- Email: [email protected]
A study of the viscosity-dependent electronic relaxation of some triphenylmethane dyes using picosecond flash photolysis
Volume 71, number
CHEMICAL
1
A STUDY OF THE VISCOSITY-DEPENDENT OF SOME TRIPHENYLMETHANE David A. CREMERS Department
PHYSICS
L April 1980
LETTERS
ELECTRONIC
RELAXATION
DYES USING PICOSECOND FLASH PHOTOLYSIS
and Maurice W. WINDSOR
of Cirremlsny, Washmgton State Umwrs~ty, Pullman, Waslrmgtorr 99164. USA
Recewed 29 October
1979
Tune-resolved absorption measurements of ground-state recovery and euclted-state tnphenylmethane dyes m solutions of different viscosity. A kmetlc model us proposed earher measurements
absorption are reported for several to explain the results of these and
1. Introduction The viscosity-dependent relaxation of photoexcited tnphenyimethane (TPM) dyes (fig. 1) has been studled in several ways. Fluorescence quantum yield 0 measurements [l-3] and direct kmetic measurements using picosecond spectroscopic techmques [S-S] indicate that increased solvent vlscoslty results m decreased rates of electroruc relaxation. Forster and Hoffmann (FH) [ I] proposed a model of this viscosity dependence which predicts the Q = $I3 relationshp found m some measurements of Q [1,3,8 J According to tis model, the photoexcited molecule undergoes a change m conformation to a new structure characterized by enhanced rates of non-radiative decay. The importanr conformatIonal change is assumed to be synchronous rotation of the phenyl rings about the bond between the ring and central carbon atoms (see fig. 1). The rotation IS driven by steric repulsion between adjacent ring orthohydrogen atoms and is hindered by viscous drag mtroduced by the solvent molecules. Increased VISCOSlty slows the rate of ring rotatron and lengthens the tune interval between excitation and the achevement of those conformations that give rise to fast non-radiative decay. In solvents of low viscosity, these conforrnatlons are reached more rapidly and fluorescence is quenched by the faster radiationless paths of relaxation. Although the FH model predicts the observed Q 0: ~$3 relationship, the time dependence of excitedstate relaxation of exp(-at3) also predicted by Uus
-I-
CI-
‘R Rg. 1. The triphenyhnethane dyes crystal violet CV (R =N(CH&. R’ = HI, ethyl violet EV [R = N
model has not been observed. Magde and Windsor [4] measured the recovery of ground-state absorption of crystal violet (CV) following picosecond excitation. The data did not distinguish between several possible functional descriptions of the recovery. They concluded, however, that internal conversion is the dominant non-radiative mechanism . Ippen et al. [6] measured ground-state recovery (GSR) of malacbte green and found it to be a single exponential for TJ< 1 poise. For 17> 1 poise the recovery was described as a double exponential. Yu et al. [7] monitored malachite green fluorescence and found it to decay as a single exponential. Hirsch and Mahr [8] aIsct 27
Volume 7 1, number 1
CHEMICAL
PHYSICS
studied malachite green emlsslon and characterized it as a double-exponentlal decay. The hfetlmes of the faster decays of Hirsch and Mahr agree with the hfetimes of Yu et al. and the slower component of CSR measured by Ippen et al. Despite the several studies referred to above, a description of the viscosity dependence has not yet been proposed which reconcdes all the kmetlc and quantum yield measurements The role of vlscoslty in the p~cosecond relaxation kmetlcs of TPM dyes has been of mterest to this laboratory for several years [4,5]. We present here a more detaded mvestlgatlon of these VIScosity effects.
2. Experimental The dyes used m tlus study are shown m fig. lCrystal violet (Baker Chemical) used m the majority of experiments was purltied via the leuco base Ethyl violet (Eastman) and parafuchsin (Tndom Fluka) were punfied by column chromatography smce smaller amounts of these dyes were needed. The tnmethyl derivative of CV was prepared and punfied accordmg to the hterature [9]. All dyes were m the form of the chlonde salt. Solvents of the desired viscosity were prepared by mcung deionized water, reagent-grade glycerol and glucose in various proportlons The \IScosltles of all dye solutions were directly measured. Unless otherwlse specified, all measurements were made on 6 X IO-5 molar solutions at 25°C. The picosecond flash photolysis apparatus used m this study 1s described elsewhere [lo] It IS based upon a single-pulse-selected. mode-locked Nd - glass laser. A smgle amphfied picosecond pulse at the laser fundamental (I 060 nm) is S-IO ps m duration and has an energy of about 15 mJ. Frequency doubhng produces a 530 nm excltatlon (pump) pulse of about 2 mJ. The remammg 13 mJ of 1060 nm radlatlon IS focused mto a cell of CC14 to generate a spectrally broad (400-900 run) picosecond probe pulse of weak mtenslty The probe light transmitted by the samp!e is analyzed by a 2 m Jarrell-Ash spectrometer and detected on an optical multvAmnel analyzer (Prmceton Apphed Research). The geometry of the pump and probe pulses at the sample [IO] permits measurements of the difference in optical density (JOD) between excited and unexcited regions of the sample. Each laser finng produces
28
1 April 1980
LETl-ERS
a smgle measurement of AOD at a specific wavelength and delay. The excitation and probmg laser pulses were imaged onto a small volume of the dye solution contamed wlthm a 2 mm path length spectrometer cell Following each laser firing, the contents of the cell were replaced by a fresh dye solution. The temperature of the sample regon was carefully controlled to prevent fluctuations in vlscoslty due to thermal effects ms was especially critical for the more viscous solutions The hygroscoplc nature of glycerol made It necessary to sheld the dye solutions from the atmosphere. The vlscoslty of the glycerol solutions 1s very sensitive to water contamination. The mtenslties of the pump and probe pulses vary Lvlth each laser firing. Variations m probe Intensity do not affect a measurement of AOD. Fluctuations in excitation power occur, and do cause changes m AOD, buth these are mmumzed by using sufficient power to saturate the sample The effect of saturatmg the CV sample IS shown m fig. 2. The changes in AOD due to vanatlons in excltatlon power are small when 100% of the power IS used Each data point of a spectrum or hnetlc curve m this study represents the average of usually six separate AOD measurements The total length of the precision bars equals twice the standard devlatlon of the AOD values used to compute the arerage. Experiments m which the relative polarizations of the excltatlon and probe pulses were adjusted either to accent or to mask the effects of molecular reonentatlon ylelded smular results. Thus our measurements were apparently not affected by onentatlonal relauatlon of the entire molecule following photoexcltatlon.
06
04
-
-
02
-0
00
d 1 0
a’ r
I 20
% Fig. 2. cerol).
P
6-B -A’ P
I 40
PUMP
I 60
1 80
I 100
POWER
Saturation curve for crystal wolet (6 X 10m5 M in gly-
Volume 7 1, number 1
CHEMICAL
PHYSICS LETTERS
Undoubtedly thrs is a result of the high solvent viscosrtres we have used which cause overall rotational motions to be very much slower than the processes we have studred.
3. Results Optical density drfference spectra of CV over the range 380-900 nm at time delays of +20, +80 and +2730 ps are shown in fig. 3. To obtain data in the 380-450 run region the picosecond apparatus was modified to permit generation of the probe pulse by the 530 run excitation pulse The continuum produced in this way has significantly greater intensity in the bIue regron than the 1060 MI generated contmuum. The effects of dispersion in the arrival time of the probe pulse at the sample have not been removed from these spectra Calculation and experience rndrcate that, over the spectral regron considered here, the drfference in amval tunes between the red and blue probe light is, at most, 20 ps. Such dispersion has no effect on the kinetrc data since these are taken wrthm a narrowly defined wavelength band The delays of +20, +80, and +2730 ps were established at 595 run, the peak of CV absorptron. Three drstmct regions are evident rn the +20 and +80 ps spectra: (500-630 nm) ground-state bleaching; (380-500 MI) excited-state absorption (ESA), and (630-860 run) stimulated emasion. The presence of stimulated emission IS indicated by apparent bleaching m a regron of no ground-state absorption. In the 630-
1 April
860 nm region the weak probe pulse is amplified in the excited region of the sample. Experiments in which stimulated emission was suppressed by filters in the probe beam before the sample, indicate that it has no effect upon the kinetrcs monitored at other wavelengths. This can be attributed to the weak intensity of the probe pulse. In the event that a single type of excited molecule is produced by excitation, the spectra of fig. 3 can be used to find its spectrum, e’(X), at one of the delay times. The AOD produced in a sample of thickness Cis AOD(t, A) = [e’(X) - ea( X)] C’(r)l
AOD(t, x) = -@)C’(f)I
~~~
,
(2)
which permits calculation of C’(r). This value of C’(t) can then be used in eq. (1) to fiid e’(X) since ~a( A) is known. We have applied this procedure at x = 595 nm to the difference spectra to obtain e’(h) at +20 ps and +80 ps. Comparison of the excired-state spectra in fig. 4 indicates that over a time interval of 60 ps an increase of about 15% occurs in the bhre absorptron. Detailed measurements show that the excitedstate extinction coefficient does increase with time in the region of 430 run. This could be interpreted either as the formation of a small amount of triplet [4] or as an actual change in the value of e’(A) with time. In the
A
4 23 -
pAog& 0
a+&-b x 0
-08
_
Here C’(t), e’(h) and es(X) are the concentration of the excited molecules, and the extinction coefficients of the excited- and ground-state molecules. if a wavelength x exists for which eg 4 d, then at x
404
1980
500
*
cl 0
‘w-,
600
WAVELENGTH
:
e &a
0
1111111111111111111111111, 400
%I
0
0 00 a
-
%
I-9
_ -2-,
0,., 400
700
800
900
(nm)
Fii. 3. Picosecond difference spectra of crystal violet in glycerol: (0) f 20 ps; (A) + 80 ps; (+) + 2730 ps.
,,
,
,
500
,
,
,
,,
,
,
,ca,,
600
WAVELENGTH
,I,
700
I,,, 800
4 900
(nm)
Fig. 4. Spectra of the excited state of crystal violet obtained by applying eqs. (1) and (2) to the difference spectra of tiig. 3: (0) + 20 ps; (A) + 80 ps.
29
Volume 7 1. number 1
CHEhllCAL
most viscous solutrons (I) > 180 parse), a small percentage of the excited dye molecules IS converted to a longlived species. possrbly the tnplet ‘Dus IS shown by the residual AOD at 595 nm at a delay of 4 8 ns. whrch corresponds to about 10% of the uutrally exerted populatron. In the least vrscous solutron (17 = 0.86 parse) there IS no evrdence at any wavelength of the presence of the longhved specres at 4 8 ns. The complete (wrthm our expenmental precrsron) decay of the absorption at 430 MI by 2.7 ns m glycerol (fig. 3) favors the mterpretatron that e’(X) Increases with time The 5-10 ps pulses of our laser limrt kmetic studres to solvents mlth vrscosttres m excess of 0.7 parse (I.e., reiaxatron times 230 ps). The vrscosrty dependence of ground-state repopulatron (GSR) of CV was momtored at 564.588 and 6 10 MI The results of some measurements at 564 nm are presented m fig 5. bra slgmficant differences were observed between GSR rates momtored at the three wavelengths. This IS corroborated by the drfference spectra of fig. 3 whrch show that the decay of bleachmg proceeds umformly over the SOO530 nm region. Ground-state recovery of CV m cyclohexanol(0 86 poise at 16°C) and of CV m a glycerolwater solutron of the same vlscosrty was srmrlar. Ilus supports the rdea that viscosrty, rather than another solvent property, IS the controlhng factor m the relaxanon process. The agreement between the experrments of FH [I] and Brey et al. [3] m whrch vrscosity was
0
500
TIME
1000
1500
(psec)
Fig. 5 Recorery of ground-state absorption III crystal violet Solvents are- Co), 4G5 gIucose/60~ glycerol (TI = 180 poise). (a). 10% glucose/SGE glycerol (17= 40 potse); (o), glycerol (n = 8 Pose), (0). 86 5% glycerol/l3 5% Hz0 (9 = 0 8 poise) In both figs 5 and 6 the residual optical density difference at 4.8 ns. AOD(-), has been subtracted from measured AOD ralues to obtain the data shown here. The precision of the points
wtthout precisron bars IS less than or equal to the size of the
point. 30
1 Apnl1980
PHYSICS LE-ITERS
n 0
a
7 (3
s
I
200
TIME
400
1
600
801
(psec)
Fig. 6 Recovery of ground-state absorption III ethyl ulolet (o), crystal uxolet (o), and parafuchsln (3) vaned by changing
solvent composrtron and by the applicatron of high pressures, respectively, also supports this Idea. The time dependence of crystal violet ESA was morutored at 430 nm The decay curves were srmilar to those for GSR in solutions of corresponding viscosity_ The time dependence of probe pulse amphfication due to strmulated emrssron was observed at several wavelengths UI the 630-860 run region. If the reverse absorptrve transrtron does not occur, this method can be used to momtor the excited-state population These measurements gave relavatron rates srmllar to those obtained m the GSR studres The recovery of ground-state absorption followmg photoexcrtatron was also momtored m glycerol solutrons of ethyl violet, parafuchsm, and the trrmethyl crystal vrolet derrvative. The results are in fig. 6. The GSR curves for crystal vrolet and the trimethyl compound were identical.
4. Discussion The decay curves m fig. 6 support the idea that electromc relaxation of TEW dyes IS affected by a conformatron change hmdered by vrscous drag. The rates of relaxatron decrease IS the same order as the size of the phenyl nng para-substrtuent increases, I e., PF, CV, EV. Thus suggests a drrect connectron between vrscosity and rates of relaxation. However, the srmilanty of the GSR CUNVZS for TCV and CV appears to refute thrs suggestron. To mvestrgate thrs we examined the structure of each
Volume 71. number 1
CHEMICAL PHYSICS LETTERS
dye wrth space fang moiecuiar models. The models indicate that the orthomethyk of TCV would not penetrate the surrounding solvent environment to the degree charactenstrc of the para-substrtuents. Hence the vrscous drag introduced by the orthomethyis would not be as great as th.rt due to large para-groups. It follows that their effect upon relaxation would not be as great as that due to large para-groups. It follows that then effect upon relaxation would not be as significant. The GSR curves of figs. 5 and 6 indicate a complex relaxation mechamsm. Ippen et al. [6] as well as Hirsch and Mahr [8] fitted srmllar curves to a doubie-exponentiai functron. We have noticed, however, that a sum of many decaymg exponent&Is f=
F
N(0, k& exp(-k,
f)
(3)
can fit our GSR and ESA curves. Functron f describes the decay of a collection of excited molecules grouped according to the decay rate Ic,. The coefficient N(0, k,) ISa constant and represents the number of molecules associated with rate k, at trme zero. If the N(0, kl) are equal and the distnbutron of decay rates between a mmrmum (11-r) and a maxlmum (kz)vaiue IS uniform eq (3) can be integrated to give fo:
[exp(--kit)
- exp(-k2f)]
/t .
(4)
Although thrs expressron represents a sum of many equally werghted decaying exponent& rt IS possible, using physically reasonable values of k, and kZ, to duplicate a double-exponenti~ function quote well. A conformational analysis of TPM dyes f 111 mdrcates they may represent a system described appropnately by eq. (3). The analysis shows (1) the ground-state potential surface 1s shallow wtth respect to asynchronous vanatton in the phenyl ring angles, and (2) the energy gap between the states So and S, narrows wrth ring angle. Point (1) would perrmt a wrde distnbution of ground-state conformations. Pomt (2) would allow the decay rate kl to vary with conformation, since rates of internal conversion are strongly dependent upon the So + S, energy gap. The quantum yield, kinetic, and conformational data suggest the following model of the vtscosity dependence: Photoexcrtatron leads to a replication of the groundstate conformation distnbution in the exerted state $5, _ In high viscosrty solvents the conformatrons are fixed and the decay of the excited-state population is described
1 April 1980
by eq. (3). Both fast ~ra~atio~e~) and slow (luminescent) relaxation processes are important. As viscosity is reduced the possrbility of a conformational change in the excited molecules prior to relaxation is increased. Excited molecules with decay rates near k, have a fiite probability of acquiring a new conformation characteristrc of the faster rates near k, _ This probabi~ty is related to the solvent vrscosity. The observed increase in the excited-state extmction coefficient at 430 nm is consistent with a change in the molecular conformation of the excited state. Reduced vtscosity then leads to a channeling of excited molecules into the faster nanradrative pathways of decay. Consequently the slower radiative processes are quenched. The accumulated kinetic data can be understood in terms of thrs model. As viscosity is decreased only the fastest-decaymg exponentials of eq. (3) remain important and in very fluid solutions eq. (3) may reduce to only a few exponent& terms. This would explain the observation of Ippen et ai. [6] that the description of GSR changes from a double exponential decay (or, as we suggest, a multrphasic decay) to a single exponential as viscostty ISreduced. Lummescence is described by the slower (radiative) components of eq. (3) This explains why the faster component of luminescence noted by Hrrsch and Mahr [St agrees with she slower component of GSR measured by lppen et al. On the other hand, the fast component of GSR is large!y the result of radratiordess decay of those molecules excited initially with large kf vahres. The above discussion shows that, on a semi-qnantitative levei, our model can describe the main features of electronic relaxation in triphenyimethane dyes. To test the model further, we have examined in some detail the effects of changing the angle between the phenyl rings and the moiecuIar plane. in this study, based upon a conformational analysis that includes molecukr orbital c~cu~ations as a function of ring angIe, we used numerical methods to solve a modified Fokker-Plan& differential equation. With this model we have been able to generate curves that show the same viscosity dependence as the experimental kinetic data. There is reason to believe that the model may also be able to account for the Q of n2j3 reiatronship. A detarled account of this work is in preparation and will be presented elsewhere.
3h
Volume 71, number
I
CHEMICAL
PHYSICS
Acknowledgement
We thank Chris Khmaier for the synthesis and purification of the trimethyl derivative of crystal violet. This work was supported in part by the Office of Naval Research and by the US Army Research Office (Durham).
References [I J Th. Fiirster and G Hoffmann, Z. Physik. Chem NF 75 (1971) 63. [2] C-J. hfastrangelo and H.W Offen. Chem. Phys. Letters 46 (1977) 588.
32
LETTERS [3]
1 Apd
1980
L.A. Brey. G B. Schuster and H.G. Dnckamer, J. Chem. Phys 67 (1977) 2648. [4] D. Magde and M.W. Windsor, Chem. Phys. Letters 24 (1974) 144. [S 1 B A. Bushaw, Ph D. Thesis, Washmgton State Utnverslty (1975). [61 E P. Ippen. C-V. Shank and A Bergman. Chem. Phys. Letters 38 (1976) 611. [71 W. Yu, F. Pellegnno, hf. Grant and R R. A&no. J. Chem. Phys. 67 (1977) 588. (81 M.D. Hirsch and H. Mahr, Chem Phys Letters 60 (1979) 299. [9] C C. Barker, M H. Bride and A. Stamp, J. Chem. Sot. (1959) 3957. [IO] D. hfagde and M.W. Wmdsor, Chem. Phys. Letters 27 (1974) 31. [ II] D.A. Cremers and hf.W. Wmdsor, to be published.
CHEMICAL
1
A STUDY OF THE VISCOSITY-DEPENDENT OF SOME TRIPHENYLMETHANE David A. CREMERS Department
PHYSICS
L April 1980
LETTERS
ELECTRONIC
RELAXATION
DYES USING PICOSECOND FLASH PHOTOLYSIS
and Maurice W. WINDSOR
of Cirremlsny, Washmgton State Umwrs~ty, Pullman, Waslrmgtorr 99164. USA
Recewed 29 October
1979
Tune-resolved absorption measurements of ground-state recovery and euclted-state tnphenylmethane dyes m solutions of different viscosity. A kmetlc model us proposed earher measurements
absorption are reported for several to explain the results of these and
1. Introduction The viscosity-dependent relaxation of photoexcited tnphenyimethane (TPM) dyes (fig. 1) has been studled in several ways. Fluorescence quantum yield 0 measurements [l-3] and direct kmetic measurements using picosecond spectroscopic techmques [S-S] indicate that increased solvent vlscoslty results m decreased rates of electroruc relaxation. Forster and Hoffmann (FH) [ I] proposed a model of this viscosity dependence which predicts the Q = $I3 relationshp found m some measurements of Q [1,3,8 J According to tis model, the photoexcited molecule undergoes a change m conformation to a new structure characterized by enhanced rates of non-radiative decay. The importanr conformatIonal change is assumed to be synchronous rotation of the phenyl rings about the bond between the ring and central carbon atoms (see fig. 1). The rotation IS driven by steric repulsion between adjacent ring orthohydrogen atoms and is hindered by viscous drag mtroduced by the solvent molecules. Increased VISCOSlty slows the rate of ring rotatron and lengthens the tune interval between excitation and the achevement of those conformations that give rise to fast non-radiative decay. In solvents of low viscosity, these conforrnatlons are reached more rapidly and fluorescence is quenched by the faster radiationless paths of relaxation. Although the FH model predicts the observed Q 0: ~$3 relationship, the time dependence of excitedstate relaxation of exp(-at3) also predicted by Uus
-I-
CI-
‘R Rg. 1. The triphenyhnethane dyes crystal violet CV (R =N(CH&. R’ = HI, ethyl violet EV [R = N
model has not been observed. Magde and Windsor [4] measured the recovery of ground-state absorption of crystal violet (CV) following picosecond excitation. The data did not distinguish between several possible functional descriptions of the recovery. They concluded, however, that internal conversion is the dominant non-radiative mechanism . Ippen et al. [6] measured ground-state recovery (GSR) of malacbte green and found it to be a single exponential for TJ< 1 poise. For 17> 1 poise the recovery was described as a double exponential. Yu et al. [7] monitored malachite green fluorescence and found it to decay as a single exponential. Hirsch and Mahr [8] aIsct 27
Volume 7 1, number 1
CHEMICAL
PHYSICS
studied malachite green emlsslon and characterized it as a double-exponentlal decay. The hfetlmes of the faster decays of Hirsch and Mahr agree with the hfetimes of Yu et al. and the slower component of CSR measured by Ippen et al. Despite the several studies referred to above, a description of the viscosity dependence has not yet been proposed which reconcdes all the kmetlc and quantum yield measurements The role of vlscoslty in the p~cosecond relaxation kmetlcs of TPM dyes has been of mterest to this laboratory for several years [4,5]. We present here a more detaded mvestlgatlon of these VIScosity effects.
2. Experimental The dyes used m tlus study are shown m fig. lCrystal violet (Baker Chemical) used m the majority of experiments was purltied via the leuco base Ethyl violet (Eastman) and parafuchsin (Tndom Fluka) were punfied by column chromatography smce smaller amounts of these dyes were needed. The tnmethyl derivative of CV was prepared and punfied accordmg to the hterature [9]. All dyes were m the form of the chlonde salt. Solvents of the desired viscosity were prepared by mcung deionized water, reagent-grade glycerol and glucose in various proportlons The \IScosltles of all dye solutions were directly measured. Unless otherwlse specified, all measurements were made on 6 X IO-5 molar solutions at 25°C. The picosecond flash photolysis apparatus used m this study 1s described elsewhere [lo] It IS based upon a single-pulse-selected. mode-locked Nd - glass laser. A smgle amphfied picosecond pulse at the laser fundamental (I 060 nm) is S-IO ps m duration and has an energy of about 15 mJ. Frequency doubhng produces a 530 nm excltatlon (pump) pulse of about 2 mJ. The remammg 13 mJ of 1060 nm radlatlon IS focused mto a cell of CC14 to generate a spectrally broad (400-900 run) picosecond probe pulse of weak mtenslty The probe light transmitted by the samp!e is analyzed by a 2 m Jarrell-Ash spectrometer and detected on an optical multvAmnel analyzer (Prmceton Apphed Research). The geometry of the pump and probe pulses at the sample [IO] permits measurements of the difference in optical density (JOD) between excited and unexcited regions of the sample. Each laser finng produces
28
1 April 1980
LETl-ERS
a smgle measurement of AOD at a specific wavelength and delay. The excitation and probmg laser pulses were imaged onto a small volume of the dye solution contamed wlthm a 2 mm path length spectrometer cell Following each laser firing, the contents of the cell were replaced by a fresh dye solution. The temperature of the sample regon was carefully controlled to prevent fluctuations in vlscoslty due to thermal effects ms was especially critical for the more viscous solutions The hygroscoplc nature of glycerol made It necessary to sheld the dye solutions from the atmosphere. The vlscoslty of the glycerol solutions 1s very sensitive to water contamination. The mtenslties of the pump and probe pulses vary Lvlth each laser firing. Variations m probe Intensity do not affect a measurement of AOD. Fluctuations in excitation power occur, and do cause changes m AOD, buth these are mmumzed by using sufficient power to saturate the sample The effect of saturatmg the CV sample IS shown m fig. 2. The changes in AOD due to vanatlons in excltatlon power are small when 100% of the power IS used Each data point of a spectrum or hnetlc curve m this study represents the average of usually six separate AOD measurements The total length of the precision bars equals twice the standard devlatlon of the AOD values used to compute the arerage. Experiments m which the relative polarizations of the excltatlon and probe pulses were adjusted either to accent or to mask the effects of molecular reonentatlon ylelded smular results. Thus our measurements were apparently not affected by onentatlonal relauatlon of the entire molecule following photoexcltatlon.
06
04
-
-
02
-0
00
d 1 0
a’ r
I 20
% Fig. 2. cerol).
P
6-B -A’ P
I 40
PUMP
I 60
1 80
I 100
POWER
Saturation curve for crystal wolet (6 X 10m5 M in gly-
Volume 7 1, number 1
CHEMICAL
PHYSICS LETTERS
Undoubtedly thrs is a result of the high solvent viscosrtres we have used which cause overall rotational motions to be very much slower than the processes we have studred.
3. Results Optical density drfference spectra of CV over the range 380-900 nm at time delays of +20, +80 and +2730 ps are shown in fig. 3. To obtain data in the 380-450 run region the picosecond apparatus was modified to permit generation of the probe pulse by the 530 run excitation pulse The continuum produced in this way has significantly greater intensity in the bIue regron than the 1060 MI generated contmuum. The effects of dispersion in the arrival time of the probe pulse at the sample have not been removed from these spectra Calculation and experience rndrcate that, over the spectral regron considered here, the drfference in amval tunes between the red and blue probe light is, at most, 20 ps. Such dispersion has no effect on the kinetrc data since these are taken wrthm a narrowly defined wavelength band The delays of +20, +80, and +2730 ps were established at 595 run, the peak of CV absorptron. Three drstmct regions are evident rn the +20 and +80 ps spectra: (500-630 nm) ground-state bleaching; (380-500 MI) excited-state absorption (ESA), and (630-860 run) stimulated emasion. The presence of stimulated emission IS indicated by apparent bleaching m a regron of no ground-state absorption. In the 630-
1 April
860 nm region the weak probe pulse is amplified in the excited region of the sample. Experiments in which stimulated emission was suppressed by filters in the probe beam before the sample, indicate that it has no effect upon the kinetrcs monitored at other wavelengths. This can be attributed to the weak intensity of the probe pulse. In the event that a single type of excited molecule is produced by excitation, the spectra of fig. 3 can be used to find its spectrum, e’(X), at one of the delay times. The AOD produced in a sample of thickness Cis AOD(t, A) = [e’(X) - ea( X)] C’(r)l
AOD(t, x) = -@)C’(f)I
~~~
,
(2)
which permits calculation of C’(r). This value of C’(t) can then be used in eq. (1) to fiid e’(X) since ~a( A) is known. We have applied this procedure at x = 595 nm to the difference spectra to obtain e’(h) at +20 ps and +80 ps. Comparison of the excired-state spectra in fig. 4 indicates that over a time interval of 60 ps an increase of about 15% occurs in the bhre absorptron. Detailed measurements show that the excitedstate extinction coefficient does increase with time in the region of 430 run. This could be interpreted either as the formation of a small amount of triplet [4] or as an actual change in the value of e’(A) with time. In the
A
4 23 -
pAog& 0
a+&-b x 0
-08
_
Here C’(t), e’(h) and es(X) are the concentration of the excited molecules, and the extinction coefficients of the excited- and ground-state molecules. if a wavelength x exists for which eg 4 d, then at x
404
1980
500
*
cl 0
‘w-,
600
WAVELENGTH
:
e &a
0
1111111111111111111111111, 400
%I
0
0 00 a
-
%
I-9
_ -2-,
0,., 400
700
800
900
(nm)
Fii. 3. Picosecond difference spectra of crystal violet in glycerol: (0) f 20 ps; (A) + 80 ps; (+) + 2730 ps.
,,
,
,
500
,
,
,
,,
,
,
,ca,,
600
WAVELENGTH
,I,
700
I,,, 800
4 900
(nm)
Fig. 4. Spectra of the excited state of crystal violet obtained by applying eqs. (1) and (2) to the difference spectra of tiig. 3: (0) + 20 ps; (A) + 80 ps.
29
Volume 7 1. number 1
CHEhllCAL
most viscous solutrons (I) > 180 parse), a small percentage of the excited dye molecules IS converted to a longlived species. possrbly the tnplet ‘Dus IS shown by the residual AOD at 595 nm at a delay of 4 8 ns. whrch corresponds to about 10% of the uutrally exerted populatron. In the least vrscous solutron (17 = 0.86 parse) there IS no evrdence at any wavelength of the presence of the longhved specres at 4 8 ns. The complete (wrthm our expenmental precrsron) decay of the absorption at 430 MI by 2.7 ns m glycerol (fig. 3) favors the mterpretatron that e’(X) Increases with time The 5-10 ps pulses of our laser limrt kmetic studres to solvents mlth vrscosttres m excess of 0.7 parse (I.e., reiaxatron times 230 ps). The vrscosrty dependence of ground-state repopulatron (GSR) of CV was momtored at 564.588 and 6 10 MI The results of some measurements at 564 nm are presented m fig 5. bra slgmficant differences were observed between GSR rates momtored at the three wavelengths. This IS corroborated by the drfference spectra of fig. 3 whrch show that the decay of bleachmg proceeds umformly over the SOO530 nm region. Ground-state recovery of CV m cyclohexanol(0 86 poise at 16°C) and of CV m a glycerolwater solutron of the same vlscosrty was srmrlar. Ilus supports the rdea that viscosrty, rather than another solvent property, IS the controlhng factor m the relaxanon process. The agreement between the experrments of FH [I] and Brey et al. [3] m whrch vrscosity was
0
500
TIME
1000
1500
(psec)
Fig. 5 Recorery of ground-state absorption III crystal violet Solvents are- Co), 4G5 gIucose/60~ glycerol (TI = 180 poise). (a). 10% glucose/SGE glycerol (17= 40 potse); (o), glycerol (n = 8 Pose), (0). 86 5% glycerol/l3 5% Hz0 (9 = 0 8 poise) In both figs 5 and 6 the residual optical density difference at 4.8 ns. AOD(-), has been subtracted from measured AOD ralues to obtain the data shown here. The precision of the points
wtthout precisron bars IS less than or equal to the size of the
point. 30
1 Apnl1980
PHYSICS LE-ITERS
n 0
a
7 (3
s
I
200
TIME
400
1
600
801
(psec)
Fig. 6 Recovery of ground-state absorption III ethyl ulolet (o), crystal uxolet (o), and parafuchsln (3) vaned by changing
solvent composrtron and by the applicatron of high pressures, respectively, also supports this Idea. The time dependence of crystal violet ESA was morutored at 430 nm The decay curves were srmilar to those for GSR in solutions of corresponding viscosity_ The time dependence of probe pulse amphfication due to strmulated emrssron was observed at several wavelengths UI the 630-860 run region. If the reverse absorptrve transrtron does not occur, this method can be used to momtor the excited-state population These measurements gave relavatron rates srmllar to those obtained m the GSR studres The recovery of ground-state absorption followmg photoexcrtatron was also momtored m glycerol solutrons of ethyl violet, parafuchsm, and the trrmethyl crystal vrolet derrvative. The results are in fig. 6. The GSR curves for crystal vrolet and the trimethyl compound were identical.
4. Discussion The decay curves m fig. 6 support the idea that electromc relaxation of TEW dyes IS affected by a conformatron change hmdered by vrscous drag. The rates of relaxatron decrease IS the same order as the size of the phenyl nng para-substrtuent increases, I e., PF, CV, EV. Thus suggests a drrect connectron between vrscosity and rates of relaxation. However, the srmilanty of the GSR CUNVZS for TCV and CV appears to refute thrs suggestron. To mvestrgate thrs we examined the structure of each
Volume 71. number 1
CHEMICAL PHYSICS LETTERS
dye wrth space fang moiecuiar models. The models indicate that the orthomethyk of TCV would not penetrate the surrounding solvent environment to the degree charactenstrc of the para-substrtuents. Hence the vrscous drag introduced by the orthomethyis would not be as great as th.rt due to large para-groups. It follows that their effect upon relaxation would not be as great as that due to large para-groups. It follows that then effect upon relaxation would not be as significant. The GSR curves of figs. 5 and 6 indicate a complex relaxation mechamsm. Ippen et al. [6] as well as Hirsch and Mahr [8] fitted srmllar curves to a doubie-exponentiai functron. We have noticed, however, that a sum of many decaymg exponent&Is f=
F
N(0, k& exp(-k,
f)
(3)
can fit our GSR and ESA curves. Functron f describes the decay of a collection of excited molecules grouped according to the decay rate Ic,. The coefficient N(0, k,) ISa constant and represents the number of molecules associated with rate k, at trme zero. If the N(0, kl) are equal and the distnbutron of decay rates between a mmrmum (11-r) and a maxlmum (kz)vaiue IS uniform eq (3) can be integrated to give fo:
[exp(--kit)
- exp(-k2f)]
/t .
(4)
Although thrs expressron represents a sum of many equally werghted decaying exponent& rt IS possible, using physically reasonable values of k, and kZ, to duplicate a double-exponenti~ function quote well. A conformational analysis of TPM dyes f 111 mdrcates they may represent a system described appropnately by eq. (3). The analysis shows (1) the ground-state potential surface 1s shallow wtth respect to asynchronous vanatton in the phenyl ring angles, and (2) the energy gap between the states So and S, narrows wrth ring angle. Point (1) would perrmt a wrde distnbution of ground-state conformations. Pomt (2) would allow the decay rate kl to vary with conformation, since rates of internal conversion are strongly dependent upon the So + S, energy gap. The quantum yield, kinetic, and conformational data suggest the following model of the vtscosity dependence: Photoexcrtatron leads to a replication of the groundstate conformation distnbution in the exerted state $5, _ In high viscosrty solvents the conformatrons are fixed and the decay of the excited-state population is described
1 April 1980
by eq. (3). Both fast ~ra~atio~e~) and slow (luminescent) relaxation processes are important. As viscosity is reduced the possrbility of a conformational change in the excited molecules prior to relaxation is increased. Excited molecules with decay rates near k, have a fiite probability of acquiring a new conformation characteristrc of the faster rates near k, _ This probabi~ty is related to the solvent vrscosity. The observed increase in the excited-state extmction coefficient at 430 nm is consistent with a change in the molecular conformation of the excited state. Reduced vtscosity then leads to a channeling of excited molecules into the faster nanradrative pathways of decay. Consequently the slower radiative processes are quenched. The accumulated kinetic data can be understood in terms of thrs model. As viscosity is decreased only the fastest-decaymg exponentials of eq. (3) remain important and in very fluid solutions eq. (3) may reduce to only a few exponent& terms. This would explain the observation of Ippen et ai. [6] that the description of GSR changes from a double exponential decay (or, as we suggest, a multrphasic decay) to a single exponential as viscostty ISreduced. Lummescence is described by the slower (radiative) components of eq. (3) This explains why the faster component of luminescence noted by Hrrsch and Mahr [St agrees with she slower component of GSR measured by lppen et al. On the other hand, the fast component of GSR is large!y the result of radratiordess decay of those molecules excited initially with large kf vahres. The above discussion shows that, on a semi-qnantitative levei, our model can describe the main features of electronic relaxation in triphenyimethane dyes. To test the model further, we have examined in some detail the effects of changing the angle between the phenyl rings and the moiecuIar plane. in this study, based upon a conformational analysis that includes molecukr orbital c~cu~ations as a function of ring angIe, we used numerical methods to solve a modified Fokker-Plan& differential equation. With this model we have been able to generate curves that show the same viscosity dependence as the experimental kinetic data. There is reason to believe that the model may also be able to account for the Q of n2j3 reiatronship. A detarled account of this work is in preparation and will be presented elsewhere.
3h
Volume 71, number
I
CHEMICAL
PHYSICS
Acknowledgement
We thank Chris Khmaier for the synthesis and purification of the trimethyl derivative of crystal violet. This work was supported in part by the Office of Naval Research and by the US Army Research Office (Durham).
References [I J Th. Fiirster and G Hoffmann, Z. Physik. Chem NF 75 (1971) 63. [2] C-J. hfastrangelo and H.W Offen. Chem. Phys. Letters 46 (1977) 588.
32
LETTERS [3]
1 Apd
1980
L.A. Brey. G B. Schuster and H.G. Dnckamer, J. Chem. Phys 67 (1977) 2648. [4] D. Magde and M.W. Windsor, Chem. Phys. Letters 24 (1974) 144. [S 1 B A. Bushaw, Ph D. Thesis, Washmgton State Utnverslty (1975). [61 E P. Ippen. C-V. Shank and A Bergman. Chem. Phys. Letters 38 (1976) 611. [71 W. Yu, F. Pellegnno, hf. Grant and R R. A&no. J. Chem. Phys. 67 (1977) 588. (81 M.D. Hirsch and H. Mahr, Chem Phys Letters 60 (1979) 299. [9] C C. Barker, M H. Bride and A. Stamp, J. Chem. Sot. (1959) 3957. [IO] D. hfagde and M.W. Wmdsor, Chem. Phys. Letters 27 (1974) 31. [ II] D.A. Cremers and hf.W. Wmdsor, to be published.