SPECTROSCOPY OF DEFECTS IN NEUTRON IRRADIATED AMMONO-THERMAL GaN BY COMBINING PHOTOIONIZATION, PHOTOLUMINESCENCE AND POSITRON ANNIHILATION TECHNIQUES

1. Introduction

GaN material and its alloys with aluminium have become widely used in fabrication of devices for biological, chemical and radiation sensing [1, 2]. Modern technology grown GaN is particularly interesting for production of radiation tolerant detectors capable of operating in extremely harsh radiation environments. The high luminosity large hadron collider (HL-LHC) will reach fluxes from 5 to 7 of the nominal LHC luminosity [3] and the future circular collider (FCC) is anticipated to reach particle energy up to 100 TeV [4, 5]. Thereby, the HL-LHC and FCC detectors should withstand the extremely high fluences up to 10¹⁷ cm⁻². Therefore, much effort has been made in search of modern materials for detectors to be used in such conditions for a long time [6]. GaN-based detectors are beneficial due to their capability to generate both the electrical and optical signals. The GaN-based detectors raise technical interest in development of the double-purpose devices capable to operate both as scintillating and charge collecting detectors for tracking of high-energy particles [7–10]. The depth and lateral homogeneity of the sensor structures are therefore necessary to get the linear response of the detector [11]. Moreover, thick sensor structures which satisfy the requirement of bulk uniformity are required to detect events of the relatively small interaction cross-sections. This situation appears in the detection of strongly penetrating radiations. As a result of irradiation, the displacement damage occurs. Then, the lattice atoms are knocked out of their sites creating thereby vacancies and interstitials [12]. Clusters of radiation defects acting as carrier traps affect the opto-electronic characteristics of the devices, reducing the luminescence intensity and efficiency of charge collection, increasing the leakage current and thereby changing the detector response [13, 14].

The GaN single-crystals (even grown by advanced technologies) contain a lot of technological defects yet. Therefore, a trade-off should be found in choice of the GaN growth technology to have a cheap, fast way to devise rather thick radiation sensors, capable of generating double responses. MOCVD GaN epilayers usually have large densities of dislocations, which lead to a fairly high leakage current. The lattice mismatch between the substrate (sapphire, Si or SiC) and GaN causes stress wafer bowing. This determines the dislocation density not less than 10⁹ cm^–2 [15]. Especially, screw dislocations strongly affect the electrical properties of the electronic devices [16]. There exists the epitaxial-lateral-overgrowth (ELO) technology, which allows making of GaN layers relatively clean from dislocations. However, this technology excludes growth of thick single crystals [17]. The most promising GaN growth technologies are hydride vapour phase epitaxy (HVPE) and ammono-thermal (AT) method in production of the double-response detectors [18, 49]. The dislocation density less than 10⁶ cm^–2 can be reached by the HVPE method. Even lower dislocation densities ≤10⁵ cm^–2 are achievable by employing the AT technology.

In this study, the AT GaN crystals doped with various impurities and irradiated with different fluences of neutrons were examined to clarify the role of the technological and radiation defects. Mg impurities are widely exploited in formation of modern devices having specific electrical characteristics [20, 21]. Moreover, Mg impurities serve as the main dopants in formation of the p-type conductivity GaN. Other dopants, namely Mn, may act as the compensating centres for unintentionally induced donor-type defects and fast carrier traps, and that can also be used to obtain semi-insulating or highly resistive material. By manipulating the ratio between the dopant density and the concentration of radiation defects, it is possible to increase the radiation resistance of the sensors. In this work, the evolution of the pulsed photoionization (PPIS), scintillation (pulsed photoluminescence – PPL) and positron annihilation (PAS) [22–25] spectroscopy means have been combined to reveal the prevailing defects of different species and scintillation efficiency with neutron irradiation fluence. The contactless PPIS is beneficial in discovery of photoactive centres. PAS is a powerful method for detecting vacancy-type defects [23].

2. Samples

The bulk GaN crystals were grown by the ammono-thermal method [26] at Ammono Company (Poland) [27]. The GaN samples were doped with Mg and Mn of concentrations 1 × 10¹⁸ cm^–3 and 10¹⁹ cm^–3, respectively. The thickness of freestanding AT GaN samples was 400–450 μm. The AT GaN crystals contained the dislocation density less than 10⁵ cm^–2 [28, 29]. In order to emulate the evolution of particle detector functionality during experiments at the LHC, it is preferable to use the reactor neutron irradiation, since homogeneous damage thereby can easily be achieved. The 1 MeV neutron equivalent of radiation damage, caused by the non-ionizing energy loss (NIEL), is widely used for the estimation of the effective fluence, because the NIEL damage becomes then similar under irradiations with different type particles [30]. The AT GaN samples were irradiated by a nuclear reactor at the Jožef Stefan Insitute (Ljubljana) TRIGA reactor [31]. Fast neutrons with the energy peak at 1 MeV and of 7.5 × 10¹² n/(cm²s) flux act as the main generator in defect formation.

3. Spectroscopy techniques

3.1. Pulsed photoionization spectroscopy

A few results on neutron-irradiated AT GaN materials and the spectral data of the prevailing defects have been reported in our previous publications [32, 33]. In this work, the evolution of the PPI [34] spectra with neutron fluence in a range of 10¹²–5 × 10¹⁶ n/cm² has been studied in more detail. The pulsed response is recorded using microwave probed photoconductivity transients (MW-PC). The peak MW-PC values were determined within the MW-PC transients to estimate the excess carrier density generated by photoionization processes. These values represent a step-like photoionization spectrum. The relaxation shape and duration of the MW-PC transients represent the carrier lifetime and trap filling/emptying processes within later carrier decay stages. The definite steps of photoionization spectra commonly correlate with the changes of the carrier recombination lifetimes, and this allows one to additionally control the filling state of the photoactive centres. The PPIS is preferable for recording of the PC signals, as it is operational in a contactless mode at room temperature and excludes contact-related effects. In this work tuneable wavelength (in a range of 210 to 2300 nm) nanosecond (4 ns) light pulses are generated by an OPO instrument Ekspla NT342B. The sample was placed on the slit-antenna of 21 GHz microwave (MW) system and excited by a laser beam. The MW-PC transients were recorded by a 2 GHz oscilloscope LeCroy Wave Runner 620 Zi on a 50 Ω load resistor connected in series with an MW detector.

Photoionization or photo-neutralization of deep levels is characterized by photoionization cross-sections. Spectral changes of the photoionization cross-section σ(hν) can be described by the Kopylov–Pikhtin model [35]:

Here E_d is the photo-activation energy, Γ is the broadening factor, which determines electron– phonon coupling, and M_ik is the matrix element of a dipole transition from the initial (i) (trap level) to the final (k) (continuum) state. The broadening factor depends on the ground and excited state vibrational frequencies and Franck–Condon shift [36]. In PPI spectra fitting, the E_d and Γ were varied as the free parameters to get the best fit estimated by the nonlinear least squares method for experimental spectra. Changes of the absorption coefficient α(hν) for hν energy photons due to photoionization of trapped carriers can then be described by

Here n_d0 is the concentration of the trapped carriers. The density of the photoemitted carriers, registered by an MW probe, comes from illumination by a light pulse of the surface density of the incident photons

where n^*_d is the density of the photoemitted carriers and F(hν) is the fixed surface density of the incident photons of varied energy hν.

3.2. Pulsed scintillation spectroscopy

The scintillation characteristics have been examined by recording of the UV laser-induced pulsed photoluminescence spectra (PPL) under the experimental conditions close to those of the PPIS measurements. An ultraviolet (UV) laser STA-3H emitting 351 nm light pulses of 400 ps was employed for the excitation of photoluminescence (PL). A PL response was filtered from a laser beam (using an interference 100% mirror at laser wavelength), collected into a UV adjusted fiberscope and transmitted to a spectrophotometer. An AvaSpec-2018L grating spectrophotometer was exploited to integrate the spectrally dispersed PL response. Over 10³ PL transients were integrated and averaged to get the scintillation spectrum in the samples irradiated with the fixed reactor neutron fluence. The PL spectra recorded on pristine and irradiated AT GaN samples provided evolution of scintillation efficiency within various spectral bands.

3.3. Positron annihilation spectroscopy

For PAS experiments, the pristine and 1 × 10¹⁶ cm^–2 as well as 5 × 10¹⁶ cm^–2 neutron fluence irradiated samples were employed. As the pristine samples contain significant concentrations of vacancy complexes, usage of rather low fluences is insufficient to produce detectable densities of radiation defects [24, 25]. Positrons implanted into the sample are captured and localized at neutral and negative vacancies. It is possible to determine the concentration of vacancies as well as to distinguish between different types of vacancies (V_Ga, V_Ga-O_N, V_Ga-H_n) [24, 25]. The Mgdoped (1 × 10²⁰ cm^–3) pristine non-annealed AT GaN sample was used as a reference – possible extraction of the annihilation parameters of the GaN lattice has been shown [24]. The pristine and neutron-irradiated AT GaN samples were examined at room temperature (RT), using a digital positron lifetime spectrometer with a Gaussian time resolution of 250 ps. Two identical pristine samples were sandwiched together with the ²²Na μCi positron source foiled in 1.5 μm thick aluminium. To have the PAS spectrum, the positron lifetime records of 10⁶ positron annihilation (PA) events were collected. The PAS measurements for irradiated samples were made using the same source, but with a different sandwich-structure (the source between the pristine and the irradiated samples). In the latter case, the 2 × 10⁶ events were accumulated to record a PA spectrum and the spectrum of the pristine sample was subtracted in the analysis. The spectrum of the positron decay P(t) (after subtracting the constant background) due to annihilations in the source material (Al foil 210–215 ns, NaCl 400 ns and positronium 1500 ns [23]) was analysed assuming the sum of the exponential components

where positrons in the state i annihilate with the lifetime τ_i and the intensity I_i. The increase of the average lifetime τ_ave [23] above the bulk lattice lifetime τ_b indicates that positrons annihilate as trapped at vacancy defects. Moreover, this parameter is insensitive to the fitting procedures, and even a change of 1 ps in its value can be resolved [23]. The τ_b parameter is the positron lifetime in the delocalized state in the lattice, and τ_v (τ₂ in a two-component spectrum) is the lifetime component that gives directly the vacancy-specific lifetime. Because the electron density at a vacancy is lower than that within the bulk, the lifetime τ_v is always longer than the bulk lifetime τ_b. The first lifetime component is τ₁ = (τ_b^{– 1}+ κ_v)^–1 < τ_b, where κ_v is the positron trapping rate at one species of vacancy defects:

Here c_v denotes the vacancy concentration and μ_v is the positron trapping coefficient, which strongly depends on the charge-state of the vacancy defect. The bulk lifetime τ_b of GaN is approximately τ_b = 160 ps, as measured in the reference sample. The parameter τ_v depends on defect species, it varies in the range 220–240 ps for V_Ga and V_Ga complexes with a single O or H atom. This is clearly less than 200 ps for V_Ga-multi-hydrogen complexes [25]. It is likely that V_Ga complexed with 3 or more hydrogen atoms do not trap positrons [23, 25]. Typical positron trapping coefficients are μ_v = 1 × 10¹⁵ s^–1 for neutral vacancies and μ_v = 2–3 × 10¹⁵ s^–1 for negatively charged vacancies at RT, respectively.

4. Evolution of the PPI spectra

The PPI spectra have been recorded at room temperature on AT GaN:Mg and AT GaN:Mn. The photoionization spectrum is obtained as the MW-PC peak-signal amplitude, proportional to the excess carrier density generated by photoionization, which depends on photon energy. The evolution of the PPI spectra with neutron fluence are illustrated in Fig. 1 for AT GaN:Mn (a) and AT GaN:Mg (b), respectively.

The step-like photoionization spectra contained four spectral peaks for AT GaN:Mn (Fig. 1(a)) and seven peaks for AT GaN:Mg samples (Fig. v(b)), obtained by fitting the recorded PPI spectra.

The PPI spectral steps E_Mn,1 and E_Mn,2 are nearly independent of neutron irradiation (Fig. 1(a)). With fluence enhancement, a few times reduction (relative to that obtained for the pristine sample) of the PPI spectral step of E_Mn,4 can be deduced from Fig. 1(a). This implies the decrease of concentration (proportional to the height of the PPIS step) of this E_Mn,4 definite trap of the same origin (with an invariable cross-section). The insignificant variation of the E_Mn,3 trap density can also be implied from the fluence-dependent evolution of Mn-doped AT GaN (in Fig. 1(a)). The decrease of concentration of the initially existing defects formed by crystal growth can be explained by defect transformations. Really, such a transformation of Mn ascribed defects has been revealed by ESR spectroscopy in our previous study [33]. On the other hand, the carrier lifetime reduction (observed on these samples [33]) with enhancement of neutron fluence seems to be the main reason for the reduced initial filling of E_Mn,4 traps.

In Table 1, the identified activation energies and values of the broadening factor are listed for the predominant centres in Mn-doped AT GaN samples. The threshold photoionization energy values, taken from literature, were compared with those, extracted from fitting of the PPIS peaks, and these values have been employed for identifying of the defects. The density of the Mn associated traps E_Mn,1 and E_Mn,2 seems to be independent of neutron irradiation fluence, despite this the impact of the carrier lifetime decrease is the same for all the resolved traps E_Mn,1–E_Mn,4. This hints that the PPIS step height highlights the initial filling state of the definite photoactive trap. Thereby, radiation defects seem to make the biggest impact on the initial filling of the E_Mn,4 centres, however, of unknown origin.

A rather different evolution (relative to Mndoped AT GaN material) of the PPIS steps was obtained in Mg-doped AT GaN samples (Fig. 1(b)). The most resolvable changes of the height of PPI E_Mg,3–E_Mg,5 spectral steps were obtained. The ESR spectroscopy, performed on these Mg-doped AT GaN samples [33], showed prevailing of the gallium vacancy and substitutional oxygen (V_Ga–O_N) [33, 39, 40] defects in the pristine material and hinted on the appearance of the nitrogen split interstitial [33, 41, 42] defects under irradiation. However, longer carrier lifetimes (relative to the Mndoped material) and existence of a small density of Mn impurities in the pristine Mg-doped AT GaN samples were evidenced [33]. A significant decrease (up to an order of magnitude) of the height of the E_Mg,3–E_Mg,4 spectral steps has been obtained in this study, while less changes of the PPIS signals have been revealed for E_Mg,1 and E_Mg,2 spectral steps. Moreover, nearly invariable height has been observed for the E_Mg,7 spectral step with enhancement of the neutron fluence. For the Mg-doped samples, appearance of the additional PPIS centre E_Mg,5 in the range of elevated fluences (>10¹⁴ n/cm²) can be deduced from Fig. 1(b). The capture cross-section values for all the mentioned traps are nearly the same. This implies that neutron irradiation governs only the initial filling (density of carriers on the definite type centres) of the technologically induced defects.

Values of the estimated activation energy and the broadening factor for different traps are listed in Table 1 for the predominant centres in the Mgdoped AT GaN samples. There, the traps associated with vacancies, interstitials, impurities and dopants are tentatively identified. The amplitudes of different components, which are proportional to the initial filling of the photoactive centres, are plotted in Fig. 2. It can be inferred that neutron irradiations do not change considerably the structure of the photoactive technological defects. In the Mg-doped sample, the radiation-induced defect E_Mg5 is only observable for the samples irradiated with elevated fluences of >10¹⁴ n/cm². The latter defect can play the role of a fast recombination centre. Its impact appears only when the carrier lifetime, linked to this radiation defect, becomes shorter than that inherent for the pristine material.

5. Evolution of the photoluminescence spectra

The neutron fluence Φ dependent evolution of the UV pulse excited photoluminescence (PL) spectra ex situ recorded in Mn (a) and Mg (b) doped AT GaN is illustrated in Fig. 3.

In Mn-doped AT GaN (Fig. 3(a)), a rather wide spectral band peaked at 2.5 eV appears in the yellow–green (YG) PL range. The low intensity blue (B) PL peaked at 3 eV can also be deduced from these (Fig. 3(a)) spectra. A clear decrease of the intensity of the YG-PL band with the enhancement of fluence is clearly observed only for large fluences >10¹⁶ n/cm². The intensity of B-PL decreases nearly monotonically with the increase of irradiation fluence.

In the PL spectra recorded for Mg-doped AT samples (Fig. 3(c)), the same PL bands appear. However, the intensity of the B-PL wing is larger in this material relative to that in Mn-doped AT GaN. Moreover, the intensity of both YG-PL and B-PL is nearly invariable up to irradiations with moderate fluences <10¹⁵ n/cm². A clear reduction of the PL intensity appears only for Φ ≥ 10¹⁵ n/cm². This intensity reduction is most expressed for the B-PL spectral wing.

It is very probable that wide PL bands (especially YG-PL) are composed of several overlapping PL peaks. These peaks can be predicted by correlating PPI and PL spectra in consideration of the conversion from absorption to emission.

The absorption centres resolved within PPIS (Fig. 1) may originate the Stokes shifted PL bands. These YG-PL and B-PL peaks were fitted (Fig. 3) by using the van Roosbroeck–Shockley (vR–S) approach [50, 51] in relating the conversion from the PPI to PL spectra. The rate ΔR(hν) of the emission via the trap of photo-activation energy E_d is linked to the photon–electron interaction cross-section σ(hν), whose value also covers the B- and YG-PL peak photon energy (assigned to PL as σ_{d PL}(hν)) and is expressed as follows:

Here n_i = 2 × 10⁻¹⁰ cm^–3 is the intrinsic carrier concentration for GaN materials [52], k_B is the Boltzmann constant, h is the Planck constant, ρ_a is the surface density of photons ascribed to the unit frequency ν within the absorption spectrum for the spectral range Δ(hν), inherent to the dedicated trap of the concentration N_d, and n_exΔ(hν) is the excess carrier density generated through photoionization in the definite spectral range Δ(hν), τ_{PL Δ(hν)} is the PL relaxation time.

The relations between the photoionization and photoluminescence spectra (Fig. 3) were simulated using Eqs. (1, 2, 3). The comparison of the simulated photoionization and photoluminescence spectra ascribed to several PPIS steps is illustrated in Fig. 3. For neutron-irradiated materials, the structure of B- and YG-PL was reproduced by varying the N_d^*value with irradiation fluence which means variation of the filling level of definite centres. Simulations were performed using the extracted values of the cross-section of the photon–electron coupling σ_d(hν) and of the photoionization energy E_d (as well as broadening factors Γ_d) and by adjusting the σ_{d PL}(hν) function to fit I_PL(hν) spectral peaks comprising a rather broad PL band. An association of the PPIS steps to the definite PL peaks is also sketched in Fig. 3 by denoting the respective deep traps (E_i, i = 2–7 according to notifications in Table 1).

A rather good reconstruction of the PL components, necessary for composing of wide PL spectral bands, was obtained by simulating (using vR–S approach) the experimental PL (Fig. 3(a, c)). The evolution of the amplitudes of different PL spectral components with neutron irradiation fluence is illustrated in Fig. 3(b, d) for GaN:Mn and GaN:Mg materials, respectively. The simulations of PIS and PL spectra show that the additional radiation-induced trap E_Mg,5 PPIS peak is very probable. Based on these fits it can be inferred that existence of technological defects in AT GaN:Mn is the main reason for the appearance of YG-PL. The richer PPI and PL spectra, recorded on AT GaN:Mg, and their modifications with neutron irradiation are determined by a variety of technological defects, the filling level of which changes with the appearance of fast non-radiative recombination centres, and radiation-induced additional photoactive traps. There, the faster reduction of the amplitudes of different PL spectral components (Fig. 3(b, d)) in comparison with the PPIS component amplitudes (Fig. 2(a, b)) can be explained by the additional (relative to the initial filling of definite traps) decrease of excess carrier concentration (the term n_exΔ(hν) in Eq. 2) through non-radiative decay channels. The B and YG bands are assigned to Mg impurities and Ga-vacancy complexes in Mgdoped AT GaN material [53]. Vacancy complexes are plausible centres responsible for luminescence variations with neutron fluence when generation of radiation vacancies is very probable. Vacancy complexes with oxygen (V_Ga-O_N) or silicon (V_Ga-Si_Ga) might be responsible for YG luminescence. Additionally, C_N-O_N technological defects may also be a reason for YG luminescence [54]. It seems that Ga vacancies are most probable non-radiative recombination centres. By generalizing the PL data, a significant reduction of the overall PL intensity is quite evident (Fig. 3(b, d)). Also the overall PL intensity appears lower in the Mn-doped material from the very beginning. The dominant traps can tentatively be estimated by using the algebra of the obtained photo-activation energy values relative to a band gap (E_G = 3.45 eV [50]) as well as valence/conduction band edges and to the identified PPIS traps as well as B and YG range PL transitions. Thereby, E_1−PPIS = 1.4 eV can be resulted from E_G−hν₂ ≈ 1.3 ± 0.1 eV.

6. Irradiation-induced vacancy defects observed by positron annihilation spectroscopy

Figure 4 shows the average positron lifetime measured at RT in the pristine and neutron-irradiated AT GaN:Mn and AT GaN:Mg samples. The positron lifetime spectra consisted of only one resolvable component, suggesting saturation trapping at V_Ga- hydrogen complexes and also a contribution from negative ions (Mn and Mg), as discussed for the case of AT GaN in detail in Refs. [23–25]. Nevertheless, important conclusions can be drawn from the average positron lifetime. First, the average positron lifetime in the pristine AT GaN:Mn and AT GaN:Mg is clearly longer than the bulk lifetime of 160 ps, indicating that positrons are trapped at vacancy defects in these samples. Moreover, the average positron lifetime is longer in pristine AT GaN:Mn than that in AT GaN:Mg, indicating that the total amount of the open volume is larger in AT GaN:Mn, suggesting that the concentration of the Ga vacancy related defects is higher. Other possible explanations, namely that the vacancy defects would be larger (that is, contain less hydrogen) in AT GaN:Mn or that AT GaN:Mg would contain more negative ions (that in GaN affect the data also at RT to some extent), are considered unlikely. Also, the hydrogen content in the samples is similar, and the Mn doping is higher (by a factor of two) than the Mg doping.

The average positron lifetime increases significantly after neutron irradiation in AT GaN:Mg, in particular at a higher fluence of 5 × 10¹⁶ cm^–2. This indicates the introduction of Ga vacancy defects in the irradiation process, similarly as observed after, e.g. electron irradiation and hydrogen implantation [37, 38]. As the vacancy defect concentrations in the pristine AT GaN samples are likely to be of the order of 10¹⁸ cm^–3 (or more due to the saturation phenomenon discussed in Ref. [25]), at least similar concentrations of vacancy defects are introduced in the neutron-irradiated samples with the higher fluence. Interestingly, the average positron lifetime is hardly affected by the neutron irradiation in the AT GaN:Mn samples, although the concentrations of the irradiation-induced vacancy defects are necessarily the same as in AT GaN:Mg. This supports the interpretation that the vacancy defect concentrations are significantly higher in AT GaN:Mn than those in the AT GaN:Mg samples in the pristine material.

7. Discussion

The positron annihilation spectroscopy (PAS) data show that the pristine AT GaN:Mn material contains a higher concentration of vacancy defects than AT GaN:Mg. This is seen both as a higher average lifetime of positrons in the pristine material, and also by the neutron irradiation having very little effect on the average lifetime in AT GaN:Mn. In contrast, in AT GaN:Mg, the irradiation (in particular with the highest fluence) causes a strong increase in the vacancy defect concentrations observed by the increase of the positron average lifetime. This is equivalent to an increase in the average size of the open volume in the vacancies. The increase of the positron average lifetime is related (as we have shown earlier [24, 25]) to the fact that the ingrown Ga vacancies in AT GaN are present in hydrogen complexes that produce a shorter lifetime than ‘pure’ V_Ga produced by the irradiation. It means that the concentration of irradiation-induced vacancies is significant compared to the concentration of the ingrown vacancies (Mg-doped) if the average open volume at the vacancies observed with positrons increases. Also, it means that the concentrations of the ingrown vacancies in Mn-doped AT GaN are significantly higher than those created by irradiation, even at the highest fluence, when no changes of the positron lifetime are observed.

The comparison of the PAS and PL data shows that the Ga vacancies produced by irradiation are non-radiative recombination centres. The significant reduction of the overall PL intensity with the enhancement of neutron fluence nicely correlates with the positron annihilation data. This effect is stronger in the Mg-doped material; as the ingrown Ga vacancy-hydrogen complexes are probably also detrimental for the PL intensity in general, the Mndoped material appears to luminesce less than the pristine Mg-doped material. Also, the stronger irradiation effect in the Mg-doped materials supports this hypothesis. Moreover, this comparison between the Mg- and Mn-doped materials evidences the smaller effect in the Mn-doped samples, where the pristine vacancy concentrations seem to be higher, as the irradiation-induced vacancies affect the positron lifetime data so little.

8. Conclusions

The decrease of concentration of the initially existing defects formed by crystal growth has been observed after neutron irradiation and it can be explained by defect transformations. This defect transformation effect of Mn-ascribed defects was revealed by ESR spectroscopy. The carrier lifetime reduction (observed for these samples) with the enhancement of neutron fluence seems to be the main reason for the reduced initial filling of E_Mn,4 traps. The density of the Mn associated traps E_Mn,1 and E_Mn,2 seems to be independent of neutron irradiation fluence; despite this the impact of the carrier lifetime decrease is the same for all the resolved traps E_Mn,1–E_Mn,4. This hints that the PPIS step height highlights the initial filling state of the definite photoactive trap. Thereby, radiation defects seem to make the biggest impact on the initial filling of the E_Mn,4 centres of unknown origin. The most resolvable changes of the height of PPI E_Mn,3–E_Mn,4 spectral steps have been obtained.

A rather different evolution (relative to the Mndoped AT GaN material) of the PPIS steps has been obtained in the Mg-doped AT GaN samples. The longer carrier lifetimes (relative to the Mndoped material) and existence of a small density of Mn impurities in the pristine Mg-doped AT GaN samples have been evidenced by EPR. The significant decrease (up to an order of magnitude) of the height of the E_Mg,3–E_Mg,4 spectral steps has been obtained with the enhancement of neutron irradiation fluence. Meanwhile, less changes of the PPIS signals have been revealed for E_Mg,1 and E_Mg,2 spectral steps and nearly invariable height has been observed for the E_Mg,7 spectral step with the enhancement of neutron fluence. For the Mg-doped samples, the appearance of the additional PPIS centre E_Mg,5 in the range of elevated fluences (>10¹⁴ n/ cm²) has been deduced. It has been concluded that neutron irradiation governs only the initial filling (density of carriers on the definite type centres) of the technologically ingrown defects.

The traps associated with vacancies, interstitials, impurities and dopants have been tentatively identified. It has been inferred that neutron irradiations do not change considerably the structure of the photoactive technological defects. The radiation induced defect E_Mg,5 is only observable for the samples irradiated with the elevated fluence of >10¹⁴ n/cm². The latter defect can play the role of a fast recombination centre whose impact appears only when the carrier lifetime, linked to this radiation defect, becomes shorter than that inherent for the pristine material.

The relations between the photoionization and photoluminescence spectra have been simulated using the van Roosbroeck–Shockley approach. For neutron-irradiated materials, the structure of B- and YG-PL has been reproduced by varying the N_d^* value with irradiation fluence which means variation of the filling level of definite centres. An association of the PPIS steps to the definite components of the PL bands has been performed. A rather good reconstruction of the PL components, necessary for composing of the wide PL spectral bands, has been obtained by simulating the experimental PL spectra. The evolution of the amplitudes of different PL spectral components with neutron irradiation fluence has been obtained. The richer PPI and PL spectra, recorded on AT GaN:Mg, and their modifications with neutron irradiation are determined by the wider variety of technological defects, the filling level of which changes with the appearance of fast non-radiative recombination centres, and radiation-induced additional photo-active traps. The faster reduction of the amplitudes of different PL spectral components in comparison with the PPIS component amplitudes can be explained by the additional decrease of the excess carrier concentration n_exΔ(hν) through the non-radiative decay channels. Vacancy complexes with oxygen (V_Ga-O_N) or silicon (V_Ga-Si_Ga) might be responsible for YG luminescence. Ga vacancies are most likely non-radiative recombination centres. This is supported by the PL data. Additionally, C_N-O_N technological defects may also be a reason for YG luminescence. The positron annihilation data support the hypothesis that Ga vacancies are non-radiative recombination centres and show that the Mn-doped material has more vacancies than the Mg-doped material.

Acknowledgements

This research is funded by the European Social Fund under Measure No. 09.3.3- LMT-K-712 -02-0012 ‘Development of Competences of Scientists, other Researchers and Students through Practical Research Activities’. The authors are indebted to Dr. M. Zając for providing the AT GaN materials. The neutron irradiations at the TRIGA Facility were supported by the AIDA-2020 Transnational Access Project funded by the EU Horizon 2020 R&D Programme under Grant Agreement No. 654168.

References

NEUTRONAIS APŠVITINTO AMONOTERMINĖS TECHNOLOGIJOS GaN SPEKTROSKOPIJA FOTOJONIZACIJOS, FOTOLIUMINESCENCIJOS IR POZITRONŲ ANIHILIACIJOS METODAIS

Santrauka

Impulsinės fotojonizacijos, fotoliuminescencijos ir pozitronų anihiliacijos spektroskopinių tyrimų rezultatai buvo derinami siekiant identifikuoti savituosius ir radiacinius defektus reaktoriaus neutronais apšvitintose amonoterminės technologijos GaN (AT GaN) medžiagose, priemaišintose Mn ir Mg. Įvertintos defektų evoliucijos charakteristikos keičiant neutronų apšvitos įtėkį plačiame 10¹² – 5 × 10¹⁶ n/cm² intervale. Atskleista, kad nespindulinę rekombinaciją nulemia galio vakansijos, kurių koncentracija yra didesnė Mn priemaišintame AT GaN. Vakansijų kompleksai su deguonies ir kitomis priemaišomis nulemia fotoliuminescencijos spektrų evoliuciją ir visų aptiktų liuminescencijos smailių intensyvumo mažėjimą didėjant apšvitos įtėkiui. Fotojonizacijos ir fotoliuminescencijos spektrų sąsajos paaiškintos Kopylovo-Pikhtino ir van Roosbroeck’o-Shockley modelių artinyje. Vakansiniai defektai identifikuoti matuojant pozitronų gyvavimo trukmę.