Magnetic field-effects in bipolar, almost hole-only and almost electron-only tris-(8-hydroxyquinoline) aluminum devices T. D. Nguyen, Y. Sheng, J. Rybicki, and M. Wohlgenannt* Department of Physics and Astronomy and Optical Science and Technology Center, University of Iowa, Iowa City, Iowa 52242-1479, USA Received 28 February 2008; revised manuscript received 1 May 2008; published 16 June 2008 We present magnetoconductivity and magnetoluminescence measurements in sandwich devices made from the -conjugated molecule tris-8-hydroxyquinolinealuminum Alq 3 and demonstrate effects of more than 25% and 50% magnitude, respectively. These effects are known to be caused by hyperfine coupling in pairs of paramagnetic species, and it is often assumed that these are electron-hole pairs. However, we show that the very large magnitude of the effect contradicts present knowledge of the electron-hole pair recombination processes in electroluminescent -conjugated molecules and that the effect persists even in almost hole-only devices. DOI: 10.1103/PhysRevB.77.235209 PACS numbers: 73.50.h, 73.43.Qt I. INTRODUCTION Organic magnetoresistance OMARis a recently ob- served 19 large, low-field magnetoresistive effect up to 10% at 10 mT and 300 Kin organic light-emitting diode OLED structures. Similar effects have also been observed in various measurements ranging from electroluminescence to photo- conductivity. 1015 OMAR poses a significant scientific puzzle since it is the only known example of large room-temp- erature magnetoresistance in nonmagnetic materials with the exception of very-high-mobility materials. 16,17 The exact mechanism causing OMAR is currently not known with cer- tainty. Three kinds of models based on spin-dynamics in- duced by hyperfine interaction have recently been suggested as possible explanations of OMAR: iElectron-hole pair EHPmechanism models 46,8,18 based on concepts bor- rowed from the so-called magnetic field-effects in radical pairs. 10,19 In this model the spin-dependent reaction between oppositely charged polarons to form an exciton “recombina- tion”is of central importance. iiThe triplet-exciton po- laron quenching TPQmodel 7 that is based on the spin- dependent reaction between a triplet exciton and a polaron to give an excited singlet ground state i.e., the “quenching” of the triplet exciton by the polaron. iiiThe bipolaron mechanism 20 that treats the spin-dependent formation of doubly occupied sites bipolaronsduring the hopping trans- port through the organic film. Whereas mechanisms iand iiare excitonic in nature, the bipolaron mechanism can exist also in unipolar devices. We anticipate that the quanti- tative modeling of OMAR will yield sensitive tests of our understanding of organic semiconductor devices. At present, however, any analysis of OMAR experiments is plagued by ambiguity: experiments must be devised that will allow one to distinguish between the three mechanism mentioned above. Specifically, if model iis correct, then measure- ments of OMAR allow determination of the singlet:triplet ratio in OLEDs, whereas if iiis correct it will yield insights into the physics of triplet excitons, and finally if iiiis cor- rect OMAR can be used to test our understanding of charge and spin transport as well as bipolaron formation. In the present paper we will study OMAR in tris-8-hydroxy- quinolinealuminum Alq 3 devices with different electrode materials to put the three models of OMAR to a test. II. EXPERIMENT Our devices used an undoped organic semiconductor layer, and consequently the carriers that result in electrical current must be injected from the electrodes. If both the an- ode and cathode are chosen suitably, both form Ohmic con- tacts and the device is bipolar and shows efficient electrolu- minescence. If one of the electrodes is chosen to enforce a large barrier to the injection of this carrier type, then the device is almostunipolar and therefore shows ideally no electroluminescence. With this in mind we have fabricated devices with a large number of electrode material combina- tions. The fabrication started with glass substrates coated with either 30 nm of Al, 40 nm of Ag, 25 nm of Cr, 40 nm of Au prepared by electron-beam evaporation at 10 -6 mbar, 40 nm of indium-tin-oxide ITO, purchased from Delta Tech- nologies, or the conducting polymer poly3,4-ethylene- dioxythiophene-polystyrenesulfonatePEDOT, purchased from H C Starckspin-coated onto ITO as the anode. Since we need to measure the electroluminescence output to assess the carrier balance in the device, the thickness of the anode electrode had to be carefully chosen; It has to be thick enough to show high conductivity but has to be thin enough to be optically semitransparent. The transmission spectrum of the electrodes was measured and was used to correct the externally measured electroluminescence intensity. The Alq 3 sublimed, HW Sands Corp.layer was thermally evaporated in high vacuum 10 -6 mbaronto the bottom electrode, yielding an organic semiconductor layer thickness of 100 nm, without breaking the vacuum. The cathode, either Ca with an Al capping layer, Al, or Au was then deposited by thermal Caor electron-beam evaporation Al, Auon top of the organic thin film. The device area was 1 mm 2 for all devices. The samples were operated in dynamic vacuum inside a cryostat located between the poles of an electromagnet, al- though the measurements were all taken at room tempera- ture. The magnetoconductance MCratio was determined PHYSICAL REVIEW B 77, 235209 2008 1098-0121/2008/7723/2352095©2008 The American Physical Society 235209-1