Evaluation of thermal evaporation as a deposition method for vacuum-processed polymer-based organic photovoltaic devices Open Access

π-Conjugated organic polymers have sparked interest in optoelectronics due to their unique electrical and optical properties as organic semiconductors.¹ They offer advantages such as low-cost synthesis and facile modulation of the optoelectronic properties by incorporating different moieties, as well as large-area and lightweight features for the fabrication of optoelectronic devices.^2,3 Their ability to delocalize charges across the structure enables the creation of intrinsic semiconductors with proper bandgaps (Eg), easing efficient light absorption and charge transport—critical factors in photovoltaic applications.⁴ Polymer-based organic photovoltaics (OPVs) have shown promising results; however, conventional approaches, such as solution-based deposition,⁵ present several limitations. Several factors, such as solution viscosity and solubility, constrain the choice of solvents. For instance, there is the possible exclusion of high-molecular-weight polymers that cannot be dissolved, along with the necessity of using orthogonal solvents to deposit additional layers, preventing the perturbation of the previously deposited films. In addition, spin coating poses challenges in precisely controlling the active layer composition and thickness. At the same time, processes such as thermal annealing can induce unfavorable morphological changes that negatively impact device performance. These limitations have driven the exploration of alternative deposition techniques, such as high-vacuum thermal evaporation,⁶ a scalable method widely used in several technologies, including organic light-emitting diodes (OLEDs) for display applications,^7,8 which offers precise control over film thickness and improved reproducibility. In addition, from a sustainability and atomic economy perspective, thermal evaporation presents significant advantages over solution-based techniques like spin coating. In spin coating, a considerable amount of material is lost: if the material does not dissolve completely, it must be filtered out, and during high-speed rotation, only a fraction of the solution adheres to the substrate, while the rest is ejected by centrifugal force, leading to material waste. In contrast, thermal evaporation allows for more efficient material utilization, minimizing losses and reducing environmental impact. These advantages instill optimism about its potential to address critical challenges faced by solution-based methods, such as for small molecules and polymers.¹ 

Most initial studies on thermally evaporated polymers focused on polyaniline (PAN) and its conductive properties.^9–12 Uvdal et al.⁹ deposited thin films of PAN using a Pyrex boat capable of reaching temperatures over 350 °C. Subsequently, these films were exposed to an NH₃ (g) atmosphere. Size exclusion chromatography (SEC) demonstrated that various molecular weights were obtained. However, there was a mismatch with the molecular weight measured before evaporation. To reconcile this discrepancy between high molecular weights and short chains, it was suggested that the chains observed in the optical spectra were conjugated segments free of localized defects within larger polymeric molecules. These chain segments could merge into larger molecules at branching or cross-linking sites. In the study by Plank et al.,¹⁰ emeraldine films were deposited at two different thicknesses, 100 Å (ultra-thick) and 1000 Å (thick), using a vapor pressure of ∼5.0 × 10⁻⁶ torr and heating a quartz cell to 325–375 °C. These films were doped with HCl, resulting in high ordering and higher electrical conductivity for the ultra-thin film. On the other hand, the thick film showed more significant branching in the polymer deposition, leading to lower conductivity. It is important to note that both studies were conducted under different conditions but presented the same optical transitions, such as material oxidation near 2.2 eV and the π–π* transition near 4.0 eV.

In a study by Kovacik et al.,¹³ the deposition of poly(3-hexylthiophene) (P3HT) was carried out using the vacuum evaporation technique with a tungsten boat. The material was evaporated at two temperatures (370 and 420 °C), close to its decomposition temperature (400 °C). During this process, a significant decrease in molecular weight was observed, from 36 000 to 1 500 g/mol, via SEC analysis, as only the shorter fractions of the P3HT polymer were deposited. In addition, UV–visible analysis revealed a blue shift in the absorbance of the evaporated fraction compared to the polymer before evaporation, which Kovacik attributed to the deposited low molecular weight. Later, Kovacik et al.¹⁴ compared polythiophene (PTh) and P3HT as donor materials in bilayer OPV devices. Surprisingly, the PTh cell showed a 70% increase in power conversion efficiency (PCE) compared to P3HT-based devices. This experiment was based on the evaporation of PTh at different temperatures (275–300 °C), below its decomposition temperature. UV–visible analysis showed that the same polymeric fraction was obtained in both depositions. However, the polymer evaporated at 300 °C showed greater amplitude and captured the two peaks seen in the absorption spectrum of the polymer before evaporation. In addition, SEC was used to analyze the soluble evaporated polymer fragments. Due to the lack of aliphatic chains in PTh, the evaporated fractions presented lower solubility than the initial material. This revealed a decrease in molecular weight and polymerization degree for both materials. In another related study, Lanzi et al.¹⁵ compared the P3HT/C₆₀ system with an octylthiophene-derived oligomer (OCT)/C₆₀, achieving PCEs of 0.35% and 1.21%, respectively. Similar characteristics to those of previous studies were seen for P3HT. However, it was noted that polymer evaporation generated a pyrolysis reaction, as evidenced by the pressure increase before the deposition temperature (350 °C). In addition, the absorption of OCT was shifted from 462 to 518 nm after the evaporation process.

Considering the low thermal conductivity of polymers, innovative boat designs are required to evaluate the evaporation of polymeric materials in a controllable manner, compared to traditional tungsten or molybdenum boat sources. Lee and Tang¹⁶ reported a new, efficient heating boat based on a conical glass boat equipped with a metal spiral inside to evaporate small molecules for OLED devices. Since the contact surface between the material to be evaporated and the heater is increased compared to commercial boats, heat transfer from the coil to the material is improved. Due to this increase in the contact area, this boat appears as an ideal option for the thermal evaporation of polymers, as it promotes more uniform and controlled evaporation. In addition, the boat features two outlets for the evaporated material: one at the top for material deposition and another on the side with a variable opening to control the deposition rate using a quartz crystal monitor sensor. This boat has also been used to fabricate small-molecule OPVs¹⁷ and perovskite materials with highly volatile precursors, like methylammonium iodide.¹⁸ 

In addition, Schiff base polymers, including poly(azomethine)s, have been positioned as promising candidates for next-generation OPV devices due to their exceptional optoelectronic properties, characterized by increased rigidity and conjugation.^19–21 Their robust thermal stability and the ability to adjust their Eg value make them particularly attractive for applications requiring precise control during the deposition process, a crucial aspect for achieving consistent and reliable device performance. The polymer PAZ2ThA2 reported by Sobarzo et al.²² was used to develop the method and fabricate OPV devices due to its thermal properties, making it an optimal candidate to evaluate the feasibility of depositing thin films of polymer materials by thermal evaporation.

Based on the previously reported procedure,¹⁷ ITO pre-patterned glass substrates underwent a batch-cleaning treatment. Initially, they were cleaned using a detergent solution (ALCONOX), followed by rinses with deionized water (DI). Subsequently, they were immersed in an ultrasonic bath for 10 minutes, alternating between DI, acetone, and isopropyl alcohol. Finally, the substrates were dried using a flow of nitrogen gas. The OPV devices were fabricated using a bilayer-heterojunction type architecture: ITO|MoO_x (8 nm)|PAZ2ThA2 (x nm)|C₆₀ (40 nm)|Bphen (8 nm)|Liq (1 nm)|Al (100 nm). All films were prepared by thermal vapor deposition in a vacuum chamber (<5.0 × 10⁻⁶ Torr). The material PAZ2ThA2 was previously synthesized.²² For all experiments, polymer-only films were made to sample the evaporated fractions of the polymer by UV–visible spectroscopy analysis, using THF as the solvent for solutions and glass substrates for thin films, and by SEC, along with a static light scattering DAWN EOS in line with an Optilab DSP interferometric refractometer (both were obtained from Wyatt Technology) using THF as the mobile phase. Tooling factor calibrations were conducted by measuring film thicknesses with a Filmetrics Pro-Film 3D optical profilometer. Current density-voltage (JV) characteristics were recorded with a Keithley 2400 source meter in the dark and under irradiation at 800 W·m⁻² generated by a Solux 3SS4736 with a 50 W 47 K halogen lamp. The light intensity was calibrated with a Hamamatsu S1787-12 silicon photodiode. External quantum efficiency (EQE) measurements were performed with an Acton Research SpectraPro-275 monochromator. The devices were manufactured in two batches of six substrates each, where two were used to collect the composition of the vapor fraction at the beginning and the end of each experiment, verified by UV–visible and SEC. The thickness dependence on the device performance was evaluated in three batches. In all experiments, a control device with no polymer was fabricated.

During the deposition process of the PAZ2ThA2, temperature, deposition rate, and evaporation time were controlled using the scheme shown in Fig. 1(a). As PAZ2ThA2 is thermally stable (T_5% = 418 °C) with a glass transition temperature (T_g) of 216 °C,²¹ the boat source was heated up to 220 °C to prevent material decomposition, especially considering the vacuum pressure of the system during evaporation (10⁻⁵ torr). As shown in Fig. 1(b), an increase of 0.1 kW from the power supplied to the source boat was initially applied every 3 min, up to 15 min, as the deposition rate started to increase. Then, a power increase of 0.05 kW was applied every 3 min as the deposition rate reached 1.00 Å·s⁻¹ to continue adjusting the power to maintain the deposition rate as constant as possible. A maximum rate of 1.4 Å·s⁻¹ was observed to decrease later despite the ongoing temperature increase. The parameters of the evaporation profile, including the chamber’s pressure, are described in Table S2. Considering a polydispersity of 1.65 for the starting material, low-molecular-weight fractions would initially evaporate before the deposition rate peaks. After this point, 20 nm of material was collected as an initial sample at the beginning of the evaporation to continue with the device fabrication. Finally, once the different thicknesses of the material were deposited, a final sample with 20 nm thickness was also collected at the end of the evaporation for further analysis. This procedure was repeated for all experiments, where deposition of the material occurred between 200 and 220 °C.

UV–visible absorption profiles of the initial and final samples were measured to analyze the composition of the evaporated fractions. As shown in Fig. 2(a), both fractions presented similar profiles in the solid state. Their optical Eg was determined using Tauc plots (Figure S3), with values of 2.68 and 2.65 eV for the initial and final samples, respectively. While both values are similar, a slightly higher value is observed for the initial sample, considering that lower molecular-weight fractions would evaporate initially during the deposition process. However, this difference is amplified when both samples are measured in THF [Fig. 2(b)], as the Eg for the initial and final samples are 2.68 and 2.76 eV, respectively. The absorption spectrum of the initial sample maintained the peak and shape around 390 nm from film to solution. Nonetheless, the profile for the final sample changed not only in the Eg value but also in the intensity and form of the band, suggesting the formation of possible new fractions during the deposition process, compared to the profile of the starting material with an Eg of 2.50 eV (Figure S4). Regarding the solid-state samples, although there is an overlap between the absorption band of the deposited fractions and C₆₀ around 400 nm, an additional band at ∼ 500 nm of C₆₀ allows the differentiation of the donor and the acceptor for the photogeneration process of the OPV devices.

The initial and final samples underwent SEC to analyze the composition of the fractions obtained after evaporation. As shown in Fig. 3, both samples presented similar elution times, confirming that the composition of the vapor remained mostly constant during evaporation. Nevertheless, the final sample presented a broader peak, consistent with the UV–visible absorption results, where more fractions were collected at the end of each evaporation. It is appropriate to highlight that although the elution time of both samples was similar, around 15 min, the SEC of the starting material presented a prominent peak at 23 min, with an estimated M_w of 14 400 g·mol⁻¹ and M_n of 8 700 g·mol⁻¹, resulting in a PDI of 1.65 (Figure S1). Meanwhile, the evaporated samples presented four-fold values of M_w and M_n at different concentrations, as observed in Figure S2. This increase in the estimated Mw and Mn could be attributed to cross-reactions that, to some degree, occurred during the evaporation process of PAZ2ThA2, mainly due to the amino and aldehyde terminal groups of the polymer, as reported for other polymeric systems.^9,10,15 In addition, it is plausible that the deposited fractions formed clusters, leading to an excessive increase in molecular weight compared to the starting materials, causing a lower interaction with the gel of the chromatographer, leading to shorter elution times.

Five devices were fabricated following the architecture ITO|MoO_x|PAZ2ThA2 (x nm)|C₆₀|Bphen|Liq|Al, varying the thickness of the donor material from 0 to 15 nm, including a control device without PAZ2Th2 (Table I). From the J–V characteristics, the short-circuit current density (J_sc) values increase as the thickness of the deposited layer increases, reaching a maximum of 1.60 mA·cm⁻² (±0.03) at 9 nm.

Regarding the open-circuit voltage (V_oc), the control device exhibits an average value of 0.65 V (±0.05). As the thickness of the donor material increases, a reduction of the V_oc is observed. This change in V_oc was expected, as the exciton dissociation is shifted from the MoO_x|C₆₀ interface—where the V_oc is proportional to the difference between the work function of the hole injection layer (HIL) and the LUMO of the acceptor material—to the PAZ2ThA2|C₆₀ interface, where the V_oc is proportional to the difference between the HOMO of the donor and the LUMO of the acceptor.²³ In contrast, the fill factor (FF) values remain primarily constant, around 0.3–0.4, as shown in Fig. 4(a) and Figure S5. However, as the thickness further increases, a decrease in the generated photocurrent is observed. This is attributed to the fact that beyond 9 nm, the excitons formed near the HIL interface are unable to reach the donor–acceptor interface, causing their excitonic quenching. This characteristic is expected in the bilayer system since the exciton diffusion length for organic materials is typically around 10–20 nm.¹ 

Moreover, the spectral response analysis reflects a current change by increasing the evaporated fractions' thickness, as illustrated in Fig. 4(b). For the control device, the photocurrent generated is dominated by the absorption of high-energy photons by the fullerene, from 350 to 500 nm, at the MoO_x|C₆₀ interface.^23,24 Although a reduction of the photocurrent is observed in that region by adding a 5 nm layer of the deposited material, an increase in the spectral response appears around 600 nm, causing an improvement of the integrated photocurrent J_EQE. This change can be associated with capturing the high-energy photons by the donor layer, which also presents a peak at 400 nm [Fig. 2(a)], allowing the absorption and dissociation of low-energy photons to C₆₀. As the thickness of the donor layer further increases, the entire spectral response improves up to 9 nm, where the evaporated fractions filter the photon absorption and exciton dissociation at the donor–acceptor interface. This is further observed by the devices' power conversion efficiency (PCE) value.

Thermally stable polymers can be exposed to vacuum-based processes, such as thermal evaporation, to deposit fractions of polymeric materials to fabricate OPV devices. Although low deposition rate values were obtained, our results demonstrate that the composition of the evaporated fractions can remain constant along with the temperature, enabling proper control over the deposited thin-film thickness, as confirmed by UV–visible absorption spectroscopy and SEC analysis. Bilayer heterojunction OPV devices were successfully fabricated, with an optimal deposited thickness of 9 nm of the evaporated fraction as donor layer, with a J_sc of 1.6 ± 0.03 mA·cm⁻², while FF and V_oc remained primarily constant. These results pave the way for fabricating vacuum-processed OPV devices in bulk configuration for higher efficiency donor–acceptor systems. However, deposition rate values must be improved, as the low deposition rate hinders the evaporated fractions' co-deposition with C₆₀ or other acceptor molecules.

Additional figures, including the physicochemical characterization of the polymeric material, optical spectroscopy, deposition parameters, and dark current–voltage characteristics of the OPV devices, can be found in the supplementary material.

The work was funded by the internal UC fund DIPOG Ciencia Aplicada 2023.

The authors have no conflicts to disclose.

Alexis F. González: Formal analysis (equal); Investigation (equal); Methodology (equal); Visualization (equal); Writing – original draft (equal). Patricio A. Sobarzo: Formal analysis (equal); Resources (equal); Writing – review & editing (equal). César Saldías: Investigation (equal); Methodology (equal); Resources (equal); Visualization (equal); Writing – review & editing (equal). Claudio A. Terraza: Funding acquisition (equal); Resources (equal); Writing – review & editing (equal). Felipe A. Angel: Conceptualization (equal); Funding acquisition (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article and its supplementary material.