(a) Photoluminescence I PL / I PL 0 at 458 nm vs. t of 190 nm thick DPVBi fi lms 

Contexts in source publication

Context 1

... spectra of oxidized DPVBi, calculated at B3LYP/6-311 þ þ G** level, for the lowest energy CFs found. There are three important conclusions to be drawn from the comparison: (a) numerous lines found by DFT fall in the region of 1250 cm À 1 feature, many of them belonging to C – O bond, while only few of them belong to pure DPVBi molecule; (b) the feature about 1700 cm À 1 consists of two nearby broad peaks and DFT results suggest that these cannot belong to a molecular pristine DPVBi since there are no vibrations in its spectra between 1600 cm À 1 and 2900 cm À 1 (see Fig. 3a). In contradistinction, molecule with two O attached to a phenyl ring has an in-plane stretching of the C 1⁄4 O vibration at 1690 cm À 1 (denoted M 1 0,2,0 in Fig. 3b). The lowest energy M 0,2,0 found has an in-plane oscillation of the C 1⁄4 O bond at 1730 cm À 1 , in a very good agreement with the frequency of the second peak of the feature (both calculated at B3LYP/6-311 þ þ G** level); and, (c) DFT results for IR spectra of the molecular DPVBi with one or two hydroxyl groups ( M 0,0,1 and M 0,0,2 ) give strong peaks around 3590 cm À 1 corresponding to an – OH stretching mode. Since there are no strong peaks above 3100 cm À 1 in the measurement, we conclude that the noncovalent interactions between new species and surrounding molecules strongly damp and broaden vibrations of – OH group. From the MS and IR spectra one sees that new species are photo-oxidized DPVBi molecules or its fragments with one, two and even three oxygen atoms added, which can be singly or doubly bound to C atoms of DPVBi. Fig. 4a shows absorption spectra as a function of wavelength λ after different times of irradiation in air of 190 nm thick DPVBi fi lm deposited onto fused silica substrate and irradiated with I UV 1⁄4 0.4 mW cm À 2 . Pristine fi lm (no irradiation) shows two absorption bands at 242 and 355 nm, corresponding to two lowest excited electronic states. The energy gap is (3.0 7 0.1) eV, esti- mated from the long-wavelength absorption edge [30], in a good agreement with our TD-B3LYP/6-311 þ þ G** result for the fi rst excitation of DPVBi in the lowest vibrational state being at 3.14 eV. Irradiation with UV light induces gradual disappearance of both bands and emergence of new peak around 255 nm, re fl ecting a gradual chemical change in fi lm composition (degradation of DPVBi) and a formation of photo-oxidized products of DPVBi. The percent of change in the value of absorbance A at λ 1⁄4 355 nm is assumed to be roughly the same as the percent of impurities present in a fi lm. To track this change, A taken at 355 nm, normalized to its initial value A 0 at time t 1⁄4 0, is plotted in the inset of Fig. 4a as a function of irradiation time t . The decrease in normalized absorbance A / A 0 is close to linear at the beginning and then gradually slows down. Nonlinear part of the curve can be a consequence of change in degradation process dynamics due to a signi fi cant loss of DPVBi material (around 60%). From the linear part the rate R of change of normalized absorbance A / A 0 is de fi ned by A / A 0 1⁄4 1 À Rt /100 (the inset of Fig. 4a and for the fi lm from Fig. 4 it is around 0.22%/min. Effects which degradation in air has on the photoluminescence (PL) are shown in Fig. 4b as a series of spectra of a 190 nm thick fi lm irradiated with UV intensity I UV 1⁄4 0.4 mW cm À 2 as function of irradiation time t . Irradiation induces no new bands in PL spectrum. Inset in Fig. 4b shows that PL intensity I PL at λ 1⁄4 458 nm, normalized to its value I PL 0 at t 1⁄4 0, has a quick, exponential decay with t . Time necessary for PL intensity to drop to half of its initial value – half lifetime t 1/2 is 45 s. During that time, according to the value of R , absorbance changed only for a little less than 0.2%, pointing to some non-trivial mechanism of PL quenching. The in fl uence of different UV light intensities on the sample degradation in air was also studied: the rate R and time evolution of I PL / I PL 0 at 458 nm were measured as function of I UV for 190 nm thick fi lm. Photoluminescence I PL has exponential decay which is faster for larger I UV (Fig. 3a), while the rate R is proportional to I UV . In the inset of Fig. 3a, t 1/2 was plotted versus I UV / I UV max , along with the fi t to the relation t 1/2 1⁄4 CI UV À 1/ n used for OLEDs [7]; C is a constant and n is so-called acceleration parameter, whose value is (1.08 7 0.06). This means that t 1/2 is, to a good approximation, inversely proportional to I UV . The product of R and t 1/2 , which is the percentage of changed absorbance or DPVBi molecules that leads to 50% decay of PL, should be then approximately indepen- dent of I UV , as R $ I UV . This is shown in Fig. 3b where the product Rt 1/2 stays approximately constant and less than 0.2% when I UV varies. Eventual dependence of Rt 1/2 on I UV would indicate that dynamics of formation of impurities is not the same for different UV light intensities Fig. 5. Photoluminescence spectra of DPVBi fi lms in high vacuum (5 Â 10 À 4 Pa) and nitrogen atmosphere (Fig. 6a) show no change even for irradiation times of 15 min with I UV max . Under the same conditions, the rate R of change of absorbance is zero (Fig. 6b). Thus, under the conditions and on the time scale of our experiments, no change in fi lm composition can be induced by UV light in vacuum . Likewise, PL does not change when fi lms are exposed only to air: PL of DPVBi fi lm was fi rst measured in situ in vacuum (brie fl y exposed only to UV light ), then the fi lm was exposed only to air (no UV light) so that oxygen enters the fi lm [12,13] and, fi nally, brought back to low vacuum (few Pa), where PL was recorded once again (the two measured values of PL were the same). An impli- cation, for practical purposes, would be that even the low vacuum is suf fi cient for extraction of oxygen from amorphous DPVBi fi lms of thickness of order of 200 nm. Additionally, both rates of change of absorbance and PL increase with the increase of oxygen pressure as shown in Fig. 6. These considerations imply that the simultaneous presence of oxygen and UV light is necessary for degradation of DPVBi fi lms and that the excited DPVBi molecules interact with oxygen to produce impurities. In general, there are two possible pathways for chemical reaction of DPVBi and oxygen, both typical for such compounds [8,31]. In one of them excited singlet molecule gives an electron to ground-state oxygen molecule to form radical cation and the superoxide anion, which can further react chemically to form new species [31]. In the other, host molecule in excited triplet state acts as a sensitizer and transfers its energy to ground-state triplet oxygen to form singlet oxygen and ground-state host molecule [31]. The energy needed for singlet oxygen formation is 0.97 eV. The fi rst excited triplet state of DPVBi, according to our TD-B3LYP/ 6-311 þ þ G** calculation, is 2.19 eV, which is in a good agreement with experimental value of 2 eV obtained for spiro-DPVBi [32]. This energy is suf fi cient for singlet oxygen formation in our fi lms. Singlet oxygen is very reactive and may further interact with other DPVBi molecules to form photo-oxidized species. We fi nd by DFT that DPVBi can form a bound state with singlet oxygen, with the binding energy of 0.17 eV, but could not fi nd any bound state in the triplet case. Oxygen is a well-known PL quencher [12,13]. However, we fi nd no evidence for direct collisional quenching with oxygen. If this type of quenching was present, PL intensity would be (partially) reversible after removal of oxygen from the fi lm. In the following experiment it was demonstrated that PL is not reversible (Fig. 7): fi rst a fi lm was UV-degraded in air for about t 1/2 , then the air was evacuated to the pressure of few Pa in a few seconds and the fi lm kept in vacuum for 50 min in dark. After 50 min several measurements of PL were carried out, which show that its value remained the same as the one at the end of the degradation in air. Thus, PL quenching is only due to new photo-oxidized species. At the level of 0.2% of impurities (that quench 50% of PL), the average distance between impurity molecules d i is around 7 nm. This value is obtained taking the density of DPVBi to be 1.2 g cm À 3 (that of monoclinic crystal with two molecules per unit cell) [33] and assuming that molecules form a simple cubic lattice. The long- range Förster resonant energy transfer (FRET) from DPVBi to impurity molecule can be ruled out due to absence of a spectral overlap between DPVBi emission centered at 458 nm and the impurity absorption at 255 nm [15,34 – 36]. Thus, only reasonable assumption on the mechanism of PL quenching is that the excitons diffuse through the fi lm at long distance and, if during their lifetime an impurity molecule is reached, quenching as the most probable outcome may happen [15,37 – 41]. Exciton diffusion length l D for amorphous DPVBi fi lms is measured by Choukri [42] to be (8.7 7 0.6) nm and satis fi es the condition d i o l D required for quenching [43]. The Förster energy transfer among DPVBi molecules occurs with the probability higher than 50% when their separation is smaller than 1.4 nm. The Förster self-radius R 0 1⁄4 1.4 nm for DPVBi is calculated from the formula R 0 1⁄4 0.211( κ 2 n À 4 Q D J ( λ )) 1/6 [44], where κ 2 1⁄4 2/3 is the geometrical factor for random orientation of the donor and acceptor transition dipole moments, Q D 1⁄4 0.45 is PL quantum ef fi ciency of amorphous DPVBi [45] and J ( λ ) 1⁄4 4 Â 10 12 (nm) 4 M À 1 cm À 1 is the overlap integral between PL and absorption spectra of DPVBi. As, to the best of our knowledge, experimental data for the refractive index n of DPVBi are lacking, we have, instead, taken the value n 1⁄4 2 of structurally similar TPD molecule [46]. Excitations are localized and move by hopping from one DPVBi molecule to another via one of two possible ...

Context 2

... to 50% decay of PL, should be then approximately indepen- dent of I UV , as R $ I UV . This is shown in Fig. 3b where the product Rt 1/2 stays approximately constant and less than 0.2% when I UV varies. Eventual dependence of Rt 1/2 on I UV would indicate that dynamics of formation of impurities is not the same for different UV light intensities Fig. 5. Photoluminescence spectra of DPVBi films in high vacuum (5 Â 10 À 4 Pa) and nitrogen atmosphere (Fig. 6a) show no change even for irradiation times of 15 min with I UV max . Under the same conditions, the rate R of change of absorbance is zero (Fig. 6b). Thus, under the conditions and on the time scale of our experi- ments, no change ...