On validity and limits of deducing the degree of charge transfer from shifts of cyano vibrations

Understanding the p-doping of organic semiconductors often relies on spectroscopic fingerprints of cyano vibrations, which strongly depend on the charge state of the dopant molecule following intermolecular charge transfer. Interpreting these vibrations can be difficult as a number of other factors can impact them. Here, we formalize the assumptions behind the determination of molecular charge from cyano vibrations and we use computational modeling to demonstrate key obfuscating factors in this process. We notably demonstrate that cyano vibrations do not necessarily shift linearly with the molecular charge and investigate which molecular parameters can explain that. Finally, we provide guidelines for the study of charge transfers involving new molecular dopants based on their cyano vibrations.


Introduction
The molecular doping of organic semiconductors (OSCs) is essential for achieving control over energy level alignment and charge carrier density. This is vital for employing OSCs to fabricate electronic devices [1][2][3]. One of the most crucial factors to understand in terms of intermolecular interactions between dopants and OSCs is the transfer of charge between the two, leading to local changes in the electron density. For p-type doping, electron density is removed from the OSC by the dopant, and the converse applies for n-type doping. Two cases of electron density transfers, for each type of doping, are typically distinguished in literature and are understood to have different origins. The remaining of this work will focus exclusively on p-type doping.
First, a partial transfer of charge can emerge (often named formation of a charge-transfer complex), where the electron density corresponding to less than an elementary charge is exchanged [4]. This is understood as some amount of charge that is effectively shared between two entities. This process occurs following the hybridization of the highest occupied molecular orbital (HOMO) of the OSC and the lowest unoccupied molecular orbital (LUMO) of the dopant [5]. It provides stabilization to the doped OSC system by the formation of a bi-molecular state, in which the HOMO and the LUMO have been pushed away in energy from each other as compared to the pristine molecules. Therefore, this process generates a complex which effectively has a reduced electron affinity as compared to the original, unaffected dopant. As a result, this phenomenon quenches the efficiency of molecular dopants to increase the number of charge carriers, and the process is generally regarded as detrimental. Molecular design strategies such as the physical shielding of the dopant core via bulky side groups have been suggested to eliminate this type of interaction [2,6,7]. Second, the transfer of a complete charge-often named integer charge transfer, or ion pair formation-can emerge, where an electron of the OSC is fully transferred to the dopant, rather than shared, leading to the generation of two radical ions of opposite charge and spin [8].
These effects have been detailed more explicitly since, and include countercharge location relative to the cyano group [15], the vicinity of hydrogen atoms [16,17], and the vibrational Stark effect from varied degrees of polaron delocalization [18], where we recently noted that the latter effect might equally result from bipolaron formation instead [19]. In some cases, the frequencies corresponding to cyano vibrations even support a wrong interpretation of the charge transfer type [20], where it was suggested that the embedding into differently crystalline structures might be the cause. Given the number of factors which can impact cyano vibrations, sometimes modes stemming from the core [21] are preferred, although it is not always possible to use them as the unambiguous assignment of vibrations in this region is difficult. Despite issues with cyano vibrations having been well discussed qualitatively, their causes still remain to be understudied on the fundamental level. This is not necessarily critical for well-known systems, for instance P3HT doped with TCNQ and its fluorinated derivatives, where the issues described above have been widely documented, but it is certainly of high value for less-well studied systems or new molecular dopants. This is particularly relevant as it has been found that both fractional and integer charge transfer can coexist within the same systems, where their impact on key observables such as conductivity strongly depends on whether the charge transfer occurs within crystalline or amorphous phases of the system [19,22]. Thus, discriminating between different charge transfer scenarios is of key importance for the advancement of the field and hinges on the ability to probe and correctly deduce charge transfer using vibrational spectroscopy. Overall, while basic principles together with experimental evidence leave no doubt that the cyano vibrations are highly sensitive to various degrees and types of charge transfer, constraints brought about by the solid state due to physical, electronic, and electrostatic interaction with the local environment are highly susceptible to obscure their interpretation and remain poorly understood.
In this work, we use computational modeling to study a selection of strong electron acceptors which comprise cyano groups to be used as molecular dopants. Three dopants have been previously studied both experimentally and computationally, and, in addition, we construct three hypothetical dopants serving as computational toy models which allow demonstrating key intramolecular factors that obfuscate the interpretation of charge transfer. We formulate general guidelines for the comprehension of cyano vibrations that we suggest be followed when characterizing new dopants, thereby, providing tools for a more rigorous interpretation of cyano vibrations and the characterization of charge transfer in general.

Key concepts
To facilitate the discussion that follows, we first introduce the key concepts on which we base our argument.
Sensitivity, in this context, refers to the fact that large shifts in the frequencies of cyano-related vibrations are expected between a neutral and a charged molecular dopant comprising cyano groups. This concept should not be confused with that of specificity, i.e. numerous factors other than molecular charge can equally impact these modes. While cyano modes are severely shifted by an excess or deficiency of charge density on the molecular dopant, that sensitivity can be an issue as other electrostatic and electronic phenomena are likely to shift these modes as well. It should be noted that other vibrational modes can be used for the quantification of charge transfer as discussed above, e.g. modes from the dopant core. While these vibrations are more specific, as they do not stem from the molecule's periphery, they are typically less sensitive. This comes in addition to the fact that interpreting these modes can be difficult as they tend to be masked by modes stemming from the OSCs to which the dopants are applied.
The underlying assumption in the quantification of charge from cyano vibrations is that a relationship between the energy of cyano vibrations and the molecular charge, more specifically molecular charge excesses or deficiencies, exists. Indeed, infrared vibrations can be understood analogously to spring and mass systems, where the frequency depends on the spring constant which is proportional the electric force between two atoms involved in the stretching vibrations. This force can be modelled as the Coulomb force between the atoms, and is accordingly proportional to the atomic charges. The relationship between molecular charge and the cyano vibration therefore originates from the relationship between molecular and atomic charges, i.e. how excess or deficiency of charge is reorganized within a molecule.
Two types of factors are distinguished in this work: electronic and geometric. Electronic factors refer to contributions stemming from the electron density, i.e. how it reorganizes within a molecule or how its localization generates electrostatic interactions, for instance. Geometric factors refer to contributions stemming from changes in the molecular geometry, for instance, a change in bond or dihedral angle, or from the rotation of non-monoatomic groups. These impact the change in dipole moment along vibration coordinates, which translates into changes in the infrared response. Additionally, geometric factors imply electronic factors as well and disentangling them might not always be possible or relevant.
Linearity is a central aspect in determining the degree of charge transfer from the frequency of cyano vibrations and has been assumed throughout literature (among others, [11,20,23,24]). Linearity refers to a behavior best explained by an equation involving only the first power of an independent variable. Non-linearity here refers to a behavior that either does not follow obvious trends or is best explained by an equation involving higher powers of an independent variable. This can occur when some variables are unaccounted for or when the assumed physical model is too simplistic. Hence, discussing linearity is highly relevant for the problem of charge transfer and cyano vibrations.

Materials and methodology
Three sets of two similar molecules are studied in this work. The well-characterized molecular electron acceptors TCNQ figure 1(a) and F4TCNQ (d), from which we construct the hypothetical molecules H2Ph2TCNQ (b) and F2Ph2TCNQ (e) respectively, with two of the hydrogen or fluorine atoms in opposite positions on the quinone core replaced by phenyl rings. CN6CP (c) is also investigated as a more recent example for a molecular dopant of high electron affinity with a different core, along with a hypothetical derivative CN3CS3CP (f), where three cyano groups are replaced by chlorosulfonyl (CS) groups, analogous to TM3CN3CP [10], but with substituents featuring higher charge density. The geometry of the molecules in their neutral, radical anionic, and dianionic forms are optimized using density functional theory (DFT), then their vibrational modes are obtained. DFT calculations are carried out using the ωB97X-D functional from Chai and Head-Gordon [25], with the 6-31+g(d,p) basis set, in Gaussian 09 rev E0.1 [26]. Thermal enthalpies are obtained following the calculation of vibrational frequencies in Gaussian as discussed in [27]. formally defined as the angle between the normal vectors of the planes formed by the first three atoms (the two golden quinone double-bonded carbons and the closest phenyl carbon) and the last three atoms (the two golden phenyl carbons and the closest quinone carbon). Carbon: gray, hydrogen: white, nitrogen: blue, fluorine: cyan, sulfur: yellow, oxygen: red, chlorine: green.
Atomic charges are calculated using a partitioning based on the quantum theory of atoms in molecules as implemented in the software Multiwfn [28]. All isosurfaces and maps are also calculated in Multiwfn; the visualization of the isosurfaces is done in Avogadro [29] and in Gaussian's visualisation interface. The color projection of the change of atomic charges on molecular structures is done using Vesta [30].

Molecular and atomic charges
First, the assumption that molecular charge is linearly related to the atomic charges involved in cyano vibrations, i.e. on the cyano's carbon and nitrogen atoms, is investigated. More precisely, atomic charges here refer to the partial atomic charges present, in average, on chemically-identical species, e.g. the average of partial atomic charges on the three nitrogen atoms. The atomic charge on the nitrogen atom, as visible in figure 2(a), scales linearly with the molecular charge, forming almost a perfect straight line for all molecules. However, atomic charges on the carbon (b) diverge from linearity, with very different trends for the six molecules. It is not surprising that the carbon atomic charge sees more variation than the nitrogen given its closer contact with the molecular core and the different substituents bound to it. This first glance already demonstrates a crucial factor in the relation between molecular charge and the cyano vibrations, that is, different molecules have different tendencies to internally reorganize molecular charge excesses or deficiencies, in the general case, in a non-linear manner. It is noted that considerable atomic charges are also found on other parts of the studied molecules and are discussed in section 4.2.3 and in figures 8 and 9; however, the relationship between these atomic charges and cyano vibrations is more subtle and therefore not investigated immediately.
Yet, atomic charges in themselves are not as relevant in determining vibrational modes as the coupling between them through a chemical bond, i.e. the bond dipole moment. In a first approximation, assuming the same cyano bond length for the three charge states (neutral, radical anion and dianion) of a given molecule, the shift of the frequency of cyano vibrations will depend on the change in the product of nitrogen and carbon atomic charges shown in figure 2(c). In particular, considering the two atoms as a spring-mass system, the force should be proportional to the product of atomic charges, and the energy associated with the oscillation scales linearly with the force constant. Still with this analogy, the curve of the product of atomic charges with respect to an implicit variable also related to the force, i.e. as presumed, the molecular charge, should display a second-degree relation. From the same panel, it seems that all the molecules studied here are well-behaved in this regard, with a trend that follows a second-degree curve, with the extent of the decay depending on other molecular properties. In short, based on atomic charges on the cyano atoms, all the molecules should see a more or less linear shift in frequency of the cyano vibrations with the molecular charges towards lower wavenumbers, if this shift depends only on the charge, as argued above.

Molecular charge and cyano vibrational frequencies
Having established that, in a first approximation, the C≡N force driving the vibration should lead to a linear shift of the cyano mode with the molecular charge, the vibrational spectra of the corresponding molecules are calculated and the frequencies corresponding to the peak maximum of the higher energy vibration (when more than one exist) are compiled. Surprisingly, even for a single, isolated molecule, some degree of non-linearity between the frequency of cyano modes and the molecular charge state can be found. While trends in TCNQ, F4TCNQ and CN6CP are indeed linear, the highest energy cyano vibration in the H2Ph2TCNQ and F2Ph2TCNQ molecules showcase mild non-linearity, and for CN3CS3CP the curve is very far from linear, as almost no shift is observed between the radical anion and the dianion (see figure 3(a)). Already, it is apparent that deviations from linearity stem from the non-planarity of the molecules, as all three perfectly flat molecules feature perfectly linear trends.
Focusing on the two most prominent cases of non-linearity, H2Ph2TCNQ and CN3CS3CP, (figure 3(b)) demonstrates different fits relating the peak position to the molecular charge. One could assume (1) linearity between all three charge states; (2) linearity between the neutral and radical anion states only; (3) the data following a second degree curve. Considering these three assumptions, (1) seems unjustified for molecules with pronounced non-linearity in the data, for CN3CS3CP in particular; (2) seems unjustified as the existence of a special regime between the neutral and radical anion states, and of another one between radical anion and dianion, appears artificial; (3) seems unjustifiable as three data points can always be perfectly fitted by a second order curve and as a linear shift with the molecular charge is expected from the simple considerations above. Yet, (3) seems unavoidable for non-planar molecules as their behavior does not strictly follow linearity, and the inclusion of a curvature term is not problematic for planar molecules as it simply goes to 0. In reality, however, the situation is more complex than described before as not only the product of atomic charges will change, but also the bond length, which adds to other effects such as substituent-cyano intramolecular interactions. A second degree curve can thus capture more diverse and complex physical phenomena that are susceptible to influencing the vibrations.
In figures 3(c) and (d), we show the fit curves that would be used in the context of determining the molecular charge from the frequency of the cyano vibrations, following the above three possibilities (illustrated in panel (b)). The area of uncertainty stemming only from the assumption of the model used, or, in other words, the lack of certainty in what assumption allows to correctly describe the shift, is shown in gray; vertical/horizontal reference lines show the degree of charge transfer corresponding to a given cyano vibration frequency according to the three different assumptions. Even for a molecule with little deviation from linearity (with a coefficient of determination R 2 > 0.99) across the three charge states, an uncertainty in the order of 10 to 20% already emerges from the assumption choice and needs to be compounded to other sources of uncertainties that might arise in experiments. Therefore, a high R 2 should not be misunderstood as a gauge for the validity of a linear assumption, as all molecules but one show R 2 > 0.99 (all R 2 are listed in table 1).
Most notably, for a molecule showcasing significant non-linearity, the area of uncertainty simply becomes almost as large as the range of values covered between the extrema defined by the neutral and radical anion molecules. We stress that the correct assumption to use cannot be known without probing the second ionization of the molecule, and that huge sources of error in interpreting cyano vibrations can emerge  from non-linearities similar to those illustrated. Our results demonstrate that it is indispensable to probe the dianion vibrational fingerprint of molecular dopants which are non-planar, in particular when the extent of their non-linearity is unknown, as it is the case when studying the charged states of new molecules comprising cyano groups.

Atomic charges and cyano vibrational frequencies
Following the results presented in the last two subsections, we now verify whether the product of atomic charges on the cyano carbon and nitrogen can indeed linearize the frequency of cyano vibrations, which is shown in figure 4. In all cases except for CN3CS3CP we find that the product of atomic charges on the nitrogen and carbon is very close to linearity with the respective cyano frequencies across the three different charge states of a given molecule.
Taking everything together, we can therefore distinguish three cases. First, both the product of atomic charges and the molecular charge linearize the frequency of the cyano vibration. These molecules are referred to as 'well-behaved' in the following, and seem to be flat and highly symmetric, in addition to featuring monoatomic substituents only. Second, there is the case where the product of atomic charges but not the molecular charge linearizes the frequency of the cyano vibration, that is, H2Ph2TCNQ and F2Ph2TCNQ. There, we suspect this is due to geometrical changes occurring between the different charge states. Finally, there is the case where neither the product of atomic charges nor the molecular charge linearize the cyano vibration, that is, CN3CS3CP. In the following section, the various causes capable of impacting the cyano vibration (other than the molecular charge) are being investigated in more detail.

Factors influencing cyano vibrations 4.2.1. Geometrical distortions
To assess the impact of geometrical distortions, we construct a toy model based on the F2Ph2TCNQ and H2Ph2TCNQ molecules by constraining the dihedral angle of the phenyl ring with respect to the quinone double bond, controlled via the rotation illustrated in figure 1(b). It is suspected that the main source of calculated non-linearity in these molecules comes from small physical changes upon ionization events, resulting in some reorganization of atomic charges across the molecule. Indeed, the amount of charge belonging to the phenyl rings strongly depends on the angle it makes with the remaining of the conjugated system. Optimized geometries for the neutral molecules have a dihedral angle higher than 75 degrees (i.e. the phenyl rings are aligned within 15 degrees of the out-of-plane direction with respect to the quinone core), while this dihedral angle decreases to 60-65 degrees for the radical anions (less out-of-plane with respect to the quinone core).
The impact of this rotation on the infrared responses is demonstrated for the neutral and radical anions of the two molecules in figure 5. The largest changes are seen for the neutral F2Ph2TCNQ molecule. There, both a shift of the frequency as well as the gradual disappearance of the lower energy cyano vibration can be seen. The latter can be explained by the gradual out-of-plane bending of the quinone when the angle decreases, which can modify dipole moment changes upon molecular vibrations as well as which molecular vibration modes are possible. As such, it should be noted that geometrical flexibility and degrees of freedom are capable of deeply impacting cyano vibrations, both by shifting them, or by impacting the amplitude of their infrared response, even to the point of creating or extincting absorption features.
To substantiate the claim that these changes are due to geometrical instead of electronic changes, we more carefully probe the relationship between twisting angle, cyano vibrations, and the atomic charges. First, it is shown in figure 6(a), that the twisting angle can vary quite dramatically for an only small energy cost, as highlighted by the shaded area marking an energy cost smaller than 0.1 eV. This energy leading to a reduced twisting angle can be provided by factors such as a favourable crystal structure or the interaction with a solvent. It should be noted that as the dihedral angle decreases, molecules start to suffer from intense steric repulsion, explaining the gradual out-of-plane bending of the quinone core as previously mentioned. This, however, comes with a great energy cost, which is also visible from the shaded area in figure 6(a). It is apparent from panel (b) that neutral molecules are impacted more significantly by these geometrical changes than charged molecules. We note that for both F2Ph2TCNQ and H2Ph2TCNQ in their neutral forms, the shift of the cyano vibrations as a function of the twisting angle follows the same direction. Panel (c) demonstrates the crucial observation that the two molecules have diametrically opposed behavior of their cyano atomic charges as a function of the twisting angle, which can be understood by different back-donation to/from the fluorine and hydrogen atoms. Finally, panel (d) shows that, as expected from panels (b) and (c),   the cyano vibrations shift in different directions with respect to the relevant atomic charges, implying that the shifts, to some surprise, find their origin somewhere else than in the atomic charges, i.e. in the geometrical changes themselves. This finally confirms the role of geometrical degrees of freedom in cyano vibrations.

Electrostatic interaction
In addition to the impact of geometrical changes, the cyano vibrations can also be impacted by electrostatic interactions which can either be of intermolecular or intramolecular origin. Here, in the CN3CS3CP molecule, substituents are in sufficient proximity to the cyano groups, and electron-rich enough, that they can have strong electrostatic interactions with them. We demonstrate these interactions in figure 7 using various techniques; technical details are provided in the figure caption.
We use maps of the total electrostatic potential ((a) to (d)) which reveals both the magnitude and localization of electrostatic interactions [31], of the reduced gradient density (e), which reveals non-covalent interactions in real space [32][33][34], as well as of the electron density Laplacian (f), which reveals, notably, regions of high and low potential energy and pathways across them [35]. While in-depth information concerning the appropriate interpretation of these techniques can be found in more specialized works, it is notable here that from all of them, significant interactions between the cyano and chlorosulfonyl groups are deduced. Furthermore, the strength of these interactions increase with the molecular charge, which is directly reflected by the increase in |isovalue| necessary to find the isosurface related to cyano-chlorosulfonyl interactions.
When the two conditions of enough proximity between substituents and cyano groups, and enough charge on the substituents, are met, strong electrostatic interactions between them are expected. We note that while this is unlikely for TCNQ-like molecules, such interactions are likely for molecules of the same type as CN3CS3CP, that is, molecules featuring a cyclopropane core on which cyano and substituent groups are attached to the same carbon. Examples include TM3CN3CP and similar molecules synthesized by Saska et al [9]. We conclude that the possibility that these interactions exist should be examined when molecules contain non-monoatomic substituents along with cyano groups. Provided that these strong interactions exist, they can impact cyano vibrations dramatically.

Non-uniformity of intramolecular charge reorganization
In general, non-uniformities in the way that excess or deficiency of charge reorganizes intramolecularly are possible and can lead to an unexpected behavior of the cyano vibrations. Here, non-uniformity refers to a change in the charge reorganization behavior as a function of the molecular charge itself. First, we probe the atomic charge change upon ionization for TCNQ and F4TCNQ in figure 8. In panels (a) and (c), the change in atomic charge for the TCNQ molecule between (a), neutral and radical anion states, and (c), between neutral and dianion states, is represented using a color scale directly onto the molecular structure (capped sticks style, where the atoms are centered at each vertex). The change of atomic charge is normalized for each panel in a way that the maximal change of atomic charge corresponds to 1 (identified by the label 'max' in figure 8) and the minimal change of atomic charge corresponds to 0 (identified by the label 'min'), allowing to map the change of atomic charge onto the full range of the color scale. Therefore, the figure acts as a comparative tool between different charge state transitions, where different colors for the same region of a molecule directly indicate a change of tendency in the charge reorganization as a function of the molecular charge. The same is shown for F4TCNQ in panels (b) and (d). For both molecules there is only little change in behavior of the quinone substituents, with hydrogens and fluorines maintaining a similar fraction of the charge at both ionization stages. Nevertheless, there appear to be differences in the interplay between the core and the cyano groups between TCNQ and F4TCNQ across the two charge state transitions, which are represented by the black arrows in figure 8. A different charge reorganization between forming the radical anion and the dianion from the neutral molecule simply gives rise to the non-linearity of atomic charge product with the cyano vibration frequency shown in figure 2(c); and the higher impact of molecular charge itself on the charge reorganization behavior for TCNQ explains its higher non-linearity with respect to F4TCNQ, which already demonstrates the utility of figure 8 to rationalize some of the observations previously made. However, these molecules pose no difficulties in interpretation because they are both 'well-behaved' (as defined earlier); their cyano groups are physically isolated and do not interact with the remaining of the molecule.
The same illustration for molecules with a cyclopropane core is given in figure 9, with CN6CP in panels (a) and (c), and CN3CS3CP in panels (b) and (d). It provides interesting insight into the more complex behavior of the latter. CN6CP behaves very similar to the TCNQs, with, for an increased molecular charge, a more pronounced accumulation of charge outside the core, at the carbon bridging two cyano groups. For CN3CS3CP, the chlorine atoms have a tendency to attract more electron density upon increasing the ionization state of the molecule, but the behavior of everything else remains similar. While this change might appear negligible, it, in fact, shines light on the behavior highlighted in the previous section, that is, the increasing interaction strength between cyano and the substituents with the molecular charge. Indeed, CN3CS3CP is the only molecule where the substituents can match the impact of the cyano group on the electron density (same color in figure 9), and more importantly, perhaps, it is the only molecule where the substituents can efficiently interact with the cyano groups (from a geometrical point of view). CN3CS3CP also happens to be the molecule featuring the highest non-linearity in the trends of the cyano vibration frequency against the molecular charge ( figure 3(a)), showing that the accumulation of charge at the substituents is responsible for the increasing strength of the cyano-substituent interactions, and thus, for the high non-linearity of the cyano vibration frequency with the molecular charge. While the existence of substituent-cyano interactions was shown in the last section (as visualized using different types of maps in figure 7), the increase in interaction strength needs to be explained by a non-linearity of the molecular charge reorganization following ionization. Overall, this demonstrates that the presence of substituents attached to the core with electron-withdrawing or electron-donating abilities comparable in magnitude to that of the cyano groups can induce significant changes in the intramolecular charge reorganization upon charging, which can have a dramatic impact on the cyano vibrations, when cyano-substituent interactions exist.

Summary and outlook
For the doping of OSCs, the need to discriminate between fractional and integer charge transfer exists because molecular dopants can form charge-transfer complexes and lose doping efficiency. This process can be prevented by tweaking the symmetry and increasing the dimensionality of new molecular dopants specifically designed for this purpose. Such novel doping agents then require reliable tools to determine their efficiency and charged state, where infrared spectroscopy is the method of choice. Confirming the efficacy of a given design strategy to prevent charge-transfer complex formation, in particular for the very likely case that such molecules feature cyano groups, requires a fundamental understanding of the factors which can impact cyano vibrations, other than molecular charge excess. These factors can either be intermolecular or intramolecular in origin, and should be critically examined prior to the determination of the degree of charge transfer based on cyano vibrations, as they can influence cyano vibrations in a similar way as molecular charge excess.
In this work, we demonstrated that cyano vibrations, while highly sensitive to an excess of molecular charge which can arise from intermolecular charge transfer have little specificity towards it, and are impacted by a number of additional factors which can obfuscate the straightforward interpretation of charge transfer phenomena. We investigated how these factors impact cyano vibrations by focusing on the concepts of molecular and atomic charges as well as on their role in defining cyano vibrations. More specifically, we explicitly showed and exemplified that, (i), cyano vibrations do not necessarily shift linearly with the molecular charge, which can lead to significant errors in deducing the latter from experimental infrared data, (ii), geometrical degrees of freedom can shift and create/extinct cyano vibrations, (iii), electrostatic interactions, even intramolecular ones, can impact cyano vibrations and, (iv), the process of reorganizing molecular charge excesses or deficiencies into atomic charges should be viewed as non-linear, with an impact on cyano vibrations to be assessed on a case-to-case basis depending on the respective molecule. We note that, while we did not consider intermolecular interactions explicitly, these findings are applicable in a more general way to all interactions with molecular dopants. Intermolecular interactions can induce effects similar to those probed here, for instance, by leading to geometrical changes or electrostatic interactions. Therefore, the next logical step required to gain a more robust understanding of cyano vibrations in doped OSCs is the systematic investigation of donor/acceptor complexes, where intermolecular interactions play a defining role in the resulting geometry and transfer of charge. While the structure of several prototypical complexes is known and can be used as starting point, molecules involved are often planar and, based on the results here, should be relatively well-behaved. As such, complexes involving potentially more problematic molecules,