J. CHEM. SOC. FARADAY TRANS., 1994, 90(21), 3267-3271 Time-resolved EPR Study of Spin Polarization and Dynamics of F+ Centres in Additively Coloured CaO Crystalst Carlo Corvaja, Lorenzo Franco, Luigi Pasimeni and Antonio Toffoletti Department of Physical Chemistry, University of Padova, Via Loredan 2,35131 Padova, Italy Optical excitation of additively coloured CaO crystals generates 'T, and 3T, excited states and F+ defects, the latter being revealed by EPR experiments. Interaction of the excited 3T,u triplet state and the F+ centre is documented by the emissive character of the transient EPR signal detected just after a laser pulse. A quenching of the initial polarization, observed at high microwave power, is ascribed to microwave saturation of the F+ spin leveIs.In additively coloured CaO crystals the defect present at the largest concentration is the F colour centre consisting of an oxygen ion vacancy at which a pair of electrons are trapped with singlet ground state. Besides F centres other defects have been identified in CaO crystals, among them the Ff centre formed by a single electron trapped at the oxygen ion vacancy.' Neighbouring F+ centres can associate giving rise to aggregates such as dimers F; + (a nearest-neighbour oxygen divacancy containing two electrons) and trimers Fi + (a self-trapped hole, adjacent to an F;' centre).' The F+ centre is paramagnetic (S = 1/2) and Henderson and Wertz' reported that its continuous wave (CW) EPR spectrum consists of a single narrow line (linewidth <20 pT), being F+ free of hyperfine splitting because of the low (0.14%) natural abundance of the 43Ca nuclide with I = 7/2.We have observed that irradiation of the crystal by UV or visible light causes a marked enhancement of the EPR line intensity. Moreover, in some conditions, by using pulsed irra- diation, the signal intensity just after the light pulse corre- sponds to microwave emission. It has been shown that when CaO crystals are excited with light within the F band (F absorption band), photoionization of the F centres takes place, with F -+ Ff centre conversion. The most detailed study of photoionization at the F centre in CaO was reported by Welch et d..' They also showed that in CaO the F absorption band, which peaks at 400 nm, arises from the allowed transition from the 'A, ground state to the 'Tlu state, rapidly converted into the 9T,, state by inter- system crossing (ISC).At temperatures above 55 K electrons excited to the 'T,, and 'TI,, levels of the F centre are involved in a thermal ionization process which raises them to the conduction band. The activation energy for the excitation from the 'Tlu is 0.1 eV. Electrons excited into the conduction band are then trapped at shallow states below the bottom of the conduction band, from whence they are subsequently slowly released to form the ground-state F centre. The enhanced CW EPR signal intensity we observed under optical excitation could be due to the increased concentration of the doublet species F generated by irradiation.+ Recently some papers have appeared in the literat~re~.~ in which a transient change in intensity of the EPR spectrum of an S = 1/2 species under illumination was observed. This effect is explained in terms of a tripletaoublet (TD) inter- action. The latter causes a change in the relative spin sublevel population of the doublet species and consequently of the EPR signal intensity. Our observation on CaO might also be explained by this mechanism. The occurrence of an inter- action between triplet and doublet states in CaO crystals was t This paper was presented at the 27th International ESR Con-ference at the University of Wales, Cardiff, 21st-25th March, 1994. suggested by Glasbeek and Hond' in order to explain the temperature behaviour of the relaxation properties of the Fi+ defect in the excited triplet state in the presence of the F+ doublet species.In order to distinguish the possible causes of EPR signal intensity variation and to have further insight into the TD interaction, we have performed two types of time-resolved EPR experiments. In the first the excitation light was pulsed, which instantaneously generates the excited species and the time-evolution of the CW EPR spectrum was recorded.6 In the second, a pulsed EPR experiment was performed7 in the dark as well as during continuous illumination. The triplet- doublet perturbation switched on by the triplet species gener- ation, was expected to produce initial spin polarization.It was possible to study the spin dynamics of the F+ centres by comparing their spin relaxation properties with and without illumination. Experimenta1 Additively coloured crystals of CaO were purchased from Spicer Ltd. CaO is face-centred cubic and the unit cell (a = 4.80 A) contains four molecules of CaO.* The crystals were orientated by the X-ray technique and attached at the bottom of a Plexiglas rod which allowed us to rotate them around an axis perpendicular to the static magnetic field, B,. The rod was inserted into a quartz tube to avoid any exposure of the crystal to humidity. EPR measure- ments were performed with the magnetic field exploring the [OlO] plane. CW and pulsed EPR spectra were recorded by using an EPR Bruker ESP 380 X-band spectrometer with a dielectric resonator.The crystal was irradiated by a xenon lamp (Cermax 300 W). Time-resolved (TR) EPR measurements were carried out with a Bruker ER 200 D EPR X-band spec- trometer equipped with a standard TElo2 rectangular cavity and nitrogen flow cryostat. Pulsed light was supplied by an excimer XeCl laser (A = 308 nm, pulse duration = 20 ns) (Lambda Physik LPX 100). The EPR signal was directly recorded without field modulation and lock-in detection. The signal was collected and averaged by a digital oscilloscope (Nicolet 4094C, maximum resolution 5 ns per point). An averaged off-resonance signal was subtracted from that on-resonance to avoid noise coming from the laser trigger. Results The room-temperature CW EPR spectrum of the CaO crystal recorded in the dark is shown in Fig.1. The spectrum consists of the superposition of two lines with different line- widths and saturation properties. The measured linewidth Fig. 1 CW EPR spectrum of the F+ defects in additively coloured CaO crystals recorded in the dark at room temperature. Experimen- tal detection conditions are: mw power 0.54 pW,field modulation frequency 6.25 KHz, field modulation strength 0.6 pT. ABPpof the narrow component is 1.3 pT which is at the limit of field homogeneity of the EPR spectrometer, superimposed on a broader component having linewidth of 4.0 pT. This extremely narrow EPR line was recorded by using a low modulation frequency (6.5 kHz) in order to avoid line broadening due to sidebands of the modulation frequency.When the crystal is illuminated, the two components merge in a broader single line (ABpp= 6.5 pT) and, most inter- estingly, the signal intensity is enhanced considerably. After the light is turned off the intensity of the EPR line regains its previous value in times of the order of several tens of minutes (Fig. 2). Spin-lattice TI and spin-spin T2 relaxation times were measured from pulsed EPR experiments, performed in the dark as well as under continuous illumination. The decay time T; was obtained from the free-induction decay (FID) signal, which monitors the time-evolution of the y component of magnetization, M,(t), after a n/2 microwave pulse. TT is defined as = 1/T2h + 1/T2inh (1) where T2h and TZinh are the homogenous and inhomogeneous contributions.At room temperature the values of 7’;mea-sured in the dark and under continuous illumination are 4.8 f0.2 and 1.1 f0.1 ps, respectively (Fig. 3). The homogeneous contribution T2h has been measured from the decay of the spin-echo intensity in a Carr-Purcell- Gill-Meiboom sequence of microwave pulses J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 0 10 20 0 1 2 t/P frequency/M Hz Fig. 3 Experimental FID (left) and its fast Fourier transform (right) of F+ centres in CaO recorded at 290 K (a)in the dark and (b)under continuous illumination of the sample. The FID consists of a damped oscillation with damping parameter T:. The frequency of oscillation represents the offset of the magnetic field from the resonance. 7r/2-(-z-n-t-echo-), .To obtain the correct TZhvalue the suitable phase cycle was used in order to compensate for the incorrect length of the pulses. At room temperature in the dark the measured value of T2h is 30 k2 ps, which would correspond to a linewidth of only 0.19 pT in the CW spec-trum. No spin-echo was detected under continuous illumi- nation. The TI values were measured by the inversion recovery technique applied to the FID signal, consisting of the sequence of a n pulse followed after a delay time, t, by a n/2 pulse.’ After the first microwave pulse, magnetization of the spin system reverts to its initial value and evolves in time to recover its equilibrium value with the characteristic time TI.We measured the peak intensity of the spectra obtained by Fourier transform of the FID as a function of the delay, t, between the pulses (Fig. 4) The value of TI, measured in the dark at room temperature, is 48 & 4 ps. In the TR-EPR experiments’signals detected with the mag- netic field, B,, set in resonance with the EPR line, exhibit a marked dependence on the microwave (mw) power. At low 1 MHz 0 100 200 T time/min Fig. 4 Stacked plot of a series of FT EPR spectra of F+ centres in Fig. 2 Decay of CW EPR signal intensity of F+ centres in CaO, CaO recorded at different delays, T, in a n-z-n/2 pulse sequence. after illumination with visible light. Experimental points are fitted Typical pulse lengths are 16 ns (n/2 pulse) and 32 ns (n pulse).The with the function I = A/(B+ t)where A = 9 min, B = 11 min. real part of the transform, after a linear phase correction, is shown. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 ~ ~~~ emission 6 12 18 ti me/ms Fig. 5 TR-EPR signal of the F+ defects in the CaO crystal recorded at 250 K and 50 pW of mw power mw power, particularly, the signal, initially in emission, con- verts into absorption and eventually disappears with a single- exponential decay function as shown in Fig. 5. On increasing the mw power the emission reduces and the initial growth of the absorption becomes faster. At sufficiently high mw power the signal displays characteristic damped oscillationsY6 whose frequency depends on the strength of the oscillating mw field.A selection of experimental transient signals is shown in Fig. 6, together with computer simulations which will be discussed later. Discussion Measurements of T,and T,by Pulsed EPR In the dark the Tz value measured from the FID signal at room temperature is 4.8 ps corresponding to a linewidth AB,, = 1.2 pT which compares nicely with the value of 1.3 pT measured for one component of the CW spectrum. The dephasing-time, TZh,obtained from the decay of the spin- echo intensity by applying the Carr-Purcell-Gill-Meiboom sequence of microwave pulses is 30 ps. This time is extremely long and corresponds to a homogeneous linewidth of 0.19 pT. (4 20 mW 20 mW 2 mW 0.2mW 0.2 mWII7 0 18 ti me/ps Fig.6 (a) Experimental and (b) simulated TR-EPR signals of F+ defects in a CaO crystal recorded after the UV laser pulse at 290 K with different values of mw power. Fitting parameters are: TI = 4 p, T, = 0.8 ps. The above results are explained by allowing that in the dark the CaO crystals may host two kinds of F+ defects: (i) isolated F+ centres; (ii) weakly interacting F centres, giving + rise to the narrow and the broad components in the CW spectrum, respectively. The observed FID and spin-echo signals are assigned to the former species whose narrow com- ponent in the CW spectrum is considered inhomogeneously broadened by the inhomogeneity of the static magnetic field. The broad line arises from the interacting F+ defects with a homogeneous linewidth.Its FID signal decays rapidly, obscured by the slower decaying FID signal of the isolated species and no spin-echo signal is expected. When the light is turned on, new Ff species are generated from the F centres and the amount of interacting F+ defects increases at the expense of the isolated ones. The CW spec- trum then shows a single broad line, the FID signal decays with a shorter time TZ and no spin-echo is detected due to the homogeneous linewidth of the interacting species. All these features have been observed experimentally. EPR Experiments with Pulsed Light Two time regions could be clearly identified in the time- evolution of the EPR signal shown in Fig. 3. In the first, the EPR signal, initially in emission, converts into an absorption signal whose maximum intensity is achieved within a few hundred ps.After that time the variation is much slower and the signal decays exponentially with the characteristic time of 5.4 ms. This decay time is much longer than the spin-lattice relaxation time TI, as measured by the pulsed EPR experi- ments. At times longer than a few hundred ps any polariza- tion effect is over and the subsequent decay of the signal has to be ascribed merely to a process which depletes the species responsible for the EPR absorption. On the other hand the EPR absorption recorded after continuous illumination shows a much slower signal decay (see Fig. 2). The results of the two experiments indicate that there are two independent decay channels. The studies of luminescence and photoconductivity carried out on CaO crystals' have shown that the light absorption by F centres is followed by thermal photoionization from the excited ITlu and 3Tlu states, with the production of F+ centres and electrons in the conduction band.The photoion- ization process causes a decrease of the excited-state lumines- cence lifetime which drops from ca. 3 ms at 4.2 K to a fraction of an ms at temperatures higher than 70 K. The con- duction electrons are eventually trapped by relatively shallow impurity levels and are successively released to combine again with the F+ centres. The EPR signal decay should reflect the recombination process. The fact that the EPR signal decays with two very different characteristic times (of the order of lo3 and s) suggests the presence of two types of electron traps having different energy depths.Concerning the short time range, we note that the mecha- nism of photoionization does not contain any spin-selective process capable of generating spin polarization. Therefore, the contribution to the EPR signal due to this mechanism is initially zero. To account for the observed initial emissive polarization one is forced to look for a different process. Very recently, we reported the transient EPR emission spectra of a free radical trapped in a single crystal of chlora- nil.' When the crystal was illuminated inside the cavity of an EPR spectrometer, the intensity of the free radical spectral lines changed.After a laser pulse, the transient EPR signal was initially in emission. This behaviour was accounted for by assuming that triplet species, produced by the light excita- tion, interact with the stable radical species, according to the following kinetic scheme. T+ D The same model could explain the initial spin polarization observed in the transient EPR spectra of CaO F+ centres. Accordingly, the excited triplet states 3Tlu generated in the crystal, interact with Ff defects in the ground state giving rise to triplet-doublet pairs. If the spin-exchange interaction between the partners of the pair is non-zero the energy levels of the quartet and doublet spin states of the pair are split. The pair doublet state deactivates by conserving its spin multiplicity and giving rise to triplet quenching, while the quartet state is unreactive.However, before it dissociates back into a triplet and a doublet, it acquires a small admix- ture with the doublet components. The doublet-quartet mixing is caused by the zero-field splitting (zfs) interaction. Such a mixing is spin selective and if J < 0 (the doublet state lies lower in energy than the quartet), one finds that the rate constant of the process populating the 1/2 spin component of the excited doublet species D* emerg- ing from the doublet state of the pair is larger than that of the -1/2 component. The D* levels decay rapidly into those of the ground-state D species by a spin-allowed transition.Hence, the doublet species D participating in the pair forma- tion acquires emissive spin polarization. Another source of doublet-quartet mixing is the hyperfine interaction, which, however, is completely absent in CaO F+ centres. It should be noted that in our crystals the initial spin polarization does not involve all the F+ centres because those formed by photoionization are not polarized. Dependence ofthe TR-EPRSignal on mw Power Fig. 5 shows the EPR signal evolution in the short time range. The time dependence of the EPR signal is affected by the mw power. On increasing the mw power: (i) as expected, the signal displays transient nutations due to the spin dynamics;" (ii) the initial spin polarization diminishes until it disappears.Indicating by o1= yB,, the Rabi frequency proportional to the strength, B,, of the oscillating mw field, by following Atkins et al." one derives, at resonance, the following rela- tion for My@)which is proportional to the EPR signal where and = &[(T,' + Ti1) T [(T,' -Ti')' -4~:]~/'] (4) where T, and T, are the spin-lattice and the spin-spin relax-ation times, respectively, while M,, and M,(O) = 2, M,, are the magnetization along z at thermal equilibrium and just after the laser pulse. At high mw power eqn. (2) reduces to the form My(t)= (al/b)exp(at)sin(bt) (5) where the parameters a and b are given by: J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 At moderate mw powers the signal decays in a single-exponential manner with a time constant T;" given by l/T"," = l/Tl + a:T2 (7) The measured values of Ttff extrapolated to zero mw yield Tl-In order to simulate the whole series of EPR signals, recorded at different mw powers, one has also to take into account how the mw power affects the initial spin polariza- tion.This decreases as the mw power is increased. In order to account for this behaviour, it is worth noting that there should be a mechanism acting on the F+ centres before the laser pulse. In fact the initial polarization is generated in a time much shorter than the reciprocal of the Rabi frequency, o;l, which is the time the mw field needs in order to produce an appreciable change in the spin sublevel population. Moreover, the EPR line of the F+ centres is very narrow, as we have shown, and it can be easily saturated. Thus, one could infer that saturation of the spin levels of the F+ doub-lets existent before the laser pulse, is responsible for the reduced initial polarization.Let us consider the spin levels of the F+ centre, those of the T-D pair in the doublet state ('[T, D]) and the two spin sublevels 4[T, D]*33/2 of the doublet-triplet pair; the latter would be in a pure quartet state in the absence of doublet- quartet mixing. They are represented schematically in Fig. 7. When an F+ centre in the spin state -1/2 forms a pair with an F centre in the triplet excited state having -1 spin component, the [T, D] pair produced is in the 4D,D]-3/2 quartet state. Because of the mixing caused by the zfs inter- action within the F-centre triplet state, the quartet substate is mixed with the + 1/2 component of the pair doublet state as indicated by the double arrows in Fig.7. Therefore the pair could evolve, conserving the total spin, into an F+ centre in the excited doublet state (spin component +1/2) and a ground-state F centre. The successive deactivation of the excited F+ centre occurs at the + 1/2 sublevel. An analogous process takes place from the + 1/2 level to the -1/2 level. The overall process due to the doublet-triplet interaction can be envisaged as a population transfer from 'D+,/, to 'Ddl/, and vice versa. By indicating with k+-and k-+ the 8 0. D" 1/2 -112 Fig. 7 Energy levels of the F+ doublet state (either in the ground state or in the excited state, D*) and of the triplet-doublet interacting pair.Continuous straight arrows indicate the population transfer path from F-to F+ through interaction with a triplet-state F centre in the I -1) spin substate (T-J. Dashed arrows show the path from F+l,2+to F- through interaction with a triplet-state F centre in the I + 1) spin substate (TI).Double arrows symbolize the mixing between quartet and doublet states of the interacting pair (the larger the arrows the greater the mixing). J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 rate constants of the two processes, the net spin polarization, P, transferred is given by P = k+-n+l12-k-+ n-l12 where n,,/, are the populations of the corresponding spin levels.When the exchange interaction J is large and negative (in the pair the quartet has higher energy than the doublet) so that the only effective mixing occurs between 4[T, D]-3/2 and ’[T, D]+ 112, the rate constant k-+ % k+ -and a net emissive polarization is delivered to the F+ species inter- acting with the 3Tlutriplets. However, if J is fairly small, the doublet-quartet leads to the result k, -x k-+ . In these conditions the transfer of polarization, P,relies merely on the difference (n+1/2 -n-lj2) that can be depleted by saturating the EPR transition by a strong mw field, before the interaction with the triplet species. In order to compute the transient signals we assumed an initial polarization proportional to the saturation factor : The simulated signals, shown in Fig.6, fit rather satisfacto- rily the experimental ones, although the fitting was achieved by considering a single set of relaxation times, neglecting the presence of two species and, furthermore, the effect due to the chemical decay was not taken into account. The above observations indicate that in [T, D] pairs in CaO the exchange interaction is rather small. This is not sur- prising because the excited triplet species 3Tlu in CaO crys- tals are generated at the localized F sites lying rather far from the F+ defects, and they do not diffuse. Conclusions We have shown that the generation of triplet states by illumi- nation with UV-visible light in the additively coloured CaO crystal produces new excited F+ besides the stable F+ species in the ground state, as revealed by the enhanced intensity of the CW EPR spectrum.We have also shown that F centres excited in a triplet state give rise to interacting tripletdoublet pairs responsible for the initial spin polarization observed in the TR-EPR signal after the laser pulse. On increasing the mw power, this polar- ization is quenched. The saturation of the EPR transition, occurring between the levels of F’, is responsible for the quenching of the initial spin polarization. This work was supported by the Italian National Research Council (CNR) through the Centro Studi sugli Molecolari Radicalici ed Eccitati and by the Minister0 dell’universita e della Ricerca Scientifica e Tecnologica (MURST). References 1 B. Henderson and I. E. Wertz, Adv. Phys., 1968,17,749. 2 L. S. Welch, A. E. Hughes and G. P. Summers, J. Phys. C, 1980, 13, 5801. 3 C. Blattler, F. Jent and H. Paul, Chem. Phys. Lett., 1990, 166, 375. 4 A. Kawai, T. Okutsu and K. Obi, J. Phys. Chem., 1991,%, 9130. 5 M. Glasbeek and R. Hond, Phys. Rev. B, 1981,23,4220. 6 R. Furrer, F. Fujara, C. Lange, D. Stehlik, H. M. Vieth and W. Vollmann, Chem. Phys. Lett., 1980, 75, 332. 7 A. Schweiger, Angew. Chem. Znt. Ed. Engl., 1991,30,265. 8 C. W. Bunn, Chemical Crystallography, Oxford Univerisity Press, London, 1961. 9 C. Corvaja, L. Franco, L. Pasimeni, A. Toffoletti and L. Monta-nari, Chem. Phys. Lett., 1993,210,355. 10 (a)P. J. Hore and K. A. McLauchlan, J. Magn. Reson. 1979, 36, 129; (b)P. J. Hore and K. A. McLauchlan, Mof.Phys., 1981,42, 533. 11 P. W. Atkins, K. A. McLauchlan and W. Percival, Mof.Phys., 1973, 25,281. Paper 4/02242D; Received 14th April, 1994