This chapter aims at reviewing magnetic field effects (MFE) in organic light emitting diodes (OLED) with an emphasis on our study under high magnetic field up to 9 T. This subject includes organic spintronics in general, which is a hot subject attracting many researchers recently. Since singlet-triplet conversion in excitons is critically important in the current efficiency of OLEDs, spintronics aspects of OLEDs should be studied in detail. However, due to the difficulty in the fabrication of stable devices, the number of the researches has been limited.
We have found two things up to now, by making very stable OLEDs and measuring them under high magnetic fields. (1) Efficiency of OLEDs decreases quadrically with the magnetic field up to 6 T and the rate of decrease becomes smaller between 6T and 9T. (2) Minority carrier conductivity decreases linearly with the magnetic field, whereas that of majority carrier is almost constant. (3) Anomalous behaviors (large magnetoresistance etc.) are only seen in bipolar injection, which agrees with previous reports.
Although the mechanisms behind these findings have not been clarified yet, some hypotheses have been made with the analogy with MFE on chemical reactions. In principle, MFE on charge transport and recombination has similarity with chemical reaction under magnetic field, and the terminology and concept should be parallel between these subjects. We will try to combine the current knowledge of organic charge transport, OLEDs, spintronics and MFEs of chemical reaction to make a unified picture of these issues.
2. Spins in organic devices
2.1. Spins in OLEDs
The roles of spins in OLEDs and other organic devices are discussed well but have not been clarified quantitatively in experiments. We will review it in terms of OLED efficiency first. Figure 1 shows the schematic mechanism of OLED with the emphasis on the spin states. Since most of the organic semiconductors are used as intrinsic, the charges are transported via HOMO (highest occupied molecular orbital, in the case of holes) or LUMO (lowest unoccupied molecular orbital, in the case of electrons) of organic semiconductor molecules, usually by hopping in amorphous devices. Electrons and holes finally meet in one luminescent molecule and make excited states or excitons. The important thing is that there are two kinds of excitons, namely singlet excitons and triplet excitons. Although singlet excitons can be relaxed radiatively, triplet excitons cannot emit light in ordinary materials due to the spin selection rule. Since the charges injected from electrodes are not spin polarized unless spintronics techniques were used, the spin polarizaion statistics is singlet : triplet = 1 : 3. The emission efficiency of OLEDs are governed by this factor and it is well known that incorporation of heavy atoms (Pt, Ir etc.) in the luminescent dye molecule greatly alleviate this burden via intersystem crossing (Baldo et al. 1999). Since the chemical synthesis of the luminescent molecules with heavy atoms is not fully developed and the heavy atoms are costly, other methods such as applying magnetic field to OLED (Kalinowski 1997) or mixing magnetic nanoparticles in the device (Hu et al. 2006, Sun et al. 2007) have been attempted. These approaches uses MFEs on carrier injection, transport and recombination, which are related with spintronics of organic semiconductors.
Pure MFE without using magnetic electrodes has been studied. Experimentally, the reports on MFE of OLEDs without ferromagnetic component qualitatively agree with each other,
2.2. Organic spintronics
Spintronics study is now extended to all kinds of semiconductor materials. Organic semiconductors are not the exception. Since organic semiconductors consist of light elements such as carbon, hydrogen, oxygen and nitrogen, lifetime of spin polarized carriers might be long in organic semiconductors. After the proposal of this concept (Dediu et al. 2002), many papers have been published on the spin injection and transport in organic semiconductors. Most of the researches have been focused on performance of spin valves and magnetoresistance of organic semiconductors. A spin valve is a two terminal device consisting of a non-magnetic layer sandwiched by two different magnetic electrodes. The coersive forces of two electrodes are different and spin-polarized carrier injection and scattering makes characteristic magnetic field dependence of the device characters (I-V curve). Various organic semiconductors have been attempted in the device structures, and the large difference in the device resistance (magnetoresistance; MR) are achieved depending upon the spin orientation of the magnetic electrodes. At first MR was only substantially observed at low temperatures, but recently great MR at room temperatures are frequently reported. A variation of this research is MR measurement of mixture of magnetic nanoparticles and organic semiconductors. Some samples were prepared by codeposition of magnetic metals (cobalt etc.) and organic semiconductors. MR corresponding to the magnetization of magnetic nanoparticles (Sakai et al. 2006, Miwa et al. 2007) can be observed and its origin has been elucidated by x-ray magnetic circular dichromism (Matsumoto et al. 2009, Zhang et al. 2010).
An important topic related to the subject in the following is MR of devices without magnetic (or spin polarized) materials. Strong increase in conductance is observed in organic semiconductors when weak magnetic field (~ 100 mT) is applied. It is becoming a consensus that this high MR is only observed with bipolar injection, i.e., both of electrons and holes are injected to one layer, as shown in the following experiment (section 4.5).
3. Magnetic field effect in chemical reactions
First, we will follow up the current understanding of MFE in chemical reactions. Some of the chemical reaction change their reaction rate under magnetic field. In a simplified picture, those reactions proceed via intermediate state whose energy can be altered by the magnetic field. The energy difference can be due to the Zeeman effect on spin triplet state, which does not work on the spin singlet state. Therefore the reaction path between the singlet to or from the triplet can change the reaction rate. It must be noted that the Zeeman energy is too small to alter the reaction path in a single molecule. This is because the energy difference between spin singlet and spin triplet is very large (0.1~1eV) compared to the Zeeman energy (< 10 meV) under easily achievable magnetic field. Therefore it is considered that the intermediate state to which magnetic field can affect is a “radical pair”, in which an anion radical and a cation radical are placed closely and about to transfer charges. Those radicals have unpaired spins and thus spin triplet and spin singlet states exist. The energy difference between the singlet and the trpilet is very small because the spin-spin interaction is small due to the large distance belonging to different (but adjacent) atoms (or molecules or ions) and can be comparable with the Zeeman energy. MFE on chemical reaction rate comes from the radical pairs.
The MFE on chemical reaction rates are complicated and are classified as follows. The dependence of these effects on the magnetic field is schematically shown in Fig. 2.
(a) Hyper fine coupling (hfc) mechanism: This effect is caused by the interaction between nuclear spin and external magnetic field. hfc mechanism causes the increase of reaction rate in low magnetic field.
(c) triplet-triplet anhilation (TTA): TTA is the collision reaction between triplet excitons to make singlet. It occurs when the density of the excited states created from radical pairs is high and they are migrating. This effect is rarely importent in MFE of solution chemistry but becomes important in solid state devices.
The contribution of these mechanisms in the actually observed MFE is still under active argument. The characteristic magnetic field strength which gives the inflection points in (a)-(c) greatly differs from each other as discussed in the following. It is expected that the contributions can be elucidated by measuring the device properties in the wide range of the magnetic field. We therefore started experimental study described in the following section.
It should be noted here that all of the above mentioned mechanisms exhibit
4. Effect of high magnetic field on organic light emitting diodes
In the previous sections, we have reviewed the MFEs on charge transport in organic semiconductor devices including OLEDs and chemical reaction kinetics. Since some of the exciton-related effects saturate at relatively low magnetic field (~ 1T), it is expected that the contribution of the above mechanisms in MFE will be separated if high magnetic field is applied. In this section, we present our experimental study of MFE on OLEDs under high magnetic field (Goto et al., 2010). It seems that the results cannot be explained by the known exciton-related mechanisms (
4.1. Preparation of compact and stable OLEDs for the measurement in high magnetic field
Since the sample space of the high field magnet is small, the sample must be as compact as possible, while maintaining the stability to warrant the reliable measurement. We made fluorescent devices and phosphorescent devices. The structure of fluorescent device is shown in Fig. 3(a). An indium tin oxide (ITO) coated glass substrate (Aldrich) with a sheet resistivity of 8-12 Ω/square was used as the substrate. 100nm N’,N’-Di(naphthalene-1-yl)-N,N’ dipheyl-benzidine (α-NPD) and 100nm Tris-(8-hydroxyquinolino) aluminum (Alq) were deposited successively as the hole transporting layer and emitting & electron transporting layer, respectively. Then a cathode was deposited, which consisted of a 2 nm Cs layer followed by 150 nm of Al. The ITO substrate was cleaned by ultrasonicating in ethanol and acetone. Following this, the ITO was treated in ozone for 20 min.. The deposition of the organic layers (α-NPD and Alq, Luminescence Technology Corporation) was performed using Knudsen-cells in a vacuum chamber with a base pressure during evaporation of ~10-7 Torr. Cs was deposited with alkali metal dispenser (SAES Getters). The deposition rate of organic materials was about 0.1 nm / s, which was measured by calibrated quartz crystal microbalances. The structure of phosphorescence device is shown in Fig. 3(b). Doping of 5% Btp2Ir(acac) in CBP was performed by controlling the evaporation rate by monitoring the quartz crystal microbalances.
The sample OLEDs and unipolar devices were transferred from the deposition chamber to glove box filled with dry N2 without exposing them to air. The electrical connection to the OLED was made using thin Cu wire with In contact. Then the OLED sample was sealed in a glass box (made of O.D. 20 mm x t 3 mm pyrex tube and two t 0.1mm glass plates) using photo-hardening epoxy (Threebond 3124) together with a zeolite desiccant (Shinagawa Kasei Co. LTD). These sealing process was essential to obtain stable devices.
4.2. Measurement under magnetic field
MFE was measured at 300K in superconducting magnet using Physical Property Measurement System (PPMS; Quantum Design). The magnetic field was perpendicular to the device plane. The magnetic field was increased from 0 T to 9 T and then was decreased from 9 T to 0 T in order to check the temporal changes. The results are shown after confirming pure MFE is observed, unless stated otherwise. The emission intensity was measured with photon counter H7155-21 (Hamamatsu) in magnetic shielding made of thick iron plates and cylinders. The shielding of photon counter was tested and it was confirmed that there was no magnetic field dependence on its output. The bias was applied by the Keithley 6487 picoammeter / voltage source in constant voltage mode.
4.3. Results of fluorescent OLED
First we show the characteristics of the fluorescent device without applying the magnetic field. Figure 4(a) shows the emission intensity and current of the fluorescent OLED as a function of voltage, together with those of a phosphorescent device (Fig.4(b)). It is reported that TTA in Alq-based fluorescent OLEDs occurs when the current density is larger than 100 mA/cm2 (Kondakov 2007) and some of our measurement exceeds this limit. However, since the magnetic field dependence of TTA appears only at low temperatures (Lei et al. 2009, Liu et al. 2009), we consider we can neglect contribution from TTA in the present measurement at 300 K.
The emission intensity, current and emission efficiency of the same device under the magnetic field are shown in Fig. 5. The emission intensity and current show different magnetic field dependence on the sweep direction (0 -> 9T / 9 -> 0T) (Fig 5 (a)(b)). However, their ratio, i.e. emission efficiency, does not show the hysteresis as shown in Fig. 5(c). It means that the hysteresis comes from the charge injection process from the electrodes to the emission layer. We found that the "hysteresis" is dependent on both of the magnetic field and the time from the start of the current flow. The time dependence is probably due to the bias stress on the device, but the magnetic field dependence might be related with MFE of the trap / detrap processes.
The emission efficiency (and also the emission intensity and the current) increase steeply as a function of
In order to see dependence on
4.4. Results on phosphorescent OLED
Figure 7 shows the MFE on the emission efficiency of the phosphorescent OLED. In contrast to the results of the fluorescent device, it did not show the magnetic field dependence. Although we changed the driving voltage (4V, 6V, 8V, 10V), the magnetic field dependence did not appear.
4.5. Magnetoconductance measurement of unipolar devices
In order to investigate the charge balance factor which might influence the EL efficiency, we measured the magnetoresistance of the majority and minority carriers in α-NPD and Alq by making the unijunction devices with different work function electrodes (Au and Cs). All of the devices showed Ohmic
Also it should be noted that steep increase around zero field was not observed in unipolar devices, which agrees with previous reports (Yusoff, 2009).
4.6. Discussions – the origin of field dependence
We found the decrease in the fluorescent efficiency in organic EL devices proportional to
The EL efficiency (
The remaining factor in eq. (2) is the charge balance factor. We examined various models to relate the MFE on the charge balance factor, and the following is the only model that can barely explain the
Since the current is constant as a function of
The recombination rate is proportional to
and the decrease of the emission proportional to
We understand that the carrier transport of unipolar devices are not the same as the bipolar devices, for example, charge injection at the electrodes might be strongly involved in the minority carriers. However, our finding of linear MFE on minority carriers has not been reported and no theoretical prediction has been made to the author's knowledge. There might exist other mechanisms which also explain these results, but we hope the present result and discussions may stimulate the study of MFE on organic semiconductors and devices. Although TTA is not likely to work at room temperature and hfc will saturate at relatively low magnetic field,
There is another mechanism which might explain the
In the range beyond 6.5T, we observed
We did not find observable MFE in phosphorescent OLEDs (Fig.7). This result indicates that intersystem crossing occurs so fast in RPs and excited state molecules that Larmor precession in
We reviewed recent studies on organic spintronics and MFE on chemical reactions in relation to the MFE on OLEDs. We measured EL efficiency of fluorescent and phosphorescent OLEDs in the magnetic field up to 9T and in the fluorescent device we found quadratic decrease as a function of the magnetic field between 0.1 ~ 6.5T. We also measured magnetoconductance of unipolar devices and observed that only minority carriers show significant magnetoconductance decreasing linearly with the magnetic field (15% at 9T in Alq). B1/2 dependence in the range beyond 6.5T can be explained by MFE on the density of singlet exciton caused by Δg mechanism. In contrast, we did not find any MFE in the phosphorescent devices.
The author is grateful to the collaboration and discussions with Mr. Yuichiro Goto, Mr. Takuya Noguchi, Mr. Utahito Takeuchi, Dr. Kunitada Hatabayashi, Dr. Yasushi Hirose, Prof. Takehiko Sasaki, Prof. Tetsuya Hasegawa (all at the University of Tokyo). He would like to thank Prof. Takayuki Uchida (Tokyo Polytechnic University) for valuable information on the fabrication of OLED.