Polaron photoconductivity in the weak and strong light-matter coupling regime Nina Krainova,1 Alex J. Grede,1 Demetra Tsokkou,2 Natalie Banerji,2 and Noel C. Giebink1 1Department of Electrical Engineering, The Pennsylvania State University, University Park, PA, 16802, U.S.A. 2Department of Chemistry and Biochemistry, University of Bern, Bern, CH-3012, Switzerland We investigate the potential for cavity-modified electron transfer in a doped organic semiconduc- tor through the photocurrent that arises from exciting charged molecules (polarons). When the polaron optical transition is strongly coupled to a Fabry-Perot microcavity mode, we observe po- laron polaritons in the photoconductivity action spectrum and find that their magnitude depends differently on applied electric field than photocurrent originating from the excitation of uncoupled polarons in the same cavity. Crucially, moving from positive to negative detuning causes the up- per and lower polariton photocurrents to swap their field dependence, with the more polaron-like branch resembling that of an uncoupled excitation. These observations are understood on the basis of a phenomenological model in which strong coupling alters the Onsager dissociation of polarons from their dopant counterions by effectively increasing the thermalization length of the photoexcited charge carrier. The notion that optical environment can change the pendence changes systematically with the polariton de- nature of a chemical reaction emerges in the strong cou- tuning, leading us to conclude that these observations pling regime when molecular electronic (or vibrational) reflect a genuine change in the underlying photoinduced transitions hybridize with light to form polariton states electron transfer process that occurs following the exci- that have different energies, coherence, and dynami- tation of each species. cal characteristics than the bare (uncoupled) molecules As in previous work [14], we study the organic themselves [1–5]. Dubbed polariton chemistry [6, 7], a semiconductor 4,4’-cyclohexylidenebis[N,N -bis(4- growing body of work has now established that it is in- methylphenyl)benzenamine] (TAPC) doped with MoO3. deed possible to alter the rate or yield of certain photo- The high electron affinity of MoO3 induces ground state chemical reactions [1, 8, 9](and possibly some reactions electron transfer from the highest occupied molecular in the dark [10–12]) by strong coupling the participant orbital (HOMO) of TAPC, yielding a large density of molecules to an optical microcavity or plasmon mode. holes (i.e. TAPC cations) that absorb strongly with Photoinduced charge transfer is arguably one of the a peak at ∼ 1.8 eV and shoulder at ∼ 2.1 eV (Fig. most important areas to explore such modification since 1a) that are due to excitation of the hole downward it underlies processes ranging from photolithography to into lower-lying molecular orbitals [14]. The same photosynthesis. In this case, strong coupling is predicted transitions are observed in the photocurrent external to alter both the free energy driving force of the reaction quantum efficiency (EQE) spectrum shown in Fig.1a for (dependent on the polariton energy) and its reorgani- a sandwich-type device consisting of indium-tin-oxide zation energy due to polaron decoupling (where polari- (140 nm)/30 vol% MoO3:TAPC (160 nm)/Ag (20 nm) tons preserve the ground state nuclear configuration)[4]. as illustrated in the inset. There, the magnitude of the Whether measurable changes in electron transfer rate can EQE spectrum (measured under chopped illumination be observed in practice is, however, an open question using lock-in detection) increases with bias and its shape since any such changes must compete with the extremely remains constant, independent of the bias polarity. The short (< 100 fs) polariton lifetime [13] in typical metal dashed lines in Fig. 1b show the component peaks that microcavity and plasmon systems. Moreover, given that make up the polaron EQE spectrum superimposed on the optical properties of polaritons in most organic ma- an exponential background thought to be associated terials are well described classically based on their nat- with photocurrent generation from the absorption tail ural dielectric function, it is reasonable to ask whether of neutral TAPC [15] (i.e. due to exciton dissociation). chemically-distinct polariton states truly emerge. Note that the photocurrent measurements are conducted Here, we explore these questions by measuring the pho- at low temperature to minimize the background dark toconductivity of a p-doped organic semiconductor in current flowing in the highly doped film (its conductivity which charged, rather than neutral molecules (i.e. po- is thermally activated)[16, 17]. The EQE data at room laronic rather than excitonic states) are strongly cou- temperature are qualitatively similar except the polaron pled to a Fabry-Perot microcavity. We confirm the exis- features are slightly broader and the overall magnitude tence of polaron polariton modes in the photoconduc- of the spectra is roughly two orders of magnitude higher. tivity action spectrum and find that their magnitude Polaron photoconductivity of this sort has not pre- evolves differently with applied bias than photocurrent viously been studied in organic semiconductors but is that originates from exciting uncoupled polarons in the probably analogous to that in traditional inorganic pho- same cavity. Importantly, the difference in functional de- toconductors such as Hg-doped Ge [18], where photon ab- 2 (a) 0.4 (a) − 5 ± 0.7V 10 0.2 ± 0.5 V Absorption 0.0 ± 0.3 V 2 − 6 EQE 10 ) 3 V ± 0.1 V -5 1.5 0.3 V ×12 1 Silver EQE (×10 0.5 MoOʥ:TAPC (b) 200 ITO Glass 0 150 Silver 1.7 1.8 1.9 2.0 2.1 2.2 MoO :TAPC Energy (eV) 3 (b) Aluminum + Chrome excited TAPC 125 Glass hole φ (º) HOMO 50 hv <1 ps 0 Energy lower-lying 1.6 1.8 2.0 2.2 MOs Energy (eV) bound TAPC hv FIG. 2. (a) EQE magnitude spectra obtained from the mi- crocavity shown in inset of (b) at varying negative (dashed) r and positive (symbols) biases. Light is normally incident on 0 the cavity and the data are collected at a temperature of 100 K. (b) Corresponding phase of the photocurrent relative to MoO3 the chopper reference signal at 389 Hz. The device structure is shown in the inset. FIG. 1. (a) Photocurrent external quantum efficiency (EQE) spectra measured at T=100 K for a 160 nm-thick, 30% MoO3:TAPC film using the sandwich-type device structure shown in the inset. The shape of the EQE spectrum remains measurements shown in the Supplementary Material [20], constant from low (0.3 V) to high (3 V) bias and consists of which includes Ref. [21]), only one or two such excited two peaks (the dashed lines are the result of a multipeak fit) state hops can realistically occur, leaving the relaxed hole that correspond with those observed in the polaron optical only slightly farther from its counterion than where it absorption spectrum of the bare film (measured at T=100 started. The resulting picture is then very similar to K accounting for transmission and reflection) shown in the a traditional Onsager-type dissociation process [22, 23], top panel. (b) Schematic of the polaron photoconductivity with drift/diffusive escape from the counterion Coulomb mechanism. potential beginning from an initial ’thermalization’ dis- tance, r0, as indicated by the blue arrows in Fig. 1b. Photocurrent in the strong coupling regime is subse- sorption liberates a trapped hole from a shallow acceptor quently explored by replacing the indium-tin-oxide con- dopant (Figure 1b). In the case of MoO3-doped TAPC, tact with an optically-thick 100 nm Al mirror and main- the vast majority of holes at room temperature and below taining the MoO3:TAPC layer thickness at d = 160 nm are trapped in the weakly-screened Coulomb potential to achieve a zero-detuned microcavity with respect to of the MoO3 counterions [16, 17]. This is evident from the peak of the main polaron transition. Strong coupling the fact that the hole concentration inferred from electri- is verified by angle-dependent reflectivity measurements cal conductivity measurements (≤ 1018 cm−3)[19] is typ- shown in the Supplementary Material [20], which yield ically orders of magnitude lower than that inferred from upper and lower polariton (UP and LP, respectively) the polaron absorption coefficient (∼ 1020 cm−3)[15], par- modes with a vacuum Rabi splitting of ~Ω ≈ 0.25 eV. ticularly at low temperature [16]. Thus, the majority of polaron absorption is due to bound, rather than free Figure 2a and 2b respectively show the magnitude and holes. Exciting a bound hole to a lower-lying molec- phase (ϕ) of the photocurrent spectrum measured for ular orbital can liberate it through one or more hops this sample at an incident angle of θ = 0◦, close to the (i.e. electron transfer events) in the excited state as il- anti-crossing point in the dispersion. In contrast to the lustrated by the red arrows in Fig.1b. However, because weakly-coupled case in Fig. 1, the shape of the spectra excited state hopping competes with ultrafast relaxation in Fig. 2a changes dramatically with bias, evolving from back to the HOMO (< 1 ps based on transient absorption a single peak near the bare polaron energy at low bias 3 (a) 200 and diffusion (shown at the top of each plot), as expected since the dynamics of the dissociation process are orders φ (º) φ 0 of magnitude faster than the chopping frequency. The −5 10 0.7 V apparent increase in diffusion current magnitude at high 0.5 V biases is likely an artifact associated with the difficulty −6 10 0.3 V inherent in attempting to extract a small difference from EQE two large numbers (i.e.