The influence of ion irradiation on hydrogen chemisorption and diffusion on gadolinium surfaces

Journal of Alloys and Compounds 688 (2016) 553e560

Contents lists available at ScienceDirect

Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom

The influence of ion irradiation on hydrogen chemisorption and diffusion on gadolinium surfaces A. Abaramovich a, Y. Eisen b, N. Shamir c, *, M.H. Mintz a, S. Cohen d, S. Zalkind d a

Department of Nuclear Engineering, Ben-Gurion Univ. of the Negev, POB 653, Beer-Sheva 84105, Israel Soreq Nuclear Research Center, Yavne 81800, Israel c Department of Materials Engineering, Ben-Gurion Univ. of the Negev, POB 653, Beer-Sheva 84105, Israel d Nuclear Research Center-Negev, POB 9001, Beer-Sheva 84190, Israel b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 May 2016 Received in revised form 13 June 2016 Accepted 14 June 2016 Available online 9 July 2016

The role of irradiation induced surface defects on the chemisorption process of hydrogen and inward penetration of chemisorbed H atoms was studied on sputtered, with 4.5 keV Ar þ ions and on annealed gadolinium surfaces. A combination of Direct Recoil Spectrometry (DRS) and Contact Potential Difference (CPD) enabled to distinct between topmost surface and subsurface processes. Simulations of irradiationinduced defects, which assume two types of defects-vacancies due to sputtering and Frenkel pairs, taking account of the annealing processes occurring at different temperatures, have reasonably reproduced the experimental Ar scattering peaks in the DR spectra. It has been concluded that the presence of surface defects has a significant effect on the binding energies of the chemisorbed hydrogen, resulting in hydrogen trapping on the surface and affects the surface to subsurface inward penetration process. © 2016 Elsevier B.V. All rights reserved.

Keywords: Gadolinium Surface Ion irradiation Simulation Hydrogen Chemisorption Diffusion

1. Introduction Metal-hydrogen systems are of interest from both a theoretical and a practical point of view. They can be utilized for energystorage systems, in hydrogen sensor applications and in catalysis. Some of the rare earth hydrides can also be used for switchable optical devises, where their optical properties can change between metallic to transparent, depending on the hydrogen load [1,2]. Due to their chemical reactivity, some of the rare-earths are used as surrogates for actinides corrosion studies [3], avoiding the need for expensive experimental and safety set-ups. The rare-earth metals have also a scientific interest due to their highly localized 4f electrons and some extraordinary electronic and surface magnetic properties. The initial stage of the massive hydrogen reaction is by hydride nucleation and growth on the surface and different families of hydride precipitates were identified and correlated with different types of defects at the oxide and oxide-metal interface [3e6].

* Corresponding author. E-mail address: [email protected] (N. Shamir). http://dx.doi.org/10.1016/j.jallcom.2016.06.130 0925-8388/© 2016 Elsevier B.V. All rights reserved.

Recently, it has been also found that hydroxyl groups adsorbed on the gadolinium oxide surface impede the dissociative chemisorption of hydrogen and inhibit this reaction [7]. Understanding and controlling the surface characteristics is therefore essential in order to understand and control the initial interactions of hydrogen with metals and alloys. In order to avoid the formation of a native oxide and its influence on the hydrogen interaction with metallic surfaces, ultrahigh vacuum (UHV) conditions have to be employed. Hydrogen adsorption on evaporated thin films of Gd (0001) using angle resolved photoemission and scanning tunneling microscopy (STM) revealed that hydrogen adsorption on gadolinium is a dissociative chemisorption process that occurs on the surface and that hydrogen atoms tend to form islands and alter the surface electronic structure [8]. Hydrogen adsorption starts at crystallographic surface imperfections, which form the initial nucleation centers of the hydrides. In the system of evaporated Gd layers on top of W(110), oriented (0001) Gd islands are formed and elastic strain in the Gd islands can also influence the hydride nucleation [8e11]. Li [8] also compared the sticking coefficient and chemisorption of hydrogen on the Gd surface at 120 K and at room temperature and attributed the differences to the possibility of hydrogen diffusion

554

A. Abaramovich et al. / Journal of Alloys and Compounds 688 (2016) 553e560

into the bulk. Sputtering with Arþ ions in the keV range is commonly used to clean the surface during experiments and can also be used to mimic radiation effects and damage at the surface of nuclear materials, influencing the corrosion behavior (see for example Refs. [12e14]). The objective of the present work was to study the influence of radiation damage, as inflicted by Arþ ions irradiation, on hydrogen adsorption and trapping on Gd surfaces, at room temperature and below. Direct recoil spectrometry (DRS) was applied to monitor hydrogen adsorption in the range 140e473 K. A Kelvin probe, measuring contact potential difference (CPD) was used to evaluate surface work function (WF) changes due to hydrogen exposure. This combination enables the distinction between topmost surface and subsurface processes. The sample on which the experiments was performed was polycrystalline Gd, probably containing some bulk oxygen (surface oxygen under the detection limit of AES and the more sensitive DRS) so the processes and calculated values might be somewhat quantitatively affected by its presence. Since the measurements are relative, i.e. sputtered vs. annealed, the effect of surface defects and the principle processes are, in our opinion, not significantly affected by them. 2. Experimental The experiments were performed in an ultra-high-vacuum (UHV) system, pumped by turbo-molecular and titanium sublimation pumps to a base pressure of ~2  1010 Torr. The pressure is monitored by a Bayard-Alpert type ionization gauge and a quadrupole residual gas analyzer (RGA). The direct recoil spectrometry (DRS) is based on grazing irradiation of the surface with a pulsed beam of 3 keV Arþ ions (at 15 to the surface) and time of flight (TOF) measurements of the surface atoms and ions, which are recoiled in a forward direction at the same angle, following the direct collision inflicted by the impinging primary ions (it is common to indicate the scattering angle with respect to the incident angle e therefore the scattering angle is 30 ). A channel electron multiplier detector, which is sensitive to both ions and fast neutrals, is mounted on a long drifting tube at the opposite direction. Typical ion current densities used are ~0.1 nA/mm2 and a DR spectrum can be collected with a total ion dose of <103 ions/surface atom, so the technique can be considered as nondestructive. The main characteristics of this technique are topmost surface sensitivity and detection of light atomic masses, including hydrogen [15e18]. The system also contains standard surface analysis instrumentation for AES and XPS and a Kelvin Probe for contact-potential-difference (CPD) measurements. Sputter cleaning of the sample surface is performed by a rastered, differentially pumped, Arþ ion gun with an energy range up to 5 keV and current of about 2 mA/cm2, located at 30 to the surface. This ion gun was also used to induce radiation damage at the sample surface. The polycrystalline Gd sample (Goodfellow 99.9%, ~1 cm2 area, 1 mm thick) was gradually polished using diamond past, down to 1 mm roughness, cleaned in distilled water and ethanol and then attached by spot-welding to two Ta wires, which are connected to copper feedthrough rods on the manipulator. The sample can be cooled down to ~140 K, by dipping the copper rods in liquid nitrogen and heated up to ~1000 K by driving an electric current through the wires. The sample temperature was monitored by a ChromeleAlumel (type K) thermocouple spot-welded to the sample back. The Gd sample surface was continually sputtered at room temperature for 24 h with 4.5 keV Arþ ions, in order to obtain a

steady state and spatially uniform roughness of the surface. Scanning electron microscope (SEM) micrographs, taken after prolonged sputtering, revealed the formation of relatively flat facets on the surface, as depicted in Fig. 1. Annealing of the surface was performed at 673 K for 30 min. The radiation damage was induced by irradiation of the annealed sample with 4.5 keV Arþ ions for different doses and sample temperatures. The ion dose was evaluated by measuring the sample current during irradiation. In order to improve accuracy and overcome the secondary electron emission, the sample was biased at þ100 V during current measurements to recollect those electrons. Two kinds of experiments were performed: 1) Radiation damage measurement The radiation damage was monitored by the decrease of the DR Ar(SS) peak. The polycrystalline gadolinium metal was bombarded uniformly (starting from an annealed surface) at various temperatures, by 4.5 keV argon ions, impinged the surface at 30 . The flux of the argon ions during irradiation was held constant at an average value of 7.4  1012 ion/s/cm2. At each temperature and for short periods of time (50sec each), DR spectra were taken and were used to probe the surface as a function of time. 2) Hydrogen adsorption measurements For each adsorption experiment, measurements were taken on an annealed and on a sputtered surface, in order to evaluate the net contribution of radiation damage. In order to equalize the initial conditions during hydrogen exposure and rule out the difference in surface damage due to ion irradiation at different temperature, all the radiation damages for the adsorption experiments were performed by sputtering at room temperature to 1.8  1016 ions/cm2, before lowering the sample temperature for hydrogen exposure. 3. Results and discussion 3.1. Analyzing radiation induced damage at the Gd surface 3.1.1. DRS results DR spectra from clean Gd surface, presenting the forward scattered Ar(SS), as a function of the 4.5 keV ion irradiation dose is presented in Fig. 2. As the radiation damage to the surface increases

Fig. 1. SEM micrograph of the gadolinium surface after a prolonged sputtering, depicting the steady state features developed at the surface.

A. Abaramovich et al. / Journal of Alloys and Compounds 688 (2016) 553e560

DRS Intensity [counts]

5000 4000 3000

annealed 6x1014 ions/cm2 1.2x1016 ions/cm2 1.8x1016 ions/cm2

Ar(SS)

298K

2000 1000 0 13.5

14.0

14.5

15.0

Time of Flight [μsec] Fig. 2. DR spectra of the elastically scattered Ar(SS) as a function of irradiating dose of the annealed surface (sputtered with 4.5 KeV Ar þ ions) at room temperature. The attenuation of the Ar(SS) corresponds to the net radiation damage accumulated at the surface.

with Arþ ion dose, there is an attenuation of the Ar(SS) signal, relative to the annealed surface. The attenuation of the Ar(SS) peak, as a function of the radiation dose at different sample temperatures, is presented in Fig 3. It is clear that there is self-annealing of the damage with rising temperature during the irradiation process and at surface temperature above 473 K the annealing process is rapid enough to eliminate defects accumulation at the surface. In contrast to isolators and even semiconductors, where radiation to similar ion energy and dose can cause amorphization of the surface region, for metallic surfaces the crystallographic nature of the surface is remained [19]. According to TRIM [20] simulation, sputtering leads to the formation of vacancies in the gadolinium lattice which are generated within a thicker region than the DRS probing depth. The vacancies can be correlated to Gd atoms ejected from the topmost surface and to Frenkel defects at the subsurface. Simulation programs, like TRIM [20], can provide a tool to estimate the nature and range of ion irradiation damage of the target, but the main disadvantage of TRIM is that it does not account for temperature driven processes as self-annealing of the induced damage and therefore the quantification of the damage may be valid only for

555

radiation at low temperatures, where diffusion is minimized. To account for temperature influence on the damage formation and annihilation, simulation was performed, using a phenomenological formalism with two kinetic parameters: a) activation energy for recombination of vacancies and b) activation energy for recombination of vacancies and interstitials (Frenkel defects). The decrease of Ar(SS) with irradiation dose is a combination of temperature independent radiation damage and temperature dependent thermal annealing. Since the 223 K and 173 K curves look the same, we can conclude that no effective thermal annealing is taking place below about 223 K and this combined curve can serve as a basis for the evaluation of the net thermal annealing at higher temperatures. Fig. 4 presents the net annealing contribution to the accumulated radiation damage, by subtraction of this base line from the Ar(SS)T lines for all other temperatures. This contribution will have to be taken into account, dealing with the influence of radiation damage with hydrogen adsorption and diffusion. 3.1.2. Simulation The interaction of ions with a surface is a complex process. An incoming ion can cause direct knockout of surface atoms and it can penetrate into the solid and slow down by electronic stopping and by nuclear collisions causing cascade damage. As a result, target atoms are ejected and displaced causing disorder and surface roughness, initially on the atomic scale and later on the microscopic and macroscopic scales. Prolonged sputtering of the surface results in the appearance of topographic features that depend on matrix crystallography, dislocations, impurities etc. [21,22] (see Fig. 1). This damaged surface has a higher energy and dangling bonds and it is usually more reactive towards adsorbed gases. Annealing the sputtered surface at high temperature can cause atoms rearrangement and “smoothing” of the surface on the atomic and microscopic scales, but less on the gross macroscopic topography. Ion scattering techniques can provide a powerful probe to follow and evaluate surface damage and recovery after ion irradiation and annealing as well as their influence on gas adsorption and reactions [12,14,21e23]. In the DRS technique, the Ar deflected from the surface is actually low angle ion scattering, and in the present work it was used to probe the radiation induced damage and vacancy formation at the surface. The damage at the surface of polycrystalline gadolinium layer resulted by the bombardment of 4.5 keV Arþ ions can be decomposed into two components:

Argon irradiation dose [ions/cm2] 0.0

15

5.0x10

16

1.0x10

1.5x10

1.0

16

2.0x10 473 K 373 K 323 K 298 K 273 K 223 K 173 K

0.8

0.5 473 K 373 K 323 K 298 K 293 K 173/223 K

0.4

Ar(SS)T-Ar(SS)173/223

0.9

Normalized Ar(SS)

16

0.7

0.3 0.2 0.1

0.6 0.5

0.0 0

10

20

30

40

Time [min] Fig. 3. The DRS Ar(SS) intensity attenuation vs. time and ion dose irradiation at different sample temperatures.

0

10

20

30

40

Time (min) Fig. 4. Thermal annealing of the radiation damage vs. time, for all measured temperatures.

A. Abaramovich et al. / Journal of Alloys and Compounds 688 (2016) 553e560

1. Generation of gadolinium vacancies e sputtering is causing removal of gadolinium atoms from the surface, leaving vacancies of gadolinium atoms in the lattice. Those vacancies are filled, with time, by atoms from deeper layers, depending on the temperature of the bombardment. 2. Generation of Frenkel defects e these defects are caused by the recoiled gadolinium atoms. Interstitials and vacancies of gadolinium atoms are created but also annealed with time, depending on the temperature. Diffusion of interstitials is faster than the diffusion of vacancies. The two mechanisms behave differently as a function of time and temperature and also might have different activation energies. The first mechanism depends on the diffusion length (or diffusion time) between the vacancy and the interstitial whereas the second mechanism depends on the reorganization of the lattice with time and temperature to obtain minimum energy and this might involve a longer diffusion length. One general statement can be made that either diffusion rates or reorganization rates of the lattice, behave with temperature T according to the Arrhenius formula:

20

15

10

5

0

0

5

10

15

20

25

30

35

Gd layer thickness (A°) Fig. 6. Number of Frenkel defects per incident Ar ion vs. Gd layer thickness.

The activation energy, E, for the annealing mechanism of Frenkel defects is different from that of the annealing mechanism of vacancies. It is the aim of the present chapter to quantify E of both mechanisms of the annealing processes and try to explain the experimental data of the DRS, as manifested in Fig. 3. The vacancies formed during irradiation are generated in a much thicker region than the DRS technique can probe. It is not clear how these vacancies are filled to form a new lattice state of minimum energy. It is assumed that the vacancies are filled first at layers near the surface. Since sputtering is continuous at each temperature, the DRS technique samples each time a different layer. Because the entrance and the exit directions of the probing and measured Ar ions in the DRS are at 15 relative to the sample plane, the time of flight spectra measured (Fig. 2) implicates that the probing thickness of the DR is about 3 Å. However, the sputtering due to argon ions of energy 4.5 keV generates vacancies also at a much deeper layer than 3 Å as can be seen in Figs. 5e8. Therefore all calculations were carried out on a layer of 18 Å thick, a thickness

2.0

1.5

1.0

0.09

0.06

0.03

0.00

5

10

15

20

25

30

0

Fig. 7. Number of Gd vacancies per Å vs. Gd thickness.

1.0

0.8

0.6

0.4

0.2

0.0

0.5

0

Gd thickness(A )

Number of Frenkel defects per A°

(1)

Number of Gd vacancies per A0

0.12

  dN E ¼ N0  exp  dT kT

Sputtered atoms per incident ion

25

Vacancies per ion

556

0

7

14

21

28

35

Gd thickness(A0) 0.0

Fig. 8. Number of Frenkel defects per Å vs. Gd thickness.

0

5

10

15

20

25

30

Layer thickness(A) þ

Fig. 5. Number of vacancies by sputtering per single 4.5 keV Ar ion at normal.

60

to surface

which reflects a maximum of the gradient in the number of Frenkel defects per thickness. It is further assumed that the results obtained in the DRS experiment, probing the first layer of 3 Å, are affected by

A. Abaramovich et al. / Journal of Alloys and Compounds 688 (2016) 553e560

the damage at a thicker layer of 18 Å. As we have indicated above, there are two interactions: The first is a continuous one using 4.5 keV argon ions that is responsible for the sputtering, and the other is momentary, a TOF measurement using 3 keV argon ions sampling the damage at the surface region during irradiation. The 3 keV argon ions interact with the surface by  Rutherford scattering and scatters Ar atoms/ions at 30 . The scattering behaves in the center-of-mass system as follows:

ds ðz1  z2 Þ2 ðmb=srÞ ¼ 1:296  dU E2 sin4 ðq=2Þ

(2)

where Z1 ¼ 20 and Z2 ¼ 64, E is the center of mass energy (2.4 meV) and q is the scattering angle of the incident argon atom in the center of mass system. The number of scattered argon ions into a solid angle dU is proportional to the following parameters: the number of gadolinium atoms in a 3 Å thick layer and the differential Rutherford cross section, ds=dU given by Equation (2). Because of damage at the surface due to sputtering by the 4.5 keV argon ions, the detected number of argon ions of energy ~2.8 keV in the DRS measurements varies with irradiation time and temperature (Fig. 2). The latter can be written as a0-N’v(t), where a0 is the number of gadolinium atoms in a layer of 18 Å prior to the 4.5 keV argon irradiation and N’v(t) is the number of vacancies due to sputtering and Frenkel defects generated after a certain time t of irradiation. To find the vacancies due to sputtering (Nv) and Frenkel defects (Nv1) the differential Equations (3) and (4) are solved using the boundary condition that at t ¼ 0, Nv ¼ 0 and Nv1 ¼ 0, can be written as follows:

dNv ¼ aI  bNv dt

(3)

and

dNv1 ¼ gI  dNv1 dt

(4)

where I is the flux of 4.5 keV argon ions, a and g are the number of sputtered gadolinium atoms and the number of recoiled Gd atoms per unit flux of incident argon ions and b and d are expressed by the Arrhenius formulas expressed as:

  E b ¼ A  exp  a kT 

d ¼ A  exp 

Eav kT

557

3.1.3. Simulation outcome The gadolinium sputtering probability of a 4.5 keV argon ion is shown in Fig. 5 as a function of the layer thickness. It can be seen that the saturation value is about 2 vacancies per incident ion. The reason for a number higher than 1, is the large straggling or multiple Coulomb scattering of each ion entering the gadolinium lattice. For a layer thickness of 3 Å, a ¼ 0.12. The number of initial atoms in a layer of 18 Å is a0 ¼ 5.43  1015. The number of Frenkel defects per incident argon ion is much larger than 1. For a layer thickness of 3 Å, av ¼ 0:3, as seen in Fig. 6. It is also seen in Figs. 7 and 8 that the maximum sputtered gadolinium atoms per unit thickness occurs at a layer thickness of 5e10 Å and the maximum number of Frenkel defects per unit thickness occurs at about 18 Å. Fig. 9 shows the theoretical results for the remaining gadolinium atoms in the 3 Å thick layer according to Equation. (6) as a function of irradiation time, where the ordinate, in Fig. 9, is given by:

YðtÞ ¼ ða0  Nv ðtÞ  Nv1 ðtÞÞ=a0

(9)

Fig. 10 presents the number of perturbed Gd atoms after 40 min of sputtering vs. temperature, assuming 1016 atoms per unit area. The total number of 4.5 keV Ar ions impinging the surface after 40 min of irradiation is 1.8  1016/cm2. The number of remaining perturbed atoms at room temperature (298 K) is 6.8  1015. The number of atoms in a layer of mean thickness 18 Å is about 6  1015. This means that at room temperature (298 K) the mean remaining perturbed atoms is about 40% of the number of 4.5 keV Ar atoms impinging the surface and on the other hand is almost equal to the total number of atoms in a mean layer of 18 A. The DRS experiment does not measure the real damage to the surface caused by the 4.5 keV Arþ ions but only the effect in a damaged 3A thick top layer. The damage occurs at much deeper thickness and there is an influence, due to diffusion, on the top 3 Å layer by those deeper layers. By comparing the simulation results obtained in Fig. 9 to the experimental DRS results (Fig. 3) in a temperature range 298e473 K, it can be seen that decent agreement to the experimental data is obtained with activation energies Ea ¼ 1:17eV and Eav ¼ 1eV. 3.2. Hydrogen surface adsorption and inward diffusion As mentioned before, the topmost surface sensitivity and the ability to “see” hydrogen makes DRS an ideal technique for

(5)

 (6)

where Ea and Eav are the activation energies for filling up gadolinium vacancies by gadolinium atoms from interior layers (not necessarily of the same thickness) and the activation energy for filling up Frenkel vacancies respectively. A is the Debye frequency (The Debye frequency contains the velocity of sound in matter, the velocity of sound in gadolinium lattice is equal to 2.68  105 cm/s). The solution of the above differential equation for Nv ðtÞ and Nv1 ðtÞ, assuming that the 4.5 keV argon current, I, remains constant at each temperature, is given by the following equation:

Nv ðtÞ ¼

 aI  1  ebt b

Nv1 ðtÞ ¼

 gI  1  edt d

(7)

(8)

Fig. 9. Normalized unperturbed Gd Atoms, at various temperatures vs. time. It can be seen that the calculations simulate well the experimental results of DRS Ar(SS) intensity attenuation vs. time (Fig. 3).

A. Abaramovich et al. / Journal of Alloys and Compounds 688 (2016) 553e560

1E16

240

1E15

200

H(DR) Intensity [counts]

Remained perturbed Gd atoms

558

1E14

1E13

1E12

1E11

300

350

400

450

160

140K 170K 220K 270K RT (a) Sputtered A

120

B

D

C

80 40

500

Temperature (K)

0

Fig. 10. Number of perturbed Gd atoms after 40 min of sputtering vs. temperature.

500

1000

1500

2000

Time [sec]

(i) For both types of surfaces (i.e. sputtered and annealed) the accumulation of topmost surface H atoms is decreasing with increasing exposure temperature. This effect can arise either by a lower adsorption rate caused by relative increase of H desorption relative to H adsorption, or by an increased inward penetration of the chemisorbed surface H atoms. This trend is much more pronounced for the annealed surface, where at room temperature and above almost no accumulation of surface hydrogen is detected, regardless the applied exposure pressure (up to the highest pressure of 2  106 Torr). For the sputtered surface, very low accumulation of surface H is detected at the lower exposure pressure, but it significantly

240

140K 170K 200K 220K 270K

200

H(DR) Intrnsity [counts]

continuous monitoring and following hydrogen adsorption on the surface. Fig. 11a presents DRS spectra, taken for a sputtered surface exposed to different doses of H2 at 140 K. For the low temperature exposures, where inward diffusion is negligible, the adsorption curves were fitted successfully to the single site (clustering) model [24], Fig. 11b, in agreement with the kinetics obtained by Li et al. [8] for low temperature adsorption. Fig. 12 depicts the H(DR) intensity vs. time, for continuous exposure of sputtered (a) and annealed (b) surfaces to hydrogen. The different sections in figure (labeled A, B, C) correspond to different exposure pressures (which have been increased gradually at some given exposure doses. Section D depicts H(DR) intensity after hydrogen evacuation. Comparing both figures (Fig. 12 a,b) several conclusions may be drown:

160

(b) Annealed

120

A

C B

D

80 40 0

500

1000

1500

2000

Time [sec] Fig. 12. H(DR) vs. hydrogen exposure for (a) the sputtered surface. (b) the annealed surface A: P ¼ 2  108 Torr, B: P ¼ 2  107 Torr, C: P ¼ 2  106 Torr, D-vacuum.

increases for higher pressures (above about 2  106 Torr). In any case, it is unlikely that at room temperature the desorption rates should be high enough to cancel totally the net H chemisorption, since it is known that gadolinium hydrides are formed in fact at these low pressure ranges [11]. Also, as

Fig. 11. (a) DRS spectra of hydrogen exposures (to various doses) on an ion irradiated (sputtered) gadolinium surface at 140 K; (b) Normalized H(DR) vs. hydrogen exposure for the sputtered and annealed surfaces, together with the fits to the clustering model [24], for the 140 K exposures.

A. Abaramovich et al. / Journal of Alloys and Compounds 688 (2016) 553e560

presented later by our CPD measurements, hydrogen atoms do adsorb by both gadolinium surfaces (sputtered and annealed), within all experimental exposure doses.

The net effect of the radiation damage on hydrogen chemisorption on the surface is depicted in Fig. 13, by presenting the difference between the H(DR) intensities of both surfaces, i.e. I(sputtered)-I(annealed) vs. exposure (Fig. 12, aeb). The net contribution of radiation damage to the processes of adsorption and diffusion (Fig. 13), displays some feature, common to all temperatures, with different temperatures dependences: a. Initially, as stated before, the excess of radiation damage sites causes enhanced adsorption, being manifested more for the lower temperatures. This increase overtakes the increase of diffusion due to the radiation damage. At a certain exposure (temperature dependent), when the radiation induced adsorption sites are close to being filled with adsorbates, the diffusion increase balances the adsorption and an intensity maximum is reached. For 298 K, it seems to be beyond the measurement range (due to slower adsorption and/or faster diffusion). For 270 K, as well as for 140 K it is at ~1000 L and for the other temperatures it is within the exposure rang. What seems to be an inconsistency in the temperature dependence of this maximum is (except for 140 K, that will be discussed later) probably an outcome of the combination of the processes of adsorption and diffusion, having a different temperature dependence.

H(DR)sputt-H(DR)ann

C

298 K 270 K 220 K 200 K 170 K 140 K

50

0

1

10

100

1000

H2 Exposure [L] Fig. 13. The H(DR) intensity difference between the sputtered and annealed surfaces vs. hydrogen exposure. A: P ¼ 2  108 Torr, B: P ¼ 2  107 Torr, C: P ¼ 2  106 Torr.

0.12

298 K

0.08

H2/Gd Annealed

0.04

Δφ

(ii) The fact that for the higher temperature experiments, the amount of accumulated surface H atoms are much lower for the annealed surface as compared to the sputtered one (for the same doses), may indicate that the inward H diffusion is more pronounced for the defected-free surface. This view is further substantiated by the decay behavior of the surface H(DR) signals displayed during the evacuation stages that conclude each exposure experiment, in Fig. 12a and b (i.e. the last section labeled D-vacuum). It is evident that for the sputtered (defectscontaining) surface, the lower temperature saturation values, attained at the higher exposures, are maintained almost unchanged during evacuation stage and only for the higher temperatures a significantly decay is observed (Fig. 12a), whereas for the annealed (defect-free) surface, a pronounced decay takes place even for the lower temperature exposures. It may therefore be concluded that the sputtering produced surface defects provide more stable (i.e. energetically favorable) surface chemisorption sites for the dissociated H atoms, leading to higher inward diffusion barriers. Hydrogen atoms thus are trapped and reside more on the topmost surface of the sputtered sample than on the annealed one. (iii) Comparing the initial hydrogen coverage increase vs. exposure, even at 140 K, clearly indicates that on the sputtered surface hydrogen accumulates much faster than on the annealed surface. According to Li [8], the dissociation barrier for this dissociative chemisorption to occur should be very low, 0.02e0.03 eV, if any, so surface defects should play a minor role in dissociation. The faster initial H accumulation on the sputtered surface probably results from reduced inward migration due to the surface defects, but it cannot be ruled out that these defects contribute also to the dissociation stage.

B

A

100

Hence, the above effect of decreasing surface hydrogen with increasing temperature is, probably, due to increased inward diffusion (surface to subsurface penetration) rather than due to desorption.

559

0.00 -0.04

160 K

-0.08 -0.12

0

50

100

150

200

250

300

Exposure [L] Fig. 14. CPD of Gd exposed to hydrogen at 298 K and 160 K.

b. For the higher temperatures (298 K and 270 K), there is a slow increase of the net hydrogen adsorption on the damaged surface (for the annealed surface, there is practically no adsorption), meaning that though the radiation damage also increases the inward diffusion, the radiation damage sites of adsorption (that are not annealed) are deep enough to not release all adsorbed species, at the higher pressure steady states. The CPD measurement for the annealed surface, at RT and 160 K, is presented in Fig. 14 and substantiates the assumption that for the annealed surface, at high temperatures, the hydrogen diffuses to the sub-surface, rather than adsorbs on the surface, as is the case for the low temperatures. For the 160 K measurement, the CPD decreases, as expected from the dipole moment created by the charge transfer of electron from the adsorbed hydrogen to the surface below it. For 298 K, on the other hand, the increase of CPD points to sub-surface migration of hydrogen, reversing the direction of the dipole moment. 4. Conclusions 1. Combined measurements by DRS and CPD may provide an efficient method for studying surface chemisorption and subsurface penetration of hydrogen on metallic surfaces.

560

A. Abaramovich et al. / Journal of Alloys and Compounds 688 (2016) 553e560

2. Irradiation-induced defects have a pronounced influence on the dissociative chemisorption of H2 on polycrystalline gadolinium surfaces. Such defects have a very pronounced influence on the binding energies of surface chemisorbed H atoms, hindering their surface to subsurface penetration kinetics. 3. Simulations of irradiation-induced defects, which assume two types of defects-vacancies and Frenkel pairs (of vacancyinterstitials), taking account of the annealing processes occurring at different temperatures, may reasonably reproduce the experimental Ar scattering peaks in the DR spectra. Such simulations yielded an estimate for the activation energies for the annealing of the two types of defects, with Ea ¼ 1.17 eV for the vacancies annealing and Eav ¼ 1.0 eV for the Frenkel defects annealing process. 4. As indicated in the introduction, the study was performed on a polycrystalline Gd sample, probably containing some bulk oxygen. However, we believe that this does not have a significant effect on the processes taking place and on the contribution of surface defects (measured differentially). It can, however somewhat affect the calculated values. Acknowledgment This work was partially supported by a grant from the Israel Council for Higher Education and the Israel Atomic Energy Commission and a grant from the Ministry of National Infrastructure, Division of R&D.

References [1] J.N. Huiberts, R. Grissen, J.H. Rector, R.J. Wijngaareden, J.P. Dekker, D.G. de Groot, N.J. Koeman, Nature 380 (1996) 231. [2] E. Nowicka, R. Nowakowski, R. Dus, Appl. Surf. Sci. 254 (2008) 4346. [3] J.P. Knowles, G. Rule, M. Brierley, Corros. Sci. 77 (2013) 31. [4] Y. Ben- Eliahu, M. Brill, M.H. Mintz, J. Chem. Phys. 111 (1999) 6053. [5] G. Benamar, D. Schweke, J. Bloch, T. Livne, M.H. Mintz, J. Alloys Compd. 477 (2009) 188. [6] G. Benamar, D. Schweke, G. Kimmel, M.H. Mintz, J. Alloys Compd. 520 (2012) 98. [7] G. Benamar, D. Schweke, N. Shamir, S. Zalkind, T. Livneh, A. Danon, G. Kimmel, M.H. Mintz, J. Alloys Comp. 498 (2010) 26. [8] D. Li, J. Zhang, P.A. Dowben, M. Onellion, Phys. Rev. B 48 (1993) 5612. [9] M. Getzlaff, M. Bode, R. Wiesendanger, Surf. Sci. 410 (1998) 189. [10] M. Getzlaff, M. Bode, R. Pascal, R. Wiesendanger, Phys. Rev. B 59 (1999) 8195. [11] H. Realpe, N. Shamir, Y. Manassen, M.H. Mintz, J. Alloys Compd. 473 (2009) 521. [12] S. Zalkind, M. Polak, N. Shamir, Surf. Sci. 513 (2002) 501. [13] J. Stultz, M.T. Paffett, S.A. Joyce, J. Phys. Chem. B 108 (2004) 2362. [14] S. Cohen, M.H. Mintz, S. Zalkind, A. Seibert, T. Gouder, N. Shamir, Solid State Ionics 263 (2014) 39. [15] J.W. Rabalais, CRC. Crit. Rev. Solid State Mater. Sci. 14 (1988) 319. [16] O. Grizzi, M. Shi, H. Bu, J.W. Rabalais, Rev. Sci. Instrum. 61 (1990) 740. [17] M.S. Hammond, J.A. Schultz, A.R. Krauss, J. Vac. Sci. Technol. A 13 (1995) 1136. [18] M. Tassotto, P.R. Watson, Surf. Sci. 464 (2000). [19] D.W. Moon, Y. Ha, H.K. Kim, K.J. Kim, H.S. Kim, J.Y. Lee, S. Kim, Appl. Surf. Sci. 150 (1999) 235. [20] T.R.I.M.,J. Ziegler, Transport Properties of Ions in Matter, 1992. [21] U. Valbusa, C. Boragno, B. de Mongeot, J. Phys. Condens. Mater. 14 (2002) 8153. [22] L.S. Dake, D.E. King, J.R. Pitts, A.W. Czanderna, Ion beam bombardment effects on solid surfaces at energies used for sputter depth profiling, in: A.W. Czanderna, T.E. Mady, C.J. Powell (Eds.), Methods of Surface Characterizations, Vol. 5, Kluwer Academic Publishers, 2002. [23] H.J. Kang, D.W. Moon, H.-I. Lee, J. Surf. Anal. 15 (2009) 216. [24] J.A. Schults, M.H. Mintz, T.R. Schuler, J.W. Rabalias, Surf. Sci. 146 (1984) 438.