X-ray assisted point defects creation in micron-size Zn, Mo oxide particles at liquid nitrogen temperature

Charge trapping processes induced by the X-ray irradiation in the heavy Mo doped ZnO and MoO3 micropowders synthesized by the hydrothermal growth method were investigated in detail. Electron paramagnetic resonance (EPR) and thermally stimulated luminescence (TSL) were applied in a correlated manner to discover the role of the Mo doping in the charge trapping processes in ZnO. Thermally unstable oxygen- and molybdenum-related charge trapping centers were studied. Molybdenum and oxygen created electron-hole trapping pairs in some cases were observed. Some part of the hole trapping centers seemed to be directly connected with the creation of Mo5+. The correlation between EPR and TSL data was found.


Introduction
ZnO can be easily synthesized as hexagonal Wurtzite nanorods by using hydrothermal method of growth [1][2][3][4][5].Random nucleation seeds can be used as well as these nanorods can be grown on the nucleation layer [6][7][8].Since ZnO-based nanostructures are intensively studied as the potential ultrafast scintillators to be implemented in the detectors ring of the time-of-flight positron emission tomography (TOFPET) [9,10], to know the details about the charge trapping processes appearing under X-ray will be of great importance for many scientists engaged in this field.Some other applications of ZnO can be also mentioned such as different catalytic activities (photo-, electrocatalysis, sensing, Li-ion batteries) [11][12][13][14][15][16][17].
The following properties making nano zinc oxide important for the TOF-PET applications can be outlined.(i) Ultrafast excitonic luminescence with the decay time of hundreds of ps as mentioned in the recent reports on the ZnO:Mo [6,18,19].The maximum of the excitonic band appears at about 380 nm, however, it is changing depending on the type of nanostructure: in the free-standing nanorods/nanocolumns it is redshifted as compared to the deposited films of aligned nanorods [20].(ii) Excitonic emission can be moderated by different treatments as reported in many works [8,[21][22][23].For example, X-ray irradiation or oxygen plasma treatment lead to the suppression of both excitonic and defects-related bands [24][25][26].In general, the X-ray irradiation may be the tool for the surface defects moderation leading to the new bonding abilities.
Great disadvantage of the ZnO nanoparticles always was the defect-related luminescence created by the e.g., zinc vacancy in the hydrothermally grown samples.From another point of view -this band exists in all hydrothermally grown ZnO samples and can be moderated by different treatments [24-Figure 1. EPR spectra measured at T = 50 K in the ZnO:Mo(10 %) powder after X-ray irradiation and annealing at Tann = 296 K after the X-ray irradiation.O1-7, O(1,2), Mo + indicate the signals produced by the X-ray induced defects.SD marks the typical shallow-donor signal in ZnO (see e.g., [31][32][33] and the references therein).
The sample was subsequently heated to 296 K.After 4 minutes spent at this temperature, it was promptly cooled to 50 K.Then, it was again measured.New signals became visible (O3,6,7) while the O(1,2),O1,2,4,5 and Mo 2 5+ disappeared (figure 1).The Mo 1 5+ remained unchanged.All the signals mentioned were also observed in the EPR spectra of the ZnO:Mo(0.25, 1 %) and described in detail in the previous work [25].There, the intensity of the O(1,2),O1-7 and Mo 1,2 5+ signals was increasing upon Mo content.Since the EPR intensity was normalized to 1 mg in work [25] and in the present case (see Experimental) the O(1,2),O1-7 and Mo 1,2 5+ signals detected in the ZnO:Mo(0.25, 1 %) from work [25] and presently in ZnO:Mo(10 %) could be compared.Indeed, they were about 1.5 times stronger in the ZnO:Mo(10 %) as compared to the ZnO:Mo(1 %).From this follows the conclusion that Mo influences X-ray assisted defects creation through charge trapping.According to [25]  5+ signals were related to Mo 5+ ions situated in different places in the host (for more details see [25]).Since the direct connection between the oxygen-based and Mo-based signals intensities as well as Mo presence was established by consulting to the previous work [25] the following conclusion can be made.The creation of the O -defects is much more probable than the O 2 − since the charge compensation of the view (two steps) should exist as reported for ZnMoO4 [34,35] In order to follow the trend with the increased Mo content, the doping level of Mo was increased to 30 %.The ZnO:Mo(30 %) sample, however, appeared to be not ZnO but the mix of complex molybdates as it was also reported in [27,28].Even in this case the knowledge of charge trapping processes is important for bonding peculiarities understanding as they are critical for phase purity of the material and charge/energy transfer there.EPR spectra of the ZnO:Mo(30 %) before and after the X-ray irradiation at LNT can be seen in figure 2A.There are resonances prior to X-ray irradiation.Four contributions (1-4 in figure 2A) appear in the magnetic field range corresponding to the g factors lower than the free-electron value 2.0023 [36].Moreover, the corresponding g factors g = 1.946 (1), g = 1.935 (2), g = 1.904 (3) and g = 1.868 (4) are typical for Mo 5+ (the outer shell is 4d 1 and the electron spin is S = 1/2) [37].However, the powder EPR spectrum of Mo 5+ can be composed of maximum 3 contributions, not four.Therefore, one may expect at least two strongly overlapped Mo 5+ related signals existence (  2A).They should originate from the Mo 5+ occupying different positions.Taking into account that according to XRD, the ZnO:Mo(30 %) is the mix of at least two material phases Zn3MoO9•H2O and Zn5Mo2O11•5H2O as reported in [28], the There is another group of signals appearing centered at the free electron g factor (corresponds to the resonance magnetic field 3342.5 G) and spreading almost symmetrically with respect to this data point covering the resonance magnetic field range from 3260 to 3425 G range (figure 2A).The character of the localization of these spectral features and the same reaction to the different microwave power led to conclusion that they should originate from d 5 shell.Since no additional pronounced (super)hyperfine structure was observed, one may expect this spectrum to be produced by Fe 3+ .The following spin-Hamiltonian has been used to approximate the EPR spectrum in figure 2A H ̂=  ̂ +  2 0  2 0 +  4 0  4 0 +  4 4  4  4 , where , ,  ̂, ,    ,    (k = 2, 4; q = 0, 4) are Bohr magneton, g factor, electron spin operator (z component, electron spin S = 5/2), crystal field parameters and Stevens operators [37].The experimental spectrum as well as the simulated ones are shown in figure 2B.The agreement between the spectra are, in general, good.Some small disagreement originates from the complexity of the EPR spectrum containing besides these resonances also the other resonance line, e.g., belonging to Mn 2+ as reported in [29,30].The spin-Hamiltonian parameters were thus determined as follows:  = 2.0023 ± 0.0005,  2 0 = 19 ± 1 MHz,  4 0 = −0.80 ± 0.05 MHz,  4 4 = 0.80 ± 0.05 MHz.X-ray irradiation causes the non-equivalent increase of the components 1-4 (contribution 2 becomes more intense as compared to 1).This supports the existence of several Mo 5+ centers.Moreover, this also confirms participation of molybdenum in charge trapping processes.Since the charge state of molybdenum is mostly Mo 6+ (according to XPS data reported in [28]), the electron trapping should be expected: Mo 6+ + e -→ Mo 5+ .This process is reversible by increasing the annealing temperature to 170 and 296 K (figure 2A).However, even the annealing at 296 K is not enough to totally recover the Mo xx + signal intensity (figure 2A).All of these also confirms that the Mo 6+ position in this case is the same as that of the

Mo
x + in the not irradiated sample.
The X-ray irradiation also causes the appearance of the totally new signals which features are stressed by C1-4 in figure 2A.The spectral position of the C1-3 is in the region corresponding to the g factor values found slightly above the g factor of the free electron, 2.0023.This is typical situation for the oxygen-related centers as mentioned above for the ZnO:Mo(10 %) sample (according to [25] The spectral position of the C4 component is characterized by the g factor slightly smaller than 2.0023.This is typical for the electron trapped at an oxygen vacancy [38].
To study this in more detail, the X-ray induced spectrum has been fitted with four contributions using the spin-Hamiltonian below.
where  ̂,  ̂,  are electron spin operator ( ̂= ( ̂  ̂  ̂) and the electron spin S = 1/2), g tensor with the three main components (g1-3) and the magnetic field vector.The experimental and calculated spectra are shown in figure 2C.The fit is very good confirming correctly chosen model.The fit parameters appear in Table 1.The linewidth in the case of the O2,3 is about 2-3 times larger as compared to the O1.This is the way to implicitly account either for the superhyperfine interaction or the local crystal field strength distribution or both.The existence of the F + center (confirmed by the g = 1.997 < 2.0023, see Table 1) after the X-ray require at least one of the O1-3 to originate from the O -defect.The same is true for the creation of Mo 5+ .The following charge compensating mechanisms can be proposed: In general, these processes appear at short distance between the donor and acceptor [34,35].Therefore, one may expect at least two of the O1-3 to be produced by the O -defects.Considering the same g3 value determined for all three O1-3 signals, the same origin must be expected, i.e., along with the proofs above, one may conclude that all the oxygenrelated signals originate from the differently perturbed O -defects.Nevertheless, the presented data are not enough to unambiguously distinguish which of at least two material phases Zn3MoO9•H2O and Zn5Mo2O11•5H2O as reported in [28] the O1-3 belong to.
The EPR spectra of the MoO3 sample before and after the X-ray irradiation with the subsequent annealing at 170 and 296 K are shown in figure 3A.They consisted of the U1-3 signals and 2-5,7 spectral components.The U1-3 components appear in the magnetic field range 3265-3440 G corresponding to the g factor 1.95-2.05.To find out the origin of these signals, they were approximated using Eq. 2 (these signals have different origin and, considering the typical powder shape [36] the S = 1/2 is expected).The fitting signals are shown in figure 3B.The fit is good.The spin-Hamiltonian parameters (Eq.2) are listed in Table 1.Considering the g tensor values a bit higher than the 2.0023 and typical for the O - defects [39,40] in the case of U1, one may refer it to as being the O -defect as well.The g tensor values of the U2 signal are just a little bit smaller than the 2.0023 value, and, therefore, this signal was assumed to be the F + center mixed with the F center, F + -F center (two electrons trapped at the same oxygen vacancy) [41].The g tensor values of the U3 signal are typical for the F + center [38].
Considering the absence of any hyperfine structure due to the 95,97 Mo nuclei [37] one may expect the U3 spectrum indeed to originate from the F + center [38].
There are other resonances stressed out by the numbers 2-5,7 (figure 3A).They appear in the magnetic field region 3438-3585 G (figure 3A) corresponding to the g factor values 1.87-1.95,typical for Mo 5+ (4d 1 outer shell, S = 1/2) in ZnO and other oxide materials as reported in [8,37].The number of contributions (5 in total) suggests at least two powder spectra overlapped (figure 3A) [36].These signals were fitted using Eq. 2 as well.The fit is good (figure 3B).The fit parameters are listed in Table 1 as well.As one can see, the parameters are indeed typical for the Mo 5+ as can also be seen for the Mo 5+ in ZnO:Mo(10 and 30 %) (see Table 1 and [37]).The hyperfine structure due to the 95,97 Mo is lost in the broadened 2-5,7 resonances [8].These were marked as (Table 1).
New spectral features marked as C 1-3  appear after the X-ray irradiation in the EPR spectrum of the MoO3 (figure 3A).They are found in the magnetic field range 3280-3350 G corresponding to the g 2.0024-2.0450which is typical for the oxygen-related centers (O 2 − or O -) [42,43].To find out the origin of these signals, Eq. 2 was used to fit them (these signals have different origin and, considering the typical powder shape [36], S = 1/2 is expected).The experimental spectrum was created by subtracting spectrum measured before the X-ray irradiation from the spectrum measured after the X-ray irradiation to avoid additional influence of the signal existing prior to the irradiation.The fit is shown in figure 3C.It agrees well with the experimental one.The fit parameters are listed in Table 1.The C 1-3  components in figure 3A thus were assigned as follows: ).Four oxygen-related centers were discovered also considering the changes in the C 1-3  components due to the annealing at 170 and 296 K as shown in figure 3A.Note, that the EPR spectra were all the time measured at 70 K.Considering the remained U2 signal (U2R in figure 3C), one may expect the participation of the origin of the U2 signal (F + -F center) in the electron trapping processes.
X-ray irradiation is also responsible for the creation of the additional signals 1,8 and the broadening of the 2-5,7 signals and increase of their intensity.This is indication of several new signals appearance strongly overlapped with the They are expected to be produced by Mo 5+ as well, considering the same magnetic field range as of the centers may appear in each of the materials as well as in both.However, based on the presented data it is hard to distinguish unambiguously.
In some cases, the linewidth was about doubled, e.g., increased from 6-8 to 15-20 G (Table 1).This is the sign of the crystal field strength variation over the positions of the specific charge trapping centers, the probable indication of the surface character of the defects.
The detected oxygen-and molybdenum-related centers thermal stability is investigated by the correlated EPR, TSL and RL experiments as discussed in the section below.

Thermal stability investigation
The TSL glow curves of the as grown ZnO:Mo(10, 30 %) and MoO3 samples are shown in figure 4. The glow curve of the ZnO:Mo(10 %) is composed of one complex and broad glow peak having maximum at 113 K. Very similar TSL peaks were measured in the undoped ZnO and ZnO:Mo(0.05,0.25 and 1 %) samples previously [25].Taking into account the same charge traps activation in the ZnO:Mo(10 %) as compared to the ZnO:Mo(0, 0.05, 0.25, 1 %) (see figure 1 and [25]), the TSL peak in the glow curve of the ZnO:Mo(10%) is expected to have the same origin as the peaks in the ZnO:Mo(0, 0.05, 0.25, 1 and 10 %) samples.
The intensity of the glow curves obtained for the ZnO:Mo(30 %) and MoO3 samples is negligibly weak (see also an inset in figure 4).The glow curve of the ZnO:Mo(30 %) consists of two broad peaks having maxima at 115 and 158 K whereas the MoO3 TSl glow curve of is composed of the broad peaks at 110 and 187 K.
The order of the trapping kinetics as well as the advanced analysis for the trap depth (Et) and frequency factor (f0) determination was impossible in all the ZnO:Mo(30 %) and MoO3 samples as the intensities of the glow curves were too weak [44].However, this was possible for the dependences of the corresponding EPR spectra on the annealing temperature.These were obtained applying commonly known pulse annealing method.There, an EPR spectrum is measured at the fixed temperature after each 4 minutes annealing cycle while the annealing temperature is increased with the certain increment.The annealing temperature dependences of the peak-to-peak intensity are plotted for the O1-3 EPR signals along with the TSL glow curve measured in the ZnO:Mo(30 %) in figure 5.The intensity of the F + signal (Table 1) was too weak in this case to be considered.The O1-3 EPR decay curves were analyzed taking into account the first order trapping kinetics using Eq. 3 (the direct proportionality between the EPR intensity and paramagnetic particle concentration [36] was considered): where Ii,i+1, f0, t, Et, kB and Ti are the peak-to-peak EPR intensity in i th and i+1 th cycle of annealing, frequency factor, annealing period (4 minutes was found to be optimal for good thermalization of the samples), trap depth or activation energy [44], Boltzman constant and the annealing temperature in the i th cycle.As on can see, the EPR decay curves could be approximated in the simple way only for the O1 signal (Eq.3).However, the decay curves of the O2,3 EPR signals were more complex, composed of two descending regions.These regions were approximated (Eq. 3) with the curves indicated as the O2,3(1,2), respectively (figure 5A).The corresponding kinetic parameters were determined and listed in Table 2.
Table 2. Trap depths (Et), frequency factors (f0), initial intensity (I0), parameter P, initial concentration (n0) of traps and the number of available traps multiplied by the R = A/Ar (A, Ar are the transition coefficients) [44] (NR) of Eqs.2-6 used in the simulations.
The O2,3(1,2) regions in the EPR decay curves should appear due to either overlap of the several signals having close values of g factors (see figure 2) or possible recombination with the electrons released by some non-paramagnetic electron traps.To study this question and to fit the TSL glow curve in figure 5B, the TSL glow peaks were simulated by using Eq. 4 with the slightly modified parameters for the O1 and O2,3(1,2) from Table 2: where  = 0.1 K/s is the heating rate for the TSL glow curves acquisition, n0 and T0 are the initial concentration of traps and starting temperature.The rest of parameters are completely the same as in Eq. 3.However, thus simulated TSL glow peaks were not enough to fit the glow curve totally (figure 5B).The other two peaks (O(1,2)) were simulated with this purpose but considering the second order kinetics in Eq. 5 where N, n0 are the total and actual/initial (filled traps) concentrations of the electron or hole traps in the material, respectively; R = A/Ar, A, Ar are the transition coefficients [44]; kB is the Boltzmann constant.All the parameters used in the simulations (Eqs.4,5) are listed in Table 2.The parameters of the O(1,2) were very close to the parameters of the oxygen-related O(1,2) charge trapping centers reported in [25,45].Therefore, the ZnO phase existence in the ZnO:Mo(30 %) can be expected.The fit of the TSL glow curve is good (figure 5B).Small differences in the Et and f0 values determined from the fitting of the EPR decay curves and used in the simulation of the TSL glow curve can be explained by the low amplitude of the glow curve and different heating method used in EPR and TSL techniques.All of this indicates that the O2,3 (1,2)  trapped with released electrons.The number of the oxygen-related, most probably, hole trapping centers (see also the section 3.1) is thus five.However, it is impossible to guess which material phase of the detected by XRD [25] they belong to.
Experimental EPR decay curves of the O1  and O3  signals measured in MoO3 (figure 3) is shown along with the calculated ones in figure 6A.where P = kBRN.The parameters of fit are listed in Table 2.Note that the trap depth and frequency factor of the O1  are the same as those of the O(2) above and the O(1) in the undoped ZnO or ZnO:Mo(0.05%) as discussed in [25,45].The decay curve of the O3  appears more complex.At least three regions can be distinguished: one ascending and two descending.The ascending (77-110 K) and the first descending (110-130 K) regions should appear due to the re-trapping of holes released from the O1  hole trap and recombination with electrons released from Mo 5+ , respectively.The second descending region (130-170 K) is thus considered to be produced by real thermal release of holes from the O3  trap.The 110-130 K and the 130-170 K regions were fitted with the calculated ones (using Eq. 4 for the first order kinetics), ( ) , respectively, with the parameters listed in Table 2. Remarkably, the O3  exhibits first order trapping kinetics as compared to the O1  .The experimental decay curve of the Mo 5+ ( 53, 4,

Mo x x y
+ in figure 3A) was fitted with the curve calculated by using Eq.6 for the second order trapping kinetics.The parameters of fit are listed in Table 2.By comparing the decay curves and the determined kinetic parameters of the O1  and Mo 5+ centers (Table 2) the existence of the electron-hole pair trapped at the O1  and Mo 5+ , respectively, can be expected.Similar electron-hole pair was reported in ZnMoO4 [34,35].To fit the TSL glow curve in figure 6B, the TSL glow peak was simulated using Eqs. 4 with the modified parameters of the O1  and ( ) . To complete the simulation and to reach the agreement between the experimental and calculated TSL glow curve in figure 6B, the other center was considered, the ( ) . Definitely, it is strongly overlapped with the other resonances in figure 3 and therefore could not be separated.Probably, this can be related to the U2 signal in figure 3A.The parameters used for the glow peaks simulation are listed in Table 2. Considering very low intensity of the TSL glow curve measured in MoO3, the fit appears good enough.Therefore, the charge trapping models proposed are corrected.

Conclusions
ZnO:Mo(10, 30 %) and MoO3 microparticles were synthesized using the hydrothermal growth method.They were exposed to the X-ray irradiation.According to [29] the ZnO:Mo(10 %) is the multiphase compound, composed of the Wurtzite ZnO and complex zinc-and molybdenum-based oxides.The latter material phases are even more pronounced in ZnO:Mo(30 %).The ZnO phase was detected only in a negligible amount the sensitivity of XRD (see [30]).The MoO3 sample was found to be the composition of anhydrous and hydrated MoO3 [29].
The existence of the oxygen-based trapping centers ( O − or 2 O − ) and the Mo 5+ -based electron trapping centers were revealed after the X-ray irradiation of all the samples.Their spectral localization as well as thermal stability were investigated as well.The g factors and kinetic parameters (trap depths and frequency factors) were determined.The oxygen-related charge trapping centers were correlated to the TSL glow peaks.The obtained results will help to understand the processes of defects creation in different, even complex structures and develop the tool for the improvement of their light yield.

Figure 2 .
Figure 2. A -EPR spectra measured at 50 K before and after X-ray irradiation at LNT and after X-ray irradiation at LNT and subsequent annealing (temperatures are listed in the legend) in ZnO:Mo(30 %).C1-4 and 5 1, 2 Mo xx + 3+ are stressed by violet arrows.B -EPR spectrum measured at 50 K before X-ray irradiation shown along with the calculated one.C -EPR spectrum measured at 50 K after X-ray irradiation shown along with the calculated ones.O1,2,3 and F + are the signals originating from oxygen centers and the electron trapped at an oxygen vacancy (VO): V O 2+ + e − → V O + (F + ).
to originate from Zn3MoO9•H2O or from Zn5Mo2O11•5H2O.Based on the presented data it is hard to unambiguously distinguish them.

Figure 3 .
Figure 3.A -EPR spectra measured in the MoO3 at T = 50 K prior to and after X-ray irradiation at LNT and after X-ray irradiation at LNT and subsequent annealing at 170 and 296 K. U1-3 indicate spectral components existed prior to the X-ray irradiation.C 1-3  indicate spectral components appeared after the X-ray irradiation.

+
and numbers 1-8 stress out the Mo 5+ centers contributions with the vertical dashed lines.B -EPR spectrum measured in the MoO3 at T = 50 K before X-ray irradiation shown along with the calculated one including fit of the U1-3 and -EPR spectrum measured after the X-ray irradiation in the MoO3 at T = 50 K.The spectrum measured prior to the X-ray irradiation was subtracted.It is shown along with the calculated one including contributions from O 1-4  oxygen-related centers.U2R indicates the remained U2 signal after the subtraction.
-F centers requires hole trapping centers for the charge compensation, i.e., the O -defects.Therefore, theO 1-4   are expected to be produced by the O -defects appearing in different local surroundings.Because the MoO3 sample is indeed the mix of the MoO3 and MoO3•0.5H2Omaterial phases, the

Figure 4 .
Figure 4. TSL glow curves measured in the as grown ZnO:Mo(10, 30 %) and MoO3 samples as indicated in a legend.Specific peaks are indicated.

Table 1 .
Spin-Hamiltonian parameters of Eq. 2. The error of the g tensor values determination was ±0.003.
regions appear due to the overlapping of signals produced by the different charge trapping centers and not due to the recombination of the holes