2gas (



From gas phase analysis.

Fig. 3. Fraction of^16;18O2in O2(



Fig. 4. However, slightly higher ratios are obtained for¹⁸O/¹⁶O. This might be due to surface water in air exposure upon transfer to the SIMS apparatus. The results in Fig.

4 imply that some oxide growth has taken place near the gas interface at all positions on the sample by metal transport from the substrate. The corresponding ⁵²Cr¹⁸O/

52Cr¹⁶O and ¹⁸O/¹⁶O values near the substrate interface (at a sputter depth where

52Cr¹⁸O has decreased to approximately 10% of its maximum) are also shown in Fig.

4. The ratios in the Pt-area are high, which means that Pt promotes oxide growth near the substrate by oxygen transport through the oxide.

In Fig. 5, counts of ¹⁹⁵Pt and⁵²Cr¹⁸O are shown vs. distance from the Pt-area.

These counts have been obtained by integrating the count rate over sputter time (depth), until the count rate of⁵²Cr¹⁸O has decreased to 1% of its highest value. By combining the data in Fig. 5, a relation between counts of⁵²Cr¹⁸O and counts of

195Pt is shown in Fig. 6. From the results in Figs. 5 and 6, a strong enhancement of the⁵²Cr¹⁸O counts in the Pt-area is found. Also some enhancement of the⁵²Cr¹⁸O counts is observed adjacent to the Pt-area, but not at distances exceeding 4–5 mm.

18O was present only in the second stage of the oxidation (Fig. 2). Therefore a depth profile of¹⁸O-containing ions in SIMS will mimic the relative distribution in depth of oxide growth. The in-depth distribution of⁵²Cr¹⁸O from four positions on the sample is shown in Fig. 7. In all four profiles the depth has been set to unity where the count rate has decreased to 0.5% of its highest value. Also the highest count rate has been set to unity in all four profiles. It is found that:

• a pronounced oxide growth takes place near the substrate on the area with Pt;

• a small, but significant, growth takes place near the substrate in oxide positions 0.1 and 2.9 mm from Pt;

• a virtually zero oxide growth takes place near the substrate in oxide positioned 4.8 mm from the Pt-area.

Fig. 4. Ratios of⁵²Cr¹⁸O/⁵²Cr¹⁶O and¹⁸O/¹⁶O near the gas interface and near the substrate interface of oxides at different positions on the sample. From SIMS measurements.

XPS gives information from the uppermost few nm of solid surfaces. A higher binding energy peak of O (1s) in Fig. 8 is present both in the Pt-area and 4.3 mm from the Pt-area compared with the energy at a distance of 6.1 mm from the Pt-area.

Hence, a major influence of Pt on the uppermost oxide surface is present as far as 4–6 mm from the Pt-area, whereas near the chromium substrate only a minor in- fluence of Pt is present outside the Pt-area (Figs. 4 and 7). These observations strongly point at a fast surface diffusion of dissociated oxygen, O^n, primarily in the external surface of the oxide, but also in internal oxide surfaces including grain boundaries. This scenario can be explained by a ‘‘spill-over’’ of O^nfrom the Pt-area, which acts as a generator for O^n. Such a spill-over effect of adsorbed species is actually a known phenomenon in the field of heterogeneous catalyses where oxide

Fig. 5. Counts of⁵²Cr¹⁸O and¹⁹⁵Pt at different positions on the sample. The counts have been obtained by integration of count rate over sputter time until the count rate of⁵²Cr¹⁸O has decreased to 1% of its maximum. From SIMS measurements.

Fig. 6. Counts of⁵²Cr¹⁸O plotted vs. counts of¹⁹⁵Pt. Data from Fig. 5.

supported noble metal islands are used. In that case the spill-over effect is predicated on rapid surface diffusion [10].

4. Summary and conclusions

The influence of porous Pt on the oxidation of pure Cr in 20 mbar O2at 800
C was studied with the¹⁸O-SIMS technique, gas phase analysis and XPS. Results from the measured emissions of negative (CrO) ions in SIMS,¹⁸O content in the gas phase and O (1s) photoelectrons in XPS at different oxide depth and positions from an area with porous Pt show that:

Fig. 8. O (1s) photoelectron peak from three positions on the sample. From XPS measurements.

Fig. 7. Normalised count rates of⁵²Cr¹⁸O vs. normalised oxide depth from four positions on the sample.

From SIMS measurements.

fluence of Pt is identified which can be explained by a fast diffusion of atomic and/

or ionic oxygen in the external surface of the oxide.

2. A main effect of a Pt-addition on the mechanisms of oxide growth is an increased transport of dissociated oxygen, which predominately takes place on external and internal oxide surfaces. This leads to increased oxide growth near the chromium substrate, which is favourable for the adherence of the oxide.

Acknowledgements

Financial support from KME (G. Hultquist), HTC (E. H€oornlund) and SFS (Q.

Dong) is gratefully acknowledged.

References

[1] B. Tveten, G. Hultquist, T. Norby, Oxid. Met. 51 (1996) 221.

[2] F.A. Golightly, F.H. Stott, G.C. Wood, Oxid. Met. 10 (1976) 163.

[3] R.J. Hussey, M.J. Graham, Oxid. Met. 45 (1996) 349.

[4] E.J. Felton, Oxid. Met. 10 (1976) 23.

[5] D.L. Douglass, P. Kofstad, A. Rahmel, G.C. Wood, Oxid. Met. 45 (1996) 529.

[6] G. Hultquist, B. Tveten, E. H€oornlund, M. Limb€aack, R. Haugsrud, Oxid. Met. 56 (2001) 313.

[7] T.
AAkermark, G. Hultquist, L. Gr
aasj€oo, J. Trace Microprobe Tech. 14 (1996) 377.

[8] E. H€oornlund, Appl. Surf. Sci. 199 (2002) 195.

[9] G. Hultquist, B. Tveten, E. H€oornlund, Oxid. Met. 54 (2000) 1.

[10] G.A. Somorjai, Introduction to Surface Chemistry and Catalysis, John Wiley & Sons, INC, 1994, p. 345.

Paper II

Concentration-dependent hydrogen diffusion in quartz related to trapping

Q. Dong and G. Hultquist

Div. Corrosion Science, Department of Materials and Engineering, Royal Institute of Technology,

Dr. Kristinas väg 51, SE10044, Stockholm, Sweden

a)

Corresponding author: gunnarh@kth.se J. Rundgren

Theory of Materials, Department of Physics, Royal Institute of Technology, SCFAB Building 11, SE10691, Stockholm, Sweden

An analysis of literature data of transport of hydrogen in vitreous silica

strongly points at a contribution of atomic hydrogen to the overall

hydrogen transport. Such a contribution explains an extraordinarily high

diffusivity and permeability of hydrogen in vitreous silica relative to

those of noble gases. In this paper transport of hydrogen through a 0.1cm

thick quartz wall was studied, where different pressures of hydrogen: 70

mbar, 460 mbar, 880 mbar and 1200 mbar were applied at the high-

pressure side at 550

C. The steady state flux of permeated hydrogen was

found to increase proportionally to the applied hydrogen pressure over the

quartz wall, while the steady state concentration of hydrogen in the quartz

wall did not. An iterative fitting procedure of the experimental data

revealed a concentration dependant diffusivity of hydrogen. Knowledge

of both flux and concentration at varying applied hydrogen pressures is

needed for correct calculation of effective diffusivity. Our observation of

a concentration dependent diffusivity can be linked to a situation where

the transport of atomic hydrogen is retarded in reversible traps. We find

that the fraction of hydrogen retarded in traps decreases when the

hydrogen concentration in the material increases.

I. INTRODUCTION

Many different processes are controlled by permeation of gases through materials, for example, oxidation of silicon in O

and/or H

O. Hydrogen has recently been found to influence the oxide growth mechanism in oxidation of many metals at high temperatures. Hydrogen is shown to increase the fraction of oxide growth at the oxide-gas interface during high-temperature oxidation of chromium and a number of alloys.

Studies of the silica-hydrogen system show that several factors can influence the transport of hydrogen itself in quartz, such as impurities, thermal history, and the amount of trapped hydrogen in the bulk.

Lee et al

investigated the diffusion of hydrogen in silica and observed two diffusion regimes, called “normal” and “abnormal diffusion”. The features were explained by hydrogen traps in the silica. Later Shelby

established that permeability and solubility vary little among different silica glasses. It is generally believed that hydrogen is trapped by forming OH-groups, and several mechanisms for the formation of OH are proposed. However, effects of trapping on the diffusivity in oxide systems can probably be further clarified, which is the aim of this paper.

II. ATOMIC HYDROGEN DIFFUSION IN VITREOUS SILICA BASED ON INTERPRETATION OF LITERATURE DATA

FIGs 1, 2 and 3 illustrate the permeability P, the diffusivity D, and the

solubility S, respectively, of He, Ne, Ar, oxygen, and hydrogen in

vitreous silica at 550°C. These data are obtained from Refs. 9 and 10. In

FIGs. 1 and 2, we plot the H

molecule on the solid line determined by

the noble gases He, Ne and Ar, later to be referred to as the noble-gas

line, at the position corresponding to the size of the molecule. For the

hydrogen atom, H, we extrapolate P and D values from the noble-gas line

with a size half that of the H

-molecule

. In FIGs. 1 and 2 the expected

permeability and diffusivity for H

and H differ by six orders of

magnitude. On the other hand, from FIG. 3 one finds that the solubility S

to a fairly good approximation can be considered constant with respect to

the size. Since P roughly equals D times S, it follows from the

extrapolation that the hydrogen atom would have high permeability in combination with high diffusivity.

In FIGs.1 and 2 we observe that the measured values of permeability and diffusivity of hydrogen are higher than those expected from an H

- molecule positioned on the noble-gas line. These differences are labeled

∆P and ∆D, respectively and can be explained by a certain contribution of diffusion of H (An analogue situation should apply for oxygen with a certain contribution of diffusion of O).

Other experiments strongly indicate a considerable contribution of H diffusion to the overall hydrogen transport in vitreous silica

, namely:

1. A porous Pt coating on the high-pressure side of a quartz wall causes an overall increased flux of hydrogen through the wall. An increased dissociation rate of H

takes place on Pt particles with a spill-over to the quartz surface.

2. Virtually all hydrogen (>95%) diffused through a quartz wall has undergone dissociation.

The observations lead to the conclusion that H diffusion gives rise to a substantial contribution to the overall hydrogen transport in vitreous silica. Two scenarios are imagined as to how H diffusion occurs:

A. H diffusion retarded by reversible traps gives rise to the contributions

∆D and ∆P in excess of P and D due to H

diffusion. A particular hydrogen atom is trapped most of the time, but when non-trapped, it moves with great speed.

B. A coupled transport of H and H

, where hydrogen atoms are bonded in H

molecules most of the time. Fast non-bonded H atoms would give rise to the contributions ∆D and ∆P.

In the following sections of this paper the hydrogen transport in vitreous silica is examined, especially in relation to the above scenarios A. and B.

The investigation is based on experiments on hydrogen diffusion through

a quartz wall at 550°C, where the applied hydrogen pressure is varied and the permeability P and the solubility S are measured. In Sec. IV a theoretical analysis of the obtained data is made, where the diffusivity is permitted to vary as a function of the local hydrogen concentration in the quartz plate. Finally in Sec. V we discuss our experimental results with respect to possible atomic hydrogen diffusion.

III. EXPERIMENT

The quartz specimen was a vitreous silica tube (outer diameter 1.2 cm, thickness 0.1 cm and length 32 cm) sealed at one end and enclosed in an outer quartz tube. The impurity levels in wt ppm for this type vitreous quartz were reported by the manufacturer as follows: Al-30; B-0.5; Ca- 3.5; Cu-0.1; Fe-2.5; Li-4; Mn-0.1; Mg-1; P-0.005; K-2.5; Ti-0.7; Na-4.

A gas handling system supplied hydrogen to the free volume between two quartz tubes at a selected pressure. Hydrogen gas permeated through the inner quartz tube wall and via a leak valve the permeated hydrogen was detected by a mass spectrometer, MS, placed in ultra high vacuum. From calibration of the MS signal and pumping speed, the flux was quantified in mol cm

s

(Ref. 13). The quartz tube was heated by a mobile tube furnace to a desired temperature in the range 25-1000

C. The details of the equipment used are described in Ref. 12.

Flux was measured as follows. The quartz tube was held at 550

C sufficiently long to provide a constant low background of hydrogen. At time zero, a known pressure of hydrogen, 70, 460, 840 or 1200 mbar, was introduced into the outer volume. The flux of gas through the specimen was MS monitored. When no further increase of MS signal was detectable, steady state flux was considered to occur.

The outgassing of hydrogen from the specimen was carried out in the

following way. After steady state flux was established, the influx gas was

evacuated down to less than 10

mbar, and the quartz tube was rapidly

heated to 900

C. A MS decay curve was recorded, and the amount of

hydrogen released from the specimen was obtained by integration. For