Resolving and Controlling Photoinduced Ultrafast Solvation in theSolid StateMilan Delor,† Dannielle G. McCarthy,† Benjamin L. Cotts,† Trevor D. Roberts,† Rodrigo Noriega,†,∇
David D. Devore,‡ Sukrit Mukhopadhyay,‡ Timothy S. De Vries,‡ and Naomi S. Ginsberg*,†,§,∥,⊥,#
†Department of Chemistry and §Department of Physics, University of California Berkeley, Berkeley, California 94720, United States‡The Dow Chemical Company, Midland, Michigan 48674, United States∥Materials Sciences Division, and ⊥Molecular Biophysics and Integrated Bioimaging Division, Lawrence Berkeley NationalLaboratory, Berkeley, California 94720, United States#Kavli Energy NanoSciences Institute, Berkeley, California 94720, United States
*S Supporting Information
ABSTRACT: Solid-state solvation (SSS) is a solid-state analogue of solvent−soluteinteractions in the liquid state. Although it could enable exceptionally fine control over theenergetic properties of solid-state devices, its molecular mechanisms have remained largelyunexplored. We use ultrafast transient absorption and optical Kerr effect spectroscopies toindependently track and correlate both the excited-state dynamics of an organic emitterand the polarization anisotropy relaxation of a small polar dopant embedded in anamorphous polystyrene matrix. The results demonstrate that the dopants are able torotationally reorient on ultrafast time scales following light-induced changes in theelectronic configuration of the emitter, minimizing the system energy. The solid-statedopant−emitter dynamics are intrinsically analogous to liquid-state solvent−soluteinteractions. In addition, tuning the dopant/polymer pore ratio offers control oversolvation dynamics by exploiting molecular-scale confinement of the dopants by the polymer matrix. Our findings will enablerefined strategies for tuning optoelectronic material properties using SSS and offer new strategies to investigate mobility anddisorder in heterogeneous solid and glassy materials.
Solvation is a stabilizing interaction of solute and solventmolecules, a fundamental process predominantly associated
with liquid-state systems. Its ubiquity in nature as well as itspotential for tuning the electronic properties of chemicalsystems has motivated highly detailed theoretical andexperimental investigations seeking to understand the factorsgoverning the properties of solvation and how to controlthem.1−6 Solvation analogues in the solid state, permitted bythe presence of rotationally mobile polar moieties in crystalline,semicrystalline, or amorphous materials, however, remaincomparatively underexplored despite their promise fordynamically controlling the properties of molecular devices.7−19
Solid-state solvation (SSS) has been shown to radically affectsystems consisting of fluorescent dyes embedded in amorphouspolymer films doped with small polar molecules, whereby anincrease in dopant concentration leads to a significant red shiftof the dye emission.7 The lower-energy emission has beenpostulated to occur due to stabilization of the emissive statethrough electronic and nuclear/orientational reorganization ofthe polar dopants, a process analogous to solvent−soluteinteractions in liquids. Such SSS has, however, never beendirectly observed because all studies have concentrated on thealready relaxed excited state of embedded emitters, thusobscuring the underlying molecular dynamics responsible forthe energetic stabilization. Here, we investigate how to better
exploit solvation to actively control the properties of functionalsolid-state materials by directly monitoring both solvent andsolute responses on ultrafast time scales in their mutual,photoinduced interaction within amorphous polymer solids.We have recently shown that SSS can be extended to control
not only the emission energy but also the yield of differentexcited states in thermally activated delayed fluorescence(TADF) emitters, a promising new class of emitters for organiclight-emitting diodes (OLEDs).21 These TADF emitters areperfect candidates to explore solvation tuning as theiroptoelectronic properties are governed by delicate interactionsbetween close-lying excited states in the singlet and tripletmanifolds. In particular, the energy gap between the lowestsinglet and triplet statesoften charge transfer (CT) and π−π*states, respectivelytunes the overall quantum yield (QY) offluorescence. The latter is dictated by a complex interplay ofthermodynamics and transition dipole coupling to the lightfield; a smaller energy gap increases the rate of thermallyactivated reverse intersystem crossing from the nonemissivetriplet to the singlet but typically accompanies a lowering of theoscillator strength of the radiative S1 →S0 transition due to the
increased polarization of S1.20,22−25 These counterbalancing
repercussions of SSS lead to a nonmonotonic dependence ofthe emission QY with respect to the singlet−triplet energygap.18 By varying the concentration of the small-moleculedopant camphoric anhydride (CA) in amorphous films of theTADF emitter 2,5-bis(4-(10H-phenoxazin-10-yl)phenyl)-1,3,4-oxadiazole (2PXZ-OXD)20 embedded in a glassy polystyrene(PS) matrix (Figure 1a,b), the lowest singlet and triplet states
are stabilized to varying degrees (with CT states displayinggreater solvatochromism, Figure 1c). In this way, the QY of thisOLED material can be enhanced by over 20%, with a maximumQY achieved between ∼3 and 7 wt % CA dopingconcentrations.18
Using the same ternary host−guest−dopant system,PS:2PXZ-OXD:CA, we now turn our attention to the initialsolvation response following a sudden change in electrondensity distribution caused by photoexcitation of 2PXZ-OXDto a singlet CT manifold (S1 electric dipole ≈ 21 D, S0 electricdipole ≈ 2 D). By monitoring the excited-state evolution of theemitter using transient absorption (TA) spectroscopy in boththe presence and the absence of the CA dopant (electric dipole≈ 6 D), we find distinct ultrafast spectral evolution signaturesthat correspond to dynamic stabilization of the emitter’s lowestCT state in the presence of CA. To confirm the equivalence ofsuch stabilization to that in liquid solvation, whereby solventmolecules reorient in response to changes in electronicconfigurations of the solute, we use the optically heterodyne-
detected optical Kerr effect (OHD-OKE); this powerfulapproach verifies that a polarization anisotropy can be opticallycreated in CA-doped amorphous polymer films, and it enablesmeasuring of the time scales associated with orientationalrelaxation of CA molecules embedded in the polymer matrix.We find that at optimized CA concentrations that maximizetheir relative contribution to the dielectric medium whileretaining sufficient mobility in the polymer free volume, thepolarization anisotropy dynamics corroborate the emitterexcited-state stabilization time scales measured using TAspectroscopy. The results explicitly reveal the intrinsicallyclose association between solid-state and liquid-state solvation.The emitter excited-state evolution is af fected by the introduction
of a solid-state solvent. Figure 2a displays the TA spectra of spin-cast ∼1 μm thin films of PS:2PXZ-OXD with 0, 5, and 10 wt %doping concentrations of CA. The system is excited using a 400nm, ∼100 fs duration laser pulse at the lowest-energy allowedelectronic transition (centered at 400 nm with an onset at 450nm18), but the expected ground-state bleach at thesewavelengths is not clearly discernible. Instead, the spectrafrom 400 to 650 nm are in all cases dominated by excited-stateabsorption (ESA) with two broad bands centered at ∼440 and∼530 nm, which decay on few-ns time scales. The introductionof CA produces differences in the ESA evolution that areimmediately visible at early time delays, most evident in theless-congested spectral region around 550−580 nm in Figure2a. To elucidate the early time dynamics of each system, thekinetics are globally fit across all wavelengths. The resultingspectral amplitudes of each exponential component, or decay-associated spectra (DAS), are displayed in Figure 2b. Lifetimesand associated errors, summarized in Table 1, are averages andstandard deviations of the mean of globally fitted lifetimesacross three independent data sets.There are several key features that highlight the different
excited-state evolution in the first few hundred picosecondsfollowing photoexcitation of 2PXD-OXD in PS in the presenceof vs in the absence of CA. First, a sum of two exponentialswith lifetimes τ1 = 97 ± 6 ps and τ2 = 1.1 ± 0.2 ns deconvolvedfrom the instrument response function (IRF, 120 fs) issufficient to fit the observed dynamics of the binary systemPS:2PXZ-OXD. These correspond, respectively, to an intra-molecular nuclear reorganization and to the excited-statedepletion of the emitter. Yet, a third, faster time componentis consistently necessary to fit the dynamics of PS:2PXZ-OXD:CA(5,10%): τ3 = 6.3 ± 0.7 ps and τ3 = 4.3 ± 0.9 ps,respectively. This third exponential is reflected in the growth ofthe differential absorbance around 430−450 nm on short timescales, as evidenced by the negative sign of the correspondingblack DAS curve (see also Figures S1 and S2 for kinetic tracesat representative wavelengths with and without CA). Thisapparent growth can arise due to both blue-shifting ESAs andred-shifting stimulated emission as the system relaxes along itsnuclear reaction coordinates following photoexcitation into theelectronic excited state. Indeed, a small but clearly resolvabledynamic blue shift of the peak ESA positions occurs with 5%CA concentration (see open circles in Figure 3c). This behavioris expected in the case of solvation stabilization because thelowest singlet excited state possesses predominantly CTcharacter while higher-lying excited states possess mixed CT/locally excited character and are thus less affected by the localdielectric (see the computational section in the SupportingInformation). A second key difference when CA is present isthat the intramolecular dynamics leading to the stabilized 1CT
Figure 1. Overview of experimental aim. (a) The oxadiazole−phenoxazine donor−acceptor−donor TADF emitter 2PXZ-OXD20 isembedded in a PS host matrix doped with the small polar moleculeCA. (b) Upon photoexcitation of 2PXZ-OXD, a charge transfer statewith a large electric dipole (∼21 D) is created; we monitor whetherand how CA molecules reorganize to stabilize this new electronicconfiguration. (c) Steady-state emission spectra showing that theemissive state (S1) is stabilized upon increasing dopant concen-tration.18
The Journal of Physical Chemistry Letters Letter
DOI: 10.1021/acs.jpclett.7b01689J. Phys. Chem. Lett. 2017, 8, 4183−4190
state are faster (τ1 = 24 ± 2 ps with 5% CA and 50 ± 12 ps with10% CA vs 97 ps with 0% CA). While PS aromatic side chainsmay also electronically interact with 2PXZ-OXD22 andpotentially affect dynamic components in the TA spectra, theaddition of CA clearly has its own effect on the system’sexcited-state dynamics.To connect these observations to solution-state dynamics, we
also investigated the excited-state evolution of 2PXZ-OXD in
toluene and chloroform under the same experimentalconditions (Figure S3). We found that the overall spectraldynamics in solution are very similar to those of the ternarysolid-state PS:2PXZ-OXD:CA system. In particular, thepicosecond time scale growth observed in the films at shortwavelengths also occurs on the characteristic solvation timescale for chloroform (2.6 ps). Furthermore, changing from anonpolar solvent (toluene) to a relatively polar one (chloro-form) leads to blue-shifted ESAs and faster excited-stateevolution toward the stabilized 1CT state.Overall, the TA data for PS:2PXZ-OXD:CA(5,10%) films
suggest that the singlet CT state stabilization is both enhancedand accelerated in the presence of CA. The characteristic earlytime spectral evolution in the presence of CA is remarkablysimilar to that observed in solution-state TA and suggests thatCA dynamically solvates the excited state of 2PXD-OXDembedded in the polymer matrix. In earlier studies of SSS,7,18
an increase in dopant concentration led to a relatively lineardecrease in emission energy and increase in electronicpermittivity of the films from 0 to 10% doping concentration.One might therefore expect that the trend in excited-stateevolution from 0 to 5% CA concentration could be extrapolatedto 10% concentration. While the early time dynamics with 10%CA do indicate solvation-like behavior according to ourassignments, the intermediate evolution time scale of tens ofps, indicative of the time required for complete CTstabilization, is considerably slower than that observed with5% CA while still faster than that with 0% CA. Furthermore,ESA dynamics blue shifts with 10% CA concentrations are toosmall to be reliably resolved within the error of the experiment.We will return to this surprising observation below.
Figure 2. Chirp-corrected TA spectra following 400 nm excitation (a) and DAS (b) extracted from global analysis of the time-resolved data forPS:2PXZ-OXD:CA(0,5,10%) films. All exponentials are deconvolved from a Gaussian IRF of 120 fs full width at half-maximum (fwhm). Twoexponentials are required to fit the 0% CA data, while three exponentials are required to fit the 5 and 10% CA concentration data. The additionalshort exponential component represents shifting ESAs and stimulated emission, assigned to excited-state stabilization by CA.
Table 1. Short (<1 ns) Lifetime Components, InPicoseconds, From TA and OHD-OKE Data
0% CA 5% CA 10% CA
TAa 97 ± 6 6.3 ± 0.7 4.3 ± 0.924 ± 2 50 ± 12
ESA blue shift − 4 ± 2 d22 ± 9
OHD-OKEb − 1.6 ± 0.2 1.5 ± 0.414 ± 2 13 ± 2
CV(t)c − 2.5 ± 0.1 0.6 ± 0.1
25 ± 0.2 3.2 ± 0.113 ± 0.1
aValues quoted for the TA fits are averages and standard deviations ofthe mean across three globally fitted data sets for each concentration.bValues quoted for the OHD-OKE fits are averages and standarddeviations of the mean across four different data sets for eachconcentration. cCV(t) is calculated using data shown in Figure 3a,b,where the thickness of the 5 and 10% CA films are equivalent. dAnydynamic blue shift of the ESA in 10% CA films is too small (<1−2nm) to be reliably extracted from our data.
The Journal of Physical Chemistry Letters Letter
DOI: 10.1021/acs.jpclett.7b01689J. Phys. Chem. Lett. 2017, 8, 4183−4190
Orientational relaxation of CA molecules occurs on the sametime scale as excited-state solvation of the emitter. While the TAdata suggest that CA indeed plays an active role in stabilizingthe excited state of 2PXZ-OXD, the molecular underpinningsof this stabilization, in particular, whether or not it is due to aliquid-like process of solvent reorganization, remains an openquestion even with the above TA results. In order to answerthis question, we also perform OHD-OKE experiments thatprobe the decay of the CA polarization anisotropy created byan intense linearly polarized nonresonant pump pulse. In theseexperiments, exponential relaxation on the order of a fewpicoseconds is a characteristic signature of orientationalrandomization of pump-aligned molecules in the sample,26,27
serving as an indicator of rotational mobility in the glassypolymer matrix.To obtain a sufficient signal-to-noise ratio in disordered,
amorphous samples such as the PS films currently investigated,we make several modifications to a typical OHD-OKEexperiment (see the Experimental Methods section). As thepump should be nonresonant and we are primarily interested inthe typically >1 ps diffusive orientational component of theresponse, we intentionally stretch the initially 40 fs laser pulsesto 0.7 ps (fwhm) to allow for more polarizing, higher powers tobe used without damaging the films.28−30 Furthermore, toincrease the sample path length, the films were drop-cast andthermally annealed (rather than spin-cast); the resulting free-standing films are optically clear with a thickness ofapproximately 80 ± 20 μm. Finally, to avoid two-photonabsorption of the 800 nm pump pulses as well as emitter
aggregation during drop-casting, we employed emitter-freePS:CA films for OHD-OKE experiments.The −1 to 25 ps OHD-OKE signal plotted on a log scale in
Figure 3a is dominated by an instantaneous electronic responsearound time zero, followed by a much weaker contributionfrom nuclear dynamics, which is shown more clearly on a linearscale in Figure 3b up to 100 ps. The nuclear response is ofprimary interest to this study. It is immediately evident fromFigure 3a,b that no longer-time signal is present in PS-onlyfilms, while in PS films doped with CA, slower exponentialcomponents are present, indicating that the CA molecules areindeed mobile and that their alignment by the linearly polarizedpump pulse leads to a transient birefringence that disappears asthe dopants rotationally randomize. The PS-only trace onlydeviates slightly from the IRF, indicating a fast-relaxing (≪1ps) inertial component due to pump-induced distortions in thepolymer matrix. Both the 5 and 10% CA traces exhibit double-exponential behavior at times > 1 ps, with a short componenton the order of τ = 2 ps and a longer component on the orderof τ = 14 ps (Table 1; see Figure S4 and Table S1 for thespread of values across samples). The longer time scale, tooslow to represent unhindered orientational relaxation of a smallquasi-spherical molecule like CA, can be readily assigned tohindered reorientation due to the confining environment of theglassy polymer matrix. On the basis of an extensive body ofwork on anisotropic relaxation times of nonwetting liquidsconfined in nanoporous sol−gel glasses,31−36 we assign theshorter component to rotational relaxation of CA moleculesthat are far enough away from the polymer pore surfaces toremain relatively unhindered, though it may also include
Figure 3. OHD-OKE data for PS:CA from 0 (black), 5 (red), and 10% CA (blue) free-standing films. (a) Normalized OHD-OKE traces from −1 to25 ps, along with the IRF measured by the pump-induced birefringence in a 0.1 mm quartz slide. (b) Un-normalized OHD-OKE traces from 2 to100 ps. Each trace is offset by 0.2 a.u. for clarity. (c) CV(t) calculated according to eqs 1−4 in the main text. For comparison, the Stokes shiftcorrelation function for the 440 nm ESA observed in the TA of PS:2PXZ-OXD:CA(5%) is plotted (open circles). The Stokes shift correlationfunction is calculated as S(t) = [ν(t) − ν(∞)]/[ν(0) − ν(∞)],1 where ν(∞)is the peak frequency of the ESA after 100 ps and ν(0) is taken as thepeak frequency of the ESA band from PS:2PXZ-OXD:CA(0%) at early time delays. The calculated S(t) and CV(t) for 5% CA films extracted fromTA and OKE correspond well to each other. (d) Total OHD-OKE signal (from the sum of the fit exponential amplitudes to the nuclear dynamics)measured in PS:CA(0, 2.5, 5, 7.5, 10%) shown in red histograms, overlaid with the estimated percentage of singly occupied polymer pores ascalculated from Poisson statistics (black squares; see the text for details).
The Journal of Physical Chemistry Letters Letter
DOI: 10.1021/acs.jpclett.7b01689J. Phys. Chem. Lett. 2017, 8, 4183−4190
contributions from collision-induced interactions.26 Further-more, although the relative contributions of the two anisotropyrelaxation time scales may change with temperature, thesechanges are not expected to follow Arrhenius behavior inamorphous glasses because trapped excess free volume remainsrelatively constant over a wide temperature range.37
In order to relate the observed polarization anisotropyrelaxation time scales to dipolar solvation, we use a well-knownprocedure that relates the solvation correlation function CV(t)to the dipole−dipole correlation function C(2)(t) obtained bynumerical integration of the nuclear coordinate response r(t)extracted from the OHD-OKE signal O(t)26,38−41
≈ αC t C t( ) [ ( )]V(2) /3s (1)
where
∫∫
≈ −′ ′
′ ′∞C t
r t t
r t t( ) 1
( ) d
( ) d
t
(2) 0
0 (2)
Here, r(t) is obtained by taking the imaginary component ofthe Fourier transform of the OHD-OKE signal O(t)deconvolved from the IRF (assumed to be equivalent to thelaser autocorrelation function G(2)(t))42−44
= −−⎪ ⎪
⎪ ⎪⎧⎨⎩
⎡⎣⎢⎢
⎤⎦⎥⎥⎫⎬⎭
r tO t
G tH t t( ) 2 Im
( ( ))
( ( ))( )1
(2) 0(3)
where H(t) is the Heaviside step function.αs from eq 1 is the dipole density of the solvent medium,
obtained from40
α πρμε
= −−
⎜ ⎟⎛⎝⎜
⎞⎠⎟⎛⎝
⎞⎠k T
43
11
s
2
B
1
(4)
where ρ is the CA number density (assumed to be statisticallydistributed throughout the film), μ is the CA electric dipolemoment, and ε is the electric permittivity of the medium,obtained from ref 7 for different CA concentrations. We obtainvalues of αs = 2.8 for 5% CA and 5.2 for 10% CA films. Thepower law exponent αs/3 relating the solvation time correlationfunction to the dipole−dipole correlation function in eq 1 is anappropriate approximation in the case that r(t) is dominated byorientational dynamics but breaks down if intermolecularinteractions contribute significantly to the signal.39
Figure 3c displays the calculated CV(t) for PS:CA films atdoping concentrations of 5 (red curve) and 10% (blue curve).For comparison, the Stokes shift correlation function, S(t),1
extracted from the blue shift kinetics of the ∼440 nm ESAobserved in the TA of PS:2PXZ-OXD:CA(5%), is plotted as afunction of time (see Figure S5 for additional detail). Althoughthe latter is noisy due to the total blue shift being only ∼2 nm,it is clear that the evolution of CV(t) for PS:CA(5%) is verysimilar to the blue shift dynamics observed in the TA. Ignoringthe characteristic26,38,39 Gaussian-like contribution in the firstpicosecond, CV(t) can be satisfactorily modeled by abiexponential function with τ1 = 2.5 ± 0.1 ps and τ2 = 24.6± 0.2 for PS:CA(5%) and with a triexponential function with τ1= 0.6 ± 0.1 ps, τ2 = 3.2 ± 0.1 ps, and τ3 = 13.2 ± 0.2 ps forPS:CA(10%). S(t) for PS:2PXZ-OXD:CA(5%) displayed inFigure 3c can also be modeled with a double-exponentialfunction (Figure S5), yielding τ1 = 4 ± 2 ps and τ2 = 22 ± 9 ps.
These lifetimes are summarized in Table 1 along with thosefrom the raw TA and OKE data.Notably, at a 5% CA doping concentration, there are
remarkable similarities between the solvation response CV(t)calculated from the dynamics of the polarization anisotropymeasured using OHD-OKE and the components ascribed toexcited-state stabilization from the TA data, which cantherefore be conclusively assigned to solvation. The keyconclusion from these observations is that CA indeed acts tostabilize the excited states of 2PXZ-OXD embedded in theglassy polymer matrix in a manner closely equivalent tosolvent−solute interactions in liquids, on a time scale thatreflects partially hindered motion.The picture is less clear at 10% CA concentrations, where
CV(t) does not follow the TA dynamics (Table 1). Moreover,while one might expect a larger anisotropy in the 10% CA filmthan that in the 5% CA film (due to a larger number of alignedmolecules in the probed volume), we observe the opposite: theamplitudes of the exponential components are consistentlysmaller in films with 10% than those with 5% CA (Figures 3band S2). To elucidate this behavior, we measure the OHD-OKE signal in 2.5% CA concentration intervals from 0 to 10%PS:CA films of the same thickness to within ±10 μm (FigureS6). We use a sum of the amplitudes of the two exponentialcomponents needed to fit the OHD-OKE response after 1 pspump−probe delay as a relative measure of the initial nuclearpolarization anisotropy created by the pump pulse in the films;the results are plotted with red bars in Figure 3d. Beyond 5%dopant concentration, the maximum achievable polarizationdrops. We hypothesize that this saturation effect signifies thatthe CA molecules become increasingly hindered at the higherCA concentrations and are thus not as readily aligned by thepump pulse.More specifically, this saturation effect can be explained by
how CA molecules are distributed in the PS matrix’s pores. Thefree volume fraction in a PS film is approximately 5.7% at roomtemperature with an average pore radius of 2.88 Å or a volumeof ∼100 Å3.45 For comparison, the volume of a CA molecule is∼82 Å3. Using the number densities of pores and CA moleculesin our films, we estimate the distribution of CA molecules perpore using Poisson statistics. The black squares in Figure 3dindicate the percentage of polymer pores that are statisticallypredicted to be occupied by a single CA molecule, peaking at37% for 5 wt % CA before rolling over for higher loading.These predictions assume that no aggregation occurs duringdrop-casting, although due to the solution-phase deposition thesolid state system is kinetically trapped out of equilibrium,making it difficult to characterize with equilibrium thermody-namics. Because the CA mainly resides within polymer pores,higher concentrations of CA imply overcrowding of pores, evenif the CA does not manifest attractive interactions that wouldinduce aggregation. The trends in Figure 3d therefore suggestthat one explanation for the observed saturation is purelymorphological. Because the average PS pore possesses a similarvolume to a CA molecule, it is likely that the observedanisotropy in the OHD-OKE signal is dominated by singlyoccupied pores. Slower, more severely hindered reorientationsare still likely to occur for CA in multiply occupied pores athigher CA concentrations, but these molecules statisticallycontribute less to the overall signal and are thus morechallenging to measure (see examples of persistent anisotropydespite low signal-to-noise ratio in Figure S7).
The Journal of Physical Chemistry Letters Letter
DOI: 10.1021/acs.jpclett.7b01689J. Phys. Chem. Lett. 2017, 8, 4183−4190
As a result of the CA-to-free-volume saturation effect, the10% CA TA data cannot be readily compared to the OHD-OKE data. Although the excited-state dynamics of the emitterare at least partly influenced by solvation from CA dipolereorientation, additional processes likely arise due to CAmolecules in more crowded environments where incompletereorientation and collision-induced contributions dominate theOHD-OKE signal. Because of these contributions, thecorrespondence between the polarization anisotropy correla-tion function and the solvation time correlation function breaksdown. On the shorter few-ps time scale, there is a reasonabledegree of correspondence between the short 4.3 ps componentin TA and the intermediate 3.2 ps component of CV(t). Thesecomponents likely arise due to relatively unhindered CArotation in singly occupied pores. Although this similarityindicates that liquid-like solvation still occurs at 10% dopingconcentrations, the contribution of dipolar rotation to solvationis reduced compared to that in films at 5% dopingconcentrations. More constrained, incomplete dipole reorienta-tion can still stabilize the energy of the emissive state but willalso lead to strong local electric fields that can lead to spectralshifts of individual molecules. Such inhomogeneous environ-ments would be reflected in a broadening of site energies as CAconcentration increases,7 as observed in steady-state measure-ments of the fluorescence spectra of 2PXZ-OXD in PS as afunction of CA concentration (Figure S8).18
The apparent reliance of SSS on the intricate interplaybetween nanoscale polymer morphology and dopant concen-tration in PS suggests that SSS and its saturation could occurmore generally in other different polymer systems. Wetherefore assess its potential generality using three morpho-logical criteria. First, the polymer free volume ratio sets anupper bound on the dopant concentration prior to saturation ofindividual pores by multiple dopants. Second, the averagepolymer pore volume within which SSS dopants must berotationally mobile limits the dopant molecule size and quantitythat can reside within each pore. A third parameter is the spatialdistribution of SSS dopants and solvated emitters, in particular,the number of SSS dopants within Coulombically coupled“solvation shells” surrounding the emitter (Figure S9). Weassembled a list of known free volume ratios and pore radii for arange of polymers (Table S2 and associated text) to assess towhat extent our findings should generalize to other systems.Because PS possesses a relatively small average pore radius andfree volume ratio, many other polymer matrices appearhospitable to mobile SSS “solvent” molecules. We also estimatethe ideal SSS dopant concentration for CA and two otherpotential polar dopants (Table S3) for a subset of the polymersin Table S2. We compute that, given a random distribution ofPS pores, the ideal CA concentration for SSS is 4.5−6 wt %, inagreement with our measurements. Furthermore, at 6 wt %concentration of 2PXZ-OXD, we estimate that approximately17 CA molecules reside within 2 nm of each 2PXZ-OXDmolecule (Figure S9). We anticipate that for polymers withmuch higher free volumes, such as those with twisted backboneconfigurations that prevent effective packing (e.g., poly-(trimethylsilyl propyne)), much higher doping ratios can beachieved prior to saturation, thus maximizing SSS interactionsand the range over which SSS can tune optoelectronicproperties. Given the large range of polymers with similar orlarger free volume ratios and pore radii than PS (Table S2), webelieve that SSS will find applications in a variety of polymer-based optoelectronic devices. While we focus on amorphous
polymers, the same principles should translate to semicrystal-line and crystalline materials. For example, the highlyrotationally mobile interstitial organic cations in hybridorganic−inorganic perovskites can be considered SSS-activemoieties that may respond to changes in free carrierdistribution in the material.8,9
Overall, we have shown that SSS by small-molecule dopantsin amorphous polymer films can be directly observed on itsnative time scales using time-resolved spectroscopies that trackboth the excited-state dynamics of embedded emitters and thepolarization anisotropy created by rotationally mobile dopants.We found that, at doping concentrations that optimize thepolymer free volume per dopant molecule, SSS is intrinsicallyanalogous to liquid-state solvation, and the same modelsrelating polarization anisotropy and dipolar solvation can beused to describe its dynamic evolution. The results presented inthis study suggest a strategy to use SSS as a dynamic probe ofmicrostructure and heterogeneity in molecular materials and toexploit it to finely tune the energy landscape of optoelectronicdevices using the ample body of knowledge generated fromsolution-state studies in the past six decades. It is alsointeresting to note that a new paradigm is emerging for solid-state materials whereby the presence of polar dynamic disordercan lead to defect-tolerant devices.8,9,17,46−49 Polymer devicescharacterized by large static disorder may benefit from the useof SSS to add a dynamic degree of freedom that could shieldexcitons and charge carriers from an inherently large density oftraps. For example, large-amplitude dielectric fluctuations couldeffectively flatten the spatioenergetic landscape of the materialto enhance neutral or charged excitation transport in complexoptoelectronic materials. Active, spatially varying optical controlof dopant orientation could even more optimally direct the flowof these excitations to further manipulate and enhance energytransport and transduction.
■ EXPERIMENTAL METHODSSample preparation procedures and the experimental setup aredescribed in more detail in the Supporting Information.All films were prepared in a sealed glovebox with a N2
atmosphere (<2 ppm of H2O, < 10 ppm of O2). TernaryPS:2PXZ-OXD:CA films for TA were spin-cast from chloro-form solutions and encapsulated in order to minimize exposureto air and moisture during measurements. PS:CA films forOHD-OKE measurements were drop-cast from chloroformsolutions and then thermally annealed in a vacuum oven. Theresulting free-standing films were brittle and optically clear.OKE traces using 40 fs pulses on nonannealed films reveal thecharacteristic C−Cl bending Raman mode (263 cm−1) ofchloroform through a prominent oscillatory feature with a∼130 fs period;43 however, annealed films did not show anyoscillatory feature, confirming that no residual solventcontaminates the signal (Figure S10).Ultrafast spectroscopy is performed using a Ti:sapphire
regenerative amplifier delivering 5W, 5 kHz, 40 fs pulsescentered at 800 nm. For TA spectroscopy, the pump is thefrequency-doubled (400 nm) output, and the probe is a whitelight supercontinuum generated in a CaF2 crystal. The pumpand probe are set at magic angle (54.7°) polarization withrespect to each other. The pump energy is 120 nJ/pulse (200μm diameter). For OHD-OKE, the 800 nm output is split intopump and probe, which are set at 45° polarization with respectto each other before being focused to 300 and 200 μm spots,respectively, in the sample. The probe is then spatially filtered
The Journal of Physical Chemistry Letters Letter
DOI: 10.1021/acs.jpclett.7b01689J. Phys. Chem. Lett. 2017, 8, 4183−4190
and passes through an achromatic quarter waveplate and aWollaston prism,50 thus splitting the probe beam into paralleland perpendicular components, which are simultaneously sentto a balanced photodetector that suppresses common-modenoise by up to 40 dB. The difference output from the detectoris sent to a lock-in amplifier referenced to the pump choppingfrequency (2.5 kHz). In order to obtain a higher signal-to-noiseratio for long-lived anisotropies (>1 ps, i.e., slow orientationalrandomization), the pulses are stretched to 0.7 ps fwhm bytuning the compressor grating in the regenerative amplifier.When stretched pulses are used, the pump power can beincreased up to 20 μJ/pulse without inducing sample damage.
■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.jpclett.7b01689.
Detailed sample preparation and experimental methods;kinetic traces for film TA; solution-state TA; additionalOHD-OKE data on several samples and spots todetermine the spread of values; detail of the TA blueshift kinetics and associated correlation function; OHD-OKE data for 2.5 and 7.5% CA films; long-lived (>20 ps)OHD-OKE decay traces observed in some higher-concentration films; emission energy and line widthsfor PS:2PXZ-OXD:CA films at different CA concen-trations; design principles for SSS; OHD-OKE tracestaken with 40 fs pulses of annealed and nonannealedfilms; computational methods; and calculated excited-state energies (PDF)
■ AUTHOR INFORMATIONCorresponding Author*E-mail: [email protected] S. Ginsberg: 0000-0002-5660-3586Present Address∇R.N.: Department of Chemistry, University of Utah, Salt LakeCity, UT 84112, United States.NotesThe authors declare no competing financial interest.
■ ACKNOWLEDGMENTSThis work has been supported by The Dow ChemicalCompany under Contract #244699. B.L.C. and T.D.R.acknowledge National Science Foundation Graduate ResearchFellowships (DGE 1106400). N.S.G. acknowledges an Alfred P.Sloan Research Fellowship, a David and Lucile PackardFoundation Fellowship for Science and Engineering, and aCamille and Henry Dreyfus Teacher−Scholar Award. Wewould like to thank Prof. Adam Sturlaugson, Prof. David Jonas,Dr. Michael Bishop, Boris Spokoyny, and Aaron Goodman fortechnical advice and Prof. David Limmer for comments on themanuscript. We would also like to acknowledge Prof. S.Ramasesha’s group for the partial development of the ZINDOcode.
■ REFERENCES(1) Jimenez, R.; Fleming, G. R.; Kumar, P. V.; Maroncelli, M.Femtosecond Solvation Dynamics of Water. Nature 1994, 369, 471−473.
(2) Maroncelli, M.; Fleming, G. R. Picosecond Solvation Dynamics ofCoumarin 153: The Importance of Molecular Aspects of Solvation. J.Chem. Phys. 1987, 86, 6221−6239.(3) Ben-Naim, A.; Marcus, Y. Solvation Thermodynamics ofNonionic Solutes. J. Chem. Phys. 1984, 81, 2016−2027.(4) Stratt, R. M.; Maroncelli, M. Nonreactive Dynamics in Solution:The Emerging Molecular View of Solvation Dynamics and VibrationalRelaxation. J. Phys. Chem. 1996, 100, 12981−12996.(5) Tomasi, J.; Mennucci, B.; Cammi, R. Quantum MechanicalContinuum Solvation Models. Chem. Rev. 2005, 105, 2999−3094.(6) Greaves, S. J.; Rose, R. A.; Oliver, T. A. A.; Glowacki, D. R.;Ashfold, M. N. R.; Harvey, J. N.; Clark, I. P.; Greetham, G. M.; Parker,A. W.; Towrie, M.; et al. Vibrationally Quantum-State-SpecificReaction Dynamics of H Atom Abstraction by CN Radical inSolution. Science 2011, 331, 1423−1426.(7) Madigan, C. F.; Bulovic, V. Solid State Solvation in AmorphousOrganic Thin Films. Phys. Rev. Lett. 2003, 91, 247403.(8) Zhu, H.; Miyata, K.; Fu, Y.; Wang, J.; Joshi, P. P.; Niesner, D.;Williams, K. W.; Jin, S.; Zhu, X.-Y. Screening in Crystalline LiquidsProtects Energetic Carriers in Hybrid Perovskites. Science 2016, 353,1409−1413.(9) Bakulin, A. A.; Selig, O.; Bakker, H. J.; Rezus, Y. L. A.; Muller, C.;Glaser, T.; Lovrincic, R.; Sun, Z.; Chen, Z.; Walsh, A.; et al. Real-TimeObservation of Organic Cation Reorientation in MethylammoniumLead Iodide Perovskites. J. Phys. Chem. Lett. 2015, 6, 3663−3669.(10) Chang, W.; Akselrod, G. M.; Bulovic, V. Solid-State Solvationand Enhanced Exciton Diffusion in Doped Organic Thin Films underMechanical Pressure. ACS Nano 2015, 9, 4412−4418.(11) Bulovic, V.; Shoustikov, A.; Baldo, M. A.; Bose, E.; Kozlov, V.G.; Thompson, M. E.; Forrest, S. R. Bright, Saturated, Red-to-YellowOrganic Light-Emitting Devices Based on Polarization-InducedSpectral Shifts. Chem. Phys. Lett. 1998, 287, 455−460.(12) Bulovic, V.; Deshpande, R.; Thompson, M. E.; Forrest, S. R.Tuning the Color Emission of Thin Film Molecular Organic LightEmitting Devices by the Solid State Solvation Effect. Chem. Phys. Lett.1999, 308, 317−322.(13) Roslyak, O.; Birman, J. L. Hybridized Quadrupole-DipoleExciton Effects in a Cu2O-Organic Heterostructure. Phys. Rev. B:Condens. Matter Mater. Phys. 2007, 75, 245309.(14) Green, A. P.; Butler, K. T.; Buckley, A. R. Tuning of theEmission Energy of Fluorophores Using Solid State Solvation forEfficient Luminescent Solar Concentrators. Appl. Phys. Lett. 2013, 102,133501.(15) Ahn, S.; Kim, J. N.; Kim, Y. C. Solid State Solvation Effect of aDonor−acceptor Type Fluorescent Molecule and Its Application toWhite Organic Light-Emitting Diodes. Curr. Appl. Phys. 2015, 15,S42−S47.(16) Bernardo, B.; Cheyns, D.; Verreet, B.; Schaller, R. D.; Rand, B.P.; Giebink, N. C. Delocalization and Dielectric Screening of ChargeTransfer States in Organic Photovoltaic Cells. Nat. Commun. 2014, 5,3245.(17) Niesner, D.; Zhu, H.; Miyata, K.; Joshi, P. P.; Evans, T. J. S.;Kudisch, B. J.; Trinh, M. T.; Marks, M.; Zhu, X.-Y. Persistent EnergeticElectrons in Methylammonium Lead Iodide Perovskite Thin Films. J.Am. Chem. Soc. 2016, 138, 15717−15726.(18) Cotts, B. L.; McCarthy, D. G. D. G.; Noriega, R.; Penwell, S. B.;Delor, M.; Devore, D. D.; Mukhopadhyay, S.; De Vries, T. S. T. S.;Ginsberg, N. S. Tuning Thermally Activated Delayed FluorescenceEmitter Photophysics through Solvation in the Solid State. ACS EnergyLett. 2017, 2, 1526−1533.(19) Haseyama, S.; Niwa, A.; Kobayashi, T.; Nagase, T.; Goushi, K.;Adachi, C.; Naito, H. Control of the Singlet−Triplet Energy Gap in aThermally Activated Delayed Fluorescence Emitter by Using a PolarHost Matrix. Nanoscale Res. Lett. 2017, 12, 268.(20) Lee, J.; Shizu, K.; Tanaka, H.; Nomura, H.; Yasuda, T.; Adachi,C. Oxadiazole- and Triazole-Based Highly-Efficient ThermallyActivated Delayed Fluorescence Emitters for Organic Light-EmittingDiodes. J. Mater. Chem. C 2013, 1, 4599.
The Journal of Physical Chemistry Letters Letter
DOI: 10.1021/acs.jpclett.7b01689J. Phys. Chem. Lett. 2017, 8, 4183−4190
(21) Uoyama, H.; Goushi, K.; Shizu, K.; Nomura, H.; Adachi, C.Highly Efficient Organic Light-Emitting Diodes from DelayedFluorescence. Nature 2012, 492, 234−238.(22) Noriega, R.; Barnard, E. S.; Ursprung, B.; Cotts, B. L.; Penwell,S. B.; Schuck, P. J.; Ginsberg, N. S. Uncovering Single-MoleculePhotophysical Heterogeneity of Bright, Thermally Activated DelayedFluorescence Emitters Dispersed in Glassy Hosts. J. Am. Chem. Soc.2016, 138, 13551−13560.(23) Santos, P. L.; Ward, J. S.; Data, P.; Batsanov, A.; Bryce, M. R.;Dias, F.; Monkman, A. P. Engineering the Singlet-Triplet EnergySplitting in a TADF Molecule. J. Mater. Chem. C 2016, 4, 3815−3824.(24) dos Santos, P. L.; Ward, J. S.; Bryce, M. R.; Monkman, A. P.Using Guest−Host Interactions To Optimize the Efficiency of TADFOLEDs. J. Phys. Chem. Lett. 2016, 7, 3341−3346.(25) Zhang, Q.; Li, B.; Huang, S.; Nomura, H.; Tanaka, H.; Adachi,C. Efficient Blue Organic Light-Emitting Diodes Employing ThermallyActivated Delayed Fluorescence. Nat. Photonics 2014, 8, 326−332.(26) Smith, N. A.; Meech, S. R. Optically-Heterodyne-DetectedOptical Kerr Effect (OHD-OKE): Applications in Condensed PhaseDynamics. Int. Rev. Phys. Chem. 2002, 21, 75−100.(27) Hunt, N. T.; Jaye, A. A.; Meech, S. R. Ultrafast Dynamics inComplex Fluids Observed through the Ultrafast Optically-Hetero-dyne-Detected Optical-Kerr-Effect (OHD-OKE). Phys. Chem. Chem.Phys. 2007, 9, 2167.(28) Gottke, S. D.; Brace, D. D.; Hinze, G.; Fayer, M. D. TimeDomain Optical Studies of Dynamics in Supercooled O -Terphenyl:Comparison to Mode Coupling Theory on Fast and Slow Time Scales.J. Phys. Chem. B 2001, 105, 238−245.(29) Gottke, S. D.; Brace, D. D.; Cang, H.; Bagchi, B.; Fayer, M. D.Liquid Crystal Dynamics in the Isotropic Phase. J. Chem. Phys. 2002,116, 360−367.(30) Sturlaugson, A. L.; Fruchey, K. S.; Lynch, S. R.; Aragon, S. R.;Fayer, M. D. Orientational and Translational Dynamics of Polyether/Water Solutions. J. Phys. Chem. B 2010, 114, 5350−5358.(31) Farrer, R. A.; Fourkas, J. T. Orientational Dynamics of LiquidsConfined in Nanoporous Sol−Gel Glasses Studied by Optical KerrEffect Spectroscopy. Acc. Chem. Res. 2003, 36, 605−612.(32) Loughnane, B. J.; Fourkas, J. T. Geometric Effects in theDynamics of a Nonwetting Liquid in Microconfinement: An OpticalKerr Effect Study of Methyl Iodide in Nanporous Glasses. J. Phys.Chem. B 1998, 102, 10288−10294.(33) Farrer, R. A.; Loughnane, B. J.; Fourkas, J. T. Dynamics ofConfined Carbon Disulfide from 165 to 310 K. J. Phys. Chem. A 1997,101, 4005−4010.(34) Loughnane, B. J.; Scodinu, A.; Fourkas, J. T. Extremely SlowDynamics of a Weakly Wetting Liquid at a Solid/Liquid Interface: CS2Confined in Nanoporous Glasses. J. Phys. Chem. B 1999, 103, 6061−6068.(35) Loughnane, B. J.; Farrer, R. A.; Scodinu, A.; Reilly, T.; Fourkas,J. T. Ultrafast Spectroscopic Studies of the Dynamics of LiquidsConfined in Nanoporous Glasses. J. Phys. Chem. B 2000, 104, 5421−5429.(36) Scodinu, A.; Farrer, R. A.; Fourkas, J. T. Direct Observation ofDifferent Mechanisms for the Inhibition of Molecular Reorientation ata Solid/Liquid Interface. J. Phys. Chem. B 2002, 106, 12863−12865.(37) White, R. P.; Lipson, J. E. G. Polymer Free Volume and ItsConnection to the Glass Transition. Macromolecules 2016, 49, 3987−4007.(38) Chang, Y. J.; Castner, E. W. Fast Responses from “‘slowlyRelaxing’” Liquids: A Comparative Study of the FemtosecondDynamics of Triacetin, Ethylene Glycol, and Water. J. Chem. Phys.1993, 99, 7289−7299.(39) Castner, E. W.; Maroncelli, M. Solvent Dynamics Derived fromOptical Kerr Effect, Dielectric Dispersion, and Time-Resolved StokesShift Measurements: An Empirical Comparison. J. Mol. Liq. 1998, 77,1−36.(40) Maroncelli, M.; Kumar, V. P.; Papazyan, A. A SimpleInterpretation of Polar Solvation Dynamics. J. Phys. Chem. 1993, 97,13−17.
(41) Cho, M.; Rosenthal, S. J.; Scherer, N. F.; Ziegler, L. D.; Fleming,G. R. Ultrafast Solvent Dynamics: Connection between TimeResolved Fluorescence and Optical Kerr Measurements. J. Chem.Phys. 1992, 96, 5033−5038.(42) Lotshaw, W. T.; McMorrow, D.; Thantu, N.; Melinger, J. S.;Kitchenham, R. Intermolecular Vibrational Coherence in MolecularLiquids. J. Raman Spectrosc. 1995, 26, 571−583.(43) McMorrow, D.; Lotshaw, W. T.; Kenney-Wallace, G. A. FsOptical Kerr Studies on the Origin of the Nonlinear Response inSimple Liquids. IEEE J. Quantum Electron. 1988, 24, 443−454.(44) McMorrow, D.; Lotshaw, W. T. Intermolecular Dynamics inAcetonitrile Probed with Femtosecond Fourier-Transform RamanSpectroscopy. J. Phys. Chem. 1991, 95, 10395−10406.(45) Liu, J.; Deng, Q.; Jean, Y. C. Free-Volume Distributions ofPolystyrene Probed by Positron Annihilation: Comparison with Free-Volume Theories. Macromolecules 1993, 26, 7149−7155.(46) Shimamura, K.; Hakamata, T.; Shimojo, F.; Kalia, R. K.;Nakano, A.; Vashishta, P. Rotation Mechanism of MethylammoniumMolecules in Organometal Halide Perovskite in Cubic Phase: An AbInitio Molecular Dynamics Study. J. Chem. Phys. 2016, 145, 224503.(47) Ziffer, M. E.; Ginger, D. S. How Hybrid Perovskites Get TheirGroove. Science 2016, 353, 1365−1365.(48) March, S. A.; Riley, D. B.; Clegg, C.; Webber, D.; Liu, X.;Dobrowolska, M.; Furdyna, J. K.; Hill, I. G.; Hall, K. C. Four-WaveMixing Reveals Weaker Many-Body Interactions and SlowerDephasing in Perovskite Photovoltaic Materials than GaAs. ACSPhotonics 2017, 4, 1515−1521.(49) Bischak, C. G.; Hetherington, C. L.; Wu, H.; Aloni, S.; Ogletree,D. F.; Limmer, D. T.; Ginsberg, N. S. Origin of ReversiblePhotoinduced Phase Separation in Hybrid Perovskites. Nano Lett.2017, 17, 1028−1033.(50) Giraud, G.; Gordon, C. M.; Dunkin, I. R.; Wynne, K. TheEffects of Anion and Cation Substitution on the Ultrafast SolventDynamics of Ionic Liquids: A Time-Resolved Optical Kerr-EffectSpectroscopic Study. J. Chem. Phys. 2003, 119, 464−477.
The Journal of Physical Chemistry Letters Letter
DOI: 10.1021/acs.jpclett.7b01689J. Phys. Chem. Lett. 2017, 8, 4183−4190
