Temperature dependence of D atom adsorption on polycrystalline tungsten

Applied Surface Science 282 (2013) 478–486

Contents lists available at SciVerse ScienceDirect

Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc

Temperature dependence of D atom adsorption on polycrystalline tungsten Sabina Markelj a,∗ , Olga V. Ogorodnikova b , Primoˇz Pelicon a , ˇ Thomas Schwarz-Selinger b , Iztok Cadeˇ za a b

Joˇzef Stefan Institute and Association EURATOM-MESCS, Jamova cesta 39, SI-1000 Ljubljana, Slovenia Max-Planck-Institut für Plasmaphysik, EURATOM Association, Boltzmannstr. 2, D-85748 Garching, Germany

a r t i c l e

i n f o

Article history: Received 15 April 2013 Accepted 30 May 2013 Available online 17 June 2013 PACS: 68.43.-h 68.43.Mn 68.47.De 82.80.Yc 28.52.Fa Keywords: In situ ERDA Tungsten Hydrogen atom Hydrogen isotope exchange Deuterium adsorption

a b s t r a c t Temperature dependence of D atom adsorption on polycrystalline tungsten was studied by in situ ion beam method Elastic Recoil Detection Analysis (ERDA). A new procedure named thermoadsorption (TA) was developed for this study, where the sample is first exposed to a deuterium atom beam at high temperature and then, while being continuously exposed to the atom beam, is slowly cooled down. H and D concentrations are determined during this cooling by ERDA. A stepwise increase of the surface areal density was observed starting from (1.2 ± 0.3) × 1015 D cm−2 at sample temperature around 750 K, to (2.2 ± 0.3) × 1015 D cm−2 when temperature was around 600–500 K and final increase to (6.8 ± 0.6) × 1015 D cm−2 when sample temperature was below 440 K. From this, three individual binding states were identified for the studied polycrystalline tungsten. We present a numerical model adequate to our experimental procedure which was developed by taking into account all relevant surface processes. The binding energies for desorption/adsorption were derived by modeling the TA data with the numerical model and were determined to be: 1.05 ± 0.06 eV, 1.7 ± 0.08 eV and >2.2 eV. Isotope exchange at 485 K sample temperature was also measured and modeled. An estimate of the reflection coefficient for 0.2 eV hydrogen atoms on polycrystalline W was obtained from modeling the isotope exchange data and was determined to be 0.96 ± 0.02. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Studies of hydrogen interaction with surfaces represent an extensive and very active research activity due to the fundamental nature of the subject, hydrogen atoms or molecules being the simplest adsorbing species, and on the other hand due to the large number of hydrogen related applications (fusion, hydrogen fuel cells, plasma processing, etc.). Main interest in this paper is the hydrogen interaction with tungsten. One can find in the literature extensive experimental [1–7] and theoretical [8–12] studies on hydrogen molecule interaction with tungsten but only few studies were done with hydrogen atoms [11]. With atoms an additional process comes into play which is not possible with molecules, i.e. Eley–Rideal type atom recombination described in more detail below. For hydrogen on tungsten reactions on the surface are expected to be dominant, since it has a positive heat of solubility and therefore diffusion of hydrogen atoms from the surface in the bulk is small

∗ Corresponding author. Tel.: +386 15885265. E-mail address: [email protected] (S. Markelj). 0169-4332/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.apsusc.2013.05.157

[13]. A polycrystalline tungsten surface consists predominantly of (1 1 0), (1 1 1) and (1 0 0) surfaces [2]. The saturation surface density obtained from the thermodesorption (TD) study [2] for these single-crystal surfaces is 1.05 × 1015 H cm−2 (0.74 monolayer (ML)), 2.6 × 1015 H cm−2 (4.5 ML) and 1.5 × 1015 H cm−2 (1.5 ML) for W(1 1 0), W(1 1 1) and W(1 0 0), respectively. Several binding states were found to exist on these single crystal W surfaces, each having a characteristic desorption energy Edes [2]. These surface properties are not known for polycrystalline tungsten but on the other hand can influence hydrogen adsorption and reemission. Namely, hydrogen molecule production by atom recombination on the surface, especially vibrational excitation of the molecules, is influenced among others by the properties of the surface adsorption sites [14,15]. The interest in the hydrogen interaction with tungsten has increased recently, since tungsten is a material of choice for the target plates in the divertor of modern fusion devices, tokamaks [16,17]. In the divertor, the bottom part of the tokamak, plasma is intentionally directed to the target plates [16] which are therefore subjected to high particle fluxes and heat loads. The prediction of hydrogen retention in W is an important issue especially due to the safety limitations on the total amount of the hydrogen isotope

S. Markelj et al. / Applied Surface Science 282 (2013) 478–486

tritium in the reactor. Therefore, the control of the in-vessel tritium inventory is of high importance [17]. The deuterium retention in W exposed to low-energy deuterium plasma and ion beams has been studied and theoretically supported [18–24]. Thermodesorption spectroscopy (TDS) and nuclear reaction analysis (NRA), using the D(3 He,p)4 He nuclear reaction, are commonly used techniques for determination of the binding/trapping sites and hydrogenic retention after material exposure to plasma or ion beams. There was no study of retention and isotopic exchange in W exposed to low energy (< eV) atoms. It is very important from the fundamental point of view to understand the hydrogen isotope migration and retention in W since thermal atoms are expected not to produce additional damage in the material but only decorate the existing trapping sites. Here we report on the first detailed study of low energy (0.2 eV) hydrogen/deuterium atom interaction on polycrystalline (PC) tungsten as a function of the sample temperature. To study hydrogen adsorption we have developed a new procedure named thermoadsorption (TA), where the sample is first exposed to a deuterium atom beam at high temperature and then, while being continuously exposed to the atom beam, the sample is gradually cooled down. In contrast to the usual temperature programmed desorption spectroscopy, where sites are being emptied during sample heating [2], we are in this case observing filling of different adsorption sites during the sample cooling. The main goal of this study is to obtain the temperature dependence of the deuterium surface areal density on polycrystalline tungsten, in the temperature range relevant for the future magnetically confined fusion device, ITER. The dynamics of hydrogen atom adsorption on tungsten are described by an extension of a kinetic model by Jackson et al. [25], which is described to some detail in Section 2. Beside TA, the isotope exchange was also studied on tungsten sample. This process is important due to its fundamental character, one pre-adsorbed isotope species being replaced by the other one and also due to its possible application in fusion for tritium removal from the tokamak wall. The isotope exchange was studied at 485 K sample temperature where different surface states were populated. The diagnostic tool used in the present study is the ion beam method Elastic Recoil Detection Analysis (ERDA). The study was performed in situ and in real time. Although the ERDA method is providing the depth profile of the hydrogen concentration up to 400 nm in tungsten, the surface part of this profile is dominant since hydrogen diffusion from surface to bulk is small allowing clear study of surface processes. Moreover, ERDA is one of the rare techniques that can give simultaneously absolute hydrogen concentrations for both isotopes, H and D. Therefore it is an excellent method for observing the above described phenomena (especially isotope exchange), since both isotopes, H and D, are followed simultaneously.

trapped on the surface but some time is needed before it is thermalized. These atoms hop on the surface and during hopping they can recombine with an adsorbed atom, producing a molecule. This is the so-called hot-atom concept of the recombination mechanism [29]. Since we were observing the temperature dependence of the atom interaction with the surface which is becoming increasingly saturated, we additionally need to consider the Langmuir–Hinshelwood (LH) recombination process [28]. This is a thermally activated process, where two adsorbed atoms diffuse on the surface [30] and if they come close to each other, they can recombine and desorb as hydrogen molecule. Atom diffusion and desorption from the surface is appreciable only if the surface temperature is high enough, above 300 K for W [2]. In order to describe the phenomena observed on the surface, a model needs to be used which has to include all relevant processes. There are two kinetic models [25,31] that include the main processes occurring on the surface exposed to H atoms with preadsorbed D atoms and considering that the LH mechanism has negligible contribution due to low surface temperature (T < 200 K). We extend a model developed by Jackson et al. [25] by including the LH mechanism, since in our case the temperature is well above 300 K. We studied hydrogen interaction with the surface by exposing the surface to a flux of a particular hydrogen isotope, where the sample temperature was either constant or varied. In the case of the isotope exchange the sample had one isotope pre-adsorbed on the surface and then the surface was exposed to the other isotope at constant sample temperature. We will describe the processes involved in isotopic exchange, which exhibits the maximum variety of process combinations. Let us consider the processes on the surface initially partially covered by adsorbed D, which is then exposed to a flux ˚H of H atoms. The surface sites that are empty, H-occupied and D-occupied are denoted as S, HS and DS , while the hot atoms (HA) are denoted by H* and D* . The free atom in the surrounding gas is denoted as H and when producing hydrogen molecules they are considered to be desorbed from the surface. The same letters in parentheses [] denote the corresponding surface concentrations expressed in monolayers or areal densities expressed in atoms per cm2 . A monolayer (ML) denotes a full coverage of adsorbing species on the surface with atom density N, typically of the order 1015 –1016 cm−2 , depending on the surface atomic structure. One has to consider the following processes: Single encounter reactions: (I) H + DS → HD + S { HD } [ER recombination] (II) H + HS → H2 + S { HH } [ER recombination] (III) H + S(HS , DS ) → H + S {Pr } [Reflection] Hot atom generation by incoming atom trapping:

2. Processes on the surface and kinetic model In order to discuss the experimental results let us first describe the processes occurring on a surface being exposed to a constant flux of hydrogen (H or D) atoms. The atoms that approach the clean surface feel an attractive potential due to the interaction between its unpaired electron and electrons from the metal gaining about 2.9 eV of energy for tungsten [2,10] (around 2.4 eV on most of the metals [26,27]). This energy is equivalent to the binding energy of a single hydrogen atom on a metal surface EH–M . Consequently, an incoming atom can be trapped on the surface with a certain probability. It can also recombine with a previously adsorbed atom, producing a hydrogen molecule which leaves the surface, the socalled Eley–Rideal (ER) recombination [27,28]. Generally, an atom cannot dissipate all the available energy immediately after being

479

(IV) (V) (VI) (VII)

H + DS → HS + D* {εHD } H + HS → HS + H* {εH } H + DS → H* + DS {εD } H + S → H* + S {εS }

HA-type reactions: (VIII) (IX) (X) (XI) (XII) (XIII)

H* + S → HS {sH } [Sticking] D* + S → DS {sD } [Sticking] H* + HS → H2 {pHH } [HA recombination] H* + DS → HD {pHD } [HA recombination] D* + HS → HD {pDH } [HA recombination] D* + DS → D2 {pDD } [HA recombination]

480

S. Markelj et al. / Applied Surface Science 282 (2013) 478–486

LH-type reactions:

Similarly the probability that H* reacts with H, forming H2 is Phh (t) = rhh h(t)/C(t). The same equations as above describe the D* reaction probability and sticking. The kinetic equations that govern the atomic concentration on the surface, taking into account Eqs. (1) and (2), are:

(XIV) HS + HS → H2 {kd-HH } (XV) HS + DS → HD {kd-HD } (XVI) DS + DS → D2 {kd-DD } The parameters relevant for the processes are denoted in curly brackets: the cross-sections  A for the ER mechanism, the reflection coefficient Pr of the impinging atoms, the cross sections εA for the formation of a hot atom, where εHD is the “knock on” cross section for the formation of a hot D atom when an incoming H collides with an adsorbed D. The probability for the hot atom thermalization and chemisorption, i.e. sticking, is denoted as sA and the probability for hot atom recombination is pAB . The sticking of gaseous atoms with immediate thermalization is negligible. Due to the relatively high sample temperatures in our experiments we also need to take into account the LH mechanism. The parameter relevant for LH-type reactions is the desorption rate constant kd-AB . The H* formation rate upon H-atom impact (processes V, VI and VII) was, considering the conservation of the flux, in [25] expressed as: d[H∗ ] = H (1 − Pr − HH [H] − HD [D] − εHD [D]) dt

(1)

and the D* formation rate as: d[D∗ ] = H εHD [D] dt

(2)

It is convenient to convert the concentrations and the crosssections into the relative quantities: h(t) = [H]/N, d(t) = [D]/N, chd =  HD N and bhd = εHD N, where N is the surface atom density. The atom flux density ˚H in at. cm−2 s is also normalized to N and being replaced by flux ˚h in units of monolayer/s. We will proceed along the line of Jackson’s model [25] by writing the averaged equations of time evolution of the reaction rates and coverage of the hot atoms, because of the large difference in the time scale for the reaction (picoseconds) and the time scale of the kinetic experiments (seconds). Every time a hot H* hops along the surface, it encounters an occupied site containing an adsorbed H or D or an empty site S, with the respective probabilities equal to the normalized surface concentrations, namely h(t), d(t) and (1 − h(t) − d(t)). The probability of H* hitting an adsorbed H atom and interacting with it to form a molecule is pHH h(t), the probability of H* hitting an adsorbed D atom and interacting with it to form a molecule is pHD d(t). The probability of H* hitting an empty site and sticking to the surface is sH (1 − h(t) − d(t)). The probability for a hot atom surviving the first surface collision is 1 − pHH h(t) − pHD d(t) − sH (1 − h(t) − d(t)). The probability that H* reacts at the second collision with an adsorbed D is (1 − pHH h(t) − pHD d(t) − sH (1 − h(t) − d(t)))pHD d(t) and so on. Similar relations hold for the other reaction processes. The average probability that a hot atom H* eventually reacts with an adsorbed D and leaves the surface as a HD molecule is: Phd (t) =

pHD d(t) pHH h(t) + pHD d(t) + sH (1 − h(t) − d(t))

Taking the ratio rhd = rHD = pHD /sH and rhh = rHH = pHH /sH we get Phd (t) =

r d(t) rhd d(t) = hd C(t) 1 − h(t)(1 − rhh ) + d(t)(1 − rhd )

The average sticking probability for H* is equal to: s¯ h (t) =

(1 − h(t) − d(t)) C(t)

dh = −h chh h(t) + h bhd d(t) − Phh (t)h˙ ∗ (t) − Pdh (t)d˙ ∗ (t) + s¯ h h˙ ∗ (t) dt − 2kd-hh h(t)2 − kd-hd h(t)d(t) = −h chh h(t) + h bhd d(t) × {1 − Pdh (t)} + h {¯sh (t) − Phh (t)}{1 − Pr − chh h(t) − chd d(t) − bhd d(t)} − 2kd-hh h(t)2 − kd-hd h(t)d(t)

(3)

dd = −h chd d(t) − h bhd d(t) − Phd (t)h˙ ∗ (t) − Pdd (t)d˙ ∗ (x) + s¯ d d˙ ∗ (t) dt − 2kd-dd d(t)2 − kd-hd h(t)d(t) = −h chd d(t) − h bhd d(t) × {1 + Pdd (t) − s¯ d (t)} − h Phd (t){1 − Pr − chh h(t) − chd d(t) − bhd d(t)} − 2kd-dd d(t)2 − kd-hd h(t)d(t)

(4)

These are the basic formulas of the model, which allow us to calculate the time variation of the H and D surface concentrations [25] for given initial conditions, here D adsorbed on the surface. In the case where H atoms are already adsorbed on the surface and the surface is exposed to D atoms, the same differential equations apply with appropriate change of indexing. A simplification of Eqs. (3) and (4) is obtained by ignoring the isotope effects. In that case the parameter set reduces to c = chh = chd , p/s = r = rhh = rhd = rdh = rdd , b = bhd and kd-hd = kd  where c is the normalized cross-section  of ER, r is the ratio between the probability for hot atom recombination (p) and sticking (s), b is the normalized “knock on” cross section εHD for formation of a hot D atom by H atom collision and kd  is the normalized desorption rate for LH mechanism. In the case when the surface is exposed only to the isotope which is also pre-adsorbed on the surface (e.g. H), the time dependent equation of the coverage is given by: dh  = −h c h(t) + h {¯sh − Phh (t)}{1 − Pr − c h(t)} − 2kd h(t)2 . dt

(5)

The last two terms in Eqs. (3) and (4) and the last term in (5) describe the thermal desorption of the molecules due to the LH mechanism. The factor 2 before the last term in Eqs. (3) and (4) comes from the fact that two hydrogen or deuterium atoms desorb from the surface during a single LH event. The pre-factor kd-hd = Nkd-AB is proportional to the desorption rate constant usually defined as [1,26,31]: kd-AB = (2) exp

 −2E 

= 2 0 exp

ch

kT

 −E

des

kT

= 2 0 exp

 ,

 −2E  ch

kT (6)

where (2) = 2 0 is a second order frequency factor, with a jump distance () and the attempt frequency (0 ) for hydrogen atoms to jump between the neighboring chemisorption sites on the surface. The desorption energy Edes is the energy needed for the molecule to desorb from the surface according to the LH mechanism and Edes = 2Ech , where Ech is the chemisorption energy. Ech is determined to be positive, if it is below the dissociation limit for the hydrogen molecule (e.g. [26]). From the desorption energy one can calculate the binding energy EH-M as EH-M = 1/2 (D0 + Edes ), where D0 is the dissociation energy for the hydrogen molecule (4.476 eV for H2 , 4.511 eV for HD and 4.553 eV for D2 [32]). Depending on the material and surface impurities, there can be an additional

S. Markelj et al. / Applied Surface Science 282 (2013) 478–486

energy barrier for desorption Ebarr . It is zero for tungsten [33], (e.g. 0.35 eV for Cu [34]). The attempt frequency is assumed to be of the order of 1013 s−1 [1,33]. The frequency factor has units of (cm2 at.−1 s−1 ] and it is experimentally determined by TDS to be in the 10−1 –10−3 cm2 at.−1 s−1 range [26]. The LH term is the only temperature-dependent quantity in the present model so that it determines the temperature variation of the saturation concentration of atoms on the surface. In addition the surface density variation as a function of temperature is flux dependent: at lower flux it will decrease toward lower temperatures. The original model without LH contribution [25] very well reproduced the experimental data by Kammler et al. [31,35]. The model and its main conclusions were further elaborated and compared with quasiclassical simulations [25,36]. The ER cross-section  is usually of the order of Å2 [28,37]. From quantum and quasiclassical calculations it was obtained to be 0.5 A˚ 2 or less [34,38]. From previous studies it also follows that b = 0.1, Pr = 0.2 [25,36] are the appropriate values for the parameters. In previous studies [25,31,39] it has been also found that absolute values for p and s are not so important, but mainly their ratio. The r = p/s ratio describes what is favorable, recombination or sticking. If p/s is small, sticking is favored over recombination, meaning that incoming atoms will prefer to stick to empty sites as long as they are available. On the other hand when p/s is close to one, the system will behave as in the ER scenario [40]. Therefore, the HA type kinetics are dominant when the atoms travel long distances before reacting, but when atoms react close to the impact site the ER type is dominant. It has been found that the HA scenario applies to substrates with strong H atom binding (EH-M > 2.5 eV, e.g. Ni), whereas ER type kinetics will apply to the weak binding materials (Cu, graphite) [35,38,40]. In the case of Cu(1 1 1) which has a surface saturation coverage  = 0.5 [33], r was obtained to be 1 [25,31]. On the other hand, when decreasing r, the concentration increases and approaches the surface coverage  = 1, as in the case of Ni [25]. Since the saturation concentration for most single-crystal W surfaces is above 1 ML and since the hydrogen atom bonding is strong, the parameter r is expected to be small [14]. Therefore for tungsten the HA type kinetics applies, as was also observed by Monte Carlo (MC) simulation [14,15]: for r < 0.01, the hot atoms will not recombine during their hopping on the surface but will lose all their energy and adsorb at some site. The atoms will first adsorb and saturate the preferred binding state with the highest binding energy. After this, atoms will adsorb in weaker binding states, with lower binding energy, having higher recombination probability compared to the atom in the preferred state, etc. This interpretation is also in accordance with the thermodesorption measurements that show sequential release from the states [1,2]. The MC simulation also showed that the two binding states should have different r parameter, for what the above equations do not account for. They assume only one binding state on the surface with one set of parameters. Generally, in a steady state when the surface saturation coverage is achieved and when the LH contribution is negligible, i.e. at low temperature, half of atoms impinging on the surface, which were not reflected, recombine (HA and ER process) and half of them stick. If an atom recombines with an adsorbed atom, it creates an empty site on which the next atom will adsorb. This steady state occurs in all cases no matter what the values of cross-section and rate coefficient are. Therefore, the parameters r, c, b and Pr can only be determined at non-steady state, such as during the isotopic exchange or increase of hydrogen surface concentration, at the start of exposure, as will be shown in the following section. 3. Experimental set-up The studies were performed in situ and in real time by the ion beam method ERDA [41,42]. ERDA is one of the rare techniques

481

which enables simultaneous hydrogen and deuterium depth profiling. 4.3 MeV 7 Li2+ was used as the probing ion beam for the present measurements. The beam was provided by the 2 MV tandem accelerator at the Joˇzef Stefan Institute (JSI), Ljubljana. The beam was collimated to 2 mm × 1 mm by a vertical rectangular shaped slit placed in front of the entrance to the experimental chamber. A conventional silicon detector was used for detecting recoiled particles (ERDA detector) and it was placed at an angle ϕ = 30◦ with respect to the incoming beam. Another detector placed at an angle ϑ = 165◦ with respect to the incoming beam (RBS detector) was used for detecting the backscattered Li ions and is here mainly used for ion dose check. The ion beam was probing the target at ˛ = 15◦ glancing angle with respect to the sample surface. A detailed description of the experimental configuration and set-up as well as a procedure for determining absolute H and D concentrations from measured spectra can be found in [42]. During the measurement the sample was exposed to a well defined beam of deuterium or hydrogen atoms. It is created by thermal dissociation of hydrogen molecules flowing through a hot capillary of an atom beam source HABS [43]. The axis of the beam of hydrogen atoms (H or D) was at an angle ˇ = 24◦ with respect to the sample normal [42]. The distance between the sample surface and capillary exit was 80 mm. The hot capillary was in all measurements heated with the same heating power of 145 W, providing a capillary temperature of 2200 K. The average driving pressure measured in the gas supply side of the tungsten capillary was approximately 0.17 mbar with H2 and 0.24 mbar with D2 . The atom flux density was determined by measuring the erosion of an amorphous-hydrogenated carbon film (a-C:H) placed as a sample and heated at 575 K (method described in detail in [44,45]). The eroded layer was analysed ex-situ by ellipsometry at the Max-Planck Institut für Plasmaphysik, Garching (IPP). From the layer thickness profile the flux density profile was obtained [42,44]. The central atom flux density at the sample was determined to be 1.6 × 1015 at. cm−2 s−1 for the H and 1.1 × 1015 at. cm−2 s−1 for the D beam. The atom beam flux density at the ion beam position needs to be averaged over the size of the probing ion beam (2 mm × 5 mm on the sample), taking into account also the precise overlap of the two beams [42]. The average flux density at the probing ion beam position was 0.63 × 1015 at. cm−2 s−1 for the D atom beam and 1 × 1015 at. cm−2 s−1 for the H atom beam. Background pressure, nitrogen equivalent, in the vacuum chamber was 3 × 10−7 mbar increasing to 10−6 when gas was introduced. The sample was mounted on a resistive heater and the temperature was computer controlled in the range between room temperature and 800 K. The sample temperature was measured by a thermocouple clamped to the sample front surface. The measurements were done on polycrystalline tungsten samples produced by Plansee AG, with grains elongated perpendicular to the surface. Samples were mechanically polished to a mirror finish at IPP, Garching. Three samples were probed in the present experiment, all having different treatment before the deuterium exposure. Different sample treatments were performed in order to study the influence of the sample annealing on the surface dynamics. The first sample (s1) was annealed for 1 h at 1360 K in a separate vacuum chamber for stress release. The second sample (s2) was heated at 473 K for 47 min and third sample (s3) was heated at 480 K for 70 min, both in the same vacuum chamber where in situ analyses were performed. The first and the second sample were then exposed to D atoms at a starting temperature of 750 K and 760 K, respectively, whereas the third sample was exposed to D atoms at the starting temperature of 480 K. The temperature treatment and exposure conditions for the three W samples are summarized in Table 1. The goal was to obtain good reference data for further studies on damaged W samples [46], where the sample temperature

482

S. Markelj et al. / Applied Surface Science 282 (2013) 478–486

Table 1 Tungsten sample pretreatment before D atom exposure and initial sample temperature TS at the beginning of the thermoadsorption procedure. Name of sample

Pre-exposure treatment

Exposure conditions: starting TS (K)

s1 s2 s3

1 h at 1360 K 47 min at 473 70 min at 480 K

750 760 480

should not exceed 500 K in order to avoid annealing of the defects. D atom interaction was studied also above this temperature in the present case as explained above in order to compare the properties of adsorption and to obtain information also above 500 K. After the start of exposure to the deuterium atom beam the sample temperature was gradually decreased from the starting temperature to around 350 K, performing the thermoadsorption study. Deuterium (D) and hydrogen (H) concentrations on and up to 400 nm below the surface were measured by ERDA in real time during deuterium exposure and simultaneous sample cooling. The advantage of the TA procedure is related to the cleaning action of atomic beam at elevated temperature. We have employed the method of chemical erosion of hydrocarbon layers by hydrogen atoms [42,44], as the cleaning method of hydrogen containing impurities on the surface. Namely, in all cases the samples had some initial hydrogen on the surface before any treatment, which is due to water and hydrocarbon adsorbates. Even the s1 sample that was annealed for 1 h at 1360 K in a separate vacuum chamber and was exposed to air only during the transfer (15 min) had water/hydrocarbons adsorbed on the surface before TA measurements. By heating and exposing the sample to deuterium or hydrogen atoms at elevated sample temperatures the hydrocarbon adsorbates on the surface were removed [42]. 4. Results and modeling 4.1. Thermoadsorption study Two typical ERDA spectra obtained on pristine tungsten and tungsten exposed to the deuterium atomic beam are shown in Fig. 1. The two spectra were obtained with a total accumulated dose of

Fig. 1. The ERDA spectra as obtained by the 4.3 MeV 7 Li2+ ion beam on tungsten sample. The spectrum with solid circles was obtained on the pristine W sample. The spectrum with open circles was obtained after the TA on sample s2 cooled down to 370 K. The surface and bulk signal of D and H are marked.

10.2 ␮C each. The peak between channels 290 and 320 is due to D at the surface and the peak between channels 155 and 185 is due to the surface H, as marked in the figure. By summing the surface peak counts and properly normalizing we obtain the so-called surface areal density from each ERDA spectrum, with the depth resolution of a few nm. The signal obtained below the respective surface peaks is attributed to D or H in the bulk, up to 400 nm deep in the present case. Surface peaks of H and D are well separated in energy due to the mass dependence of the recoil kinematic factor. The first ERDA spectrum which was obtained before any sample treatment is showing a hydrogen surface peak (13.5 × 1015 H cm−2 ) with some hydrogen in the bulk (0.5 at.%). This kind of spectrum was obtained always on tungsten samples ‘as received’. By heating the sample to 750 K the initial H on the surface decreased by about 80% and after exposing the sample to deuterium atoms, H additionally decreased to final 1–2 × 1015 H cm−2 . By heating the sample to 500 K, surface H decreases by 50% and only by exposure to D atoms, H decreases to 2–3 × 1015 H cm−2 . We interpret this decrease of hydrogen on the surface by removal of hydrocarbons on the surface as explained above and demonstrated in [42], efficiently removing the impurities by heating and exposing to hydrogen atoms at elevated temperatures. The second spectrum in Fig. 1 was obtained after TA at final sample temperature 370 K, starting temperature being 760 K (s2). Here D on the surface (5 × 1015 D cm−2 ) and H in the bulk (0.4 at.%) are observed. The H surface signal is smaller as compared to the first spectrum in Fig. 1, obtaining H surface density of 1.8 × 1015 H cm−2 . Since it has the same intensity as the H signal in bulk, we can assume it is due to H in the subsurface and it contributes to the surface signal due to the limited ERDA depth resolution. Hydrogen concentration in the tungsten bulk was found to be approximately 0.2–0.5 at.% in all studied polycrystalline W samples. The concentration was the lowest (0.2 at.%) in s1 which was heated for 1 h at 1360 K. There is also some D signal below the surface peak (near the detection limit) for all three samples, meaning that there is small D concentration in the bulk. The D signal does not significantly increase after 1–2 h of exposure. Moreover, the H and D bulk concentrations do not significantly change during the exposure to D atom beam even after long term exposure (fluence 4 × 1023 D/m2 ) at 500 K, they stay constant within the accuracy of the measurement [46]. Since we saw that D migration into bulk is almost negligible, orders of magnitude smaller, we have neglected this loss channel of D atoms for the modeling of the hydrogen atom interaction with W. In further discussion we will restrict ourselves only to the time variation of the H and D surface areal densities. The duration of each ERDA spectrum recording was about 2.5 min and the accumulated dose was 1.7 ␮C. The surface areal density data points for each spectrum are plotted at the spectrum measurement’s mid-time in the respective graphs. The D surface density (a), H surface density (b) and sample temperature (c) as a function of time during the TA for sample s2 are shown in Fig. 2. The error bars of surface density were derived from the counting statistics of integrated surface signal. The spectrum number is marked on top in Fig. 2(a) indicating individual data points. The D areal density increased with decreasing temperature. The mean initial value of the D areal density for s2 (Fig. 2(a)) at the beginning of TA (data points 1–4) is (1.1 ± 0.3) × 1015 D cm−2 . After this, an increase of areal density is observed, being stationary for another four spectra (5–8), with mean areal density (2.1 ± 0.5) × 1015 D cm−2 . Then another increase of D areal density is observed, which stabilizes between data points no. 12–no. 15 (6.4 ± 0.8) × 1015 D cm−2 . Even though the samples s1 and s2 had different thermal pre-treatment, see Table 1, the TA result, with staring temperature 750 K, was similar for both samples. The following sequence of D mean surface areal densities were obtained for s1: (1.4 ± 0.4) × 1015 D cm−2 ;

S. Markelj et al. / Applied Surface Science 282 (2013) 478–486

Fig. 2. Thermoadsorption on sample s2. (a) The experimental surface deuterium areal density is shown as dots during sample cooling and exposure to the atom beam. The modeled D areal density is shown as line. (b) Experimental H areal density variation during D atom exposure. (c) Temperature decrease and its polynomial fit.

(2.4 ± 0.5) × 1015 D cm−2 ; (6.4 ± 0.9) × 1015 D cm−2 . The H areal density is shown in Fig. 2(b) to check for adsorption of hydrogen containing surface impurities. One can observe that it starts to increase when the temperature reached 380 K and also the D areal density starts to increase further from data point 15 on. This increase is due to the interaction of D atoms with hydrocarbons from the background vacuum and their adsorption on the surface as studied in detail in [42]. The TA result for sample s3, with a starting temperature of 500 K is shown in Fig. 3. The D areal density shows only one increase from the initial mean areal density (2.1 ± 0.3) × 1015 D cm−2 to (7.6 ± 0.7) × 1015 D cm−2 . As before, we observe an increase of the

Fig. 3. Thermoadsorption on sample s3. (a) The experimental surface deuterium areal density is shown as dots. The modeled D areal density is shown as line. (b) The surface H areal density as a function of time. (c) Temperature decrease and its polynomial fit are shown.

483

H areal density at 340 K, just after D stabilizes. The starting D areal density in the case of s3 agrees well with the value obtained for s1 and s2 in this temperature range. The final D areal density is somewhat higher in this case, what could be due to the fact that the sample was not heated to high temperatures as the first two samples, leaving more impurities on the surface and part of D density was due to hydrocarbon formation. The measurements indicate the existence of three adsorption states with similar saturation areal densities for the different states on all samples. They were calculated as difference between the consecutive surface density rises at different temperatures that were averaged over all the measurements: (1.2 ± 0.3) × 1015 D cm−2 , (2.2 ± 0.3) × 1015 D cm−2 and (6.8 ± 0.6) × 1015 D cm−2 . At the D flux of 6.3 × 1014 D cm−2 the strongest binding state (ˇ3 ) is occupied already at 750 K with surface saturation density (1.2 ± 0.3) × 1015 D cm−2 , the second state (ˇ2 ) gets occupied between 600 and 500 K and has saturation density (1.0 ± 0.4) × 1015 D cm−2 and the weakest binding state (ˇ1 ) becomes occupied bellow 440 K with saturation density (4.6 ± 0.9) × 1015 D cm−2 , also given in Table 2. The modeled curves in Figs. 2 and 3, describing the variation of the surface areal density with temperature, were obtained by solving Eq. (5) for each binding state. Since the surface W atom density is not known on polycrystalline surface, we have set the surface atom density being equal to D surface areal density obtained in the experiment. The saturation areal densities Ni for each state i were determined to be: 1.2 × 1015 D cm−2 , 1 × 1015 D cm−2 and 5 × 1015 D cm−2 for states ˇ3 , ˇ2 and ˇ1 respectively (Table 2). The atom flux was normalized as ˚i = ˚H /Ni . The reflection coefficient put in Eq. (5) was determined from the isotope exchange and the time dependences of H and D increase after start of the exposure to the atom beam, presented below, and is equal to Pr = 0.96 ± 0.02. The experimental temperature time dependence was fitted with a polynomial function which was then put as T(t) into Eq. (5). The corresponding temperature fit for each sample is shown in Figs. 2(c) and 3(c). The modeled areal densities during the TA are shown in Figs. 2(a) and 3(a) as lines and are in good agreement with the experimentally obtained areal densities. The desorption energies Edes (Edes = 2Ech ) and the desorption rate (2) are the input parameters for each binding state. Both Edes and (2) can in usual temperature-programmed desorption spectroscopy be determined by applying different heating rates and different initial hydrogen coverages on the sample [1,47]. For the ˇ3 and ˇ2 state we have two measurements (sample s1 and s2), that have very similar cooling rates, ≈14.4 K/min for s1 and 12.5 K/min for s2. Therefore we assume the desorption rate (2) to be constant (from the literature [2]) and compare the obtained desorption energies for individual samples. For the case of ˇ3 state we can determine only the lower limit of desorption energy at which this state is still occupied and we have obtained that it should be above 2.2 eV. Since the surface density increase for the ˇ2 state occurs between two spectra and the temperature is changing rather quickly in that range, the desorption energy cannot be determined very precisely. Namely, a difference of a few degrees can make quite a difference in the determined desorption energy. Therefore, the accuracy of the desorption energy is dependent on the temperature fit and the way data points are plotted, in our case at the spectrum recording mid-time. For this state the (2) was taken to be 0.007 cm2 at.−1 s−1 [2]. This is the value obtained for ˇ2 state on W(1 1 0). This is justified by an X-ray diffraction analysis of the W sample used in the present experiments, that showed that the (1 1 0) single crystal surface is the dominating one. With this (2) we obtain Edes (s1) = 1.76 ± 0.14 eV for s1 and Edes (s2) = 1.64 ± 0.08 eV for s2. If the (2) is varied, the difference of 0.12 eV for the two measurements stays constant. If one would take (2) to be 0.002 cm2 at.−1 s−1 then Edes (s1) = 1.7 ± 0.14 eV

484

S. Markelj et al. / Applied Surface Science 282 (2013) 478–486

Table 2 Parameters for the three binding states, ˇi as obtained by modeling experimental temperature variation of D surface concentration: Edes – desorption energy, (2) – desorption rate constant, Dsatur – saturation areal density and sequential surface density levels averaged over all measurements. Binding state ˇi

Edes (eV)

(2) (cm2 at.−1 s−1 )

Dsatur (1015 D cm−2 )

D surface density levels (1015 D cm−2 )

ˇ3 ˇ2

>2.2 1.7 ± 0.08 1.64 ± 0.08 1.05 ± 0.06

– 0.007 0.002 0.001

1.2 ± 0.2 1.0 ± 0.1

1.2 ± 0.3 2.2 ± 0.3 (ˇ3 + ˇ2 )

5±1

6.8 ± 0.6 (ˇ3 + ˇ2 + ␤1 )

ˇ1

for s1 and Edes (s2) = 1.58 ± 0.08 eV for s2. The characteristics of the ˇ1 state we deduced from more measurements, with different cooling rates. In all cases similar values for Edes were obtained taking (2) = 0.001 cm2 at.−1 s−1 : Edes (s1) = 0.98 ± 0.06 eV, Edes (s2) = 1.07 ± 0.06 eV and Edes (s3) = 1.1 ± 0.04 eV. The mean values of the Edes for each ˇi state are given in Table 2. 4.2. Isotopic exchange Beside the TA we also studied the isotope exchange by measuring the time variation of H and D concentrations, while switching the atom beam from one hydrogen isotope to the other one at the constant sample temperature of 485 K. At this temperature the lowest binding state is not occupied and also there is no hydrocarbon deposition. The result is shown in Fig. 4. The D areal density is shown in Fig. 4(a) and H areal density in Fig. 4(b). Both combinations of isotopic exchange are presented and start and stop of individual gas inlet is shown as pressure variation in Fig. 4(c). First, the D atoms were offered to adsorb on the surface. Then the surface was exposed to H atoms. After this the H atom beam was turned off, having only H atoms adsorbed on the surface. Then the D atom beam was turned on and adsorbed H atoms were exchanged by D impinging atoms. During the gas exchange typically two spectra were recorded in order to see if there is any surface density change after stop of the atom exposure, i.e. atom desorption from the surface by LH mechanism, and to have the starting conditions before the following exposure. One can notice that even before the H atom exposure there is about 2 × 1015 H cm−2 already present, which is

Fig. 4. Isotope exchange at 485 K sample temperature. Both sequences (D by H and H by D) of isotope exchange were measured (solid circles) and modeled (lines). (a) D areal density during the isotope exchange and (b) the H areal density are shown. (c) The driving pressure to the HABS. The modeling data for first isotope exchange: b = 0.002, c = 0.005, r = 0.001, Pr = 0.98, d(0) = 0.8, Hsat = 2.2 × 1015 at. cm−2 , h = 1/2.2. The modeling data for second isotope exchange: b = 0.002, c = 0.005, r = 0.001, Pr = 0.95, h(0) = 1, Dsat = 1.4 × 1015 at. cm−2 , d = 0.63/1.4.

due to hydrogen in the subsurface, as was discussed above. Namely, due to the limitations of the ERDA depth resolution, which is few 1016 W at. cm−2 , part of the integral peak signal marked in Fig. 1, attributed to the surface integral peak, is also partially due to the hydrogen below the surface. During the H atom exposure the H areal density increased by approximately 2.0 × 1015 H cm−2 , and at the same time D decreased to zero. The scattering of H data and also the error bars are quite large. This is because the integral signal for a single spectrum is small (≈10 counts). Better statistics could be obtained if one would collect the spectra for longer times, but in that case one would lose the time resolution and the investigated phenomena could not be followed in real time. On the other hand, the scattering of D data is smaller due to larger cross section and shows more clearly the trend of the isotopic exchange. One can observe that the time needed for D being replaced by H is about 900 s, needing a fluence of 9 × 1017 H cm−2 for this isotope exchange. The time needed for H being replaced by D is about 650 s, needing a D fluence of 4.1 × 1017 D cm−2 for exchange. Eqs. (3) and (4) were solved numerically to model the isotope exchange shown in Fig. 4. Even though we have shown that in the present case there are two occupied binding states at 485 K, we have employed the above equations to fit the experimental isotopic exchange. This means that the input parameters obtained from the modeling will represent effective values for the two binding states which in principle could have different sticking and recombination coefficients. The incident atomic flux was normalized to the experimental saturation areal density, of 2.2 × 1015 at. cm−2 . The main parameter which dominates the time dependences is the reflection coefficient Pr . From the isotope exchange and the time dependences of H and D increase after start of exposure we have determined the reflection coefficient to be Pr = 0.96 ± 0.02. The other parameters determine the shape and the rate of the isotope exchange, which are not so well defined at the given time resolution and we can only give some estimates. Namely, the average saturation areal densities obtained from all measurements for ˇ3 and ˇ2 states, Table 2, are (1.2 ± 0.3) × 1015 at. cm−2 and (1.0 ± 0.4) × 1015 at. cm−2 , respectively, meaning that steady state is attained in less than a minute at the D flux of 6.3 × 1014 D cm−2 s−1 . Therefore, the steady state is achieved already during recording of the first spectrum after switching on the atom beam, since recording of one spectrum takes about 150–250 s. The weakest binding state has a saturation areal density of (4.6 ± 0.9) × 1015 at. cm−2 therefore steady state is achieved in 200 s. The modeled areal densities for both combinations of the isotope exchanges at 485 K are shown in Fig. 4 as lines and are in reasonable agreement with the measurements. The exact input parameters given in the model for the isotope exchange are given in the figure caption. From the modeling of the isotope exchange and thermoadsorption the cross-section for ER recombination was estimated to be in the range of 10−19 –10−18 cm2 . The ratio between recombination and sticking coefficient for HA is r = 0.001–0.01 for states ˇ2 + ˇ3 and r = 0.05 ± 0.02 for the ˇ1 state. The normalized “knock on” cross section b was taken to be 0.002. These values of the parameters are mainly informative. Better time resolution or working at lower atom fluxes to the surface would be needed in order to obtain more exact numbers.

S. Markelj et al. / Applied Surface Science 282 (2013) 478–486

Fig. 5. Hydrogen surface areal densities at 363 K as a function of time after being exposed to H atom beam at time zero. Time zero denotes the start of the exposure. Modeled H areal density is shown as line. Modeling parameters: h = 1/5.5, Pr = 0.94 c = 0.008, r = 0.02, Hsat = 5.5 × 1015 .

4.3. Single isotope exposure The adsorption dynamics on the sample exposed to an atomic beam of only one isotope species was studied as well. An example of H areal density increase after start of H atom exposure on s3 sample at 363 K is shown in Fig. 5. This sample was first exposed to H atom beam at 485 K, where hydrogen atoms occupied the two highest binging states. Then the atom beam was turned off and the sample was cooled down to 363 K and after that the sample was again exposed to H atom beam. The H increased during one spectrum indicating that the time needed for H adsorb is around 130 s, needing fluence of 1.3 × 1017 H cm−2 to attain saturation. The modeled H areal density increase shown in Fig. 5 was obtained by solving Eq. (5) and is in a reasonable agreement with the measured one. The exact fitting parameters are given in the figure’s caption. 5. General discussion and conclusion We have obtained the temperature variation of the saturation concentration from the present TA measurements. Three D saturation areal density levels were determined being (1.2 ± 0.3) × 1015 D cm−2 if the temperature is around 750 K, (2.2 ± 0.3) × 1015 D cm−2 if temperature is below 600–500 K and (6.8 ± 0.6) × 1015 D cm−2 , if temperature is below 440 K. The areal densities are higher, but not much than those obtained on single crystal surfaces exposed to hydrogen molecules at 78 K: 1.05 × 1015 H cm−2 on W(1 1 0), 2.6 × 1015 H cm−2 on W(1 1 1) and 1.5 × 1015 H cm−2 on W(1 0 0) [2]. This might be expected for the polycrystalline W having higher roughness. The most intriguing is the high D areal density at lowest temperature. Such a high surface areal density was observed also in other experiments, where the samples were exposed at low temperature (400 K) to D plasma or D ion beam [23,24,48]. Beside this, there were almost no differences in the surface areal densities observed on W sample being either non-recrystallized or recrystallized [24,49]. It was usually speculated that this comes from the damaging effect during the plasma/ion exposure, but since the D concentration in that narrow surface region, around 10 nm thick (the minimum NRA resolution at the surface), was much higher than one would expect for W

485

with low solubility, it is a question whether such explanation is adequate. Here we show that this could be the property of W, to attain such high D concentration on the surface at temperatures below 400 K, simply by adsorption of neutral D atoms, which are always present also in the exposure experiments with plasma. One can also not exclude that part of this high concentration at low temperature is due to adsorption of oxide or hydrocarbons on top of the surface, since the experiments are not performed in ultra high vacuum. Namely, it was shown that hydrocarbons such as acetylene are being adsorbed at this temperature range, around 400 K [50]. The high surface areal density at low temperatures observed here is important for the D retention studies in fusion since it can significantly contribute to the total D amount retained in W at temperatures below 400 K [23,49]. From the TA studies the desorption energies were also determined for the different states. They can be compared to the ones obtained by “Tamm and Schmidt” [2] on single crystal surfaces. In the present experiment we have observed the strongest state ˇ3 with desorption energy above 2.2 eV, where in the measurement by Tamm and Schmidt there was no desorption energy above 1.6 eV. For the ˇ2 state we obtain a desorption energy between 1.64 and 1.7 eV. In the measurements by Tamm and Schmidt [2], the strongest binding state for W(1 1 0) has Edes = 1.42 eV. For the other two single crystal surfaces the desorption energy of the strongest state is 1.59 eV for W(1 1 1) and 1.4 eV for W(1 0 0) [2]. Therefore from the present experiment we have obtained about 0.3 eV higher desorption energy. The desorption energy obtained for the weakest binding is about 1.1 eV in the present experiment, which is in good agreement with literature [2]. They obtained Edes = 1.17 eV for W(1 1 0) and 1.14 eV for W(1 0 0), whereas the W(1 1 1) has two binding states in this energy range with energies 0.94 eV and 1.32 eV. One could conclude that even though the experiment was not performed in ultra high vacuum we have obtained a reasonable agreement with the available literature data. Isotope exchange was studied at 485 K. The main parameter determined from the isotopic exchange and the increase dynamics with time is the reflection coefficient and is Pr = 0.96 ± 0.02. To the authors knowledge there is no other experimental determination of the reflection coefficient for hydrogen atoms on tungsten. Molecular dynamics calculations of hydrogen atoms impinging on W(0 0 1) at normal incidence were performed by Henriksson et al. [11]. They obtained a reflection coefficient of 0.75 at 0.2 eV which decreases at higher energies. Since the present experiment was performed at an angle of 24◦ to the surface normal and on a polycrystalline surface, one could expect the reflection coefficient to be higher in the present experiment. Moreover, the hydrogen molecule sticking coefficient on W(1 1 0) and W(1 0 0) obtained by calculations [51] and experiment [52] is lower in the case of W(1 1 0), which is the dominant surface on the present polycrystalline W. The study of isotope exchange gave a prediction of the estimated fluence needed for exchange with the hydrogen isotopes on the surface; one process important for the tritium removal technique in the fusion devices. This isotopic exchange on the surface will take place first, once the surface is exposed to plasma, whereas bulk exchange is slower as it is limited by the diffusion of atoms in the material.

Acknowledgments The authors acknowledge the financial support by the Slovenian Research Agency, by Association EURATOM-MESCS (Slovenian Fusion Association), by Association EURATOM-IPP and by EU EFDA IPH. The content of the publication is the sole responsibility of its authors and it does not necessarily represent the views of the Commission or its services.

486

S. Markelj et al. / Applied Surface Science 282 (2013) 478–486

References [1] (a) T.E. Madey, Surf. Sci. 29 (1972) 571; (b) T.E. Madey, Surf. Sci. 36 (1973) 281. [2] (a) P.W. Tamm, L.D. Schmidt, J. Chem. Phys. 51 (1969) 5352; (b) P.W. Tamm, L.D. Schmidt, J. Chem. Phys. 54 (1971) 4775. [3] I. Stensgaard, L.C. Feldam, P.J. Silverman, Phys. Rev. Lett. 42 (1979) 247. [4] P. Alnot, A. Cassuto, D.A. King, Surf. Sci. 215 (1989) 29. [5] M. Balden, S. Lehwald, H. Ibach, Phys. Rev. 53 (1996) 7479. [6] T.U. Nahm, R. Gomer, Surf. Sci. 375 (1997) 281. [7] R.D. Kolasinski, N.C. Bartlet, J.A. Whaley, T.E. Felter, Phys. Rev. B 85 (2012) 115422. [8] A. Forni, M.C. Desjonqueres, D. Spanjaard, G.F. Tartardini, Surf. Sci. 269/270 (1992) 201. [9] H.F. Busnengo, A.E. Martinez, J. Chem. Chem. C 112 (2008) 5579. [10] B. Jackson, M. Persson, J. Chem. Phys. 96 (1992) 2378. [11] K.O.E. Henriksson, K. Vörtler, S. Dreissigacker, K. Nordlung, J. Keinonen, Surf. Sci. 600 (2006) 3167. [12] M. Rutigliano, M. Cacciatore, Phys. Chem. Chem. Phys. 13 (2011) 7475. [13] Y. Fukai, The Metal–Hydrogen System, Springer-Verlag, Berlin, Heidelberg, 2005. ˇ [14] S. Markelj, I. Cadeˇ z, J. Chem. Phys. 134 (2011) 124707. [15] S. Markelj, PhD Thesis, Univesity of Ljubljana, Ljubljana, 2010, 195 pp. [16] J. Roth, E. Tsitrone, A. Loarte, Th. Loarer, G. Counsel, R. Neu, V. Philips, S. Brezinsek, M. Lehnen, P. Coad, Ch. Grisolia, K. Schmidt, K. Krieger, A. Kallenbach, B. Lipschultz, R. Doerner, R. Causey, V. Alimov, W. Shu, O. Ogorodnikova, A. Kirschner, G. Federichi, A. Kukushkin, J. Nucl. Mater. 390/391 (2009) 1. [17] J. Roth, E. Tsitrone, Th. Loarer, V. Philips, S. Brezinsek, A. Loarte, G. Counsel, R. Doerner, K. Schmidt, O. Ogorodnikova, R. Causey, Plasma Phys. Control. Fusion 50 (2008) 103001. [18] A.A. Haasz, J.W. Davis, M. Poon, R.G. Macaulay-Newcombe, J. Nucl. Mater. 258–263 (1998) 889. [19] O.V. Ogorodnikova, J. Roth, M. Mayer, J. Appl. Phys. 103 (2008) 034902. [20] O. Ogorodnikova, B. Tyburska, V.Kh. Alimov, K. Ertl, J. Nucl. Mater. 415 (2011) S661. [21] M. Poon, A.A. Haasz, J.W. Davis, J. Nucl. Mater. 374 (2008) 390. [22] R. Causey, K. Wilson, T. Venhaus, W.R. Wampler, J. Nucl. Mater. 266–269 (1999) 467. [23] W.R. Wampler, R.P. Doerner, Nucl. Fusion 49 (2009) 115023. [24] A. Manhard, K. Schmid, M. Balden, W. Jacob, J. Nucl. Mater. 415 (2011) S632.

[25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52]

B. Jackson, X. Sha, Z.B. Guvenc, J. Chem. Phys. 116 (2002) 2599. K. Christmann, Surf. Sci. Rep. 9 (1988) 1. S. Caratzoulas, B. Jackson, M. Persson, J. Chem. Phys. 107 (1997) 6420. C.T. Rettner, M.N.R. Ashfold, Dynamics of Gas–Surface Interaction, The Royal Society of Chemistry, Cambridge, 1991. J. Hariss, S. Kasemo, Surf. Sci. 105 (1981) L281. R. Gomer, Rep. Prog. Phys. 53 (1990) 917. Th. Kammler, D. Kolovos-Vellianitis, J. Küppers, Surf. Sci. 460 (2000) 91. G. Herzberg, Molecular Spectra and Molecular Structure, D. Van Nostrand Company, Toronto, 1950. O.V. Ogorodnikova, J. Nucl. Mater. 277 (2000) 130. M. Persson, B. Jackson, J. Chem. Phys. 102 (1995) 1078. Th. Kammler, J. Küppers, J. Chem. Phys. 111 (1999) 8115. Z.B. Guvenc, X. Sha, B. Jackson, J. Chem. Phys. 115 (2001) 9018. Th. Kammler, J. Lee, J. Küppers, J. Chem. Phys. 106 (1997) 7362. B. Jackson, D. Lemoine, J. Chem. Phys. 114 (2001) 2378. Th. Kammler, S. Wehner, J. Küppers, J. Chem. Phys. 109 (1998) 4071. T. Zecho, A. Güttler, X. Sha, D. Lemoine, B. Jackson, J. Küppers, Chem. Phys. Lett. 366 (2002) 188. ˇ S. Markelj, I. Cadeˇ z, P. Pelicon, Z. Rupnik, Nucl. Instrum. Methods Phys. Res., Sect. A 259 (2007) 989. ˇ S. Markelj, T. Schwarz-Selinger, P. Pelicon, W. Jacob, I. Cadeˇ z, J. Vac. Sci. Technol. A 30 (2012) 041601–41611. K.G. Tschersich, J.P. Fleischhauer, H. Schuler, J. Appl. Phys. 104 (2008) 034908 http://www.mbe-components.com/products/gas/habs.html T. Schwarz-Selinger, A. von Keudell, W. Jacob, J. Vac. Sci. Technol. A 18 (2000) 995. M. Schlüter, C. Hopf, T. Schwarz-Selinger, A. von Keudell, W. Jacob, J. Nucl. Mater. 376 (2008) 33. S. Markelj, O. Ogorodnikova, T. Schwarz-Sellinger, P. Pelicon, K. Sugyama, I. ˇ Cadeˇ z, J. Nucl. Mater. 438 (2013) S1027. D.P. Woodruf, T.A. Delchar, Modern Techniques of Surface Science, 2nd ed., Cambridge University Press, Cambridge, 1994. J.P. Roszell, J.W. Davis, A.A. Haasz, J. Nucl. Mater. 429 (2012) 48. A. Manhard, PhD Thesis, Univesity of Augsburg, Augsburg, 2012, 199 pp. B.D. Barford, R.R. Rye, J. Chem. Phys. 60 (1974) 1046. H.F. Busnengo, A.E. Martibez, J. Phys. Chem. C 112 (2008) 5579. H.F. Berger, Ch Resch, E. Grösslinger, G. Eilmsteiner, A. Winkler, K.D. Rendulic, Surf. Sci. Lett. 275 (1992) L627.