Transient spectral hole burning observed on the single-molecule level in terrylene-doped biphenyl

Journal of Luminescence 152 (2014) 121–124

Contents lists available at ScienceDirect

Journal of Luminescence journal homepage: www.elsevier.com/locate/jlumin

Transient spectral hole burning observed on the single-molecule level in terrylene-doped biphenyl M. Pärs 1, V. Palm n, J. Kikas Institute of Physics, University of Tartu, Riia 142, 51014 Tartu, Estonia

art ic l e i nf o

a b s t r a c t

Article history: Received 30 July 2013 Received in revised form 28 October 2013 Accepted 7 November 2013 Available online 14 November 2013

We use the method of fluorescence correlation spectroscopy to analyze the single-molecule (SM) spectroscopy data earlier recorded for a special type of terrylene SM impurity center (referred as “spectrally confined unstable molecule”, SCM) in an incommensurate single crystal of biphenyl. The SCM's SM line seems to be chaotically jumping around within a broad “spectral envelope” and was first considered being subject to a peculiar spectral diffusion behavior. However, our correlation analysis reveals that all the features observed for SCM at 1.8 K are consistent with an assumption that this SM center participates in a process of reversible (transient) spectral hole burning (THB) earlier observed for terrylene-doped polycrystalline biphenyl. No observations of THB processes on SM level have been so far reported for this impurity system, partially due to a low concentration of relevant impurity centers. Another reason making searching for such centers experimentally challenging is an unusual SM line behavior: the photoinduced transition to a metastable “dark state” leads to the SM line saturational broadening, which is much stronger than the triplet broadening. Hence required prolonged observation is often prevented by an SM act of persistent spectral hole burning. & 2013 Elsevier B.V. All rights reserved.

Keywords: Biphenyl single crystal Incommensurate biphenyl Terrylene Transient spectral hole burning Single-molecule spectroscopy Saturational line broadening

1. Introduction About a decade ago, the first spectral hole burning and singlemolecule (SM) spectroscopy experiments with terrylene (TR) molecules embedded in a low-temperature biphenyl (BP) matrix have been reported by our group [1]. The chosen impurity system (TR:BP) is of a special interest due to the incommensurate modulation in the host crystal [2], which breaks the translational symmetry and causes the spatial variation on nanometer scale in local environments. This results in a very broad distribution of spectral and dynamical properties of individual (photochemically stable, but very sensitive to the local environment) TR impurity molecules (further referred as SM TR impurity centers). In particular, effective non-photochemical persistent hole burning (PHB) processes with strongly non-exponential kinetics (indicating a very large variety of photoinduced transition probabilities among the impurity centers), as well as processes of transient (reversible) hole burning (THB) with hole lifetimes on the order of a second, have been observed for polycrystalline TR:BP samples at temperatures T below 2.2 K across the whole
10 nm broad inhomogeneous band of the purely electronic TR transition [1]. n

Corresponding author. Tel.: þ 372 7374664; fax: þ 372 7383033. E-mail address: viktor@fi.tartu.ee (V. Palm). 1 Present address: Experimental Physics IV, University of Bayreuth, 95440 Bayreuth, Germany. 0022-2313/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jlumin.2013.11.033

SM spectroscopy experiments on single-crystalline low-concentration TR:BP samples have revealed the broad variety of SM dynamical behavior, including irreversible spectral jumps of SM lines of the centers involved in PHB processes [3]. Although in certain conditions a reliable identification of an SM center as one being involved in a THB process should be possible [4–8], obtaining such a result for TR:BP system has not yet been reported, remaining an experimental challenge. Among several observed behavioral types of SM TR impurity centers, a specific type called as “spectrally confined unstable molecule” (SCM) has been described in Ref. [3], referring to an SM line that seemingly chaotically jumps around within a broad “spectral envelope” and was initially considered being subject to a peculiar spectral diffusion (SD) behavior. Among several potential SCM-type SM centers we observed for TR:BP, only one (further referred as SM1) was not involved in any PHB process for several hours, which allowed performing different types of measurements in order to collect data necessary to analyze its presumed SD dynamics. In addition to fluorescence excitation spectra obtained for SM1 with various scanning rates at temperatures between 1.8 and 2.1 K, the fluorescence intensity has been recorded within extended time periods with various time resolutions for several laser excitation frequency positions within the nearly 2.5 GHz broad “spectral envelope”. Here we present results of the correlation analysis of the data recorded for SM1 at T ¼1.8 K with time resolution of 13.5 ms,

122

M. Pärs et al. / Journal of Luminescence 152 (2014) 121–124

demonstrating that this TR SM impurity center is subject to a THB process rather than to a (spontaneous) SD process as it was first considered. Other data recorded for SM1 using different time resolutions and temperatures do not contradict this conclusion and will be presented elsewhere.

2. Experimental Few-micrometers-thick single-crystalline BP flakes doped with TR molecules at estimated concentrations of 10  10 mol/mol have been used in our SM experiments. Sample preparation procedures [3] and the experimental setup [1,9] are in more detail described elsewhere. A single-frequency ring dye laser CR-699-29 Autoscan used as a source of narrow (jitter-determined linewidth about 2 MHz) frequency-tunable excitation, had an absolute frequency positioning accuracy of 60 MHz, but the relative spectral precision within a single scan was limited only by a slow laser frequency drift, which is specified not to exceed 100 MHz/h. The lengths of the signal-registration time periods were limited by the used home-made software, which allowed continuous recording of up to 11,000 fluorescence intensity data points with time resolution up to 5.1 ms. The excitation laser intensity was stabilized at a certain level Ie (estimated as
100 mW/cm2 on the sample), which at T¼ 1.8 K did not cause any noticeable TR triplet saturation [10]. To calculate the autocorrelation function (ACF) curves, a special algorithm was created using a MathCad package.

It should be mentioned that the SM spectral profiles we obtained for SM1 are quite similar to those obtained in Ref. 11 for the case of a strongly excited triplet-saturated TR SM in a pterphenyl crystal. The spikes in SM spectral profiles are interpreted there as quantum jumps due to the intramolecular intersystem crossing (ISC) transition. In our case of a weaker excitation, similar behavior could be caused by a different process driven by an SM– matrix interaction. According to our new THB hypothesis, the “spectral envelope” actually represents an unusually broad SM line of an SCM center, which is subject to a THB process—a photoinduced transition to a non-radiative metastable “dark state” (DS). To evaluate this assumption, we performed the fluorescence correlation analysis of the data recorded for several fixed excitation frequencies as indicated in Fig. 1. Short fragments demonstrating the temporal behavior of the fluorescence signal recorded with 13.5 ms time resolution are shown in Fig. 2. It is clearly seen that we have a stationary process, which can be investigated using the technique of intensity correlation analysis. To determine the rates of assumed quantum jumps the second order autocorrelation functions (ACF) of the fluorescence intensity I(t) has been calculated according to the following equation [12]: g ð2Þ ðτÞ ¼

〈IðtÞIðt þ τÞ〉 ; 〈IðtÞ〉2

where the average was taken over several thousands of time points t. The calculated ACFs for the three fixed excitation frequencies indicated in Fig. 1 are shown in Fig. 3. It appears that all the calculated ACFs can be well fitted by single-exponential curves according to the formula: g ð2Þ ðτÞ ¼ a þb  expð  λ  τÞ;

3. Results and discussion Considering the spontaneous SD hypothesis, a relatively narrow SCM-type SM line quickly jumps between different spectral positions within a broader “spectral envelope”, so during a slower excitation frequency scan over this region the fluorescence emission appears at random spectral positions. An example of such an excitation spectrum obtained for SM1 using our slowest available laser scanning rate is shown in Fig. 1. Surprisingly, no symmetrical Lorentzian lines can be found among the spikes filling the contour, even for the fastest scanning rates. Another feature hard to explain with the SD interpretation is that the average amplitudes and widths of the spikes clearly depend on the position within the spectral contour. Consequently, a better interpretation is needed.

g ð2Þ ðτÞ ¼ 1 þC expð λ τÞ;

Fluorescence intensity, x103 Counts/s

(c)

(a)

ð4Þ

We should actually treat our hypothetical photoinduced DS transition as a parallel decay channel to the ISC transition, thus having now a four-level system. But taking into account that our

T=1.8 K 14

ð3Þ

where the decay rate λ is related to the population (k23) and depopulation (k31) rates of the triplet state, and C ¼ ðλ  k31 Þ=k31 , thus the triplet state lifetime is found as 1

(b)

ð2Þ

where a, b, and λ are the constants used as fitting parameters. Thus we can consider the three-level system approach developed for calculations of SM ISC decay parameters when the splitting of the triplet level can be neglected [13,14]. According to this approach, for sufficiently long time scales the analytical shape of the fluorescence intensity ACF can be expressed as a monoexponential curve:

t 31 ¼ k31 ¼ ðC þ1Þ=λ 15

ð1Þ

13 12

(d) 11 10 9 8 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Laser detuning, GHz Fig. 1. An example of SM excitation spectrum obtained for SM1 with a laser scanning rate of 10 MHz/s, 10 MHz resolution (1 s time bin). Arrows indicate fixed excitation laser frequencies where the fluorescence intensity temporal evolution has been recorded with 13.5 ms resolution: (a) þ800 MHz, (b) þ 1636 MHz, (c) þ 2054 MHz, (d) þ3309 MHz. Zero laser detuning corresponds to a wavenumber of 17,274.4 cm  1 (wavelength
578.89 nm).

Fig. 2. Short fragments of the fluorescence signal temporal evolution recorded with a time bin of 13.5 ms. Excitation frequencies for intensity traces (a), (b), and (c) correspond to the spectral positions shown in Fig. 1.

10

Fluorescence, x 103 Counts/s

M. Pärs et al. / Journal of Luminescence 152 (2014) 121–124

(b)

g(2)(τ)

8 6 4

(a) (c)

2 0 10

100

τ, ms

1000

40

T=1.8 K

30

FWHM: 0.48 GHz

20 FWHM: 2.8 GHz

10 0.0

10000

123

0.5

1.0 1.5 2.0 2.5 Laser detuning, GHz

3.0

Fig. 3. Calculated ACF values (data points) of the fluorescence signal emitted by SM1 center under laser excitation at three fixed frequencies shown in Fig. 1, recorded with 13.5 ms time resolution (see Fig. 2). In all three cases the ACF data can be well fitted with single-exponential decay curves (shown as solid lines); the obtained values of the decay rate parameter are: (a) λ¼ 9.43(56) s  1, (b) λ ¼ 32.8 (14) s  1, (c) λ ¼8.01(46) s  1. We consider the weak modulation features appearing in the ACF curves at longer delay times (τ 4150 ms) as likely artifacts of our experimental technique: the “noise eater” with negative feedback that was used for the laser intensity stabilization [1] could cause a weak periodic modulation of the excitation intensity.

Fig. 4. Visualized explanation of the SM1 saturational broadening. The black solid line (linewidth 2.8(3) GHz) is the Lorentzian fit to data points (solid squares) obtained by time-averaging of the fluorescence signal recorded for different laser excitation frequencies. The dotted line is the Lorentzian fit to the average values of the higher signal level (“bright state”) intensities (solid triangles) obtained from the signal recorded with 13.5 ms time resolution (see Fig. 2). The low level (DS) intensities are shown as empty circles—their average is shown as a dashed line, forming the common baseline of the both Lorentzians, which also have the same central frequency. (Zero laser detuning corresponds to a wavenumber of 17,274.4 cm  1).

excitation intensity is far from saturating the triplet transition, that our registration time bin is much longer compared to the triplet lifetime (for TR molecules t31 is less than 1 ms [15]), and assuming that the DS lifetime t41 is much longer than t31, the ISC transitions can be ignored and we again end up with a three-level kinetic scheme with DS instead of the triplet state. According to Eqs. (2)– (4), the DS lifetime can now be calculated from our ACF singleexponential fitting results:

Obtained linewidth of 480 MHz is quite a common value for TR:BP at T¼1.8 K [10]. An SM line broadening of 5–6 times indicates that the applied excitation intensity exceeds the saturation intensity by about 30 times [16]. Taking into account that the same excitation intensity is lower than the saturation intensity for ISC transition, one can conclude that due to the much longer DS lifetime the DS bottleneck of our SM1 center is much “narrower” than the triplet bottleneck. Finally, it has to be emphasized that, being in good accordance with our experimental data, the introduced model of photoinduced transitions to a metastable state DS does not reveal the physical nature of this long-lived state. Due to the photostability of TR molecules we have to assume a matrix-assisted (non-photochemical) mechanism of THB, e.g. matrix-assisted shifts of the resonance frequency of our SM center. Such “spectral jumps” are usually explained in terms of the two-level system (TLS) model [17,18], which is widely used to describe anomalous properties of lowtemperature amorphous solids, while SM spectroscopy experiments have revealed TLS signatures in low-temperature crystals as well. In our experiments we were not able to detect the second position of the SM1 line within our scanning range, which indicates that the splitting should be larger than 4 GHz (see Fig. 1). Within the frame of the TLS model this would correspond to a large coupling constant or relatively short distance between our SM and a neighboring TLS. One should be cautious with too direct application of this model, which was elaborated for glasses, to our incommensurate system. Our results, however, firmly demonstrate the existence of some kind of structural instability in the host matrix. It should be mentioned that according to our analysis [19], the line broadening of TR SMs in a low-temperature incommensurate BP matrix cannot be well described by the same TLS model, which works quite well for several amorphous solids. The reported [10] remarkable SM line broadening in BP at 1.8 K is at least partially caused by other dephasing processes acting in this crystal, which are tentatively related to the specific dynamics of incommensurate systems. The THB-related saturational broadening can be one of such processes. Although the ubiquitous PHB processes make it very hard to find a suitable for prolonged examinations SCM-type SM center, our results [1] demonstrate that THB is quite a common process for the TR:BP system, in which up to 5% of TR impurity centers may participate.

1

t 41 ¼ k41 ¼ ðb þ aÞ=aλ

ð5Þ

It can be seen from Fig. 3 that the ACF decay rate parameter λ is higher when the excitation frequency is closer to the maximum of the spectral contour (curve (b)) and significantly decreases on the wings (curves (a) and (c)). This observation supports the photoinduced mechanism as the main cause of the observed fluorescence fluctuations. On the other hand, calculating according to Eq. (5) the corresponding DS lifetimes demonstrates that within the fitting accuracy they are clearly independent of the excitation frequency, having the mean value of t41 ¼ 0.64(10) s. This indicates that, in accordance with our THB model, transitions from DS back to the “bright state” are spontaneous and do not depend on the absorption cross section. By analogy with the case of ISC transition [11] and triplet bottleneck, one can assume that observation of abundant photoinduced quantum jumps in a single-scan SM excitation spectrum readily indicates that this spectrum is significantly saturationally broadened. A smooth broadened spectral line should be observed in a case of sufficiently slow scan, when the time bin of signal accumulation is much longer compared to the DS lifetime. In the case of our SM1 center we were not able to produce such slow scans with our laser software. In order to assess the saturational broadening we constructed such a smoothed broadened spectrum of just 11 spectral points, by time-averaging of the fluorescence signal recorded for different laser excitation frequencies (solid squares in Fig. 4). These points are well fitted by a Lorentzian profile with a 2.8(3) GHz linewidth (solid line in Fig. 4). For comparison, we also estimated the non-saturated fluorescence emission rate by taking the mean intensity of the bright intervals from the fluorescence time traces recorded for several laser frequencies (solid triangles in Fig. 4). An approximate contour of the non-broadened SM line (shown as dotted line in Fig. 4) was obtained by fitting Lorentzian to these data points.

124

M. Pärs et al. / Journal of Luminescence 152 (2014) 121–124

4. Conclusions

References

An instance (referred as SM1) of SCM-type terrylene (TR) singlemolecule (SM) impurity center in incommensurate single crystal of biphenyl (BP) has been observed and investigated with SM spectroscopy technique during an unusually long period of several hours. Our correlation analysis of the data recorded at T¼ 1.8 K reveals that SM1 undergoes photoinduced transitions to a non-radiative metastable “dark state” (DS) of unknown origin. Due to the long DS lifetime of
0.64 s, a very strong saturational broadening (over fivefold for our experimental conditions) of the SM1 spectral line occurs, indicating a possible mechanism contributing to the anomalous SM line broadening for TR:BP [10,19]. All the features observed for SM1 at 1.8 K are consistent with an assumption that the SCMtype centers participate in a process of transient hole burning (THB) earlier observed for TR-doped polycrystalline BP [1].

[1] V. Palm, M. Pärs, J. Kikas, J. Lumin. 107 (2004) 57. [2] P. Launois, F. Moussa, M.H. Lemee-Cailleau, H. Cailleau, Phys. Rev. B: Condens. Matter 40 (1989) 5042. [3] V. Palm, N. Palm, M. Pärs, J. Kikas, Proc. SPIE 6029 2006, 60291N. [4] W.E. Moerner, T. Plakhotnik, T. Irngartinger, M. Croci, V. Palm, U.P. Wild, J. Phys. Chem. 98 (1994) 7382. [5] Th. Basché, W.E. Moerner, Nature 355 (1992) 335. [6] Th. Basché, W.P. Ambrose, W.E. Moerner, J. Opt. Soc. Am. B: Opt. Phys. 9 (1992) 829. [7] S. Kummer, Th. Basché, C. Bräuchle, Chem. Phys. Lett. 229 (1994) 309. [8] L. Kador, A. Müller, Mol. Cryst. Liq. Cryst. 314 (1998) 149. [9] V. Palm, Rev. Sci. Instrum. 70 (1999) 2957. [10] V. Palm, M. Pärs, J. Kikas, M. Nilsson, S. Kröll, J. Lumin. 127 (2007) 218. [11] M. Vogel, A. Gruber, J. Wrachtrup, C. von Borczyskowski, J. Phys. Chem. 99 (1995) 14915. [12] R. Loudon, The Quantum Theory of Light, Oxford University Press, New York, 2000. [13] J. Bernard, L. Fleury, H. Talon, M. Orrit, J. Chem. Phys. 98 (1993) 850. [14] R Brown, J. Wrachtrup, M. Orrit, J. Bernard, C. von Borczyskowski, J. Chem. Phys. 100 (1994) 7182. [15] A.C.J. Brouwer, E.J.J. Groenen, J. Schmidt, Phys. Rev. Lett. 80 (1998) 3944. [16] T. Basche, W.E. Moerner, M. Orrit, U.P. Wild (Eds.), Imaging and Spectroscopy, Weinheim, New York, 1997. [17] W.A. Phillips, J. Low Temp. Phys. 7 (1972) 351. [18] P.W. Anderson, B.I. Halperin, C.M. Varma, Philos. Mag. 25 (1972) 1. [19] M. Pärs, V. Palm, J. Kikas, Low Temp. Phys. 36 (2010) 448.

Acknowledgements Authors are grateful to N. Palm for preparation of terrylenedoped biphenyl single crystals. Acknowledged is support from Estonian Science Foundation Grant 8167 and EU project TK114.