At the beginning of this century, organic semiconductor (OSC) materials were introduced to the field of spintronics,1, 2 which then merged into a new research direction called organic spintronics.2, 3 A variety of OSCs including both polymer4–6 and small molecule2, 7–10 have been demonstrated to perform well as spin transport layers in organic spintronic devices. The long spin diffusion time in OSC is generally attributed to low spin-orbital coupling and week hyperfine interaction, though other factors such as paramagnetic impurities may also induce spin-scattering and hence influence spin relaxation.11 Furthermore, OSCs seem to be a good candidate for spin injection as well. For inorganic semiconductor spintronics, efficient spin injection from ferromagnetic electrodes directly into inorganic semiconductor has been regarded as one of the major challenges,12, 13 in part due to the well-known conductance mismatch.14 In contrast, a direct contact of ferromagnetic electrodes and OSC seemingly can provide a very efficient spin injection, even at room temperature.5, 9 However, direct evidence of spin transport in OSC from, for example, Hanle-effect-type measurements has not been demonstrated, despite otherwise impressive performance of OSC materials in spintronic devices. The field of organic spintronics is currently focused on trying to reconcile this apparent contradiction, and it may suggest that the mechanisms for spin injection and transport may significantly differ from the situation in inorganic semiconductor spintronics. At any rate, new improved models of spin injection and spin transport in organic spintronics need to be established. The understanding of spin-related processes at ferromagnetic electrode/OSC interfaces will be crucial in this regard, and in this article, we will give a short summary on recent results regarding characterizing and tailoring such interfaces.
In general organic electronic devices, the electrode/OSC interface are very critical for device performance. A large amount of work about the interface properties has been published in the past decades, and reviewed from different aspects.15–17 At the beginning of organic spintronics research, knowledge about ferromagnet/OSC interfaces was lacking. Some ferromagnet/OSC interfaces had been studied in the context of organic electronics, but the spin information was neglected.18 With the rise of organic spintronics, however, the spin information of ferromagnet/OSC interfaces has begun to attract much more interest. This article mainly focuses on the recent research progress of ferromagnet/OSC interfaces. Since this type of interface not only determines the charge injection/extraction but also the spin injection/detection, it critically affects the performance of organic spintronic device, and therefore often is called spinterface.13 A series of techniques which can measure spin related information are reviewed in this article, including X-ray magnetic circular dichroism (XMCD), spin-polarized photoemission spectroscopy (SPPES), spin-polarized metastable de-excitation spectroscopy (SPMDS), spin-polarized scanning tunneling microscope (SPSTM), though other techniques such as two-photon photoemission19 and low-energy muon spin rotation20 have proven to provide useful information as well. We will compare experimental data from a selection of studies, and draw some conclusions based on their results. The readers who are interested in the general information on organic spintronic devices should consult other excellent review articles.3, 21, 22
ORGANIC SPINTRONIC DEVICES
Before we go into the details of spinterface studies, first we briefly will introduce organic spintronics. The most studied prototype organic spintronic device is the spin valve. The architecture of an organic spin valve is very simple: it typically includes two ferromagnetic electrodes with different coercive force and an OSC layer in between. But, if the thickness of OSC layer is thin enough so that electron can tunnel through, then the device is in the tunneling regime, called tunneling magnetic junction (TMJ).23, 24 The typical thickness of such an organic tunneling barrier is around ∼2–3 nm or even less. For devices with slightly thicker OSC layers in between, models that include transport of charges in the film, for example, multistep tunneling models,9, 25, 26 must be applied. For the spin injection type of spin valve, the OSC layer is at least one magnitude thicker.2
For TMJs and multistep tunneling devices, mainly the energy alignment at the ferromagnetic electrode/OSC interfaces will affect the device performance. But for injection spin valves, more spin related properties of the ferromagnetic electrode/OSC interface (and of the bulk of the OSC film) play critical roles in the device performance. The lack of the knowledge about these interfaces has severely limited the understanding of device physics. In this article, we hence will focus on the studies of the ferromagnetic electrode/OSC interface in injection spin valves (for spin transport in OSC, see, for example, the following recent reviews, refs.21 and27).
The first report on vertical injection organic spin valve was published in 2004,2 as shown in Figure 1(a). In the device, a 100–200 nm of tris(8-hydroxyquinoline)aluminum(III) (Alq3) layer is deposited as a spin transport OSC layer on La0.7Sr0.3MnO3 (LSMO) substrate, covered by a 3.5-nm thick Co film on top. A sizeable magnetoresistance (MR) of 40% was observed at 11 K [Fig. 1(b)].
Since then, many groups have confirmed the similar MR from the devices with the same architecture, using both Alq328–31 and other OSC materials as spacer layer. However, even with the same LSMO/Alq3/Co structure, the observed MR is often very different, not only the size30 but also the sign.28 All these differences point to discrepancies in the ferromagnetic electrode/OSC interface properties, which can be induced by material intrinsic properties, interface interactions and even different fabrication conditions.
SPINTERFACES IN ORGANIC SPINTRONICS
Energy Level Alignment of Ferromagnetic Electrode/OSC
It is well known that the energy level alignment of electrode/OSC interfaces plays an important role in organic electronic devices. The injection barrier based on the measured energy level alignment information can help device scientists decide which interface is suitable for injecting electrons or holes. The energy levels at an interface in most cases do not align with vacuum level, although many device scientists often treat it in this way because of lacking knowledge about the interfaces. This in turn may generate misconceptions regarding device performance.
Because of the same reason, the studies of energy level alignment at ferromagnetic electrode/OSC interfaces are also critical for the understanding of organic spin valve behaviors. Using the first organic spin valve Co/Alq3/LSMO as example, there are two ferromagnetic electrode/OSC (Co/Alq3 and Alq3/LSMO) interfaces in this device. When the device scientists first published the spin valve in Nature, they gave a schematic energy level diagram which was aligned with the vacuum level (Fig. 2). With such an alignment, they suggested the holes are the main carriers in the spin valve, since the electron injection barriers are too high.
However, UPS measurements gave totally different results when Zhan, et al, studied Co/Alq3 (cobalt on Alq3) interface by UPS with the assistance of a peel-off technique.32 The existence of 1.4 eV interface dipole results in a 2.1 eV hole-injection barrier (using the standard if not totally accurate practice of estimating the ionization potential from the leading edge of the frontier feature of the UPS spectrum15). By choosing an Alq3 HOMO-LUMO gap determined from STS data, an electron injection barrier of around 1 eV is then obtained. If taking band gap obtained by inverse photoelectron spectroscopy (IPES 4.6 eV33) or ballistic-electron-emission spectroscopy (BEES 4.8 eV34), the electron injection barrier is also more than 2 eV. Nevertheless, both the electron- and hole-injection barriers are not negligible. The Alq3/LSMO interface has the similar properties, UPS results show a 0.9 eV interface dipole, yielding a 1.7 eV hole-injection barrier.35 This suggests that the Fermi level of LSMO sits right in the middle of the Alq3 HOMO-LUMO gap, as shown in Figure 3. After mapping out such an energy level alignment, the question is raised how an electron or hole can be injected into Alq3 at low bias (a few millivolts) and low temperature (11 K) with this big injection barrier?2
For the device scientists' convenience, here we list the results of energy level alignment at a series of ferromagnetic electrode/OSC interfaces, which have recently been measured. As we can see from Table 1, most of the injection barriers are significant. Again we can question how is it that charge can be injected through this barrier at low temperature and low bias? We propose that the answer can be found in interfacial hybridization-induced states (HISs)42, 43 and also in the local-order dependence of the energy of the charge injection/transport levels of the OSC.44 When discussing these effects, we will first make the distinction between indirect contact44, 45 between the FM and OSC, where a native oxide layer, hydrocarbon contamination or thin barrier layers such as Al2O3 or LiF separate the OSC from the FM and thus prevents strong interaction like hybridization of the π-conjugated orbitals with the metal d-band states or covalent bond formation between the OSC and the FM involving specific atom of the OSC molecule, and direct contact42, 43 between the OSC and FM which leads to hybridization or chemisorption.
Table 1. The Results of Energy Level Alignments for the Investigated Ferromagnetic Electrode/OSC Interfaces
Indirect contacts between the OSC and the FM electrode occur, for example, when the interface formation is carried out under ambient conditions (FM native oxide and/or hydrocarbon surface layer) or when a barrier layer is intentionally inserted to separate the OSC from the FM. In organic spin valve applications, typically Al2O329 or LiF46 are used as “engineered” barrier layers. The barrier layer (native or engineered) isolates the FM surface from the OSC interface layer preventing chemical interaction and interdiffusion (mixing). Under such conditions, neither chemical bonds can form between the OSC and the FM nor can there be hybridization of the π-orbitals due to interaction with the FM continuum of states. Hence, spin polarization of π-conjugated orbitals of the OSC due to interaction with the spin-polarized d-bands of the FM also cannot occur, at least not through direct overlap between wave functions across the interface such as interlayer exchange coupling. However, barrier layers can be used to tune the energy level alignment of the FM/OSC interface so as to minimize the injection barriers, of particular importance to devices operating in the (multistep) tunneling regime as mentioned earlier. Indeed, the use of barrier layers at metal electrodes for tuning of energy level alignment is well established in the general field of organic electronics.47, 48 It is important to note that under indirect contact conditions, the work function of the FM electrode always will be strongly modified due to the so-called push-back effect43 even if the barrier layer is an insulator and has no intrinsic dipole (e.g., hydrocarbon contamination), hence the work function values of atomically clean FM surfaces should never be used to estimate the energy level alignment.49
If the barrier layer (natural or engineered) is thin enough to allow charges to tunnel from the FM into the OSC (or vice versa), the Integer Charge Transfer model15, 44, 45 can be applied. The energy level alignment at such interfaces depend on the position of the Fermi level of the (modified) FM surface and the so-called EICT+,− energies that correspond to the ionization potential and electron affinity (hole and electron polaron formation energies) of the OSC molecule at the interface, that is, taking account the intermolecular and intramolecular order at the interface as well as the electrostatic interaction with the substrate.44 The situation at the interface is important to consider, as the energy needed to create a hole (or electron) in a particular molecular orbital of an OSC molecule strongly depends on the electrostatic screening from its environment,50 and furthermore, surface energy mismatch between the (modified) FM electrode and OSC interface layer can introduce an intermolecular order very different from that in the bulk of the OSC film. Indeed, it has been shown that the EICT+,− energies can vary as much as ∼1 eV depending on the local intermolecular order.51 Hence, a situation where the Fermi level of the FM electrode is pinned to a polaron state of the OSC layer at the interface, yet an applied field is needed to move an injected charge away from the interface into the bulk of the OSC film, can easily occur.15 On the flip side, a large variation in local intermolecular order in an OSC film will lead to a wide distribution of polaron energies, see Figure 4, that can be accessed in the injection and (multistep) hopping process, thus enabling low bias injection and transport even if the polaron energies of the “majority” intermolecular order distribution are far removed from the Fermi level of the spin valve device, as seemingly is the case, for example, the ALq3-based devices discussed earlier.
HISs have been observed at many OCS/metal interfaces and there are several mechanisms suggested for the chemical interaction at the interface that causes the appearance of HISs. One possibility is that the direct chemical interaction between metal surface and some specific atoms from organic molecule leads to the reconstruction of molecular structure through covalent bonding to the surface and thus creation of HIS. For example, in the case of absorbing F4TCNQ on Cu (111), the N atoms from F4TCNQ molecule chemically bond to Cu (111) surface and form CuN chemical bonding, eventually create HISs.52 The other possibility is that π-conjugated orbitals of the molecule (not any specific atom) strongly interact with the metal orbitals, which in turn causes hybridization. For example, Kawabe et al.53 recently have found that the metal d-band plays an important role of the HISs at OSC/metal interfaces. They claim that “the hybridization between the LUMO of organic molecule and the metal d-band states is dominant for determining the interfacial interaction.”
One important consequence for the appearance of HISs is that one of the HISs is located at around the Fermi level of OCS/metal interfaces and remains partially unoccupied, as shown in Figure 5. This will transform the OCS/metal interfaces into ohmic-like contacts.54 For several other ferromagnetic metal/OSC interfaces, both NEXAFS41 and UPS55 has shown clear evidence for the appearance of HISs, that are located around the interface Fermi level and hence turn the interfaces into an ohmic-like contacts. Hence, carrier injection and transport can be realized in direct contact organic spin valves with only a few millivolts bias voltage without the aid of thermal energy in a similar way as for well-optimized indirect contact devices. In a first step, the carrier will be injected into the HISs with no injection barrier at all and from there on, the carrier has to overcome the energy differences between HIS and OSC bulk polarons, step by step to the opposing electrode. Again, the majority bulk polarons do not necessarily have to be situated at the Fermi level as long as variations in the local intermolecular order creates enough sites with polaron energies at the Fermi level that can be accessed in the tunneling transport through the device.
Energy level alignment, at the interfaces and in the bulk, is only part of the story, however. Direct contact also may allow for spin polarization of the HIS, which in turn should affect the spin injection. Therefore, researchers have investigated the spin information of direct contact type ferromagnetic electrode/OSC interfaces to fully understand spin injection in organic spin valves, as we will describe below.
Spin-Polarized Hybridization-Induced State
The ferromagnetic metals or alloys which have been used in the organic spintronic devices are mainly 3d ferromagnetic transition metals (Fe, Co, and Ni) and their alloys. The ferromagnetism of these materials is from the spin-polarized electron population on their 3d band, as shown in Table 2. Will the HIS induced by the interaction between the 3d ferromagnetic transition metal d-band and the LUMO of the organic molecule be spin polarized as well?
Table 2. Electron States of 3d Ferromagnetic Transitions Metals (Reproduced from ref.56, with Permission from [Academic Press].)
Recently, a number of research groups have tried a series of techniques which can directly or indirectly to confirm the existence of spin-polarized hybridization-induced state (SP-HIS) at the ferromagnetic electrode/OSC interfaces. We will summarize some of the main results in the following paragraphs, categorized by the different investigating techniques.
Starting in 2002, Suzuki et al. has obtained spin information of OSC/ferromagnetic electrodes by the means of SPMDS. They first measured the spin polarization of metal (Mn, Fe, Cu, and Mg) and metal-Free phthalocyanines on Fe (100),57, 58 and continued a study of pentacene on Fe (100).59 All the spin asymmetry of these interfaces are positive (shown in Fig. 6), except for a back-donated orbital (π*4b1u) of pentacene on Fe (100), in which negative spin asymmetry is induced. As mentioned by the authors, positive (negative) spin asymmetry in SPMDS indicates negative (positive) spin polarization.59 Therefore, the spin polarization of both metal and metal-free phthalocyanines is antiparallel to Fe (100) substrate, while the magnetization of pentacene needs further investigation.
Recently, a well-established technique called XMCD has been adopted for the studies of ferromagnetic metal/OSC interfaces. Because of the unique element-resolved function, XMCD is regarded as an ideal method to probe the magnetization of different materials at an interface. To our knowledge, the first XMCD study of OSC/ferromagnetic metal interfaces is reported by Scheybal et al.,60 where they demonstrated with XMCD results the existence of exchange coupling between a large organic adsorbate manganese (III)–tetraphenylporphyrin chloride (MnTPPCl) and a ferromagnetic cobalt substrate. Although the authors did argue that the distance between Mn atom and cobalt substrate is too far for direct exchange coupling through orbital overlap between Co 3d and Mn 3d, there was no direct evidence that proved the exchange coupling occurs through delocalized electrons on the phenyl rings. Nevertheless, the work has initiated a series of XMCD studies of OSC/ferromagnetic metal interfaces.61–65
Recently, direct evidence for the spin polarization of organic molecular π-orbitals was obtained by N K-edge XMCD measurement of Alq3 sub-monolayers on Fe surfaces,41 as shown in Figure 7. The authors concluded that they have demonstrated hybridization and exchange coupling between π-conjugated orbitals in Alq3 and the Fe 3d of the substrate. More recent publications have followed, including C K-edge XMCD on C60/Fe interfaces where the mixture of C60 π/π* orbitals and Fe 3d wave functions were found to create a distinct, oscillatory magnetic moment of C60-derived hybridized interfacial state.66 The frontier π-orbitals are the main actors in charge injection and transport on OSC materials, making the demonstration of hybridization a key finding.
Spin-polarized UPS (SPUPS) has also been introduced to study the spin polarization of valance levels of OSC/ferromagnetic metal interfaces. With in situ growth mode, it is possible to measure the surface spin polarization of ferromagnetic metals and the modification resulting from OSC/ferromagnetic metal interface formation. So far, there are only a few SPUPS results published, mainly focusing on metal phthalocyanines/ferromagnetic metal interfaces.55, 67, 68 SPUPS measurements directly demonstrate the formation of HIS, induced by chemisorption of CuPc, CoPc, and FePc at the Co surface.55 Among these three interfaces, CuPc/Co interface shows the strongest spin polarization directly at the Fermi level, as shown in Figure 8. The authors suggested the differences are originated from both the molecular intrinsic properties and the interface interaction.
As mentioned by Sanvito,69 the above introduced experimental method for OSC/ferromagnetic metal interface are based on a relatively large area (micrometer scale), in which the microscopic details of the interfaces between the organic and the magnetic materials from which the spins are injected are averaged out. SP-STM could study the interface in the scale of a single molecule, and bring us more detailed information about the exchange coupling occurring at the interface. One of the main discoveries is that the spin polarization of the conducting electrons will be filtered and reversed through the OSC/ferromagnetic metal interface,70, 71 as shown in Figure 9. This finding may explain why the MR of organic spin valves is generally negative. The authors suggested that the pz-d exchange type mechanism70 could explain the modification of spin polarization of a ferromagnetic surface. An admixture of both the substrate's d orbitals and the molecule's π orbitals is believed to form the SP-HIS.69 Furthermore, the SP-STM measurement also showed the possibility of improving spin injection efficiency by the interface engineering of organic spintronics.
Spin Injection and Transport Model
Based on the listed experimental results, the process of spin injection from Ferromagnetic metal into OSC can be divided into two parts: (a) from ferromagnetic metal to the first OSC monolayer and (b) from the first OSC monolayer to OSC bulk. For the first step, the induced spin polarization of OSC interface layer's HIS plays an important role of spin injection. From the SP-STM and SPMDS results, the spin polarization of HIS is antiparallel to the ferromagnetic metal substrate. These results are in agreement with the spin hybridization transport model suggested by Barraud et al.,72 once a strong coupling occurs at the interface, the spin polarization of injected carrier is determined by spin polarization of HIS at the interface. After the HISs of OSC interface layer are populated, the spin polarized will continue to hop into bulk OSC. As the HIS is generally lower than the LUMO (majority molecular order n-polaron) and higher than the HOMO (majority molecular order p-polaron) of Bulk OSC, the electrons/holes will have to overcome the energy difference then hop to LUMO/HOMO, as suggested by many researchers.29, 55, 72 As we have argued, there is another possibility that also should be considered. At low temperature, the injected electrons/holes might never reach the LUMO/HOMO and only hop through “discontinued deep trap levels”, as described by the famous Mott's variable range hopping (VRH) model,73 where the deep trap levels are in fact not defects as per se but a simple consequence of variations in the intermolecular order in the OSC film that in turn creates a spatial and energetic distribution of polaron states (see Fig. 4). (Taken to the extreme to make the point: if every molecule in the OSC film has a different local environment, then every molecule in the OSC film has a different n- and p-polaron energy). Indeed, for the noncrystal medium, as almost all OSC films are, “at a sufficient low temperature, the phenomenon of variable range hopping is always to be expected.”73 A consequence of this proposed model is that it, at least for (multistep) tunnel devices, should be advantageous to choose molecules that feature large variations in local intermolecular order in films (i.e., form disordered films) and have a strong polaron energy dependence on the molecular order (typically but not exclusively molecules with large intrinsic dipoles). The model has other interesting consequences for devices as well. For example, as suggested by Yoo et al., transport according to the VRH model could reduce efficient spin-polarized carrier transport at high temperature.74
Spinterface Engineering: Affecting the FM Contact
We have discussed how the FM/OSC interface can affect the spin injection into the OSC film by modifying the energy level alignment and/or inducing spin polarization of (hybridized) π-orbitals in molecules at the interface. We have further discussed how the two main types of contacting, direct and indirect, affect these processes. Also of interest is how direct and indirect contact between FM and OSC molecules affect the (surface) magnetic properties of the FM itself, as this too will affect device performance. We illustrate this point with a couple of recent examples from literature.
It is well known that the spin properties of top ferromagnetic electrodes in organic spin valves depend on the surface it is deposited on. Pernechele et al. have systematically studied cobalt contacts deposited onto a variety of films, including Alq3, Al2O3/Alq3, and Si.75 From Figure 10, we can easily see that the cobalt film on top of Alq3 has the largest coercive field. By inserting a 3-nm Al2O3 barrier layer, the coercive field of cobalt film becomes nearly identical to the case for Co on Si. Hence, by preventing chemical interaction between Co and Alq3 and reducing the surface roughness, the Al2O3 barrier layer succeeds in improving the magnetic properties of the FM contact in this case.
For bottom contacting, OSC films deposited onto a FM contact, the effects on the FM magnetic properties also can be substantial. For example, XMCD results from Tran et al. (Fig. 11) indicate that a slight reduction in the spin magnetic moment (of about 1–2%) of Fe at the interface occurs when depositing C60 due to partial electron transfer from the Fe substrate to the C60 interface layer.66
Modification of the magnetic properties of the FM electrodes upon direct/indirect formation of FM/OSC interfaces has so far received less attention than, for example, energy level alignment and spin polarization in OSC molecules, but should be considered in the design of the “perfect” spinterface.
The recent research on interfaces featuring ferromagnetic metals and organic semiconducting molecules has shed new light on several important issues for organic spin valves. In particular, the ability to drive organic spin valves at millivolts bias can be explained by the access to polaron states around the Fermi level due to a combination of energy level matching between the ferromagnetic metal electrodes and organic semiconducting molecules through direct or indirect contact effects and the existence of a variation of local molecular order in the organic films. Furthermore, direct contact between a ferromagnetic metal electrode and an OSC molecule can lead to formation of spin-polarized hybridized interface states involving the metal d-band and π-orbital of the organic molecules. Finally, the interface formation also can affect the surface magnetic properties of the ferromagnetic electrodes. Taken altogether, it is clear that the richness of interface effects and the strength with which they impact device performance constitute a challenge in terms of reproducibility in fabricated devices. On the other hand, the very same richness of the spinterface provides near unlimited possibilities for designing new and improved organic spintronic devices with optimized performance.
This work was supported by EU Integrated Project OFSPIN (EU-FP6-STREP), National Natural Science Foundation of China (NSFC) Grants No. 11104037, and the Swedish Research Council Grant No. 2011–7307.
Yiqiang Zhan obtained his PhD in physics from Fudan University, China, in 2005 before moving to ISMN-CNR Bologna, Italy, as a postdoc. From 2007, he continued his research in Linkoping University, Sweden, initially as a postdoc and then as an assistant professor. Recently, he joined Fudan University as an associate professor. His research interests include the use of Synchrotron based techniques together with device study to investigate the organic spintronics-related topics.
Mats Fahlman obtained his PhD in Surface Physics and Chemistry at Linkoping University in 1995 and is currently the Professor of Surface Physics and Chemistry at Linkoping University. His research interests include the study of organic/organic and hybrid organic interfaces, their role in organic electronic and organic spintronic devices, as well as thin film synthesis of hybrid organic spintronic materials.