Highly Ordered Small Molecule Organic Semiconductor Thin-Films Enabling Complex, High-Performance Multi-Junction Devices

Abstract

Organic semiconductors have opened up many new electronic applications, enabled by properties like flexibility, low-cost manufacturing, and biocompatibility, as well as improved ecological sustainability due to low energy use during manufacturing. Most current devices are made of highly disordered thin-films, leading to poor transport properties and, ultimately, reduced device performance as well. Here, we discuss techniques to prepare highly ordered thin-films of organic semiconductors to realize fast and highly efficient devices as well as novel device types. We discuss the various methods that can be implemented to achieve such highly ordered layers compatible with standard semiconductor manufacturing processes and suitable for complex devices. A special focus is put on approaches utilizing thermal treatment of amorphous layers of small molecules to create crystalline thin-films. This technique has first been demonstrated for rubrene—an organic semiconductor with excellent transport properties—and extended to some other molecular structures. We discuss recent experiments that show that these highly ordered layers show excellent lateral and vertical mobilities and can be electrically doped to achieve high n- and p-type conductivities. With these achievements, it is possible to integrate these highly ordered layers into specialized devices, such as high-frequency diodes or completely new device principles for organics, e.g., bipolar transistors.

1. Introduction: Why High Mobility?

Organic semiconductors have seen a rapid development in recent years as they enable novel electronic applications beyond the possibilities of classical crystalline inorganic semiconductors.¹ Their key advantages are thin-film deposition onto almost any substrate, mechanical flexibility, transparency, and environmentally friendly materials, to name a few. Applications encompass both electronic and optoelectronic devices and systems.

The most successful applications so far are in the field of optoelectronic devices: Displays based on organic light-emitting diodes (OLEDs)²⁻⁷ have already obtained a multibillion $ market, mainly for smartphone and TV applications. Due to their efficient self-emitting nature and favorable spectra, OLED displays offer superior color space and contrast with respect to conventional LED displays. Moreover, they have been realized on flexible substrates, making foldable and rollable displays possible.

Organic solar cells (OSCs)⁸⁻¹⁴ are currently entering mass production. In particular, they are attractive for building-integrated photovoltaics since their light weight and flexibility allow for easy mounting of the modules onto roofs and facades.

Besides these two optoelectronic applications, there has been limited further application of organic semiconductors. OLED lighting has not found widespread application, mainly due to cost. Even less utilized are electronic applications based on organic thin-film transistors (OTFTs),¹⁵⁻²² which aside from some use in flexible e-ink displays are not commercially applied as of today.

What limits a broader application of organic semiconductors? To what extent do the morphology and electronic properties of organic materials, such as charge carrier mobility, correlate to their application potential, and to what degree are these materials compatible with standard mass production capable processes? For this discussion, the following distinctions are helpful:

• small molecule organic semiconductors vs polymer materials,

• lateral vs vertical devices and

• single- vs multilayer (junction) devices.

1.1. Small Molecules vs Polymers

It should first be mentioned that the term ”organic semiconductor” can refer to diverse classes of materials and processing methods: One must then differentiate small molecules from polymers. While small molecules can be processed via solution or vacuum deposition, polymers are usually restricted to solution processing. Currently, OLED applications and OSC mass production are based predominantly on vacuum processing. The only commonly known transistor application, e-ink displays, uses polymers and, therefore, solution processing.

To better distinguish between polymers and small molecule organic semiconductors (SM-OSCs), it is beneficial to first differentiate the electronic structures of organic and inorganic semiconductors: Organic semiconductors are usually composed of molecules that do not possess dangling bonds in their electronic system. This structure has the significant advantage that the tolerance for disorder is higher than for inorganic semiconductors, since the disorder does not necessarily create dangling bonds and, subsequently, trap states within the energy gap. On the other hand, covalent bonds are missing between individual molecules/polymers. Instead, organic materials are held together by a rather weak van-der-Waals coupling. Consequently, the transfer integrals and, therefore, the bandwidths are significantly smaller than those in inorganic materials. As an example, the band structures of some SM-OSCs were determined by photoemission experiments, observing typical electronic bandwidths of a few hundred meV.²³⁻²⁵ These bandwidths are about 1 order of magnitude smaller than in typical inorganic semiconductors. Accordingly, the effective masses of charge carriers in organic small molecule semiconductors are significantly larger than those of inorganic semiconductors and can be well above the free electron mass.²³ In the most simple transport model, the Drude model, this mass enters inversely into the mobility; thus, for fundamental reasons, the mobility of charge carriers in organic semiconductors seems to be reduced by roughly 1 to 2 orders of magnitude compared to that in inorganic semiconductors.

The situation is different for polymer materials at first glance: Here, the binding mechanism is covalent along the chain, resulting in a 1D band structure with larger bandwidths. Thus, better transport can be expected along the π-conjugated polymer backbone if specific requirements are met: First, the transport along the individual backbone of each molecule must be efficient. To achieve this, the polymer chains need to be free of chemical defects to a certain degree, such as those emerging during the synthesis or caused by irreversible oxidation. Furthermore, the chains should be reasonably ordered and aligned along the transport direction (e.g., no twisting of the backbone). Second, the interchain transfer of charge carriers must be reasonably efficient for the advantage of the efficient transport along the chain not to be lost. It has turned out that the benefits of polymers (i.e., efficient charge transport along the chain) do not necessarily translate well to improved device properties. For example, while several reports are claiming high mobilities for holes in conjugated polymers, reaching values of up to 100 cm² V^–1 s^–1,^26,27 the transistor devices with the highest switching speed (normalized by the driving voltage) are based on small molecules.^20,28 The highest switching speed for polymer-based transistors has been achieved with an 2D semi-paracrystalline.^29,30 Since the formation of polymer-based poly- or even single crystals is much more complex and has not been rigorously proven advantageous for devices, we will focus our review on small-molecule-based systems. For these materials, a clear connection between structural order (including static and dynamic) and device performance has been demonstrated.^31,32

1.2. Lateral vs Vertical Devices

A transistor’s switching speed (usually referred to as the unity-gain cutoff frequency) is, in essence, given by the transconductance across the channel divided by the capacity to be charged when the device is switched.¹⁷ The transconductance is directly proportional to the charge carrier mobility in the semiconducting material if contact resistance can be ignored. Furthermore, parasitic overlap capacitances might also limit the unity-gain cutoff frequency, as they do not contribute to the transconductance (although some overlap might be needed to enable efficient charge carrier injection). The most investigated structures are planar thin-film transistors, where charge transport is in the lateral direction, parallel to the substrate. Comparably high mobilities have been achieved in this device type based on organic materials (≥20 cm² V^–1 s^–1 for bulk single crystals³³ and ≈10 cm² V^–1 s^–1 in thin-film crystals^34,35). However, the transconductance is governed not only by the charge carrier mobility but also by the channel length, requiring downscaling of transistors to reach higher frequencies. Furthermore, if the channel length is scaled down, other effects, such as contact resistance, start influencing the transconductance (typically at a channel length of ≤10 μm). The impact of the channel mobility and contact resistance in lateral OTFTs on the final device performance has been thoroughly discussed by Klauk et al.¹⁹ However, the promise of low-cost fabrication imposes constraints on the lateral channel length to sizes in the micrometer range. An alternative to circumventing this limitation is to employ vertical organic transistors whose channel length is in the nanometer range. Such devices can surpass lateral OTFTs in terms of switching speed (often compromising other transistor properties, such as on/off ratio).

Optoelectronic devices such as OLEDs and organic solar cells are usually vertical devices, with current flowing perpendicular to the substrate. The current densities are much smaller than in transistor channels and on the order of tens of mA cm^–2. The electronic transport properties needed for carrying these currents without significant voltage loss are moderate due to the short distance of transport in vertical devices (≈100 nm). Typically, mobilities in the range of 10^–5 cm² V^–1 s^–1 to 10^–3 cm² V^–1 s^–1 are sufficient to achieve acceptable voltage losses. These mobilities are easily reached even for highly disordered organic materials. For organic solar cells, the mobilities required are higher than those for OLEDs because the tolerable voltage losses are smaller.

However, in certain situations, higher mobilities for optoelectronic devices are desirable: If OLEDs should be switched on and off quickly, e.g., when used as emitters for optical data transmission, higher mobilities are crucial to allow rapid carrier injection.³⁶ Similarly, mobility becomes crucial when organic solar cells are used as photodetectors, and high speeds are required to obtain rapid carrier sweep-out.³⁷ Finally, high charge carrier mobility is seen as a crucial factor in order to realize electrically pumped organic lasers.³⁸

1.3. Designing Crystalline Films for Complex, Multi-Junction Organic Electronic Devices

One of the key properties of an organic semiconductor determining the speed and efficiency of corresponding devices is, thus, the charge carrier mobility. This property is intricately linked to not only the energy landscape and overlap of orbitals of the individual molecules but also the arrangement of the molecules with respect to one another. In general, the higher the static order (crystallinity) and the lower the dynamic disorder in the film, the higher the expected charge carrier mobility.³¹

The conductivity of the semiconductor film, however, does not exclusively depend on the mobility but also on the density of free charge carriers. This property can be manipulated via the introduction of doping. For this reason, doping has been adopted by several industrial players to improve the conductivity of transport films, reduce the impact of injection resistance, or improve light emission and absorption in organic devices, such as light-emitting diodes or solar cells. However, not every organic semiconductor can be doped efficiently or at all.³⁹ Dopability is thus an important material property that is necessary for more complex device types.

Furthermore, most vertical devices and even some advanced types of lateral transistors require several separate layers of organic thin-films, each serving a specific function. These can be layers of different materials or simply layers with different doping concentrations and doping types. Such devices containing several layers and junctions are considered ”complex devices” in this review. Although their fabrication is more challenging and possibly expensive compared to simpler device stacks, they have been widely adopted on an industrial scale as their inherent complexity ensures excellent device performance, stability, and tunability of properties.

To achieve the next step in the implementation of organic semiconductors for complex multi-junction devices, an increase in mobility in the vertical and lateral direction is required. As stated earlier, fundamental properties suppress the maximum possible mobilities in organic semiconductors compared to inorganic materials. It is thus of particular importance to maximize the order of organic films to reach as high mobilities as possible and alleviate this disadvantage. Introducing structural order in the form of organic polycrystals and single crystals, enabling band-like charge transport, is an obvious strategy to increase the charge carrier mobility.

The most common solid form of many organic semiconductors is, however, amorphous due to the weak forces resulting from van-der-Waals bonds between the individual constituents. Under specific circumstances, solids can be made to form polycrystals or even single crystals. A large assortment of methods are described in the literature to achieve this goal. However, beyond the simple growth of crystals, it is challenging to meet the requirements for use in various types of complex organic electronic devices: We suggest the following list of requirements for material systems and processes to ensure suitability for the manufacturing of advanced electronic devices while maintaining the distinct advantages of organic electronics. Additionally, compatibility with standard semiconductor manufacturing is seen as an advantage.

• 1.The resulting crystals must be in thin-film form and preferably be compatible with standard thin-film technology (requiring, for example, thermal and mechanical stability for lithography, printing or similar processes).
• 2.The growth processes should allow for the reproducible formation of high-quality crystals with ideally 100% surface coverage or the possibility of growth at defined positions.
• 3.Crystals must be dopable with both n- and p-type dopants.
• 4.Several layers of different (crystalline) semiconductors or identical but differently treated semiconductors must be stackable to create complex, multi-junction vertical or lateral devices.
• 5.Growth processes should not add an undue amount of additional complexity to manufacturing.

In this topical review, we provide an overview of materials and crystallization methods enabling the fabrication of highly ordered thin-film crystals based on small molecule organic semiconductor materials (SM-OSCs) with high charge carrier mobility (described in Chapter 2). Following our bullet point list above, we focus on thin-film crystals grown from films created under vacuum conditions, which enable the fabrication of complex, multi-junction devices in both the lateral and vertical direction. As we discuss in Chapter 3, efficient device concepts can be implemented, allowing, e.g., to reduce contact resistance, design junction properties, and increase the speed of operation. The most commonly used material for such thin-film crystals is rubrene, due to its high charge carrier mobility and ease of handling. Hence, we focus mainly on this material. However, we also review other material classes that form large thin-film crystals under similar conditions, proving the universal character of the crystallization method. In particular, this procedure might enable complex multi-junction (homo- or heterojunction) devices such as light-emitting diodes, photodetectors, or transistors, opening a pathway toward higher device efficiency. We do not rule out crystallization from solution, as it would probably help to reduce fabrication costs. However, in the context of doped films for complex devices, such methods have not yet been fully developed.

2. Approaches toward Highly Ordered Thin-Films

Organic crystals can be formed via three pathways: from a melt, solution, or gas phase. However, techniques like the Bridgman and Czochralski processes that are standard for inorganic semiconductors are rarely useful in the scope of crystallization of organic materials; the temperatures needed to melt most semiconductors of interest require inert or even vacuum processing to avoid immediate oxidization.⁴⁰ However, growth from the gas phase requires the same systems and techniques but offers better results.

Growing crystals from solution is an appealing and possibly cost-efficient process. Critical parameters of the process are the type of solvent, speed of solvent evaporation, substrate temperature, surface tension, and speed of motion (e.g., in the case of blade or spin-coating). Extensive reviews are available covering the vast amounts of parameters that influence crystallization from solution (e.g., refs (41−44)). Techniques like blade- or shear-coating can imprint a specific orientation for solvent flow and thus enhance crystal growth in specific directions, resulting in high-performance devices.^34,35 However, also for rubrene, which is the material mainly discussed in this article, crystal growth from solution has been analyzed. For example, Matsukawa et al.^45,46 investigated the growth of bulk rubrene crystals from different solvents. Depending on the solvent and the conditions, monoclinic, triclinic, and orthorhombic free-standing bulk crystals are grown. However, in terms of mobility, the performance is worse than for vacuum-processed devices. Similar results are shown by Huang et al. for rubrene microcrystallites.⁴⁷

While it is relatively easy to create crystals from solution with comparably high quality, certain demands cannot yet be met. In particular, the integration of doping and multilayer stacking is challenging. The creation of homo- or heterojunction devices is hence almost impossible.

Growing organic crystals from the gas phase is a common technique since it produces high-purity crystals. The two methods used are growth via (re)sublimation and molecular beam epitaxy/vacuum evaporation. In either case, the material is evaporated or sublimed within an ultrahigh vacuum or carrier gas flow (see Figure 1a). The substrate to be deposited onto is at a temperature below the sublimation point of the material. The closer the temperature of the target to the sublimation point of the semiconductor, the slower the deposition/growth takes place and the more time individual molecules have to find their position and orientation within the growing crystal lattice. The resulting crystal is thus of higher quality at these higher temperatures (see Figure 1b).

Figure 1.

(a) Illustration of a carrier-gas-assisted sublimation of rubrene with a thermal gradient. Reproduced with permission from ref (48). Copyright 2018 American Chemical Society. (b) Photographs of rubrene crystals grown from the vapor phase. Reproduced with permission from ref (48). Copyright 2018 American Chemical Society. (c) Molecular structure of rubrene (top left), orientation of rubrene molecules in an orthorhombic crystal structure (right), and triclinic structure (bottom left) (only the tetracene core of the rubrene molecule is depicted). Rubrene has a herringbone structure with π–π stacking. Reproduced with permission from ref (49). Copyright 2016 American Chemical Society. (d) Air gap TFT based on a rubrene single crystal with (e) Hall effect mobility in different directions. Reproduced with permission from ref (33). Copyright 2004 American Physical Society. (f) Bulk doping of a rubrene single crystal with FeCl₃ and the density of ionized dopant sites. Reproduced with permission from ref (50). Copyright 2017 Wiley-VCH GmbH.

Most macroscopic (mm-sized) organic crystals are grown through sublimation or from solution, the former enabling some of the highest-performing devices presented in literature, which are, to a large extent, devices based on rubrene single crystals^33,51−53 (see Figure 1c for the molecular structure of rubrene and representations of its triclinic and orthorhombic unit cells). Despite the high charge carrier mobility exhibited by these large single crystals (record values measured in vacuum-gap field-effect transistors exceed ≈30 cm² V^–1 s^–1, cf. ref (33) or Figure 1d and e), their use is extremely limited due to their poor handling.

Following the previously defined list of requirements regarding the use in complex devices, the question arises of how the charge carrier density can be manipulated in bulk organic single crystals. Doping has been implemented by Ohashi et al.⁵⁰ by coevaporation of FeCl₃ into a homoepitaxial layer on the surface of a rubrene bulk crystal (Figure 1f). Kim et al.⁴⁸ demonstrate surface doping with fluoroalkyltrichlorosilane (FTS); Krupskaya et al. does so with fluorinated tetracyanoquinodimethane (Fx-TCNQ) as well.⁵⁴ Furthermore, Zhang et al.⁵⁵ demonstrated how the surface oxidization state of bulk rubrene crystals influences conductivity and mobility. However, all examples of doping in bulk crystals show doping homogeneously on the entire surface and not in the bulk. All examples feature only one carrier type - holes. Since bulk crystals are also exceedingly difficult to handle, they seem unfavorable for complex devices and larger-scale manufacturing.

2.1. Abrupt Heating and Thermal Annealing for Fabrication of Thin-Film Crystals Suitable for Complex Devices

Thin-films with a high degree of crystallinity can be made via vapor phase deposition under high or ultrahigh vacuum conditions. For example, pentacene grows into microcrystallites on a suitable substrate⁵⁶ under such conditions. However, sub-μm crystallites are of little use in, for example, OTFT devices with several μm of channel length.

One way to grow macroscopically large thin-film crystals is by recrystallizing an initially amorphous or low-crystallinity film by introducing energy in the form of heat or irradiation. This process works particularly well with rubrene and was documented for the first time by Park et al., initially by accident (see Figure 2a) and later through in situ heating during and after evaporation.⁵⁷ In these initial partially crystallized films, only relatively low hole mobilities of 1.23 × 10^–4 cm² V^–1 s^–1 were measured.

Figure 2.

(a) (top) First demonstration of initially amorphous rubrene films by abrupt heating by Park et al. Reproduced with permission from ref (57). Copyright 2007 American Physical Society. (bottom) Illustration of the homoepitaxial growth scheme. Reproduced with permission from ref (58). Copyright 2021 Wiley-VCH GmbH. (b) Polarized microscopy image of amorphous (I), orthorhombic (II), and triclinic phases (III) of rubrene (scale bar 200 μm). Reproduced with permission from ref (49). Copyright 2016 American Chemical Society. (c) Polymorph selection by choice of annealing temperature. Each temperature represents three 20 nm films on indium–tin-oxide annealed for 3 min. Reproduced with permission from ref (49). Copyright 2016 American Chemical Society.

However, this procedure results in an increase in the charge carrier mobility of over an order of magnitude compared to that of the fully amorphous film. A method to control crystal growth by rapidly heating a previously amorphous layer of rubrene was later introduced by Lee et al.^59,60 By heating at the correct temperature with specific external parameters, arrays of large single-crystal platelets can be grown. These publications focus on the orthorhombic phase of rubrene, which is well suited for lateral thin-film transistors (see Section 3.2). A closer investigation of the crystallization procedure and a clear distinction between the triclinic/spherulitic and other polymorphs of orthorhombic/platelets was later presented by Fielitz et al.⁴⁹ (see Figure 2b and c for an illustration of the crystallization process as a function of temperature and Figure 3a for a polarization microscopy image of the different polymorphs). The polymorphs of a thin-film created by this abrupt heating method can be chosen to a certain degree via the heating temperature and the abruptness of the heating step. Some publications mention that the recrystallization must take place in darkness and in an inert atmosphere.^58,60,61

Figure 3.

(a) Polarization microscopy images of triclinic spherulites, orthorhombic spherulites, and orthorhombic platelets of rubrene grown on a silicon wafer under different process conditions. Unpublished data from the authors of this article. (b) “Hole” defects (solid arrows) and line defects (dashed arrows) observed by atomic force microscopy (AFM) on a substrate with an 80 nm adlayer of rubrene. (c) AFM image of a 15 nm adlayer grown under the same conditions as those of b showing the onset of screw dislocations with a hole starting to form in the center (inset). Both panels are reproduced with permission from ref (64). Copyright 2017 American Chemical Society.

A combination of this rapid heating method with deposition from solution is shown by Jo et al.⁶² Here, rubrene is dissolved and mixed with a binding polymer. Films are deposited via spin-coating. Abrupt heating is performed immediately after deposition. Using this method, at least 125 °C is required to trigger the crystallization into triclinic crystals, while a minimum of 170 °C is necessary for orthorhombic. However, compared to those of devices made with the simple heat treatment of evaporated films, the resulting field-effect mobilities are lower. It might nevertheless be a suitable method for low-cost fabrication in the future.

These thin layers of rubrene are well ordered, although the quality is not on the level of free-standing single crystals grown in a furnace (quality refers here to the highest charge carrier mobility reached in thin-film transistors). In this form, however, their application in devices is mainly restricted to lateral thin-film devices, i.e., thin-film transistors. An advanced technique to extend the use of these thin-film crystals to vertical devices is described first by Verreet et al.⁶³ and then further by Fusella et al.^64,65 They used the initially formed crystal arrays as a template for a sequential homoepitaxial growth step. By adding additional rubrene molecules, it is possible to create void-free films of any desired thickness while maintaining a high degree of order and with growth occurring via a simple method.

The solar cells shown by Verreet et al.⁶³ are thus the first use of these rubrene layers in stacked vertical devices. The devices themselves are discussed further in section 3.5.

A detailed analysis of the epitaxy process itself is given by Fusella et al.⁶⁴ They investigated the growth method under structural considerations focusing on growth defects. The growth mechanisms in these films are found to be of Stranski-Krastanov type. Counterintuitively, the surface roughness does not increase with a greater layer thickness. The transition into island growth and the prevalence of line and screw defects suggest a strain within the crystal structure that should not occur in a crystal grown via homoepitaxy (see Figure 3b and c). Even layers grown on a bulk rubrene single crystal, eliminating the substrate mismatch, can still feature significant defects if growth conditions are not ideal. A deeper analysis about homoepitaxial growth in general is provided in a recent publication by Dull et al.⁶⁶ for which growth modes and surface properties of several organic molecules were investigated in experiment and simulation. They discovered that the property of whether molecules resemble 3D objects rather than rodlike or flat molecules is a suitable predictor for growth-related behavior.

Another molecule that can be grown as thin-films of orthorhombic crystals is anilino squaraine with isobutyl side chains (SQIB), as presented by Balzer et al.⁶⁷ These films are deposited from solution via spin-coating and subsequently annealed at 180 °C. The analysis is focused mainly on optical characteristics. A subsequent publication by Funke et al.,⁶⁸ demonstrates an elaborate analysis of the optical properties of such birefringent thin-film crystals.

In a further publication by the group of Rand,⁶⁵ an additional underlayer is introduced to facilitate reproducible crystallization of the seed crystals on different substrates. They found that the glass transition temperature T_g of the underlayer is a strong predictor for successful crystallization of the platelet-type crystals. They argue that the decisive factor for proper crystallization is the spatial orientation—hence, rotation—of the rubrene molecules. This theory is supported by results from Dull et al.⁶⁹ These findings are, however, in contrast to experiments of Sinha et al.,⁷⁰ who suggest that crystallization is governed by the transport of individual molecules across the surface, which would be influenced by the hydrophilic/hydrophobic nature of the substrate.

Based on the spatial orientation argument, Verreet et al. conclude that too high of a transition temperature hinders the proper orientation of the individual molecules due to the rigidity of the substrate.⁷¹ On the other hand, when the transition temperature is too low, thermal noise of the underlayer causes too much distortion, and the arrangement of already-positioned molecules is not stable. They proposed a sweet spot of the bulk glass transition temperature at around 70 to 90 °C for the growth of rubrene platelet crystals.

The micro- and macroscopic crystal order of the resulting thin-film depends on the temperature during sintering and the energetic landscape of the substrate. At low temperatures of 80 °C⁵⁷ up to 140 °C,⁵⁸ amorphous rubrene films morph into triclinic spherulite films. The macroscopic arrangement of the spherulites changes strongly with temperature. Temperatures above approximately 150 °C and below 170 °C lead to the formation of orthorhombic platelet-type crystals.^58,65 In contrast, temperatures above 170 °C can lead to the formation of orthorhombic spherulites.^58,65

Covering a substrate with rubrene crystals of the triclinic polymorph can usually be realized with relative ease. Depending on the surface quality and, thus, the density of growth-inducing seed defects, a certain amount of time is needed to cover the entire surface. This time depends on the properties of the substrate and the annealing temperature. Given an inert atmosphere, only a few competitive processes hinder the complete coverage, due to the low temperature. In contrast, for orthorhombic platelet-type thin-films, additional parasitic processes may occur, causing the formation of undesired polymorphs and molecular oxidation. It is possible to significantly increase the surface coverage by giving the desired polymorph a growth advantage via the introduction of a TAPC (1,1-bis[(di-4-tolylamino)phenyl]cyclohexane) underlayer (see Figure 4b). This method can then be generalized to a plethora of various substrates and surface types (Figure 4c), given that specific constraints are met. The resulting average crystal size is then mostly governed by surface properties and, to a lesser extent, by process parameters (see Figure 4a). The charge transport properties of different rubrene polymorphs are studied in a recent publication.⁷² It has been shown that the conductivity can be increased by over 4 orders of magnitude from amorphous to orthorhombic (platelet) rubrene thin-films. The molecular packing has been identified as the main cause for the different transport properties in various rubrene polymorphs. The charge transport in polycrystalline orthorhombic rubrene is limited by the energetic barriers introduced by the grain boundaries. Tan et al. used polarized scanning transmission X-ray and 4D-scanning transmission electron microscopy to elucidate azimuthal orientational discontinuities at smaller radii are erased in spherulite rubrene morphology and suggested a more complex temperature-dependent processing protocol can be used for obtaining hybrid crystalline films with the desirable transport tailored for specific applications.

Figure 4.

Crystal size (a) and surface coverage (b) of orthorhombic rubrene crystals grown on different substrates. The error bars are the standard deviations from three different films. Both panels are reproduced with permission from ref (73). Copyright 2021 Elsevier Publishing. (c) Polarization micrograph of triclinic rubrene crystals on various surfaces. Reproduced with permission from ref (58). Copyright 2021 Wiley-VCH GmbH.

A significant step toward utilizing these thin-film rubrene crystals is introducing doping into the crystallization step and the epitaxially grown layers. In ref (58) we showed that the amorphous seed film can be doped to a high degree with the p-dopant 1,3,4,5,7,8-hexafluorotetracyanonaphthoquinodimethane (F6-TCNNQ) without hindering the following crystallization significantly if spherulitic crystals are targeted. For orthorhombic platelets, only a rather low doping concentration can be used inside the seed film.⁷³ However, epitaxially grown adlayers of rubrene can be doped to a significantly higher degree. Due to the suitable energy levels of rubrene, n-doping was realized as well via the addition of tetrakis(hexahydropyrimidinopyrimidine)ditungsten(II) (W₂(hpp)₄) into bulk layers.

For both p- and n-doped films, the conductivity is significantly increased with higher doping concentrations. Since hole conduction is natively more efficient in rubrene, the electron conductivity in the n-doped films is, on average, significantly lower (more details on conductivity and mobility are provided in Sections 3.1 and 3.2).

Via GIWAXS and AFM measurements of doped and undoped crystals, it is shown that the crystal structure of thin-films stays intact at low and medium doping concentrations. Since the crystallization process and the crystals themselves are remarkably immune to defects, it seems plausible that the overall procedure can be combined with other typical thin-film processing techniques (e.g., photolithography^60,61,74,75) in the future.

2.2. Other Thin-Film Crystals—Beyond Rubrene

The progress achieved with these thin-film rubrene crystals in recent publications is promising. Despite it being a molecule that has been studied for several decades, new methods for synthesis and derivatives of rubrene are still being discussed.⁷⁶ However, why these specific molecules can grow thin-film crystals with such a relatively simple procedure is still insufficiently understood. Despite its apparent success as a basis for many high-performance devices, finding alternative molecules that can grow crystals with the same or a similarly simple method would certainly be advantageous.

However, this turns out to be a difficult task, since the orientation of the molecules within the crystal (twist) can change, as well as the orientation of the molecules toward the substrate. Due to the low internal interaction energy and the comparably stronger interaction with the substrate, the formation of well-ordered films is often suppressed.

Another point of discussion is the structure of the underlying rubrene molecules themselves. Sutton et al.⁷⁷ found two distinct configurations for the tetracene core of rubrene based on simulations. The energetically favored configuration features twisting in the central backbone, resulting in a smaller orbital overlap with its neighbors than its straight counterpart. They suggested the introduction of phenyl side chains to stabilize the molecular configuration in the electrically favorable state. As a rule, molecular candidates should ideally have a flat, stackable core while also featuring side chains to avoid rotation and anchor molecules within the crystal lattice. However, the influence of conformers on the crystallization process is still controversially debated.^69,78

There are not many publications featuring these types of thin-film poly- or single crystals used in complex multi-junction devices. Yang et al.⁷⁹ show thin crystals of Alq₃ in OLEDs in a so-called weak epitaxy process. However, these films are grown via slow evaporation onto a heated substrate, not recrystallization. No doping is involved, and the conductive properties of Alq₃ are rather poor.

Muhieddine et al.⁸⁰ present experiments growing triethylsilylethynyl anthradithiophene (TESADT) from solution via a thermal annealing step after deposition. While the method creates mm-sized crystals, the mismatch between the substrate and the organic crystals induces cracks within the film, significantly reducing the potential performance. In contrast to the crystallization of rubrene thin-films, light seems to improve crystallization for this material system.

Zhou et al.⁸¹ presented a method for growing thin-film crystals of C₁₀-DNTT in an epitaxial style similar to the heat treatment. Here, an initial monolayer of semiconductor is deposited via an ”ultraslow” shear-coating technique, resulting in clean and flat arrays of comparably large crystals. The thickness of these films is then increased via standard vacuum deposition. The already high mobility and stability of C₁₀-DNTT-based devices could be improved this way. While this is a promising method to expand epitaxial growth procedures to materials that cannot be crystallized via the abrupt-heating techniques used for rubrene, the use of solution coating and, in particular, shear-coating in combination with vacuum deposition creates a rather complicated process.

In the publication by Dull et al. of the group of Rand⁶⁹ a set of various molecules is investigated concerning their suitability for crystallization via the rapid heating procedure. They argue that a molecule must have a balance between a tendency to crystallize on the one hand and to form amorphous glass films on the other hand. One important parameter for this is the number of single bonds within the molecule. Molecules like tetracene (the core of a rubrene molecule) have no single bonds and thus tend to crystallize easily into microcrystals upon normal deposition under high vacuum conditions. It is thus much more difficult to rearrange them into large-scale crystals. The presence of single bonds leads to additional degrees of freedom, which can lead to the tendency to form amorphous glass-like films. Dull et al. proposed a sweet spot of four to seven single bonds.

The second requirement for a molecule to form crystalline platelets is a high absolute crystallization driving force, ΔG (change of Gibbs free energy), with the crystal phase being the more stable state. This is the case when there is a large enthalpy difference between the amorphous and the crystalline states ΔH_m, a large difference between the melting temperature and the crystallization temperature (T_m – T_C), and a larger T_m overall. When a higher temperature is needed to rearrange the molecules of the film, the resulting crystal is more resistant to random thermal disturbances and, thus, more likely to form.

The use of an underlayer, shown in one of their previous publications, significantly improves the yield and reproducibility of platelet crystallization. However, it does not influence whether a molecule will crystallize at all. Including rubrene, six different molecules were identified that can be grown into platelet-type crystals, while nine could be formed into spherulitic crystals (see Figure 5 for an overview of these materials).

Figure 5.

Polarized optical microscopy images and molecular structures of the materials considered for thin-film crystallization via abrupt heating. The materials have been categorized into three groups depending on the crystal morphology: platelet-forming, spherulite-forming, and those that resist crystallization. Materials that resist crystallization show little to no crystal growth and, therefore, have no associated microscope images. Reproduced with permission from ref (69). Copyright 2020 American Chemical Society.

Bangsund et al.⁸² made use of 2,2′,2″-(1,3,5-benzinetriyl)-tris(1-phenyl-1-H-benzimidazole) (TPBi) and N,N′-bis(naphthalen-1-yl)-N,N′-bis(phenyl)-2,2′-dimethylbenzidine (α-NPD) to form thin-film platelet crystals with embedded periodic features to be used in photonic devices. Aside from rubrene, no other organic semiconductor that can be grown into platelet crystals with this method has been reported to be used in any type of electronic device until now.

The discovery of this convenient crystallization method in rubrene and, in particular, its use in actual organic devices are comparably recent developments. Only limited research has gone into replicating and surpassing rubrene-based devices so far. However, Bussolotti et al.⁸³ showed that the suppressed interaction between phonons and holes in bulk crystals of rubrene might be responsible for the extraordinarily high mobility measured for these crystals. It might thus be the coincidental result of rubrene’s specific molecular properties, leading to a specific crystal and molecular arrangement. Thus, even if an alternative material system can be found that can be formed into similar thin-film crystals and can be doped in similar ways as rubrene, the resulting devices might not exceed the performance of rubrene in terms of electrical conductivity. However, in contrast to rubrene, the other materials might be of great interest for optoelectronic devices due to their specific triplet energies and emission/absorption properties.

3. Complex, Multi-Junction Devices

3.1. Conductivity of Doped Highly Ordered Rubrene Thin-Film Crystals

Electrical doping in organic semiconductors is a commonly used method to increase the charge carrier concentration and, hence, electrical conductivity of charge transport layers. Furthermore, it effectively lowers the injection/contact resistance at the semiconductor-metal interface by narrowing the charge depletion zone. For small molecule semiconductors, doping is achieved via coevaporation of dopant and host molecules, where mobile holes (p-type) or electrons (n-type) are provided to the host molecule by a charge transfer process between the dopant and host. The transfer efficiency depends on the relative position of the electronic energy structure of the dopant and host molecule.³⁹ Despite the conductivity improvement by charge carrier generation, doping in organic semiconductors has been mainly utilized in amorphous films. Hence, the carrier mobility in these films is limited by disorder, which is not favorable for high-frequency applications of diodes and transistors. To achieve high-frequency device operation, contact resistance, channel length/device thickness, and charge carrier mobility must be optimized in lateral/vertical devices. Therefore, even if ultrahigh-frequency operation can be achieved with nm-scale planar diodes and amorphous semiconductor materials,⁸⁴ highly ordered doped organic thin-films would be desirable, as they offer higher charge carrier mobility compared to amorphous films. Unfortunately, for polycrystalline films of SM-OSCs such as pentacene, it has been shown that doping and conductivity go hand in hand with a reduction in charge carrier mobility caused by impeded crystal formation.^85,86 Eventually, conductivity is even reduced with an increasing doping concentration.

However, for rubrene crystals, it is possible to incorporate a small fraction of dopants, up to around 5 mol % (either organic or inorganic dopants), into initially amorphous films without affecting the crystal growth using the abrupt heating method presented above. Furthermore, the homoepitaxial growth mode of rubrene on a seed layer also continues if dopant molecules are coevaporated. With this doping method, the crystalline rubrene thin-films can be both p-type and n-type doped, where the carrier concentration is increased without significantly compromising the crystalline order (see Figure 6a or ref ⁵⁸). The electrical conductivity of bulk-doped crystalline rubrene can reach up to 0.1 S cm^–1 for both the orthorhombic and triclinic phases in the lateral and vertical directions, respectively (see Figure 6b–c). Although, the average size of crystallites does not differ substantially for doped and undoped thin-films of rubrene, the charge carrier mobility in the doped films is slightly lower⁷³ (see also Figure 6e).

Figure 6.

(a) Energy level diagrams of TAPC, rubrene, F6-TCNNQ, and gold. An electron is transferred from the highest occupied molecular orbital (HOMO) of rubrene to the lowest unoccupied molecular orbital (LUMO) of F6-TCNNQ (orange arrow), creating a mobile hole in the HOMO state of rubrene. Reproduced with permission from ref (73). Copyright 2021 Elsevier Publishing. (b) Lateral conductivity dependence on the doping concentration for p-type doping. The charge transfer is created by modulation doping via a highly doped layer of BPAPF. The commercial p-dopant NDP-9 is used for the modulation doping. Values are similar for F6-TCNNQ doping (at 30 °C). Reproduced with permission from ref (89). Copyright 2022 American Association for the Advancement of Science. (c) Doping efficiency (η_d = N_A^–/N_A, where N_A and N_A are the density of neutral and ionized dopants), as a function of dopant molar concentration determined from Mott–Schottky analysis. The green and orange circles denote modulation-doped triclinic and orthorhombic rubrene thin-film crystals with BPAPF:NDP-9, respectively. The red circle represents bulk-doped orthorhombic rubrene thin-film crystals with NDP-9. Reproduced with permission from ref (89). Copyright 2022 American Association for the Advancement of Science. (d) Contact resistances of the orthorhombic rubrene transistor with 5 nm TAPC with and without 10 nm 5 mol % F6-TCNNQ:rubrene contact doping extracted using a four-point method. Reproduced with permission from ref (73). Copyright 2021 Elsevier Publishing. (e) Charge carrier mobility for intrinsic, doped, and contact-doped orthorhombic thin-film crystals (measured in a TFT configuration): pristine rubrene on SiO₂ (A), 0.5 mol % F6-TCNNQ-doped rubrene on SiO₂ (B), 1.5 mol % F6-TCNNQ-doped rubrene on SiO₂ (C), pristine rubrene on TAPC (D), and pristine rubrene on TAPC with 5 mol % F6-TCNNQ contact-doped rubrene (10 nm) (E). Reproduced with permission from ref (73). Copyright 2021 Elsevier Publishing. (f) Polarized microscopy image of a lateral transistor device with voltage probes for four-point and Hall measurements. The semiconductor material is orthorhombic rubrene. The film contains grain boundaries between differently oriented crystallites. Reproduced with permission from ref (90). Copyright 2020 Wiley-VCH GmbH.

Modulation-type doping represents an alternative to the direct doping method, avoiding a compromise in charge carrier mobility. The modulation doping approach uses a heavily doped wide-gap material in contact with the channel material⁸⁷ or dopants directly on the surface of the highly ordered film (surface doping). The energetic offset between the two materials facilitates the charge transfer from the wide gap material to the channel material. This doping method has the advantage of preserving the order of the channel material, and it is commonly used in inorganic semiconductors for high-electron-mobility transistors.

The conductivity resulting from electrical doping is typically orders of magnitude greater than that of undoped organic semiconductors. While in amorphous semiconductor layers the conductivity and mobility are isotropic, the crystalline order in rubrene thin-films dictates the preferred direction of transport. In particular, as shown in Figure 6b, the lateral conductivity is significantly higher in orthorhombic compared to triclinic rubrene. This is caused by the higher charge carrier mobility in the orthorhombic phase in the lateral direction. The remarkable conductivity enhancement in electrically doped organic semiconductors has been the key to achieving efficient organic electronic devices through lowering the driving voltage, improving charge extraction and reducing the contact barrier.⁸⁸ However, as shown in Figure 6c, bulk doping seems to be less efficient compared to modulation-type doping, which might be related to the narrow density of states of the rubrene crystals, reducing the efficiency of charge transfer from dopant to host.³⁹ In contrast, modulation doping enables high doping efficiency from over 20% up to almost 100% in rubrene crystals, as shown in Figure 6c. The reason behind this is presumably the highly efficient charge transfer due to the well-suited energetic offset. As a result, the narrow density of states in crystalline rubrene does not play a major role in determining the doping efficiency. As shown in Figure 6b, the electrical conductivity of modulation-doped orthorhombic rubrene can exceed 1 S cm^–1, which is at least ten times higher than the bulk-doped crystalline rubrene discussed earlier.

3.2. Lateral Device-Based Doped Rubrene Thin-Film Crystals

The organic thin-film transistor is a simple device concept that can act as a switch to drive flexible and conformal circuits. Moreover, it is also a valuable device for characterizing the charge transport properties of a semiconductor material. It has been shown that abrupt heating of amorphous rubrene thin-films helps to improve the field-effect mobility, albeit the crystal phase and growth have not been optimized.⁵⁷ Orthorhombic rubrene thin-film crystals with good surface coverage are interesting for thin-film transistor applications due to their nearly isotropic charge transport properties.^65,73 Wang et al. have reported thin-film transistors based on orthorhombic rubrene crystalline films with a maximum carrier mobility exceeding 5 cm² V^–1 s^–1 in transistors with a channel length of 20 μm.⁷³ The epitaxially grown doped rubrene crystals were used to reduce the contact resistance down to 1000 Ω cm (Figure 6d). It is worth mentioning that the size of the transistors was miniaturized, such that individual transistors are located within a single-crystal domain. Choi et al. studied combined thin-film transistors and AC Hall effect on polycrystalline rubrene, specifically exploring the effect of the grain boundaries on Hall effect measurements⁹⁰ (see also Figure 6f). The grain boundaries affect the Hall effect measurement, reducing the measured Hall mobility compared to the field-effect mobility. The field-effect mobility in polycrystalline rubrene films is also around a factor of 2 lower than the field-effect mobility within a single-crystal domain.

Another interesting application for doped thin-films is thermoelectric elements. In particular, the low intrinsic thermal conductivity of organic semiconductors renders such materials ideal for thermoelectric applications. Using highly efficient modulation-doped orthorhombic rubrene films with electrical conductivity greater than 1 S cm^–1, the Seebeck coefficient was measured to determine the thermoelectric power factor. The resulting power factor is over 20 μW m^–1 K^–2 at 80 °C, which is among the best values reported for organic thermoelectrics.^73,89 However, the enormous advantage of doping for design becomes more obvious when turning to more complex structures such as ultrahigh-frequency diodes, transistors, photodetectors, or light-emitting diodes.

3.3. Organic Diodes and Rectifiers

Thin-film transistors relying on field effect are the dominant device class in modern microelectronics. However, diodes are still frequently used, as they are easy to integrate and may outperform TFTs in some performance measures such as switching frequency. Furthermore, they excel due to their versatility in terms of functionality. To name a few functions, diodes can be used for rectification, demodulation of radio signals, overvoltage protection, tunable capacitors (varactor diode), or tunable RF resistors. This versatility is certainly connected to the fact that diodes can be built in different architectures, e.g., pn-diodes, pin-diodes, Schottky-type diodes, etc., and their device properties can be adjusted by changing the doping level, the thickness of intrinsic layers, and the semiconductor material. Hence, diodes are complex vertical devices that often possess more than one junction, and tuning the materials and junction properties is an efficient way to tailor device performance.

As a diode should rectify incoming current, its most important static device parameters are the rectification ratio (ratio between forward and reverse current at a given voltage), reverse saturation current, reverse breakdown voltage, and forward resistance. However, the dynamic behavior of a diode is more complex, and the most important parameters are the forward differential resistance, the reverse capacitance (or the space charge layer capacitance), the diffusion capacitance, and the time constants of recombination and charge carrier transit. It should be noted that in contrast to the static parameters, the dynamic parameters are small-signal quantities, as they depend on the DC bias applied (e.g., the depletion capacitance). Furthermore, it is important to note in the context of high-frequency applications that the dynamic parameters are not directly accessible for measurement. Instead, the small-signal response of the system is characterized by the scattering parameter (S-parameters) describing the reflection and transmission of an incoming power wave.

The main motivation for the development of vertical diodes on flexible substrates is to rectify or demodulate radio frequency signals for wireless communication of autonomous electronic systems, e.g., smart labels or sensors.⁹¹ In this context, research on flexible diodes is focused on improving the dynamic behavior of the device and making such diodes operate in the high-frequency (3 to 30 MHz), very-high-frequency (30 to 300 MHz), or ideally ultrahigh-frequency regime (≥300 MHz). Today, rectifiers (e.g., half wave rectifiers connecting a diode to a capacitor) based on organic semiconductor materials are able to efficiently rectify incoming signals above the GHz threshold (see Figure 7 showing literature data on organic diode-based rectifiers). This successful development has been enabled by a continuous improvement of device architectures and semiconductor material properties, as well as interface engineering (see e.g., articles by Kang et al.⁹² or Facchetti⁹³).

Figure 7.

Comparison of results from the literature: charge carrier mobility (obtained in the SCLC regime) vs the 3 dB cutoff frequency measured in organic diode-based rectifier circuits (half wave rectifiers). As shown, the cutoff frequency linearly scales with the mobility. Values for the charge carrier mobility are estimated from the current–voltage current for refs (100−102). Sawatzki et al.⁵⁸ only estimate the cutoff frequency to be >20 GHz, but the devices have been measured only until 1 GHz. Except from refs (98 and 103), only devices with a similar layer thickness have been used for comparison. Data from refs (58, 98, 99, and 101–103).

The ability of a diode to rectify a sinusoidal signal is described by the 3 dB cutoff frequency f_3dB (defined as the frequency at which the output voltage drops to half the initial output voltage at low frequency). This technical quantity is closely connected to the diode’s response time, which by itself is the sum of the forward transit τ_TT and reverse recovery time τ_RR. The forward transit time is a measure of how quickly the diode turns on, which is the time it takes for charge carriers to travel to the opposite side of the junction (e.g., holes from the p-type doped to the n-type doped side of a pin-junction). When turning the diode off, the charge carrier concentration returns to its original distribution during the reverse recovery, e.g., by recombination of minority charge carriers. In fact, the reverse recovery time is the sum of the charge storage time and the relaxation time, which both depend on the recombination dynamics.

A diode’s electrical device parameters (and in particular τ_TT and τ_RR) can be tuned by the device structure, i.e., the thickness of the intrinsic semiconductor layer, dopant concentration, and choice of metallic contacts, as well as interlayers for charge carrier injection/morphology modification. Furthermore, as the reverse recovery time in pn-diodes is governed by the minority charge carrier recombination rate, switching to Schottky diodes is an efficient way to increase the speed of operation as the reverse recovery time in these junctions is dominated by the discharging of the depletion capacitance in the absence of minority charge carriers. The thickness of the intrinsic layer and the dopant concentration govern the depletion layer capacitance⁹⁴ and the reverse saturation current, as well as the breakdown voltage.^95,96 Furthermore, the thickness of the intrinsic layer determines the forward resistance as well as the forward transit time. Hence, it is a very important parameter for optimizing the rectification behavior of the diode. Often, current in the forward direction is described as a space-charge-limited current (SCLC) (cf. e.g., refs (58 and 97−99)), and hence, the on-current is limited by the charge carrier mobility μ in the intrinsic layer and its thickness L. As derived by Steudel et al.,⁹⁷ for such an SCLC behavior, the maximum frequency f_max up to which the diode can be used as a half-wave rectifier is given by

1

where V_AC is the amplitude of the input AC voltage, V_DC is the harvested DC voltage, and V_F is the transition voltage of the diode when it enters the SCLC regime. The relationship between the cutoff frequency and the charge carrier mobility is described by this equation as linear, which is a trend that has been confirmed by experimental results (see Figure 7). Note that the two outliers on the curve are due to the fact that very thin layers were used for these devices. Furthermore, it gives a clear roadmap for the future development of materials for rectifier diodes. In this context, it should be highlighted that charge carrier mobility is a quantity that is strongly connected to the degree of energetic disorder/crystalline perfection in a film. Hence, using underlying materials such as self-assembled monolayers to improve the structural perfection of the films (cf. e.g., refs (100−103)), the charge carrier transport and injection can be greatly improved as a result of higher crystalline perfection and matching of energy levels. In consequence, the 3 dB cutoff frequency of pentacene-based Schottky diodes has been improved from 1 MHz to almost 1 GHz in the last couple of years.^100−103

Improvements on the charge carrier mobility in organic semiconductors achieved in the past decade mainly refer to the in-plane (or lateral) charge carrier mobility¹⁰⁴ as it is most important for the development of thin-film transistors. Organic diodes, though, require high out-of-plane (vertical) charge carrier mobility, and values for materials such as pentacene are only in the range of 0.1 cm²/(V s) to 1 cm²/(V s), which according to Figure 7 restricts the cutoff frequency of rectifiers to below 1 GHz.

Recently, Sawatzki et al.⁵⁸ have adapted rubrene thin-film crystals to build monolithic pin-diodes with doped contacts (see Figure 8a and compare Section 3.1). Most importantly, they developed a process to grow spherulitic dendrites of the triclinic polymorph with a diameter of several hundreds of micrometers which can be fabricated on a large variety of different underlying materials (metals or insulating polymers), including flexible substrates (see also Figure 4c). The merit of developing this growth method is the high out-of-plane hole mobility in these dendrites, which can reach values above 10 cm2/(Vs). Moreover, they have demonstrated that molecular dopants can be incorporated into the rubrene thin-film crystal up to a load of several percent without compromising the charge carrier transport properties. In fact, molecular doping improves the charge carrier injection and, hence, is an essential technological step for the fabrication of diodes with low forward resistance. Sawatzki et al. integrated these thin-film diodes into a half-wave rectifier circuit (see Figure 8b), where they obtained a loss of only 1.4 dB at 1 GHz (see Figure 8c). Unfortunately, the output of the rectifier has not been measured at higher frequencies due to experimental limits. However, they estimate the 3 dB cutoff frequency assuming an SCLC-like behavior to be as high as 37 GHz.

Figure 8.

(a) Material stack of a rectifier diode (doping in weight percent). The crystallization of triclinic rubrene is achieved by rapid annealing of an amorphous seed and consecutive epitaxy. (b) Rectifier circuit including an organic diode and setup components. The diode has an active area of 50 μm × 50 μm. (c) Rectified DC output voltage as a function of frequency at 2.7 V amplitude. Further details on the measurement can be found in ref (58). Reproduced with permission from ref (58). Copyright 2021 Wiley-VCH GmbH.

This estimation, as well as the equation developed by Steudel et al.,⁹⁷ assume that the limiting factor for switching is the forward transit and not the reverse recovery time. Looking at Figure 7, this assumption seems to be satisfied for organic rectifier diodes that have been developed so far, which is probably a consequence of the fact that trap-assisted recombination is very efficient in weakly ordered semiconductors. Hence, transport restricts the dynamic response rather than the recombination dynamics. Whether this still holds true for high-mobility materials, such as triclinic rubrene spherulites, is not clear. Recently, Wang et al.¹⁰⁵ studied the minority charge carrier recombination in organic bipolar junction transistors based on the same material system. Based on their results, the minority charge carrier lifetime is estimated to be in the range of 10^–10 s to 10^–9 s. Hence, recombination dynamics might provide an upper limit for the speed of organic rectifier diodes. However, whether these recombination dynamics reflect the properties of the intrinsic semiconductor material or are a result of imperfections (chemical traps, lattice defects, or dopants) still needs to be investigated. Presumably, doping might be helpful in designing faster devices, as a high density of dopants will result in more efficient recombination dynamics of minority charge carriers if charge carrier mobility is not affected.

3.4. Organic Bipolar Junction Transistors

As discussed in the previous section, understanding the charge carrier recombination in organic semiconductors is important for advancing organic rectifier performance and better understanding organic solar cell recombination physics. Numerous works have been carried out to quantify and address exciton diffusion length in a range of organic semiconductors with varying degrees of order. In disordered organic films, the exciton diffusion length has been limited to the nanometer range.^106−108 In contrast, it has been observed that exciton diffusion in single crystals can reach a few micrometers based on a triplet-assisted diffusion process with a long lifetime.¹⁰⁹ Nevertheless, the exciton diffusion length generally scales with the degree of order of the organic semiconductor.¹⁰⁷

Apart from exciton diffusion length measurements, there have been several reports on the majority diffusion length in organic semiconductors deduced from organic photovoltaic or light-emitting diode devices. Both electron and hole majority diffusion lengths have been shown to reach the centimeter scale, with diffusivity ranging from 1 × 10^–5 cm² s^–1 to 0.1 cm² s^–1.^110,111 On the other hand, reports on the minority carrier diffusion in organic semiconductors are rare. Existing approaches employed indirect measurement techniques, and the results from different measurements yielded very different diffusion lengths for common organic semiconductors. For example, the Dember effect shows minority diffusion over 10 μm for C60, while extraction from transient electroluminescence gives less than a nanometer for Alq₃.^112,113 Minority carrier diffusion length is an essential physical parameter, as it provides access to the recombination coefficient and can be used to obtain a more in-depth understanding of recombination mechanisms in organic semiconductors and devices.

Despite the advantage of thin-film transistors in terms of device miniaturization and process integration, bipolar junction transistors offer both low capacitance and contact resistance, which are desirable for achieving higher operational speed than comparable thin-film transistors. In addition, bipolar junction transistor operation is based on minority carrier diffusion. Hence, it can be a powerful device to probe the minority carrier diffusion process and recombination processes in a semiconductor material.

For a bipolar junction transistor to work, the minority charge diffusion length in the semiconductor must be greater than the base layer thickness. Even though the minority charge diffusion length has not been measured directly, it is expected to be short due to the disorder in typical organic films. Therefore, it is important to introduce order in organic semiconductors to allow the minority carrier diffusion length to be long enough to enable device types that are based on long-range diffusion. As discussed in Chapter 3.2, the minority charge carrier lifetime in such doped rubrene thin-film crystals is estimated to be 1 × 10^–10 s to 1 × 10^–9 s. Hence, a diffusion constant of at least 1 cm² s is needed to get a minority charge carrier lifetime in the range of 50 to 100 nm, which can be seen as a lower limit for the thickness of a thin-film layer sandwiched between two metallic electrodes without imposing the risk of a short circuit.

Recently, Wang et al. reported the first organic bipolar transistor, which is made of rubrene thin-film crystals¹⁰⁵ and has a vertical architecture, where the thickness of the base layer can be precisely controlled in the nm range. Moreover, the highly ordered rubrene thin-films can be both p-type and n-type doped with molecular dopants to control the carrier type in individual regions of the device, which is the key to enabling the operation of a bipolar transistor (Figure 9). The high vertical carrier mobility of around 3.7 cm² V^–1 s^–1 estimated from space-charge-limited current in orthorhombic rubrene crystalline film, short channel length of around 1 μm and low device capacitance allow the organic bipolar transistor to show rapid device operation up to 1.6 GHz, as estimated from the transconductance and capacitance of the device. The GHz operation is a milestone in developing organic thin-film transistors, which shows that the bipolar transistor is a promising device architecture.

Figure 9.

(a) Vertical stack configuration of the rubrene-based organic bipolar junction transistor. (b) The device under a polarized microscope (scale bar shows 100 μm). (c) Differential amplification for the device shown in b for different biasing conditions. (d) Normalized differential amplification as a function of effective base width (thickness) and n-type doping (dopant denoted here as W₂). A hole diffusion length in the n-type doped rubrene of 50 nm has been determined from this data. Reproduced with permission from ref (105). Copyright 2022 Springer Nature.

Since the operation of a bipolar transistor is based on minority carrier diffusion, it is a valuable experimental tool to directly probe the minority carrier diffusion length in organic semiconductors. The minority carrier diffusion length was estimated to be 50 nm in orthorhombic crystalline rubrene using thickness dependence measurements and Technology Computer-Aided Design (TCAD) simulations. Furthermore, it was found that the diffusion length does not depend on the dopant concentration in the investigated range, providing a strong hint about the dominance of an intrinsic recombination pathway. Overall, the organic bipolar transistor is crucial for understanding the minority carrier diffusion and recombination mechanisms in high-mobility organic semiconductors.

Despite the fast operation speed of the organic bipolar transistor, the current gain could be further improved, as it is currently limited by reverse leakage current in the diodes. The Technology Computer-Aided Design simulations show the method for further optimization by changing the device geometry to reduce the leakage current and improve base field control. Furthermore, the rectifying behavior of the diode might be further improved by tuning the junction properties concerning the dopant concentration and layer thickness.

3.5. Other Devices

Only a few examples of complex vertical devices utilizing highly crystalline thin-films have been reported in literature. Presumably, this is a result of the more complex processing of such devices compared to simpler amorphous devices. However, the combination of doping and layer stacking with crystalline films is a comparably new development that opens the gate to adapting many more device types beyond just diodes and transistors.

One early example is the solar cell architecture shown by Verreet et al.⁷¹ (see Figure 10a). They utilized orthorhombic rubrene crystals grown via abrupt heating to create organic solar cells with thick absorption layers (≈400 nm). This approach targets the compromise in organic solar cells between efficient absorption in a thick absorption layer and efficient transport through poorly conductive, low-mobility materials. As expected, the short-circuit current in these planar heterostructures increases with the increasing thickness of the rubrene. The power conversion efficiency might be significantly improved using a bulk heterostructure system of, e.g., fullerene-doped rubrene. In addition, it has been shown that solar cell devices comprised of a crystalline rubrene and C60 heterojunction exhibit band-like charge photogeneration with minimal energetic losses, highlighting the potential for highly crystalline organic photovoltaic devices.

Figure 10.

(a) Short-circuit current of a rubrene thin-film crystal planar heterostructure solar cell as a function of the layer thickness. Reproduced with permission from ref (71). Copyright 2013 Wiley-VCH GmbH. (b) Comparison of device lifetimes under high current stress conditions for amorphous and crystalline thin-film OLEDs composed of Alq₃. Reproduced with permission from ref (79). Copyright 2018 Royal Society of Chemistry. (c) Layer structure and AFM image of a crystalline pin-type OLED with high quantum yield. Reproduced with permission from ref (115). Copyright 2019 Elsevier Publishing.

Using rubrene crystals as a vehicle for fundamental studies, Kara et al.¹¹⁴ investigated the impact of different crystal structures (thin-film and bulk crystals) on the properties of solar cells. Efficiency, V_OC and j_sc can vary strongly based on the used crystals. Of particular interest is the comparably large effort needed to produce solar cells based on bulk crystals compared to thin-films being transformed into crystals by abrupt heating. However, the absolute performance of these devices is comparably poor due to the inferior absorption and optoelectronic properties of rubrene in itself. Alternative materials with better optical properties could alleviate this issue.

There are reports using single crystals in bulk form from several organic molecular crystals^116,117 to implement their high charge carrier mobility and low defect densities into OLED devices. While some of the benefits can be transferred, resulting in low-voltage operation due to more efficient transport, the overall efficiency of most of these devices is unfortunately still low due to the lack of proper doping and the large thickness of bulk crystals. Yang et al.^79,115 investigated the use of highly crystalline films of Alq₃ to create robust, high-current OLEDs to be used in electrically pumped organic lasers. Due to the higher mobility in comparison to devices based on amorphous Alq₃, the stability under high current was significantly improved, enabling the operation of the devices at high current densities (see Figure 10b and c). Furthermore, Wang et al. recently presented highly crystalline triclinic rubrene light-emitting diodes using an epitaxial growth method where functional additives are incorporated in the films, defining different zones in the device for charge injection and recombination.¹¹⁸ The highly crystalline rubrene light-emitting diodes show high current and luminance at a low driving voltage due to the significantly improved carrier mobility as well as high operational stability under a high driving current density, which is desirable for electrically driven laser and lighting applications. The crystalline rubrene light-emitting devices also show a high singlet fission and a triplet–triplet annihilation rate constant in comparison to their amorphous counterparts, which could be explained by the enhanced electronic coupling and exciton diffusivity in the crystalline rubrene. The enhanced singlet fission and triplet–triplet annihilation properties of crystalline rubrene films could be utilized for enhancing solar cell performance. Research on organic lasers has been an active topic for the past few decades due to their potential as low cost, tunable, and compact laser sources. However, achieving an electrically driven organic laser has proven to be challenging, mainly due to the lack of materials systems possessing both excellent charge transport and efficient emission properties. 4,4′-Bis[(N-carbazole)styryl]biphenyl (BSB-Cz), which displays excellent optical properties and a low threshold for achieving lasing, has been shown to be a promising candidate for electrically driven lasers, with which the Adachi group has observed indication of current-induced lasing using amorphous BSB-Cz thin-films.^116,123 Wang et al. has reported on the growth of an organic laser gain material, BSB-Cz thin-film crystals, and studied their optical properties.¹¹⁹ For the crystalline BSB-Cz thin-films, both the photoluminescence quantum yield and optical stability are improved relative to those of its amorphous counterpart. It would therefore be interesting to integrate BSB-Cz thin-film crystals into laser devices. In addition, there has been a report on using organic light-emitting diodes as an optical wireless data link with a Gbit s^–1 data rate, showing the potential of OLEDs as a fast data transmitter source.³⁶ For data transmission applications, it is important to improve the charge transport property of the organic semiconductor to enable the high-speed operation of OLEDs. Highly ordered OLEDs with enhanced carrier mobility⁷⁹ are clearly beneficial in this regard and might help advance the frequency performance of OLEDs even further.

Another interesting application in which highly ordered organic semiconductors could be of value is in thermoelectric cooling technology. The Peltier effect in organic semiconductors has rarely been seen. One key reason is their conductivity anisotropy, in which the vertical conductivity is limited, and the Joule heating associated with the device, which overshadows the Peltier effect. Wang et al. have demonstrated Peltier cooling in a compact vertical device using doped fullerene with a high vertical conductivity,¹²⁰ and this concept may be advanced further using highly ordered organic semiconductors with improved vertical charge transport properties.

4. Conclusion and Outlook

Organic bulk-type single crystals, such as rubrene or tetracene, were among the first material systems studied in the field of organic electronics, and they have remained a model system to study, e.g., transport properties or optical properties with minimal impact of structural disorder. However, these crystals have not received much attention concerning application; instead, amorphous layers for thin-film devices such as solar cells, photodetectors, light-emitting diodes, etc. have already been adapted for industrial production. The performance measures of these devices (e.g., power efficiency, lifetime, color tunability) have been greatly improved over the course of the last two decades by introducing dopants and utilizing complex multi-junction structures (e.g., pin-type OLEDs, tandem solar cells, etc.) as well as advancements with material synthesis and morphological optimization.^121,122 Most peculiarly, the increased complexity of the fabrication of such multi-junction structures did not impede their commercial success, as in the case of OLEDs. Evidently, the performance and robustness of such devices are more favored parameters compared to the cost of fabrication.

However, thin-film devices as described above suffer from the low charge carrier mobility typically observed in amorphous organic semiconductor films. To address this issue, novel crystallization methods have been developed to enable large-scale and uniform crystallization of initially amorphous films. Furthermore, these crystallization methods allow for the adaptation of controlled doping to tune, for example, the electrical conductivity or optical properties of the thin-film crystals. Now, the path is open to building and exploring complex, multi-junction devices based on highly ordered high-mobility organic semiconductor materials. From a scientific perspective, this will allow researchers to investigate topics such as recombination and charge separation dynamics for optoelectronic devices without the influence of static disorder. Furthermore, from an application perspective, these high-mobility layers might help to further boost the performance of organic electronic devices and target applications that have been thus far out of reach for organic electronics. For example, rectifier diodes, vertical bipolar transistors, or even OLEDs might operate at GHz frequencies, which enables fascinating new applications: wireless communication, energy harvesting, optical communication, and many more. However, since only a small set of materials has yet been identified to be crystallizable in this manner and fully utilized in complex devices, further investigation is needed to better understand and predict crystallization behavior in general for organic thin-film materials.

• 1.Many more molecular alternatives are clearly necessary to widen their application to additional complex device types like organic photodetectors, solar cells, OLEDs and even organic lasers.
• 2.Extending this crystallization method with more advanced techniques and combining it with solution processing is an already emerging field and would allow combining the best of two worlds in terms of quality and cost-effectiveness. The difficulties regarding doping have been the largest roadblock yet. Other important areas that are still poorly covered are the influence of process parameters on the surface coverage of growing crystals and the possibility to precisely position and align crystal growth.
• 3.Another possibility is to, on one hand, use these crystal films as a platform for device performance analysis but, on the other hand, also to utilize existing models and develop new models to optimize device structure with the new opportunities in mind. Here, a feedback loop of consecutive enhancement could accelerate progress extensively.
• 4.Lastly, a vast amount of new physics in organic semiconductors is now available to be studied with comparable ease, such as minority diffusion and recombination processes. It is thus clear that this review is only a brief overview of a large new field that is on the brink of fully emerging. The foreseeable future promises a wealth of topics that are yet to be explored.

Author Contributions

CRediT: Michael Franz Sawatzki conceptualization, data curation, formal analysis, investigation.

The authors declare no competing financial interest.

Special Issue

Published as part of the Chemical Reviewsvirtual special issue “Emerging Materials for Optoelectronics”.