Physics and chemistry in cryomatrices is an active research field and rare-gas matrices have been used frequently in studies over many years.[1–7] Defects of structure are of substantial importance in the kinetics of solid-state chemical reactions and have a pronounced effect on the efficiency of formation and stability of radicals and ionic species in cryogenic solids. The part of the defects in the accumulation of matrix-isolated species has already been discussed in earlier studies[1,8] and the well-known “matrix-site” effect often involves lattice imperfections. Whereas in the case of “matrix-site” effect the defects adversely affect guest optical spectra causing a band broadening, the presence of defects, which can be seen as charge traps, is favourable for stabilization of ionic species in cryomatrices. To cite an example, the stabilization effect of structural defects on the intrinsic ionic species Rg2+, so-called self-trapped holes (STH), in cryomatrices was recently found in the experiments carried out at Bondybey’s Laboratory.[9,10] Another interesting example to mention is the influence of local matrix morphology on the formation of neutral ground-state chemical compounds involving rare-gas atoms.[11]
A special type of lattice defects is an electronically induced point defect–Frenkel pair. These defects are created in different classes of wide-gap insulators[12] and even in narrow-gap materials.[13] The basis for the physics of electronically induced lattice rearrangement is a concentration of the excitation energy released in the relaxation process within a volume about that of a unit cell followed by the energy transfer to the surrounding matrix. The rare-gas solids are characterized by extremely small binding energies in comparison with the energies of electronic excitations. The difference in energies is of the order of 102–103. This peculiarity in conjunction with the strong electron–phonon interaction followed by the self-trapping of excitons and holes[12,14,15] results in a high quantum yield for the electronically induced processes, making the rare-gas matrices excellent objects for studying the topic under consideration.
Despite the aforementioned attractive properties of the rare-gas solids the experimental study into electronically induced defect creation was undertaken not so long ago. The first efforts were concentrated on defect creation via self-trapping of excitons into atomic states that is the case for solid Ne.[16,17] These high-resolution spectral studies revealed a creation and accumulation of point defects in the Ne lattice under irradiation by slow electrons. The self-trapping of excitons into atomic states is suggested to be a stimulating factor. The direct experimental proof of this mechanism of defect creation was provided by a recent study carried out with synchrotron radiation.[18] The results obtained in combination with the activation spectroscopy study[19,20] supported the theoretical model[21,22] of the evolution of atomic self-trapped excitons (A-STE) into permanent lattice defects—Frenkel pairs of the second neighbour configuration.
The spectroscopic investigation of lattice defect creation in the case of exciton self-trapping into molecular states, as occurred in solid Xe, Kr and Ar, is a rigorous task. In contrast to the case of exciton self-trapping into atomic states the luminescence spectra of molecular self-trapped excitons (M-STE) have a shape of broad bands stemming from the electronic transition from the 1,3Σ+u to a repulsive part of the ground state 1Σ+g (the well-known M bands.[12,14,15]) The design of a special cryogenic cell[23] for sample growth enabled us to grow samples of high crystal quality and to reveal the internal structure of the M bands. It has been shown that the M band in all rare-gas solids consists of two components—M1 (related to the molecular centers in the defect sites) and M2 (related to the molecular centers formed in the regular lattice at the exciton self-trapping). The interpretation was supported by the recent experiments on selective excitation by synchrotron radiation.[24] The existence of these two components in the M bands provides a way for application of the sensitive VUV luminescence spectroscopy method to characterize the samples and to study the electronically induced defect creation. Even in our first studies[23,25] slow electrons were used for the excitation in order to prevent the knock-on defect creation—formation of defects by elastic collisions. The accumulation of radiation-induced defects was detected in all heavy rare-gas solids—Ar, Kr and Xe.[23,25–27] The comprehensive study performed in pure and doped Kr matrices[28] permitted us to suggest mechanisms of defect creation via exciton self-trapping into molecular states.
However, the direct proof of the excitonic mechanism requires experiments with state-selective excitation. The synchrotron radiation providing an intense VUV light tunable through the range of excitonic absorption of rare-gas solids is best suited to such an experiment.
This paper reports the results of a synchrotron radiation study into permanent lattice defect creation in the Xe matrix via excitonic mechanism. The experimental findings are compared with the results obtained under electron beam irradiation, and discussed in relation to the suggested model of defect creation.
Investigation of electronically induced lattice rearrangement requires definite experimental conditions—first, high quality samples with low concentration of initial defects and impurities, then subthreshold excitation of the samples using slow particles of energy deficient to produce defects via knock-on mechanism or photons of appropriate energy, and finally, excitation of the samples at low temperatures to exclude the conventional thermally-activated defect creation. The study to be described was performed at the SUPERLUMI station of the Hamburger Synchrotronstrahlungslabor HASYLAB at DESY. The set-up designed for luminescence experiments in the VUV range is given elsewhere,[29] so only essential details are mentioned here. The samples of required quality were grown in a special closed cell mounted on a He-flow cryostat holder in an ultrahigh vacuum (UHV) environment (10−10 mbr). High-purity Xe gas (99.9996%) was used, and the gas handling system was operated under UHV conditions. The samples were condensed from the vapour phase at elevated temperature under isobaric conditions (P = 1000 Pa) with a constant cooling in the temperature range 125–110 K. Then the temperature was controllably reduced at a constant rate of 0.1 Ks−1. After the sample preparation and cooling, the cell was removed and the free-standing cryocrystal of 1 mm thickness was studied. The polycrystalline samples thus produced were optically transparent. The intense luminescence of free excitons indicated their high quality. The samples were then subjected to prolonged irradiation by the VUV light. The selective excitation was carried out with a bandpass of Δλ = 0.25 nm. The excitation energy was chosen in the range of the n = 2 term of the Γ(3/2) excitonic series. The luminescence analysis was performed with Δλ = 2 nm by using a 0.5 m high-flux Pouey VUV monochromator equipped with a multisphere plate detector. The entire M band was periodically recorded during exposure. The following decomposition of the M band into M1 and M2 subbands yielded the dose dependence of the defect component and the regular one. The influence of the annealing at 60 K on the intensity distribution in the M band was examined. To probe the defects we used thermally stimulated luminescence (TSL). The electron-hole pairs were generated by a short-term exposure to the VUV radiation. On completion of the irradiation the spectrum of the TSL was recorded.
For comparison, we replicated the experiments with excitation by an electron beam following the identical procedure of the sample growing. The experiments were performed at the Laboratory of Condensed System Spectroscopy at the Verkin Institute for Low Temperature Physics & Engineering NASU. The samples were grown in a closed cell with an electromagnetic shutter mounted on a L-He cryostat holder in the vacuum chamber. After the sample preparation and its cooling, the shutter was open providing the free access for the electron beam and emerging luminescence. We used electrons of 1 keV energy, deficient to create defects by knock-on collisions. The irradiation was performed under steady state conditions with a current density of 50 μA cm−2. The dose was increased with exposure time. The sequence of luminescence spectra was recorded using a normal incidence 1 m VUV monochromator with a bandpass of Δλ = 0.16 nm. The irradiation and measurements were performed at low temperature (15 K) in order to exclude the conventional thermal mechanism of defect creation and to avoid the annealing of the defects produced.
Fig. 1 demonstrates the internal structure of the M band in solid Xe and its temporal evolution under excitation by slow electrons. The shape of the M band profoundly varies with exposure time. These changes are caused by the intensity redistribution between the two components of its internal structure—M1 and M2. Each of the subbands approximated by the Gaussian is depicted in Fig. 1.
The dose dependence of the subband intensity is shown in the insert of Fig. 1. The intensity of the M2 subband related to the exciton self-trapped in the regular lattice remains constant. The distinct enhancement of the defect subband M1 upon irradiation indicates an increase in the number of defect sites, i.e. accumulation of permanent lattice defects. The existence of the two subbands points to a branching of the relaxation channels. Ionizing radiation (electrons in our case) creates primary excitations—free electron-hole pairs. Three channels of “indirect” exciton creation were found under selective excitation above the band gap energy Eg: (i) recombination of electron-hole pairs, (ii) inelastic scattering of electrons, and (iii) excitonic side bands of valence or inner-shell excitation.[30] Free excitons can be then self-trapped in the regular lattice or trapped in some defect site. The method of lattice defect detection developed in our studies[16,17,23] is in fact a modification of the sensitized luminescence method. The excitation energy is transferred to the defects by free excitons followed by their trapping and radiative decay yielding the M1 subband. Apparently, both kinds of defects—the initial structural defects and the defects created via electronic excitation—contribute to the M1 subband. The contribution of the initial defects can be easily found by extrapolating the dose dependence of the M1 subband intensity to zero.
Let us take up possible channels of electronically induced defect creation keeping in mind that the key point for this process is the relaxation energy release in a confined region of the crystal. As was already mentioned, two efficient channels of electronic excitation self-trapping exist in the lattice of heavy rare-gas solids–formation of STEs and STHs. The secondary electrons produced under excitation well above Eg for the most part remain in the solid, and hence each hole has, in the mean, its counterpart—an electron elsewhere. Taking into account the fact that the electrons are characterized by free-like behavior[12] in rare-gas solids, in other words, they are highly mobile, one can expect an efficient recombination on the electron thermalization. If a part of the electrons is trapped in some defect sites, STHs remain stable in the lattice at temperatures low enough to prohibit the release of the electrons from their traps. The exciton self-trapping caused by exciton–phonon interaction was found in all rare-gas solids and reported in a number of books, reviews and papers.[12,14,15,31–35]
The relevant relaxation processes involving the self-trapping of holes and excitons can be given by the reactions:RgH+ + Rg → Rg2+ + ΔE+(STH) → Rg2++ e → Rg* + Rg + ΔE(DRec)RgE* + Rg → Rg2* + ΔE*(M-STE) → hν (M-STE) + Rg + Rg + ΔE(D)We take RgH+ and RgE* to denote a free hole and free exciton, respectively. The first stage of these reactions corresponds to the self-trapping of holes (1) and excitons (2) followed by the formation of vibronically hot charged (STH) and neutral (M-STE) dimers. Note that STH and M-STE might be considered as “matrix isolated” molecular centers within its own host crystal. While relaxing through the vibronic levels, the positively charged dimer STH can recombine with a thermalized electron into a repulsive electronic state yielding two Rg atoms separated energetically (a so called dissociative recombination DRec). At both relaxation stages an essential amount of energy ΔE+(STH) and ΔE(DRec) is released in the vicinity of STH, which can be spent on the lattice defect creation. The efficiency of this defect creation channel via recombination depends on the rate of electron thermalization and is not expected to be high.
In the case of exciton self-trapping by reaction (2) the relaxation energy is transferred to the surrounding atoms at both stages—in the excited state of the M-STE and in the ground state. The energy release during the M-STE vibronic cooling ΔE*(M-STE) is about 0.45 eV [36] which is higher than the binding energy per atom in solid Xe (172.3 meV .[12]) After the radiative transition (hν (M-STE)) of the relaxed M-STE to a repulsive part of the ground state potential, the excess energy ΔE(D) = 0.86 eV is shared between two dimer atoms, in other words, there appear the two “hot” atoms with a 0.43 eV kinetic energy available for the defect creation. In accordance with this consideration, the mechanisms of exciton-induced defect creation can be classified as an “excited state” mechanism and a “ground state” one.[28]
In order to distinguish clearly the recombination (1) and excitonic (2) channels of defect creation the direct experiments with selective excitation into the exciton band were performed. First of all, the origin of the M1 and M2 subbands was verified using the advantages of the selective excitation by synchrotron radiation. The shape of the M band in the luminescence spectrum of solid Xe measured at different excitation energies—below the bottom of the lowest exciton band n = 1 Γ(3/2) and in the range of the exciton absorption—is shown in Fig. 2 where the difference in the position and halfwidth FWHM of the M band is clearly seen. The excitation below the exciton band gave rise to a defect M1 subband with a maximum at 7.05 eV. The excitation above the Γ(3/2) exciton band bottom resulted in the appearance of two subbands with the predominance of a M2 subband at 7.22 eV.
In order to shed more light on the origin of the internal structure of the M band, we restored the excitation spectra of the M1 and M2 subbands. The earlier data on the excitation spectra obtained by the “red–blue-side” separation method[31] supported the assignment of the M2 subband as originated from the M-STE in the regular lattice whereas the M1 subband was assigned to the exciton of molecular configuration trapped in the defect site. However, because of the strong overlapping of the two subbands the direct measurement of the separate excitation spectra of the M1 and M2 subbands is impossible. To restore the excitation spectra we used the following procedure:[24] a series of luminescence spectra were recorded with a successive increase of the excitation energy, then the luminescence spectra recorded were resolved into the M1 and the M2 components and the intensities of each subband were plotted as a function of the excitation energy. The resulting excitation spectra are shown in Fig. 3. The solid line represents the total yield of the luminescence in the M band as a function of the excitation energy. The procedure applied enabled us to estimate the low energy excitation thresholds E1 and E2 for the M1 and M2 subbands, respectively: E1 = 8.18 eV, E2 = 8.28 eV. The careful check of the spatial distribution of the centers emitting M1 and M2 subbands, performed with the electron beam of different energy, and hence, of different penetration depth, demonstrated the bulk origin of both centers.[24] The results obtained provided solid evidence of the “defect” nature of the M1 subband and the intrinsic nature of the M2 subband—emission of the true M-STE center in the regular lattice.
Based on this assignment we measured the dose dependence of the M1 and M2 subbands to probe the defect creation upon excitation in the exciton band. The primary monochromator was tuned to the energy of the n = 2 Γ(3/2) band in the absorption spectrum of solid Xe. Due to the low absorption coefficient the exciting light penetrated deep inside the sample, and the M-STE dimers were efficiently formed in the bulk. A sequence of spectra were measured upon exposure to the VUV light. The internal structure and the intensity distribution were analyzed. The position and FWHM of the M1 and M2 subbands were obtained from the numerical fit of the band shape assuming a Gaussian shape of the subbands. The result of this treatment—the corresponding points of the curves—yields the dose dependence of the subband intensity shown in Fig. 4. The intensity of the M2 subband associated with the M-STE in the regular lattice remained constant during exposure to the VUV light. At the first stage of VUV irradiation (up to 100 min) we observed a pronounced enhancement of the M1 subband intensity. Note that the exposure time is not corrected for the synchrotron radiation pulse duration, so it is in fact a real time of the measurements. During the next 100 min we observed a saturation of the dose dependence of the M1 intensity followed by a slow decrease with long irradiation time. The dose dependence of radiation-induced point defect (e.g. vacancies) concentration Cv is described by the following equation∂Cv/∂t = Kdc − Kr (CiCv) − Ci0Cv0) − Kt (Cv − Cv0)Here Kdc is the coefficient that characterizes the rate of permanent defect creation via exciton self-trapping, whereas Kr and Kt are the coefficients of the vacancy-interstitial recombination and trapping by some kind of sinks, respectively. At low temperatures and small doses, the reactions of recombination and trapping are extremely slow, and we observe a linear increase of the M1 subband intensity indicating the accumulation of defects in the sample. With increasing the concentration of the radiation-induced Frenkel pairs (interstitials and vacancies) the contribution of the defect annihilation processes (second and third terms of the equation) increases. Note that at low temperatures the thermally activated diffusion of vacancies to sinks is negligible and only the radiation-induced defect motion comes to play. One can suggest that on long-term exposure there occurs an aggregation of defects. The details of the defect transformation on long-term irradiation invite further investigation.
Let us discuss some possible models of exciton-induced defect creation via self-trapping into molecular states—M-STE. The self-trapping occurring during the lifetime of the excited state results in the formation of a molecular dimer aligned along the 〈110〉 crystallographic directions with an interatomic distance of 0.31 nm.[35] This stage followed by the M-STE formation in the on-center position is shown in Fig. 5 (a and b). The center formed resembles a “dumb-bell” configuration of the interstitial atom. Its displacement along the 〈110〉 direction to an off-center configuration (position c in Fig. 5) cannot stabilize the center. It was shown[37] that the split 〈100〉 “dumb-bell” form is the only stable form of the interstitial atom in the Xe lattice. The separation between the “dumb-bell” atoms is found to be 0.86d (d is the distance between the nearest neighbours), which only slightly exceeds the interatomic distance between the atoms in the Xe2* dimer (0.31 nm). Bearing this in mind, one can assume that the short-lived defect of the off-center configuration can be stabilized by the reorientation of the dimer axis to the 〈100〉 direction (d in Fig. 5). The Frank–Condon transition of this dimer to the ground state will correspond to the transition of the molecular center to the permanent defect level with almost no change in the interatomic distance. The energy needed for the reorientation can be released in the lattice in the course of the vibronic relaxation. As the theory shows,[38] in a system with a strong local vibration the energy release proceeds in a jump-like multiphonon process. It seems to be the case for the M-STE relaxation which offers the basis for the development of the “excited state” mechanism.
The model of defect creation by the “ground state” mechanism is not as straightforward as it may seem at first sight. The matter is that during self-trapping Rg atoms are brought closer together displacing along the 〈110〉 direction, as shown in Fig. 5b. However, there are no stable interstitials of the “dumb-bell” configuration aligned along the 〈110〉 crystallographic direction.[37] The recent theoretical studies[39,40] showed that in the fcc lattice of rare-gas solids the weak coupling between close-packed atomic rows and the crystal matrix may give rise to an interstitial atom of a specific configuration—a smeared clump called a crowdion. The vacancy also becomes delocalized, forming a smeared sparsed region (anticrowdion). The computational results were presented for Ar and Kr cryoscrystals and it was mentioned that the formation of crowdions is also expected in the Xe lattice.
We performed the experiment with detection of a thermally stimulated luminescence (TSL) spectrum of the samples irradiated by VUV photons. The recorded TSL spectrum is shown in Fig. 6. In view of the low TSL intensity the spectrum was collected over the heating time. The recombination of STHs with electrons gives rise to the self-trapped excitons emitting the M2 subband. If the hole is trapped in a defect site one can expect the appearance of the M1 subband in the TSL spectrum. The comparison with the luminescence spectrum excited by photons of 8.86 eV energy (in the band of intrinsic exciton absorption of solid Xe) shows that the TSL band consists of two components M1 and M2. Because of low intensity, reliable decomposition of the M band in the TSL spectrum is impossible. However, the contribution of both subbands can be ascertained. TSL of solid Xe at the photon energy 7.2 eV after irradiation by X-rays was observed in [ref. 41] but the M1 and M2 subbands in the TSL band were not discriminated. Note that the TSL from solid Xe grown at low temperature yielded a 7.6 eV band.[42] The recent study suggested this band to be originated from trimer Xe3*[43].