WorksPhysics and Astronomy
Electrical Scanning Probe Microscope Measurements Reveal Surprisingly High Dark Conductivity in Y6 and PM6:Y6 and Non-Langevin Recombination in PM6:Y6
Abstract
We used broadband local dielectric spectroscopy (BLDS), an electric force microscopy technique, to make non-contact measurements of conductivity in the dark and under illumination of PM6:Y6 and Y6 prepared on ITO and PEDOT:PSS/ITO. Over a range of illumination intensities, BLDS spectra were acquired and fit to an impedance model of the tip-sample interaction to obtain a sample resistance and capacitance. By comparing two descriptions of cantilever friction, an impedance model and a microscopic model, we connected the sample resistance inferred from impedance modeling to a microscopic sample conductivity. A charge recombination rate was estimated from plots of the conductivity versus light intensity and found to be sub-Langevin. The dark conductivity was orders of magnitude higher than expected from Fermi-level equilibration of the PM6:Y6 with the substrate, suggesting that dark carriers may be a source of open-circuit voltage loss in PM6:Y6.
Introduction
The power conversion efficiency of donor–acceptor solar cells has been increasing rapidly since the introduction of non-fullerene acceptors [1]. Solar cells built from the non-fullerene, small-molecule acceptor Y6, the polymer donor PM6, and related molecules, have shown consistently high efficiency [2, 3, 4, 5, 6, 7], reaching 19% power conversion efficiency recently [8]. It remains puzzling why the best donor–acceptor blends perform so well. Charge recombination in the best blends is 10’s to 1000’s of times slower than predicted by Langevin theory [7, 9, 10, 11, 12, 13] [14, 15, 16, 17, 18, 19, 20, 21, 22]. This anomalously slow recombination could be due to an improperly estimated local charge density or mobility [9]; charge trapping [10, 12] [23, 24]; the inhomogeneous nanoscale structure of the donor/acceptor interface [13] [15]; or a built-in electrostatic potential gradient [25]. Understanding the anomalous charge recombination can potentially reveal new opportunities for further improving efficiency. With an eye towards microscopically testing competing theories of non-Langevin recombination [15, 26–28], here we introduce a scanning probe measurement of conductivity in donor–acceptor thin films.
The measurement is an electric force microscopy (EFM) technique, broadband local dielectric spectroscopy (BLDS) [29–32]. BLDS was introduced by Labardi et al. to probe the frequency-dependent dielectric function of insulating polymers [29]. It was subsequently applied by Tirmzi and coworkers to examine photo-induced electronic and ionic conductivity in lead-halide perovskite films [30–32]. Tirmzi’s work was aided by a new theory for computing electric force microscope signals from the sample’s complex electrical impedance [33]. To more quickly identify better organic solar-cell materials, Menke et al. argue that donor–acceptor compounds should be screened using techniques that, in contrast with the widely used photoluminescence quenching measurements, probe charge concentration directly [18]. Such techniques include microwave conductivity [12], optical dielectric constant measurements [34], phase-kick electric force microscopy [35], and BLDS [30–32]. Here we apply BLDS to study conductivity in an organic photovoltaic material for the first time. We report studies of the conductivity of PM6:Y6, PM6, and Y6 (chemical structures shown in Figure 1a) films in the dark and under illumination.

Figure 1. A broadband local dielectric spectroscopy experiment probes photoconductivity in a donor–acceptor solar cell film. (a) Experimental setup, showing the structures of the PM6 and Y6 molecules studied here. (b) Applied tip–sample voltage and the resulting cantilever frequency shift versus time ; the sinusoidal on–off modulation is depicted as a square wave for simplicity. (c) Impedance model of the tip–sample interaction, with the applied tip–sample voltage, the tip capacitance, the sample capacitance, the sample resistance, and the sample impedance.
Results
Estimating resistance
The BLDS measurement is sketched in Figure 1. A metal-coated atomic-force microscope cantilever was brought near a sample surface, Figure 1a; the sample was illuminated from above with light; and the cantilever’s resonance frequency was recorded as a function of time. A voltage modulation of amplitude and oscillation frequency were applied to the cantilever, shifting the cantilever’s resonance frequency by an amount on average, Figure 1b. The voltage was additionally on-off modulated at 20 Hz to allow lock-in detection of . The 20 Hz Fourier component of the cantilever frequency, , is plotted versus voltage-modulation frequency to yield a BLDS spectrum, Figure 2. The spectrum is similar to that of an circuit, with the time constant light-dependent. Spectra were collected at various visible light intensities.

Figure 2. Representative broadband local dielectric spectra collected for PM6:Y6 (left), Y6 (middle), and PM6 (right) films with a PEDOT:PSS/ITO contact (top, red) or an ITO only contact (bottom, blue). Films were illuminated from above with a nm laser at the indicated intensity (right, legend). The data were fit to Equation (1) (lines).
BLDS spectra were collected for PM6:Y6, Y6, and PM6 prepared on ITO and PEDOT:PSS/ITO. Representative data are shown in Figure 2 and all data are shown in Figure S3–Figure S4. With the sample modeled as a resistor and capacitor operating in parallel, Figure 1c, these data were fitted to [30, 33]

Figure S3. All BLDS spectra collected for samples containing the hole transport layer, PEDOT:PSS. Numbers in the upper right corner of each spectra correspond to the cantilever number that was used to collect those data.

Figure S4. All BLDS spectra collected for samples without the hole transport layer. Numbers in the upper right corner of each spectra correspond to the cantilever number that was used to collect those data.
with the cantilever frequency; the modulation frequency; the cantilever spring constant; the tip–sample voltage; and , two derivatives derived from the tip–sample capacitance; and
a complex-valued transfer function function that depends on the tip capacitance and sample impedance . The transfer function relates the steady-state tip charge to the applied tip–sample voltage, .
It is convenient to define a sample response time and a tip–charge response time as and , respectively, with the sample resistance, the sample capacitance, and the tip capacitance. The Equation (2) transfer function can be written in terms of these charge response times as follows:
According to Equations (2) and (3), BLDS spectra have a roll-off frequency and high frequency plateau determined by the time constants and . The high frequency plateau, , gives information about the ratio of sample capacitance to tip capacitance. The roll-off frequency is . At fixed tip–sample separation, if then the conductivity is simply proportional to the roll-off frequency. This was the case in Reference 30. In contrast, here we find . In this limit we can nevertheless obtain by fitting the entire BLDS spectrum. The Figure 2 data were fit to Equations (1) and (3) with , , , and as fit parameters. All fit parameters at each light intensity can be found in Tables S4–S16.
Sample resistance was estimated by computing the ratio . To estimate the tip capacitance, the cantilever tip would usually be modeled as a sphere plus a cone. However, Hoepker et al. [36] showed that the cone capacitance does not contribute significantly to measured friction or frequency noise, so in our analysis we set , neglecting . We estimate for a radius located 120 nm over a semi-infinite ground plane. The estimated resistance of the PM6:Y6 and Y6 films is shown as a function of light intensity in Figure 3. Values for sample resistance and capacitance are listed in Tables S17–S20.

Figure 3. Sample resistance versus light intensity for all samples. Each symbol corresponds to a different physical sample. Red data points are samples that contain PEDOT:PSS, and blue data points are samples without PEDOT:PSS. Data were collected in triplicate. All error bars are .
Connecting macroscopic resistance to microscopic conductivity
We wish to compute sample conductivity from the resistance obtained from BLDS experiments. Conductivity is an intrinsic quantity that depends on sample and contact materials. Conductivity is the product of charge density and charge mobility , , with the unit of charge. Resistance is inversely proportional to sample conductivity,
with the proportionality constant, , having units of length. For a slab of area and length , . Because it depends on sample dimensions, the resistance is an extrinsic quantity.
In our experiment, depends on the effective area and thickness of the sample probed by the tip, which in turn depend on tip radius, tip–sample separation, and sample thickness. Computing for our experiment requires a microscopic theory of the BLDS signal. Describing the response of free charges and molecular dipoles in the sample to an oscillating tip charge in the BLDS experiment involves coupling a transport equation for free charges with Maxwell equations and the fluctuation–dissipation theorem; this work is beyond the scope of the present study. Lekkala, Marohn, and Loring, however, have developed a microscopic theory for friction over a semiconductor [37, 38], and Dwyer et al.’s impedance treatment of the tip–sample interaction [33], used above to interpret the BLDS spectrum, also yields an expression for cantilever friction. Let us therefore estimate by comparing the friction predicted by these two treatments.
Friction was calculated for a cantilever oscillating perpendicular to the surface of a semi-infinite semiconductor using Equations (8), (19), and (20) in Reference 38 and the parameters listed in Table 1. In the language of Reference 38, we computed (noncontact friction) for a Model I sample, with the dielectric constant of the semi-infinite substrate set to to mimic a metal. Computations were carried out by Numba-optimized Python code publicly available in the dissipationtheory package; this code was validated by comparing to a low-density analytical expansion and the friction versus charge density plots in Figure 7(b) and 9(b) of Reference 38. In Figure 4 we plot the friction versus charge density computed for the Table 1 sample. The sample mobility was taken to be the average of the electron and hole mobilities given for PM6:Y6 in Table 2. The friction rises at low charge density, reaches a maximum near a charge density of [38] , and decreases at high charge density. The presence of a friction maximum is in qualitative agreement with Dwyer’s model [30, 33].
| parameter | symbol | value |
| sample thickness | 110 nm | |
| dielectric constant | 3.4 | |
| charge density | to m | |
| charge mobility | cm V s | |
| cantilever resonance frequency | kHz | |
| tip radius | 38.4 nm | |
| tip–sample separation | 120 nm | |
| tip–sample voltage | 1.0 V |
Table 1. Sample and cantilever parameters used to compute the voltage-normalized cantilever friction using Model I in Reference 38.

Figure 4. Friction versus charge density. Friction was computed numerically using Model I in Reference 38 and the parameters in Table 1 (lower, circles). The computed friction was fit to Equation (7) (lower, lines; upper, fit residuals).
| sample | [cm V s] | [cm V s] | ref. |
| PM6:Y6 | [22] | ||
| PM6:Y6 | [39] | ||
| PM6:Y6 | [2] | ||
| PM6:Y6 | [40] | ||
| PM6:Y6 | [41] | ||
| Y6 | [39] |
Table 2. Literature estimates for mobilities used in the calculations below. The electron and hole mobility in PM6:Y6 was taken to be the average of the five reported values.

Figure 7. for PM6:Y6 and Y6 with PEDOT:PSS/ITO and ITO only contacts. Each symbol corresponds to a different physical sample. Red data points are samples that contain PEDOT:PSS, and blue data points are samples without PEDOT:PSS. Data were collected in triplicate. All error bars are .
Dwyer et al. predict [33]
with the cantilever frequency, the tip capacitance, the first derivative of the tip capacitance, the tip–sample voltage, and the transfer function given by Equations (2) and (3), computed assuming the tip–sample impedance model sketched in Figure 1(c). Both and were computed from an analytical formula for the capacitance of a sphere over a metallic half plane, Equations (52) and (53) in Reference 37 taken in the limit .
plane, Equations (52) and (53) in Reference 37 taken in the limit .
With given by Equation (4) and treating and as free parameters, the calculated versus curve in Figure 4 was fit to Equation (5). The fit was poor unless was allowed to be negative, which is unphysical. In other words, the fit required to be less than . Given this observation, let us write the capacitance as
with the proportionality constant, , having units of length. The Equation (6) parameterization is useful because it allows to be directly compared to the capacitance of a sphere of radius , ; if , then is less than the tip capacitance at infinite tip–sample separation. Substituting Equations (4) and (6) into Equation (5) we obtain the empirical expression
with the sample conductivity. The numerically calculated friction in Figure 4 was fit to Equation (7) assuming proportional errors of one percent. The fit result is shown as a solid line in Figure 4(bottom), with the fit residuals displayed above as a percentage error. The best-fit parameters are
Considering the simplicity of the Equation (7) ansatz, the global fit is remarkably good. Over ten decades of charge density and nearly three decades of friction, Equation (7) predicts the friction within 17% at low density and within 30% at high density. Conductivity was computed from the BLDS-inferred resistance using
In Figure 4, the conductivity is indicated as a second axis, and the Debye length
is shown above the residuals as a third axis. We will see below that most experiments were done in the high-conductivity region of Figure 4. In this regime, the Debye length is 10 nm or less, consistent with our approximation of treating the sample as a metal when computing and . The short Debye length implies a shallow region of accumulated or depleted charge at the sample surface, which can be modeled as a capacitor operating in parallel with , lowering the effective tip capacitance. The predicted lowering of the tip capacitance is consistent with the observed .
The charge recombination rate can be estimated from conductivity versus light intensity
With conductivity obtained from Equation (10), the versus light intensity data were plotted and fit to
with a slope and a dark conductivity, Figure 5. The dark conductivity is listed in the first column Table 3. From the Figure 5 data we can estimate the charge recombination rate and compare it to Langevin theory. We model the carriers in our sample following Burke et al. [15], who posited an equilibrium between free charges, charge-transfer states, and the ground state. The associated kinetic scheme is sketched in Figure 6. The coupled equations governing species concentrations, including background or dark carriers, are

Figure 5. Conductivity versus light intensity for PM6:Y6 samples. Each line/symbol corresponds to a different physical sample. Red data points are samples that contain PEDOT:PSS, and blue data points are samples without PEDOT:PSS. Data were collected in triplicate. All error bars are .
| dataset | dark conductivity [mS/m] PEDOT:PSS/ITO | dark conductivity [mS/m] ITO | reduction factor [unitless] PEDOT:PSS/ITO | reduction factor [unitless] ITO | background charge density [m] PEDOT:PSS/ITO | background charge density [m] ITO |
| ◦ | ||||||
| △ | ||||||
| □ | ||||||
| ♢ | — | — | — |
Table 3. Measured dark conductivity , estimated Langevin reduction factor , and background charge density for each PM6:Y6 sample. Symbols in the dataset column correspond to the symbols in Figure 5.

Figure 6. Kinetic scheme showing the rate of charges splitting , meeting , and recombining in an organic solar cell. Figure adapted from Reference 15.
where and are the background electron and hole concentrations, respectively; and are the change in electron and hole concentrations due to light; is the charge transfer (CT) state density; , , and are the rates at which carriers meet, split, and recombine, respectively; and is the charge-generation rate, proportional to the irradiation intensity and absorption coefficient. A more detailed explanation of how we arrived at these equations is given in Section S-4 of the SI. At steady state, . Comparing Equations (12a) and (15), we see that
at steady state. Plugging Equation (16) into Equation (12a) and setting the result equal to zero, since we are at steady state, yields
where is the Langevin rate, the rate at which carriers meet, and is the Langevin-reduction factor in the Burke picture [15].
The conductivity is given by
We see in Figure 5 and Table 3 that is non-zero, from which we conclude that background carriers are present in the dark. Let us assume for simplicity the background electron charge density is non-zero and that the background hole density is much less than the change in hole concentration due to light , i.e., . In the limit that , we can rearrange Equation (17) to obtain
In this limit
with , the charge-generation rate, [W m] the light intensity, [m] the absorption coefficient, [J] the energy per photon, and () [m V s] the electron (hole) mobility. Equation (20) predicts a conductivity that is linearly proportional to light intensity, consistent with the Figure 5 experiment.
Comparing Equations (12) and (20) we find
We obtained and from the best-fit and , the measured , known , and literature estimates for and (Table 2). From Equations 18a and 18b,
which can be compared to the Langevin rate [42]
with the relative dielectric constant (3.5, [22]) and the vacuum permittivity. For PM6:Y6 and Y6, we calculate to be m s and m s, respectively. Dividing Equation (23) by (24) we get the Langevin reduction factor,
Estimating the dark conductivity
Let us estimate the charge density expected in Y6 and PM6 near the ITO and PEDOT:PSS interfaces. The energy levels in a molecular semiconductor like Y6 and
PM6 follow a Gaussian distribution. The hole and electron density can be computed by multiplying a Gaussian density of states by the Fermi–Dirac distribution and integrating over all possible energies:
In eq:22, , with Boltzmann’s constant and temperature; is the molecular density, and are the mean energy and energetic disorder of the lowest unoccupied and highest unoccupied molecular orbitals, respectively; and is the electron chemical potential in the molecular film. Relevant parameters are listed in Table 4. The table also lists the Fermi level, . For simplicity we have assumed . In this limit,
| property | Y6 | PM6 | ITO | PEDOT:PSS |
| [eV] | ||||
| [eV] | ||||
| [eV] | ||||
| [meV] | ||||
| [] |
Table 4. Literature energy-level parameters and molecular density for Y6 and PM6 and Fermi level for ITO and PEDOT:PSS [3] [22, 43, 44, 45, 46, 47]. See Tables S1 and S2 for individual references. was calculated based on Reference 48.
for undoped Y6 and PM6. When a Y6 or PM6 molecule is brought near a contact, the molecule and contact will reach a common chemical potential or Fermi level. To assess the sign of the charge transfer, it is helpful to compute . When is positive, electrons will flow from the contact to the molecule, whereas when is negative, electrons flow from the molecule to the contact. Charge densities and were computed from Equations 22 with , using the parameters in Table 4, and a total charge density computed. Results are summarized in Table 5.
| molecule | contact | [mV] | [] |
| Y6 | ITO | ||
| PM6 | ITO | ||
| Y6 | PEDOT:PSS | ||
| PM6 | PEDOT:PSS |
Table 5. Computed chemical potential difference and charge density for various organic–metal contacts.
Discussion
PM6:Y6 was found to be highly conductive, even in the dark. The conductivity increased linearly with light intensity, on both PEDOT:PSS/ITO and ITO, indicative of PM6:Y6 being a good solar cell material.
All Y6 samples had much lower conductivity than the blend, indicating that the blend is a better solar cell material, as expected. Some Y6 samples showed modest photoconductivity, while in most samples the conductivity was dominated by background carriers. The conductivity was highly variable, with one sample showing almost no conductivity; this lack of conductivity might be due to material degradation, since the reagents were older as that sample was prepared about two months after the other samples. These observations are in qualitative agreement with Sağlamkaya et al., who showed that charge generation readily occurs in neat Y6 [49].
The PM6 BLDS spectra were independent of modulation frequency. The conductivity roll-off must therefore be outside our measurement limits, either below 100 Hz or above 3 MHz. Based on the moderate conductivity observed in the PM6:Y6 blend, it seems unphysical that the PM6 control would have higher conductivity than the blend. We conclude that must be less than the lowest conductivity measured in our experiment, .
Values obtained for in each PM6:Y6 sample are listed in Table 3. For PM6:Y6/PEDOT:PSS/ITO, we find to and to . These values are in order-of-magnitude agreement with obtained by Hosseini et al. [22] for a Ag/PDINO/PM6:Y6/PEDOT:PSS/ITO solar cell using bias-assisted charge extraction measurements and obtained by Zhang et al. [7] in a Ag/PFNDI/PM6:Y6/PEDOT:PSS/ITO solar cell using transient photovoltage and photocurrent measurements.
The large random error in our determination of arises primarily from the relatively poor fits of the BLDS spectra. In our fits we assumed that the sample dielectric constant was frequency-independent, which is likely a poor approximation for the small molecule and polymeric semiconductors studied here. To determine the proportionality constant connecting the best-fit bulk resistance to a microscopic conductivity, we compared a macroscopic theory to a microscopic theory for friction, and assumed that the same proportionality constant held in a BLDS measurement. Future work should focus on developing a proper microscopic theory for the BLDS measurement that incorporates a frequency-dependent dielectric constant.
We observed that was independent of light intensity, as seen in Figure 7. Given that all experiments were carried out at fixed tip–sample separation, was constant. We conclude that was must also be independent of light intensity. Since decreased with increasing light intensity, must be likewise decreasing. In the organic photovoltaic films studied here, light changed , not as is universally assumed in EFM experiments [35, 50, 51]. The standard description of EFM simply ignores the sample resistance. By acquiring BLDS spectra and interpreting them using an impedance model of the tip–sample interaction, we were able to quantify both the resistance and capacitance of our thin-film sample.
We found that if data were collected more than a week after PM6:Y6 or Y6 samples were prepared and stored in a nitrogen box in the dark, they did not display the same high conductivity as was measured immediately after preparation; PM6:Y6 and Y6 samples need to be measured within a day or two after being prepared. Consistent with this observation, Zhu et al. found that nonfullerene acceptors can aggregate in bulkheterojunction films, decreasing device performance, just by storing in the dark in a nitrogen-filled glove box [52].
Although every precaution was taken to ensure that sample preparation was consistent, we saw distinct differences in the dependence of conductivity on light intensity, Figure 5. The reproducibility of organic photovoltaic devices between labs, and within labs, remains poor [53–55]. Our study suggests that BLDS measurements can be a useful non-destructive tool for monitoring, and perhaps improving, the consistency of film conductivity prior to fabricating a full solar cell.
Conclusions
We report new evidence for significant dark conductivity in PM6:Y6 and corroborate recent studies indicating non-Langevin recombination in this material. The underlying concentration of dark carriers is many orders of magnitude larger than expected from Fermi-level equilibration. Sources of the anomalous dark carriers could be chemically oxidized/reduced PM6 or Y6 present due to reactions of these molecules with air or impurities present due to imperfect synthesis, or interface dipoles absent from our model of Fermi-level equilibration.
That the observed dark conductivity is comparable to the light-induced conductivity implies that either the electron or hole pseudo-Fermi level is partially pinned by the associated dark carriers in PM6:Y6. Such pseudo-Fermi level pinning due to dark carriers is a potential source of voltage loss and deserves further study.
Methods
The polymer donor PM6 (Ossila) and the molecular acceptor Y6 (Ossila) were stored in a nitrogen glove box and used as received within 8 months of receipt. A small amount was removed from the glove box to prepare each solution in air. Chloroform (Macron Fine Chemicals) and 1-chloronaphthalene (Sigma-Aldrich) were used as received.
PM6:Y6 samples were prepared on both ITO and PEDOT:PSS/ITO substrates as follows. The PEDOT:PSS preparation was adapted from Reference 56, with spin coating done statically instead of dynamically. The PM6:Y6 preparation followed Reference 3 as closely as possible. ITO coated glass slides (10 /sq., Nanocs) were cleaned by sonicating in isopropanol and acetone (1:1 volume ratio) for 10 minutes, followed by scrubbing with detergent (Aquet Liquid Laboratory Detergent) in DI water, rinsing with DI water, and drying with N. Slides were then UV-ozone cleaned (UVO-Cleaner Model No. 12, Jelight Company Inc.) for 10 minutes. PEDOT:PSS (AI 4083, Ossila) was filtered through a 0.45 m PTFE filter and 100 L was statically spin coated at 4000 rpm for 30 s. Films were annealed on a hot plate at 150 °C for 20 min in air and left in a nitrogen flow box to cool before spin coating the active layer. PM6:Y6 (1:1.2 weight ratio, 16 mg mL) in chloroform:1-chloronaphthalene (CF:CN, 99.5:0.5 volume ratio) was stirred for 3 h using a new PTFE stir bar prior to statically spin coating 100 L at 3000 rpm for 60 s. Control samples were prepared with PM6, 7.3 mg mL in CF:CN, and Y6, 8.7 mg mL in CF:CN. All solutions were stirred in air in the dark. Three to four replicate samples were made for each formulation. Samples were kept in the dark in a nitrogen box until they were loaded into the microscope under red light. Sample thickness, measured by profilometry (Tencor AlphaStep 500), was 110 nm, 40 nm, and 80 nm for PM6:Y6, Y6, and PM6, respectively.
Scanning probe measurements were performed under high vacuum ( mbar) in a custom-built scanning Kelvin probe microscope. The cantilever used (HQ:NSC18/Pt conductive probe, MikroMasch) had a typical resonance frequency kHz, force constant N m, tip radius nm, and cone angle . Specific cantilever parameters used in each Figure 2 experiment are given in Table S3. Data were collected at a tip–sample separation of 120 nm. Cantilever motion was detected using a fiber optic interferometer operating at nm (Corning model SMF-28 Ultra fiber; Applied Optoelectronics Inc. model DFB-1310-BF-10-A3-FA laser; New Focus Model 2053-FC photodetector). The sample was illuminated from above with a variable intensity continuous wave nm diode laser (QPhotonics QFLD-635-30SAX), and the estimated intensity at the sample was 0 to 154 mW cm (details on the laser spot size measurement are given in Section S-2 in the SI).
| Cantilever Number | [kHz] | [nm] | Q |
| 1 | 60.437 | 226.27 | 20202 |
| 2 | 63.670 | 157.44 | 19999 |
| 3 | 62.149 | 135.41 | 27778 |
Table S3. Details of all cantilevers (MikroMasch HQ:NSC18/Pt) used for the described experiments.
The cantilever was driven into self oscillation via positive feedback to an amplitude of 100 to 200 nm. Tip–sample separation was determined by first approaching the sample surface until the amplitude decreased to 80% of its initial value and then backing up to the desired separation, 120 nm, using a Thorlabs piezo controller. Tip–sample separation was checked before and after each BLDS spectrum was collected. If the tip–sample separation drifted more than 3 nm, the minimum DC step in our microscope, the spectrum was discarded.
In a BLDS measurement, the tip voltage was sinusoidally on–off modulated at a fixed frequency, 20 Hz, and sinusoidally modulated at frequencies ranging from Hz to 3 MHz at various light intensities. The cantilever frequency was measured using a commercial phase-locked loop (RHK Technology PLL-Pro), and the 20 Hz Fourier component of the cantilever frequency, , was obtained from the phase-locked loop output using lock-in detection (Perkin Elmer Instruments 7265 DSP Lock-in Amplifier). The lock-in output was divided by , with V, and the resulting voltage-normalized frequency shift was plotted versus the modulation frequency to give a BLDS spectrum.
ACKNOWLEDGMENTS
Research primarily supported by the National Science Foundation under Award DMR-2113994 (studies of organic semiconductors, electric force microscope development) with additional funding from the U.S. Department of Energy, Office of Science, Basic Energy Sciences, under Award DE-SC0022305 (electric force microscope development and calibration). The authors acknowledge the use of facilities and instrumentation supported by the National Science Foundation through the Cornell University Materials Research Science and Engineering Center DMR-1719875. Software used in this work includes, in part, h5py [57], lmfit [58], Matplotlib [59], Numba [60], NumPy [61], pandas [62, 63], Pint, SciPy [64], and SymPy [65].
AUTHOR CONTRIBUTIONS
R.L.C. carried out the experiments and data analysis. C.A.P. designed the microscope probe head, implemented microscope improvements, and measured the visible light spot size. V.E.M. implemented microscope improvements. J.A.M. executed friction simulations and conceived and directed the project. R.L.C., C.A.P., and J.A.M. wrote the manuscript, and all authors reviewed the manuscript.
FULL CHEMICAL NAMES
PM6:
Poly[(2,6-(4,8-bis(5-(2-ethylhexyl-3-fluoro)thiophen-2-yl)-benzo[1 2-b:4,5-b’]dithiophene))-alt-(5,5-(1’,3’-di-2-thienyl-5’,7’-bis(2-ethylhexyl)benzo[1’,2’-c:4’,5’-c’]dithiophene-4,8-dione)]
Y6:
2,2’-((2Z,2’Z)-((12,13-Bis(2-ethylhexyl)-3,9-diundecyl-12,13-dihydro-[1, 2, 5]thiadiazolo[3 4-e]thieno-[2”,3”:4’,5’]thieno[2’,3’:4,5]pyrrolo[3 2-g]thieno-[2’,3’:4,5]thieno[3 2-b]indole-2,10-diyl)bis(methanylylidene))-bis(5,6-difluoro-3-oxo-2,3-dihydro-1H-indene-2,1-diylidene))dimalononitrile
PEDOT:PSS:
Poly(3,4-ethylenedioxythiophene) polystyrene sulfonate
Visible Laser Spot Size and Light Intensity
The samples were illuminated from above using a variable intensity fiber-coupled red nm continuous wave diode laser (QPhotonics QFLD-635-30SAX). The laser output was directed to the sample through a 50 m core diameter, 0.22 NA multimode fiber (Thorlabs FG050LGA). The cleaved end of the fiber was placed 11 mm away from the sample and directed toward the cantilever tip at a 23° angle relative to the sample surface. The laser spot size was measured using a CMOS sensor, utilizing a method similar to the literature [1]. In brief, a Raspberry Pi camera module (Seeed Studio model 114992442 with a Sony IMX477 CMOS sensor) was loaded and approached as a sample would be. With the laser output set to 13 W, an image of the laser spot (Figure S1) was captured and saved as an array. The lmfit Python package [2] was used to fit the short axis of the elliptical spot to a Gaussian and the long axis (due to the 23° angle of incident light) to a skewed Gaussian. The SciPy Python package’s [3] signal module was used to determine both the width and the full width at half max (FWHM) of each fit. The percent of the total light contained within each spot was estimated by integrating the fits using the Simpson function from SciPy’s integrate module. We chose to use the FWHM spot size (0.026 cm) for our light intensity calculations as we expect that the cantilever falls near the center of the laser beam. The laser light intensity was estimated by multiplying the measured laser power (Coherent FieldMate 1098297 Laser Power Meter with OP-2 VIS 1098313 Si sensor) by the relative area under the curve for FWHM and dividing by the FWHM ellipse spot area.

Figure S1. Image of the laser spot used to illuminate the sample showing the (shown in blue) and full width at half max (FWHM) (shown in orange) spot sizes. The FWHM spot size was used to determine the incident light intensity.
ABSORPTION COEFFICIENT
UV-Vis spectra of polymer films (Figure S2) were collected using an Agilent Technologies Cary 8454 UV-Vis. Absorption coefficients match those found in literature within 25% [4, 5].

Figure S2. Absorption coefficient spectra for PM6:Y6, Y6, and PM6 calculated from the measured UV-Vis spectra.
Charge Density Dependence on Generation Rate
Table S1. Energy levels with references found in main text Table IV.
Table S2. Disorder parameter values used to calculate average disorder found in main text Table IV.
How would charge density at steady state depend on the generation rate, , and therefore light intensity or laser power?
A. , no background charge density
Following Burke et al. [12], we have the rate equations
with and , where and are the background electron and hole concentration and and are the change in electron and hole density due to light; , , and are the rates at which carrier meet, split, and recombine respectively; and the generation rate , where is the generation rate due to light and is the thermal generation rate.
Expanding Eqs. S1a–S1c, we have
Cancelling like terms and assuming and , and Eqs. S2a–S2c reduce to
At steady state, the above rate equations are set equal to zero and equate Eq. S3a and Eq. S3c,
Solving Eq. (37) for , we get
The generation rate can be calculated using
where is the visible light intensity, is the absorption coefficient, and is the photon energy. We get from the wavelength of the visible laser, 639.7 nm, eV. is estimated from the laser power and spot size.
We plug Eq. (38) into Eq. (37) to get
Dividing both sides by , we have
and , where is the Langevin reduction factor. When we are in the Langevin limit. Letting we get
Note that , the rate at which carriers meet, is equal to , the Langevin rate. We know that conductivity is
with and the electron and hole mobilities, respectively.
Plugging Eq. (42) into Eq. (43), we find
, finite background charge density
Starting with Eqs. S2a–S2c, and assuming , we have
At steady state, Eqs. S12a–S12c equal zero, and we get
If ,
In computing conductivity, let us neglect . This assumption gives,
In this case, conductivity is directly proportional to light intensity, which is what we see in our data.
Error Estimation
The error bars for the Langevin reduction factor were estimated as follows,
where is the Langevin reduction factor; and are the slope and intercept for the linear fit of conductivity versus light intensity; , , and are the error for the Langevin reduction factor, slope, and intercept, respectively; and is the correlation coefficient.
Broadband Local Dielectric Spectroscopy
Below are all of the BLDS spectra collected for the PM6:Y6 blend and controls, Y6 and PM6, both with (Figure S3) and without (Figure S4) the hole transport layer, PEDOT:PSS. The cantilever details for each spectra are listed in Table S3. The BLDS experiment is described by Tirmzi et al. [13], with the amplitude modulation frequency Hz and the peak-to-peak voltage applied V using a waveform generator (Keysight 33622A).
Figure S5–Figure S8 show BLDS fit parameters , , , and versus light intensity. Table S4–Table S16 list the fit parameters as well.
| Light Intensity | ||||
| 0 mW/cm | 5.45e-02 ± 3.06e-02 | -6.38e-02 ± 1.77e-02 | 7.10e-06 ± 9.41e-07 | 1.24e-06 ± 1.87e-07 |
| 0.84 mW/cm | 4.30e-02 ± 2.06e-02 | -3.71e-02 ± 1.10e-02 | 5.49e-06 ± 1.01e-06 | 4.66e-07 ± 1.05e-07 |
| 5.4 mW/cm | 7.40e-02 ± 4.05e-02 | -6.35e-02 ± 2.21e-02 | 4.58e-06 ± 1.05e-06 | 4.87e-07 ± 1.38e-07 |
| 54 mW/cm | -2.56e-01 ± 1.95e-02 | 1.06e-01 ± 1.19e-02 | 7.47e-07 ± 2.72e-08 | 5.79e-07 ± 5.88e-08 |
| 103 mW/cm | -1.40e-01 ± 8.95e-03 | 5.58e-02 ± 5.28e-03 | 7.31e-07 ± 2.52e-08 | 4.31e-07 ± 4.89e-08 |
| 154 mW/cm | -1.48e-01 ± 1.12e-02 | 5.74e-02 ± 6.46e-03 | 6.75e-07 ± 3.27e-08 | 3.70e-07 ± 5.57e-08 |
Table S4. Fit parameters for PM6:Y6/ITO, ○ dataset.
| Light Intensity | ||||
| 0 mW/cm | -6.23e-02 ± 1.32e-01 | 1.76e-02 ± 6.72e-02 | 1.64e-04 ± 3.83e-05 | 2.61e-06 ± 7.38e-07 |
| 0.84 mW/cm | -1.63e-02 ± 6.59e-02 | -6.09e-03 ± 3.36e-02 | 1.04e-04 ± 2.38e-05 | 2.14e-06 ± 5.68e-07 |
| 5.4 mW/cm | 1.10e-02 ± 6.01e-02 | -2.02e-02 ± 3.07e-02 | 8.30e-05 ± 2.28e-05 | 1.94e-06 ± 6.07e-07 |
| 54 mW/cm | 7.81e-03 ± 6.09e-02 | -1.86e-02 ± 3.12e-02 | 7.65e-05 ± 2.36e-05 | 1.86e-06 ± 6.45e-07 |
| 103 mW/cm | 1.82e-02 ± 4.64e-02 | -2.34e-02 ± 2.37e-02 | 5.65e-05 ± 1.84e-05 | 1.06e-06 ± 3.82e-07 |
| 154 mW/cm | 7.05e-02 ± 1.55e-01 | -5.04e-02 ± 7.89e-02 | 1.25e-04 ± 5.05e-05 | 1.99e-06 ± 9.49e-07 |
Table S16. Fit parameters for Y6/PEDOT:PSS/ITO, □ dataset.

Figure S5. vs. light intensity for all samples obtained from the BLDS fits. Each symbol corresponds to a different physical sample. Red data points are samples that contain PEDOT:PSS, and blue data points are samples without PEDOT:PSS. Data were collected in triplicate. All error bars are .

Figure S8. vs. light intensity for all samples obtained from the BLDS fits. Each symbol corresponds to a different physical sample. Red data points are samples that contain PEDOT:PSS, and blue data points are samples without PEDOT:PSS. Data were collected in triplicate. All error bars are .
| Light Intensity | ||||
| 0 mW/cm | 2.65e-02 ± 4.40e-02 | -5.00e-02 ± 2.44e-02 | 4.19e-06 ± 8.46e-07 | 5.63e-07 ± 1.14e-07 |
| 0.84 mW/cm | -1.94e-01 ± 1.84e-02 | 7.72e-02 ± 1.10e-02 | 8.42e-07 ± 4.09e-08 | 5.09e-07 ± 6.62e-08 |
| 5.4 mW/cm | -1.86e-01 ± 1.71e-02 | 7.39e-02 ± 1.02e-02 | 8.12e-07 ± 3.61e-08 | 4.92e-07 ± 6.49e-08 |
| 54 mW/cm | -1.94e-01 ± 1.36e-02 | 7.64e-02 ± 8.03e-03 | 7.54e-07 ± 2.44e-08 | 4.60e-07 ± 5.05e-08 |
| 103 mW/cm | -2.07e-01 ± 1.20e-02 | 8.03e-02 ± 6.97e-03 | 6.81e-07 ± 2.11e-08 | 4.04e-07 ± 4.16e-08 |
| 154 mW/cm | -1.88e-01 ± 2.13e-02 | 6.73e-02 ± 1.17e-02 | 6.00e-07 ± 4.65e-08 | 2.77e-07 ± 6.65e-08 |
Table S5. Fit parameters for PM6:Y6/ITO, △ dataset.
| Light Intensity | ||||
| 0 mW/cm | 1.60e-02 ± 7.43e-03 | -1.47e-02 ± 3.90e-03 | 8.44e-06 ± 1.28e-06 | 4.64e-07 ± 8.10e-08 |
| 0.84 mW/cm | 1.97e-02 ± 8.78e-03 | -1.70e-02 ± 4.61e-03 | 5.70e-06 ± 9.32e-07 | 3.34e-07 ± 6.61e-08 |
| 5.4 mW/cm | 2.29e-02 ± 9.47e-03 | -1.89e-02 ± 4.97e-03 | 5.22e-06 ± 8.85e-07 | 3.06e-07 ± 6.37e-08 |
| 54 mW/cm | -4.88e-02 ± 2.68e-03 | 1.86e-02 ± 1.52e-03 | 7.53e-07 ± 2.33e-08 | 3.15e-07 ± 3.72e-08 |
| 103 mW/cm | -4.91e-02 ± 1.92e-03 | 1.83e-02 ± 1.07e-03 | 7.16e-07 ± 1.81e-08 | 2.62e-07 ± 2.55e-08 |
| 154 mW/cm | -3.31e-02 ± 1.57e-02 | 9.83e-03 ± 8.00e-03 | 4.33e-07 ± 8.89e-08 | 7.80e-08 ± 8.58e-08 |
Table S6. Fit parameters for PM6:Y6/ITO, □ dataset.
| Light Intensity | ||||
Table S7. Fit parameters for PM6:Y6/ITO, ◇ dataset.
| Light Intensity | ||||
Table S8. Fit parameters for Y6/ITO, ○ dataset.
| Light Intensity | ||||
Table S9. Fit parameters for Y6/ITO, △ dataset.
| Light Intensity | ||||
| 0 mW/cm | -3.07e-02 ± 2.34e-02 | 8.66e-04 ± 1.23e-02 | 1.91e-04 ± 2.89e-05 | 1.06e-05 ± 1.98e-06 |
| 0.84 mW/cm | -3.97e-02 ± 3.13e-02 | 5.43e-03 ± 1.65e-02 | 2.11e-04 ± 3.87e-05 | 1.12e-05 ± 2.56e-06 |
| 5.4 mW/cm | -3.42e-02 ± 3.79e-02 | 2.51e-03 ± 1.98e-02 | 2.11e-04 ± 4.73e-05 | 1.02e-05 ± 2.86e-06 |
| 54 mW/cm | -4.54e-02 ± 4.53e-02 | 8.37e-03 ± 2.37e-02 | 2.28e-04 ± 5.65e-05 | 1.07e-05 ± 3.33e-06 |
| 103 mW/cm | -4.14e-02 ± 5.66e-02 | 5.82e-03 ± 2.95e-02 | 2.88e-04 ± 6.92e-05 | 1.22e-05 ± 3.83e-06 |
| 154 mW/cm | -4.66e-02 ± 5.74e-02 | 8.70e-03 ± 2.97e-02 | 3.05e-04 ± 7.15e-05 | 1.02e-05 ± 3.15e-06 |
Table S10. Fit parameters Y6/ITO, □ dataset.
| Light Intensity | ||||
| 0 mW/cm | 1.37e-02 ± 3.09e-02 | -2.77e-02 ± 1.66e-02 | 3.35e-05 ± 6.76e-06 | 2.39e-06 ± 5.22e-07 |
| 0.84 mW/cm | 5.57e-02 ± 3.91e-02 | -5.40e-02 ± 2.10e-02 | 5.92e-06 ± 1.25e-06 | 5.18e-07 ± 1.29e-07 |
| 5.4 mW/cm | 5.94e-02 ± 4.04e-02 | -5.57e-02 ± 2.16e-02 | 6.21e-06 ± 1.34e-06 | 5.05e-07 ± 1.29e-07 |
| 54 mW/cm | 6.41e-02 ± 4.98e-02 | -5.82e-02 ± 2.63e-02 | 5.13e-06 ± 1.38e-06 | 3.31e-07 ± 1.06e-07 |
| 103 mW/cm | -1.84e-01 ± 1.55e-02 | 7.25e-02 ± 8.99e-03 | 8.18e-07 ± 3.98e-08 | 3.80e-07 ± 6.34e-08 |
| 154 mW/cm | -1.81e-01 ± 9.77e-03 | 7.14e-02 ± 5.69e-03 | 8.15e-07 ± 2.69e-08 | 3.71e-07 ± 4.18e-08 |
Table S11. Fit parameters for PM6:Y6/PEDOT:PSS/ITO, ○ dataset.
| Light Intensity | ||||
| 0 mW/cm | 2.13e-02 ± 1.03e-01 | -2.78e-02 ± 5.39e-02 | 1.27e-04 ± 2.86e-05 | 5.75e-06 ± 1.54e-06 |
| 0.84 mW/cm | 4.65e-02 ± 2.52e-02 | -4.46e-02 ± 1.36e-02 | 7.81e-06 ± 1.29e-06 | 6.51e-07 ± 1.25e-07 |
| 5.4 mW/cm | 5.22e-02 ± 2.55e-02 | -4.87e-02 ± 1.37e-02 | 6.80e-06 ± 1.06e-06 | 5.93e-07 ± 1.09e-07 |
| 54 mW/cm | 6.11e-02 ± 3.78e-02 | -5.21e-02 ± 2.00e-02 | 4.49e-06 ± 1.16e-06 | 3.14e-07 ± 9.71e-08 |
| 103 mW/cm | -1.49e-01 ± 9.80e-03 | 5.92e-02 ± 5.76e-03 | 8.10e-07 ± 2.83e-08 | 4.24e-07 ± 4.91e-08 |
| 154 mW/cm | -1.44e-01 ± 7.91e-03 | 5.65e-02 ± 4.58e-03 | 7.39e-07 ± 2.47e-08 | 3.69e-07 ± 4.14e-08 |
Table S12. Fit parameters for PM6:Y6/PEDOT:PSS/ITO, △ dataset.
| Light Intensity | ||||
| 0 mW/cm | 1.44e-02 ± 2.21e-02 | -2.04e-02 ± 1.14e-02 | 3.32e-05 ± 7.54e-06 | 1.08e-06 ± 2.68e-07 |
| 0.84 mW/cm | 5.01e-02 ± 3.11e-02 | -4.10e-02 ± 1.61e-02 | 5.63e-06 ± 1.42e-06 | 2.37e-07 ± 7.37e-08 |
| 5.4 mW/cm | 5.31e-02 ± 3.42e-02 | -4.37e-02 ± 1.77e-02 | 5.50e-06 ± 1.43e-06 | 2.25e-07 ± 7.24e-08 |
| 54 mW/cm | -1.22e-01 ± 7.22e-03 | 4.68e-02 ± 4.10e-03 | 7.90e-07 ± 3.23e-08 | 2.92e-07 ± 4.36e-08 |
| 103 mW/cm | -1.23e-01 ± 5.31e-03 | 4.70e-02 ± 3.00e-03 | 7.62e-07 ± 2.80e-08 | 2.59e-07 ± 3.25e-08 |
| 154 mW/cm | -1.12e-01 ± 7.49e-03 | 4.08e-02 ± 4.08e-03 | 6.48e-07 ± 5.44e-08 | 1.81e-07 ± 4.36e-08 |
Table S13. Fit parameters for PM6:Y6/PEDOT:PSS/ITO, □ dataset.
| Light Intensity | ||||
| 0 mW/cm | -1.24e-02 ± 7.86e-02 | -1.86e-02 ± 4.01e-02 | 5.57e-05 ± 1.74e-05 | 1.11e-06 ± 3.86e-07 |
| 0.84 mW/cm | -3.76e-03 ± 6.52e-02 | -2.26e-02 ± 3.33e-02 | 4.48e-05 ± 1.37e-05 | 9.29e-07 ± 3.14e-07 |
| 5.4 mW/cm | -1.15e-03 ± 8.00e-02 | -2.55e-02 ± 4.10e-02 | 5.60e-05 ± 1.69e-05 | 1.38e-06 ± 4.66e-07 |
| 54 mW/cm | 2.09e-02 ± 5.69e-02 | -3.83e-02 ± 2.91e-02 | 3.27e-05 ± 9.05e-06 | 7.74e-07 ± 2.32e-07 |
| 103 mW/cm | 6.49e-02 ± 1.26e-01 | -6.17e-02 ± 6.41e-02 | 8.53e-05 ± 2.38e-05 | 1.49e-06 ± 4.79e-07 |
| 154 mW/cm | 5.36e-02 ± 8.47e-02 | -5.66e-02 ± 4.31e-02 | 5.32e-05 ± 1.56e-05 | 9.75e-07 ± 3.20e-07 |
Table S14. Fit parameters for Y6/PEDOT:PSS/ITO, ◦ dataset.
| Light Intensity | ||||
| 0 mW/cm | -2.02e-02 ± 5.41e-02 | -7.10e-03 ± 2.78e-02 | 6.71e-05 ± 1.73e-05 | 1.89e-06 ± 5.48e-07 |
| 0.84 mW/cm | -1.19e-02 ± 4.78e-02 | -1.12e-02 ± 2.46e-02 | 5.70e-05 ± 1.52e-05 | 1.75e-06 ± 5.20e-07 |
| 5.4 mW/cm | -3.43e-03 ± 4.25e-02 | -1.54e-02 ± 2.19e-02 | 4.87e-05 ± 1.33e-05 | 1.53e-06 ± 4.58e-07 |
| 54 mW/cm | 5.05e-03 ± 7.30e-02 | -2.08e-02 ± 3.78e-02 | 7.42e-05 ± 2.24e-05 | 2.62e-06 ± 8.95e-07 |
| 103 mW/cm | 2.17e-02 ± 7.44e-02 | -2.92e-02 ± 3.84e-02 | 7.40e-05 ± 2.30e-05 | 2.29e-06 ± 8.05e-07 |
| 154 mW/cm | 4.58e-02 ± 3.44e-02 | -4.10e-02 ± 1.76e-02 | 1.22e-05 ± 3.26e-06 | 3.05e-07 ± 8.84e-08 |
Table S15. Fit parameters for Y6/PEDOT:PSS/ITO, △ dataset.
| Dataset | ||||||||
| Light Intensity | [Ω] | [F] | [Ω] | [F] | [Ω] | [F] | [Ω] | [F] |
| 0 mW/cm | 2.55E+11 | 2.79E-17 | 1.16E+11 | 3.62E-17 | 9.56E+10 | 8.83E-17 | 9.42E+10 | 7.82E-17 |
| 0.84 mW/cm | 9.59E+10 | 5.73E-17 | 1.05E+11 | 8.04E-18 | 6.88E+10 | 8.28E-17 | 6.32E+10 | 7.97E-17 |
| 5.4 mW/cm | 1.00E+11 | 4.57E-17 | 1.01E+11 | 8.02E-18 | 6.29E+10 | 8.31E-17 | 5.52E+10 | 8.11E-17 |
| 54 mW/cm | 1.19E+11 | 6.27E-18 | 9.47E+10 | 7.96E-18 | 6.49E+10 | 1.16E-17 | 7.26E+10 | 1.06E-17 |
| 103 mW/cm | 8.86E+10 | 8.25E-18 | 8.31E+10 | 8.20E-18 | 5.40E+10 | 1.33E-17 | 6.18E+10 | 1.13E-17 |
| 154 mW/cm | 7.61E+10 | 8.87E-18 | 5.69E+10 | 1.05E-17 | 1.61E+10 | 2.70E-17 | 5.07E+10 | 1.24E-17 |
Table S17. and at each light intensity for all PM6:Y6/ITO samples.
| Dataset | ||||||
| Light Intensity | [Ω] | [F] | [Ω] | [F] | [Ω] | [F] |
| 0 mW/cm | 4.79E+11 | 5.39E-18 | 6.28E+11 | 1.02E-16 | 2.19E+12 | 8.75E-17 |
| 0.84 mW/cm | 3.05E+11 | 5.29E-18 | 3.68E+11 | 9.72E-17 | 2.30E+12 | 9.18E-17 |
| 5.4 mW/cm | 2.53E+11 | 4.91E-18 | 3.66E+11 | 9.04E-17 | 2.10E+12 | 1.00E-16 |
| 54 mW/cm | 1.18E+11 | 6.61E-18 | 1.71E+11 | 9.02E-17 | 2.20E+12 | 1.04E-16 |
| 103 mW/cm | 1.14E+11 | 3.44E-19 | 8.82E+10 | 1.16E-16 | 2.51E+12 | 1.15E-16 |
| 154 mW/cm | 1.10E+11 | 2.31E-19 | 5.62E+10 | 1.47E-16 | 2.09E+12 | 1.46E-16 |
Table S18. and at each light intensity for all Y6/ITO samples.
| Light Intensity | Dataset: ◦ [Ω] | [F] | Dataset: △ [Ω] | [F] | Dataset: □ [Ω] | [F] |
| 0 mW/cm² | 4.91E+11 | 6.83E-17 | 1.18E+12 | 1.08E-16 | 2.23E+11 | 1.49E-16 |
| 0.84 mW/cm² | 1.07E+11 | 5.56E-17 | 1.34E+11 | 5.83E-17 | 4.87E+10 | 1.16E-16 |
| 5.4 mW/cm² | 1.04E+11 | 5.98E-17 | 1.22E+11 | 5.58E-17 | 4.63E+10 | 1.19E-16 |
| 54 mW/cm² | 6.81E+10 | 7.52E-17 | 6.45E+10 | 6.96E-17 | 6.00E+10 | 1.32E-17 |
| 103 mW/cm² | 7.82E+10 | 1.05E-17 | 8.72E+10 | 9.29E-18 | 5.32E+10 | 1.43E-17 |
| 154 mW/cm² | 7.63E+10 | 1.07E-17 | 7.59E+10 | 9.73E-18 | 3.72E+10 | 1.74E-17 |
Table S19. and at each light intensity for all PM6:Y6/PEDOT:PSS/ITO samples.
| Light Intensity | Dataset: ◦ [Ω] | [F] | Dataset: △ [Ω] | [F] | Dataset: □ [Ω] | [F] |
| 0 mW/cm² | 2.27E+11 | 2.45E-16 | 3.89E+11 | 1.73E-16 | 5.36E+11 | 3.06E-16 |
| 0.84 mW/cm² | 1.91E+11 | 2.34E-16 | 3.61E+11 | 1.58E-16 | 4.41E+11 | 2.36E-16 |
| 5.4 mW/cm² | 2.84E+11 | 1.97E-16 | 3.15E+11 | 1.55E-16 | 4.00E+11 | 2.08E-16 |
| 54 mW/cm² | 1.59E+11 | 2.06E-16 | 5.38E+11 | 1.38E-16 | 3.82E+11 | 2.00E-16 |
| 103 mW/cm² | 3.06E+11 | 2.79E-16 | 4.71E+11 | 1.57E-16 | 2.18E+11 | 2.59E-16 |
| 154 mW/cm² | 2.01E+11 | 2.65E-16 | 6.27E+10 | 1.94E-16 | 4.10E+11 | 3.05E-16 |
Table S20. and at each light intensity for all Y6/PEDOT:PSS/ITO samples.

Figure S6. vs. light intensity for all samples obtained from the BLDS fits. Each symbol corresponds to a different physical sample. Red data points are samples that contain PEDOT:PSS, and blue data points are samples without PEDOT:PSS. Data were collected in triplicate. All error bars are .

Figure S7. vs. light intensity for all samples obtained from the BLDS fits. Each symbol corresponds to a different physical sample. Red data points are samples that contain PEDOT:PSS, and blue data points are samples without PEDOT:PSS. Data were collected in triplicate. All error bars are .
Atomic Force Microscopy
Atomic force micrographs (AFMs) of polymer films were collected in air on a commercial instrument in tapping mode (Asylum Research MFP-3D-BIO) using a Olympus AC160TS-R3 probe.
The AFMs (Figure S9–Figure S10) of the PEDOT:PSS/ITO samples give some insight into the conductivity. We observe that samples with higher conductivity have a lower rms roughness. This is noteable in the PM6:Y6/PEDOT:PSS/ITO samples, where the sample with lower conductivity () has a roughness of 30 nm, and the sample with higher conductivity () has a roughness of 6 nm.

Figure S9. AFMs of PM6:Y6, Y6, and PM6 samples on PEDOT:PSS. PM6:Y6/PEDOT:PSS/ITO “low ” was the second sample prepared (), with medium conductivity and “high ” was the third sample prepared (), with the highest conductivity of the 3 samples.

Figure S10. AFMs of PM6:Y6, Y6, and PM6 samples on ITO. Y6/ITO “low ” is the lowest conductivity sample (□, collected on two spots on the sample), and “high ” is one of the higher conductivity samples (△).
References
- [1]M. Bonnett Del Alamo, C. Soncco, R. Helaconde, J. L. Bazo Alba, and A. M. Gago, Laser spot measurement using simple devices, AIP Adv. 11, 075016 (2021).DOI
- [2]M. Newville, R. Otten, A. Nelson, T. Stensitzki, A. Ingargiola, D. Allan, A. Fox, F. Carter, Michał, R. Osborn, D. Pustakhod, Ineuhaus, S. Weigand, A. Aristov, Glenn, C. Deil, mgunyho, Mark, A. L. R. Hansen, G. Pasquevich, L. Foks, N. Zobrist, O. Frost, Stuermer, azelcer, A. Polloreno, A. Persaud, J. H. Nielsen, M. Pompili, and P. Eendebak, lmfit/lmfit-py: 1.2.2, Zenodo (2023).
- [3]P. Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, E. Burovski, P. Peterson, W. Weckesser, J. Bright, S. J. van der Walt, M. Brett, J. Wilson, K. J. Millman, N. May- orov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, I. Polat, Y. Feng, E. W. Moore, J. VanderPlas, D. Laxalde, J. Perktold, R. Cimrman, I. Henriksen, E. A. Quintero, C. R. Harris, A. M. Archibald, A. H. Ribeiro, F. Pedregosa, and P. van Mulbregt, SciPy 1.0: fundamental algorithms for scientific computing in Python, Nat. Methods 17, 261 (2020).
- [4]M. Nyman, O. J. Sandberg, W. Li, S. Zeiske, R. Kerremans, P. Meredith, and A. Armin, Requirements for making thick junctions of organic solar cells based on nonfullerene acceptors, Sol. RRL 5, 2100018 (2021).DOI
- [5]S. M. Hosseini, N. Tokmoldin, Y. W. Lee, Y. Zou, H. Y. Woo, D. Neher, and S. Shoaee, Putting order into PM6:Y6 solar cells to reduce the Langevin recombination in 400 nm thick junction, Sol. RRL 4, 2000498 (2020).DOI
- [6]S. Kim, J. Jeong, Q. V. Hoang, J. W. Han, A. Prasetio, M. Jahandar, Y. H. Kim, S. Cho, and D. C. Lim, The role of cation and anion dopant incorporated into a ZnO electron transporting layer for polymer bulk heterojunction solar cells, RSC Adv. 9, 37714 (2019).DOI
- [7]J. Lv, H. Tang, J. Huang, C. Yan, K. Liu, Q. Yang, D. Hu, R. Singh, J. Lee, S. Lu, G. Li, and Z. Kan, Additive-induced miscibility regulation and hierarchical morphology enable 17.5% binary organic solar cells, Energy Environ. Sci. 14, 3044 (2021).DOI
- [8]H. Tang, Y. Shang, W. Zhou, Z. Peng, and Z. Ning, Energy level tuning of PEDOT:PSS for high performance tin-lead mixed perovskite solar cells, Sol. RRL 3, 1800256 (2018).DOI
- [9]L. Perdigón-Toro, L. Q. Phuong, F. Eller, G. Freychet, E. Saglamkaya, J. I. Khan, Q. Wei, S. Zeiske, D. Kroh, S. Wedler, A. Köhler, A. Armin, F. Laquai, E. M. Herzig, Y. Zou, S. Shoaee, and D. Neher, Understanding the role of order in Y-series non-fullerene solar cells to realize high open-circuit voltages, Adv. Energy Mater. 12, 2103422 (2022).
- [10]L. Perdigón-Toro, H. Zhang, A. Markina, J. Yuan, S. M. Hosseini, C. M. Wolff, G. Zuo, M. Stolterfoht, Y. Zou, F. Gao, D. Andrienko, S. Shoaee, and D. Neher, Barrierless free charge generation in the high-performance PM6:Y6 bulk heterojunction non-fullerene solar cell, Adv. Mater. 32, 1906763 (2020).
- [11]S. M. Hosseini, S. Wilken, B. Sun, F. Huang, S. Y. Jeong, H. Y. Woo, V. Coropceanu, and S. Shoaee, Relationship between energetic disorder and reduced recombination of free carriers in organic solar cells, Adv. Energy Mater. 13, 2203576 (2023).
- [12]T. M. Burke, S. Sweetnam, K. Vandewal, and M. D. McGehee, Beyond Langevin recombination: How equilibrium between free carriers and charge transfer states determines the open-circuit voltage of organic solar cells, Adv. Energy Mater. 5, 1500123 (2015).DOI
- [13]A. M. Tirmzi, J. A. Christians, R. P. Dwyer, D. T. Moore, and J. A. Marohn, Substrate-dependent photoconductivity dynamics in a high-efficiency hybrid perovskite alloy, J. Phys. Chem. C 123, 3402 (2019).DOI