Mapping the Density of States Distribution of Organic Semiconductors by Employing Energy Resolved–Electrochemical Impedance Spectroscopy

Although the density of states (DOS) distribution of charge transporting states in an organic semiconductor is vital for device operation, its experimental assessment is not at all straightforward. In this work, the technique of energy resolved–electrochemical impedance spectroscopy (ER‐EIS) is employed to determine the DOS distributions of valence (highest occupied molecular orbital (HOMO)) as well as electron (lowest unoccupied molecular orbital (LUMO)) states in several organic semiconductors in the form of neat and blended films. In all cases, the core of the inferred DOS distributions are Gaussians that sometimes carry low energy tails. A comparison of the HOMO and LUMO DOS of P3HT inferred from ER‐EIS and photoemission (PE) or inverse PE (IPE) spectroscopy indicates that the PE/IPE spectra are by a factor of 2–3 broader than the ER‐EIS spectra, implying that they overestimate the width of the distributions. A comparison of neat films of MeLPPP and SF‐PDI2 or PC(61)BM with corresponding blends reveals an increased width of the DOS in the blends. The results demonstrate that this technique does not only allow mapping the DOS distributions over five orders of magnitude and over a wide energy window of 7 eV, but can also delineate changes that occur upon blending.


DOI: 10.1002/adfm.202007738
consequence, charge transport in OSs is slowed down as compared to that in coun terpart perfect molecular crystals. It is well established that charge carriers move via incoherent hopping within a density of state (DOS) distribution. [8][9][10][11] The broader the DOS is, the lower is the charge carrier mobility and the higher is the associated activation energy as well as the time after which a dynamic process equilibrates, such as the motion of a sheet of charge carriers injected from an electrode. Map ping the DOS to determine the energies of the electron and hopping transporting states is therefore crucial for material characterization. [12,13] Disorder in OSs is manifested in the broadening of their absorption and photo luminescence spectra of OSfilms. It reflects (i) the local variation of the van der Waals coupling of a singlet or triplet state to the polarizable environment, [8] (ii) structural variations of a chromophore, for example, the variation length of the effective conjugation length of a conjugated polymer, [14] and (iii) dynamic effects such as thermally activated rotational or vibrational motion within the chromophore. [15,16] Such contributions toward energetic dis order will also affect the DOS of valence and conduction states that control hole and electron motion. Moreover, DOS distribu tions are likely to change when going from neat to blended films. A further source of broadening of the tail of an absorption of a

Introduction
Organic semiconductors (OSs) are the key active element in todays' photocopiers and are gaining increasing importance in optoelectronic devices such as organic lighting emitting diodes, [1][2][3] organic solar cells (OSCs) [4][5][6] and organic field effect transistors. [7] For practical reasons they are amorphous or polycrystalline films. This implies structural disorder. As a semiconductor arises when absorption occurs not from the 0-0 vibrational state but-thermally activated-from phonon coupled states. This gives rise to exponential Urbach tails of the absorp tion spectra. [17,18] However, since the energies of Urbach tails are typically around a few meV only, this effect is relevant only in fairly ordered semiconductors such as molecular or inorganic crystals, or hybrid materials such as perovskites [19] and would otherwise be buried under static and dynamic level broadening.
Unfortunately, the DOS of valence and conduction states in OSs is not amenable to direct optical absorption as it is case in inorganic semiconductors. The reason is that-owing to the weak electronic coupling in OSs-photoexcitation creates exci tons rather than charge carriers, so that the direct transition from the valence (HOMO) DOS to the conduction (LUMO) DOS is not observed. A simple way to estimate the degree of dis order in OSs is to measure the temperature dependence of hole and electron transport. However, this yields only upper values for the widths of relevant DOSs because both dynamic disorder and structural relaxation may contribute to broadening, beyond the already existing static distribution in the DOS. [10,11] More over, it does not provide information on the detailed structure of the DOS. An alternative method toward DOS mapping is photoemission (PE) [20] and inverse PE (IPE). [21,22] It is experi mentally demanding and data analysis may be complicated because electrons emitted from different layers of the OS have slightly different energies since the polarization energies of molecules next to vacuum are diminished. [23,24] In this work we apply the technique of energy resolvedelectrochemical impedance spectroscopy (EREIS) to determine the DOS of selected OSs. EREIS is a novel electrochemical impedance technique bordering on voltammetry. [25,26] Usually, the organic semiconductor probed by voltammetry is dissolved in solution. In an EREIS measurement, in contrast, the OS is deposited as a film, where solid state effects prevail (Figure 1). Thus, the molecules are probed in the local environment of the film, so that the EREISspectrum is a direct reflection of the DOS of the hole (HOMO) and electron (LUMO) transporting states. In the current work we will use MeLPPP and P3HT as a donortype conjugated polymer and PCBM and SFPDI 2 as rep resentative electron acceptors in the form of neat films as well as in blends. We will demonstrate that the EREIS technique does not only allow to map the DOS functions over as many as five decades but also to delineate changes which occur upon blending. We find that the examined DOS distributions are usually of Gaussian character but in case of PCBM there is an exponential tail.

The ER-EIS Experiment
EREIS is a spectroscopic method to map the electronic struc ture of an organic solid in the contact with an electrolyte via a redoxreaction. [27][28][29][30] It evolved out of the electrochemical impedance spectroscopy and advanced general voltammetric techniques such as square wave voltammetry [31][32][33] or other reported voltammetric modulation approaches. [31,34] We briefly outline its operational principle and refer to refs. [26,28] for fur ther details. For the measurement, a 3electrode electrochem ical cell is used. Thus, a thin film, typically about 100 nm, of the OS is deposited by spincoating onto a conducting substrate, in our case indiumtinoxide (ITO) covered glass or doped Si. The OS film is covered by a liquid electrolyte that is contained by an inert, insulating frame. An Ag/AgCl reference and a Pt auxiliary wire electrode are inserted into the electrolyte, while the ITO or Si serves as working electrode. A DC voltage ramp between reference and working electrode is swept, modulated by an AC voltage of suitably chosen frequency, and the resulting cur rent is recorded. The measured impedance Z meas is a result of the Helmholtz layer that forms at the electrolyte/OS interface when a voltage is applied. Reversible charge transfer from ions of the electrolyte to countercharges in the OS in the vicinity of the interface will occur once the applied voltage compensates the difference between the energies of the relevant states, and this gives rise to the real component of Z meas . Under steady state conditions this interfacial recombination current has to be balanced by a current flowing through the OS that carries an exit contact, in our case the ITO or doped Si electrode. A crucial condition is that the rate limiting step must be charge exchange at the electrolyte/OS interface rather than the charge transport toward to the exit contact. This implies that the voltage drop across the OS bulk be negligible relative to the voltage required to drive the injection at the interface. In turn, this requires a percolating network of transport states and implies charge transport via drift and diffusion under space charge limited (SCL) conditions.
The setup for a EREIS measurement may appear similar to that of an electrochemical fieldeffect transistor. [35] In both cases, the organic semiconductor film is covered by an elec trolyte with a Ag/AgCl reference electrode and a Pt auxiliary electrode inserted into it. However, there are several differences between the two methods. In the electrochemically gated field effect transistor, the recorded signal is typically the direct cur rent flow from source to drain electrode while a potential is applied at the gate electrode. In these measurements, counte rions that diffuse into the semiconductor from the electrolyte play a role. The electrochemical impedance measurement, in contrast, is conducted using an alternating current signal such as to prevent significant counterion effects. Further measures to avoid intercalation of ions include measuring the whole spectrum in two steps, always in a new electrochemical cell, with one measurement being used to obtain the branch for hole transporting states and one for the electron transporting states, as detailed in refs. [26][27][28].
Experimentally, the charge transfer resistance at the elec trolyte/OS interface is measured by superimposing a peri odic perturbation of the applied potential. Upon scanning the applied voltage one can assess the energetic position, and thus is the differential charge transfer resistance and is the differential spacecharge capacity in the bulk of the polymer. [28] L is the sample thickness, S is the area of the working electrode, and [A] is the concentration of the OS redox species dissolved in the solvent. V f is the volume frac tion of respective charge carriers near the surface with an effective thickness of the acceptor/donor layer next to the interface of typically 1-2 nm. [36] The k et (E) is the electron transfer rate. [37] The use of the chargetransfer resistance data R ct (E F ) (Equation (2)) and the spacecharge capacitance data C sc (E F ) (Equation (3)) constitutes two complementary approaches, which probe different parts of the conjugated polymer film, both providing information on the DOS func tion g(E F ). The R ct reflects the redox process that takes place near the surface of the OS and the electrolyte whereas C SC traces the bulk of the conjugated polymer film. For the data presented below, we evaluated the chargetransfer resistance data. We confirmed that the same results are obtained for the DOS function when evaluating the differential spacecharge capacitance.

The P3HT DOS Probed by Photoemission and by ER-EIS
In order to determine the electrical gap of a P3HT film Deibel et al. [22] had measured the spectra of PE as well IPE. They provide reference information on the DOS distribution. It was straightforward to apply the EREIS technique to study P3HT and compare the results, thereby extending our earlier work. [25,27,38] In Figure 2a we present the spectrum we obtain from our impedance measurements for the DOS function g(E F ) while Deibel et al.'s results are shown in Figure 2b. For the sake of easy comparison the original PE and IPE spectra have been replotted on a semilogarithmic scale. The edges to the spectra have been fitted to Gaussian lineshapes (see Table 1). It is gratifying that the tails of the Gaussians approxi mately coincide (see also the dashed lines in Figure 2). How ever, the PE and IPE spectra are significantly broader and the separation between their maxima has increased from 3.18 to 3.9 eV. We also note the smaller dynamic range of the PE/IPE measurement.

The Donor Polymer MeLPPP with a Fullerene and a Non-Fullerene Acceptor
In order to characterize the DOS distributions of a conjugated donor polymer and, importantly, to delineate changes that can occur upon blending we use the laddertype MeLPPP. It is one of the least disordered polymers as evidenced by the narrow absorption and emission spectra. [16] This facilitates to uncover morphologyrelated changes of the DOSdistributions. As representative filmforming acceptors we use PC(61)BM and SFPDI 2 . Figure 3 shows the compilation of the EREIS spectra for the DOS function g(E F ) for the hole and electron transporting states of radical cation and anion states. We attribute the high Figure 2. The DOS function g(E F ) for HOMO and LUMO states of a P3HT film inferred from either a) ER-EIS or b) photoemission and inverse photoemission (data from Deibel et al. [22] ). The colored dashed lines indicate Gaussian fits with the solid colored line indicating the data points considered in the fit. The arrows indicate the center position of the Gaussians. The black dashed lines serve to ease comparison between DOS tails obtained from both methods. Table 1. The center energy E 0 and the standard deviation σ obtained from the Gaussian fits to the edges of the HOMO and LUMO attributed parts of the DOS function for a P3HT film. Also given is the energy gap Photo (low) energy edge of DOS function for the hole (electron) states to the HOMO (LUMO). The data derived from the spectra are summarized in Table 2. The spectra for a neat MeLPPP film are presented in Figure 3a. Both the parts of the DOS func tion attributed to the HOMO as well as the part attributed to the LUMO are of Gaussians extending over five decades in amplitude, recognizing, though, that it is principally difficult to distinguish whether a low energy tail has a Gaussian or an exponential shape. The standard deviations (σ) of HOMO and LUMOattributed g(E F ) functions are 50 and 60 meV (see Table 2).
The HOMOattributed part of the DOS function g(E F ) of a SFPDI 2 neat film is of Gaussian shape over two orders of magnitude but it carries a weak tail (Figure 3b) while the LUMOattributed part of the DOS function g(E F ) is also a per fect Gaussian with a standard deviation of 55 meV. The gap between the centers of the HOMO and LUMO attributed parts is 2.54 eV, that is, 0.17 eV higher than a literature value inferred from cyclic voltammetry. [39] Remarkably, the standard deviation of the optical absorption is about 150 meV, that is, significantly larger than the DOS values for the charge transporting states.
In Figure 3c we show the DOS function for a neat PC(61)BM film. Over two orders of magnitude the HOMOattributed part is of Gaussian shape with σ = 95 meV with an exponential tail while the LUMOattributed part is a pure Gaussian with σ = 65 meV. The electrical gap, defined by the separation between the maxima of the parts, is 2.6 eV.
In blends with MeLPPP and either PC(61)BM or SFPDI 2 the high energy edge of the lower energy part of g(E F ) is associated with the HOMO of the donor (MeLPPP) (Figure 3d,e)). The Gaussian character of the HOMO of MeLPPP is preserved but there is additional state broadening σ (70 and 63 meV instead of 50 meV) and tail states appear at the high energy side of the HOMO roughly 0.3 eV above the center of the bulkDOS. Their relative concentrations are roughly 0.001 (MeLPPP:SFPDI 2 ) and 0.01 (MeLPPP:PCBM). In the blend, the lowest energy feature in the higher energy part of g(E F ) is the LUMO of the acceptor, while the higher energy feature around −2 eV can be associated with the MeLPPP LUMO. In both cases there is a significant broadening of the LUMO of MeLPPP. Moreover, in the MeLPPP:PCBM blend, additional features can be discerned, centered at −3.60 and −3.10 eV, with σ = 75 and 110 meV, respectively, that that are barely visible in neat PC(61)BM (see also Figure 4b below for ease of comparison, and Supporting Information).

Comparison Between DOS Distribution Inferred from ER-EIS and Photoemission
The DOS distributions of HOMO and LUMO distributions of a neat P3HT film inferred from EREIS are of Gaussian shape with σ parameters of 63 meV (HOMO) and 168 meV (LUMO). The electrical gap is 3.18 eV and-since the energy of the singlet exciton is about 2.1 eV, the exciton binding energy is about 1.1 eV, that is, close to that of MeLPPP and classic molecular crystals. [24] The fact that the width of the LUMO distribution is unusually broad can be ascribed to the broader distribution of effective conjugation length in a P3HT film. [40] However, the PE and IPE spectra are significantly broader than the EREIS spectra, and the gap between the peak positions are increased from 3.18 to 3.9 eV. The observation that upon going from EREIS to PE/ IPE the HOMO distribution increases from 60 to 260 meV and the LUMO distribution increases from 170 to 360 meV demon strates that EREIS and PE/IPS monitor different phenomena. An EREIS experiment probes the electronhole transfer directly from an electrolyte to the OS. On the other hand, PE probes the ejection of a highly excited electron into the gasphase and IPE probes the dissipation of electrons upon entering the solid. [41] Therefore the respective response functions in PE and IPE are the convolutions of the DOS functions and the escape/dissipa tion functions. In the case of PE, electron ejection is likely to be affected by the electron scattering and the decrease of the polar ization energy of an electron at the interface between the OS and vacuum. [23] Both effects will broaden the response function and increase the apparent ionization energy. A PE spectrum is, therefore, unable to probe the width of the DOSdistributions for HOMO states. Studies on charge transport and theoretical studies support this reasoning. [12] Suppose the width of the HOMODOS distribution of a P3HT film were as broad as 260 meV, then the hole mobility would-based upon the Gaussian disorder model (ref. 8)-be about 10 −17 cm 2 Vs −1 , in striking disagreement with experiment. [42,43] There is indeed consensus that width of distributions of hole and electron trans porting states are typically around 100 meV. [44]

Neat Films
In all cases the DOS functions of HOMO and LUMO states shown in Figure 3a-c are either full Gaussians or Gaussians that extend into weak tails with a characteristic energy of about 100 meV, that is, at least one order of magnitude larger than those of typical Urbach tails. Provided that the DOS function indeed reflects the DOS, this confirms that the DOS widths are due to static and dynamic disorder. The simplest approach to analyze the DOS distributions is the pointsite concept. [8] It implies that either a point charge or a localized exciton polarize their environment via van der Waals coupling with coupling energies featuring an r −4 (charge) and r −6 (exciton) dependence on distance, respectively. Imposing a random distribution of the intermolecular separations translates into a distribution of Gaussian character because the polarization energies depend on a large number of coordinates each varying randomly and therefore the central limit theorem applies. It is easy to show that in this case the ratio of the standard deviations of the DOSs for charges (σ c ) and excitons (σ exc ) is σ c /σ exc = 4P c /6P exc where P c and P exc are the polarization energies of charges and exci tons, respectively.
A vapor deposited PC(61)BM film is a system in which the pointsite model may provide a reasonable estimate for ana lyzing the DOS distributions of charge carriers and excitons. The electron affinity (EA) of PC(61)BM in the gasphase and in the solid are 2.63-2.65 eV [45,46] and 3.84 eV, [47] respectively. Table 2. The center energy E 0 and the standard deviation σ obtained from the Gaussian fits to the edges of the HOMO and LUMO attributed DOS functions g(E F ) for neat films of MeLPPP, SF-PDI, 2 and PC(61)BM, as well as for the donor-acceptor blends. Also given is the energy gap E g = E 0 (HOMO) − E 0 (LUMO). For the blends, center energies to multiple Gaussian fits are given for the LUMO, corresponding to donor or acceptor or different acceptor phases (c.f. Figure 3d,  The difference between the EA values in the gasphase and the solid, that is, the polarization energy of a radical anion, is 1.2 eV like in many organic solids. [24] The value for the EA in solid PC(61)BM, −3.86 agrees well with the average value meas ured by different techniques. The standard deviations (σ) of the HOMO/LUMODOSs inferred from EREIS spectra are 95 and 90 meV, respectively. They are significantly lower than the value 130 meV that Tummals et al. [13] obtained from molecular dynamics simulations. On the other hand, the σ value for the singlet excitons in PC(61)BM, inferred from the delayed fluo rescence spectra at 295 K, is only 40 meV. [48] To be consistent with the point site model (see above) polarization energy of the singlet exciton in the solids had to be 350 meV. This is indeed the difference between the energy of the singlet state of the anthracene molecule in the gasphase (3.45 eV) [49] and in the solid (3.1 eV), [24] which can be taken as a reference value (in lack of corresponding data on PC(61)BM).
The width of the LUMODOS should be reflected in the temperature dependence of the electron mobility. From tem perature dependent SCL electron transport in PC(61)BM diodes Mihailetchi et al. [50] concluded that electron transport is consistent with the extended Gaussian disorder model and extracted a σ value of 77 meV. This value is 15% lower than the width of the LUMODOS. A possible reason for this discrep ancy is that in the SCL transport mode tail states of the LUMO DOS are partially filled. This should diminish the experimen tally determined width of the electron DOS. Moreover, the film investigated by Mihailetchi et al. and our film have not been prepared using precisely the same deposition conditions, so that the film morphology can vary slightly.
The HOMODOS of a PC(61)BM film is of Gaussian shape over only two orders of magnitude and features a broader tail at higher energies. Such exponential tails have occasionally been observed. [51] They are a signature of traps that exist in systems in which the intrinsic HOMODOS is beyond 6 eV. [52] The posi tion of the peak of the HOMODOS is 6.4 eV and is consistent with literature values. [53][54][55][56] Since the ionization energy in the gas phase is 7.59 eV, [57] the polarization energy of the radical cation in PBCM is close to 1.2 eV, that is, the same as the polari zation energy of the radical anion. This yields an electrical gap of 2.6 eV while PE and IPE yield 2.37 eV. [56] From the action spectrum of intrinsic photogeneration we obtained E g = 2.45 ± 0.05 eV. [58] Combined with the fact that hole transport in PCBM is traplimited we conjecture that the E g value inferred from the maxima of HOMOLOMO DOSs of EREIS spectra refer to the intrinsic gap while the electrical gap determined from the tail of PE as well as photoconduction spectra are affected by the contribution of hole traps. Summing up, we argue that the point site model provides a reasonable basis for analyzing the DOSs for exciton as well as charge transporting states in a small molecule system such as vapor deposited PCBM film.
Let us now consider a SFPDI 2 film. SFPDI 2 is a more extended molecule as compared to PC(61)BM. The LUMODOS of SFPDI 2 is a perfect Gaussian over 4 decades with a vari ance of 55 meV while the HOMODOS is somewhat broader (σ = 83 meV) and carries a weak tail. Remarkably the σ value for the absorption spectrum, (shown in the SI) is 0.15 eV. This demonstrates that the DOS distributions for charge carriers can be narrower than those of neutral excitations. Even this also indicates that in this case the pointsite model is unsuitable for estimating the DOS distribution for charge carriers based upon absorption or PL spectra because the exciton is more spread out than in a spherical molecule like PC(61)BM. [59] Next we turn to the EREIS spectra for the HOMO and LUMO DOS distributions of MeLPPP films. For MeLPPP the LUMO DOS is a perfect Gaussian with a standard deviation of 60 meV while the HOMODOS is a somewhat distorted Gaussian with 50 meV. The absorption and fluorescence spectra are also Gauss ians with a variance around 50 meV somewhat depending on film preparation. [60] It seems that the σ values for charge carriers and singlet excitons are comparable. The point site model is clearly inappropriate for estimating the ratio of the spectral widths. This is because in conjugated polymers, the main contribution of disorder broadening stems from local variation of conjugations. This conju gationinduced variation translates into the site energies. A grati fying test of internal consistency is that the separation between the centers of the HOMO and LUMODOSs is 3.8 eV and agrees with the threshold energy for intrinsic photogeneration. [61] Since the energy of the singlet exciton is 2.72 eV, the binding energy of exciton is about 1.2 eV like in conventional molecular solids.
In this context it is worth recalling earlier work on thermally stimulated luminescence (TSL) on thick films of MeLPPP. In such a TSL study one excites a sample at lower temperature to generate electronhole pairs. They are metastable because elec trons are deeply trapped. Upon raising the temperature from 5 K onward one observes delayed emission. It originates from the thermally activated release of holes from shallow intrinsic traps, that is, the HOMODOS, and subsequently recombines with trapped electrons. Therefore, the temperature depend ence of the TSL signal reflects the HOMODOS of the MeLPPP host. By analyzing the TSL signal the shape and the width of the HOMODOS are recovered. The result confirms that at low temperatures (≈50 K) the DOS of a thick MeLPPP film has a Gaussian shape with σ = 54 meV. [62] This is in favorable agree ment with the value of 50 meV inferred from the EREIS study.
We note that we do not observe any indication of trap states in the spectrum shown in Figure 3a. This is remarkable since electrontransport in MeLPPP has been found to be strongly traplimited, with trap concentrations expected to be around 10 18 cm −3 . [63] It is not fully clear why these traps are not evident in the EREIS spectra, and this is subjected to further investiga tion. Possibly this is related to the dominance of the counter balancing hole current in MeLPPP, or poor injection of charges directly into the trap states.

Blended-Films
It is instructive to compare the DOS distributions of MeLPPP:PCBM and MeLPPP:SFPDI 2 blends with those of the parent neat films ( Figure 4) 3) The weak, barely noticeable shoulder at −3.60 eV in the LUMO of PCBM becomes stronger in the blend with MeLPPP, so that it appears as a peak with the same intensity than the −3.85 eV peak of the LUMO in the neat PCBM. We tentatively assign the −3.85 eV feature to crystallike domains and the −3.60 eV feature to more disordered PCBM. The ra tionale behind this reasoning is that upon progressive order ing the ionization energy of a solid decreases. Formation of ordered domains of PCBM in bulkhetero junction OSCs in addition to more amorphous and possibly even intercalated structures is a wellestablished phenomenon. [64][65][66] In passing we note that the feature at −3.10 eV has also acquired more intensity compared to the neat PCBM film. Since this is more than half of an eV above the features attributed to LUMOs of differently ordered PCBM morphologies, we consider the −3.10 eV feature is more likely to arise from some different orbital.

Conclusions
Since that for PCBM both the ionization energy and the EA have been measured in the gasphase employing PE tech niques, the EREIS spectrum of a PCBM film provides values for the polarization energies of radical cations and anions as well as their standard deviations of the pertinent DOS distri butions. Combined with PL spectroscopy we conclude that the simple pointsite model, which rests upon the notion that disorder originates from of fluctuations of van der Waals coupling, is enough to understand disorder effects in PCBM. This is in contrast with films of conjugated polymers because there is an additional spreading of site energies due statistical variations of the effective conjugations length. The mutual consistence between different probes of disorder phenomena demonstrates that impedance spectroscopy is indeed a valu able technique to map the DOS distributions for charge car riers. Its particular advantage is that it allows mapping the DOS over five orders of magnitude. Its dynamic range is thus greater than that of PE spectroscopy and it is much cheaper and easier to employ. We find that in all systems we looked at, the central portion of measured HOMO and LUMO distributions are of Gaussian shape, occasionally carrying broader tails that are associated with extrinsic defects and can act as charge carrier traps. Impor tantly, the EREIS technique is able to interrogate the DOS dis tributions in neat films as well as in blends. It is therefore a method to probe changes in the HOMO/LUMO distributions upon blending, albeit subject to the condition that there is a percolation path of charges across the film. An example is the broadening of the LUMODOS of MeLPPP upon blending with PCBM.
By comparing DOS distributions of a P3HT film inferred from PE and electrochemical impedance measurements, respectively, we find that widths of the DOS distributions deter mined using PE and IPE are typically by a factor 2-3 larger than those inferred from EREIS spectra The likely reason is that in the former case there is additional spectral broadening in the course of electron ejection from the OS to vacuum (PE) and dissipation of the excess energy in the injected electron (IPE). Therefore, PE and IPE overestimate the disorder in an OS.

Experimental Section
For the ER-EIS method, the electrochemical microcells had a volume of about ≈200 µl. The active organic semiconductor was deposited on top of either an ITO covered glass or highly doped Si (n+ or p+) substrates with deposited organic semiconductor thin film. The solution of 0.1 m TBAPF6 in anhydrous acetonitrile was used as the supporting electrolyte. The active organic semiconductor electrode area was 12 mm 2 . The potential of the working electrode with respect to the reference Ag/AgCl electrode was controlled via a potentiostat. A Pt wire was used as the counter electrode. The potential recorded with respect to the reference Ag/AgCl electrode was recalculated to the local vacuum level assuming the Ag/AgCl energy versus vacuum value of 4.66 eV. An impedance/gain-phase analyzer, Solartron analytical, model 1260 (Ametek, Berwyn, USA), run in the usual three-electrode regime. The AC harmonic voltage signal frequency was usually 0.5 Hz, its rms value was 100 mV, and the sweep rate of the DC voltage ramp was 10 mV s −1 . Bode and Cole-Cole diagrams in the frequency range of 0.01-1 MHz were used for the preliminary ER-EIS frequency adjustment.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.