Pure Appl. Chem., Vol. 71, No. 11, pp. 2095–2106, 1999.
`Printed in Great Britain.
`q 1999 IUPAC
`
`Phosphorescent materials for application to
`organic light emitting devices*
`
`M. A. Baldo1, M. E. Thompson2 and S. R. Forrest1,†
`
`1Center for Photonics and Optoelectronic Materials (POEM), Department of Electrical
`Engineering and the Princeton Materials Institute, Princeton University, Princeton, NJ
`08544, USA
`2Department of Chemistry, University of Southern California, Los Angeles, California
`90089, USA
`
`Abstract: Organic phosphors have demanded the attention of the organic electroluminescence
`community because they enable efficiencies quadruple that of fluorescent materials. In this
`work, we review the categories of organic phosphors:
`lanthanide complexes, organic
`phosphors and metal-organic complexes. The characteristics necessary for efficient phosphor-
`escence are considered and conclusions are drawn as to the most promising materials.
`
`INTRODUCTION
`
`The application of phosphors to light emitting devices has a well-established precedent: the cathode ray
`tube, where luminescence is obtained from atomic transitions within carefully selected inorganic
`materials. Similar techniques can be used to exploit the advantages of phosphorescence in the flourishing
`field of organic electroluminescence, but the flexibility of organic materials allows for a variety of
`additional approaches. We shall examine these approaches in this review.
`
`Phosphorescence is distinguished from fluorescence by the speed of the electronic transition that
`generates luminescence. Both processes require the relaxation of an excited state to the ground state, but in
`phosphorescence the transition is ‘forbidden’ and as a consequence it is slower than fluorescence, which
`arises from allowed transitions. Indeed, phosphorescence may persist for several seconds after a
`phosphorescent material is excited, whereas fluorescent lifetimes are typically on the order of nanoseconds.
`
`Interest in phosphorescence, and the phosphorescence of organic materials in particular, arises from
`the application of these materials to organic light emitting devices [1–5] (OLEDs), where it is found that
`the luminous efficiency may be improved by up to a factor of four over that obtained using fluorescence.
`This increase is fundamental to organic materials and arises during the formation of an excited state (or
`exciton) from the combination of electrons and holes. Typically these excitons are localized on a small
`molecule, or localized to a region of a polymer chain, and hence it is often convenient to describe excitons
`as particles. For example, an exciton possesses spin, which must be conserved during the emission of a
`photon. As the ground state is generally spin antisymmetric with a total spin of S (cid:136) 0, the decay of S (cid:136) 0
`excitons is allowed. In contrast, the decay of S (cid:136) 1 excitons is not allowed. This poses an obstacle to
`efficient luminescence because the combination of an electron and a hole with uncorrelated spins is three
`times as likely to result in a spin-symmetric as opposed to a spin antisymmetric state [6]. From the
`multiplicity of the exciton spin states, S (cid:136) 1 excitons are known as triplets, and S (cid:136) 0 excitons are singlets.
`Thus, if the energy contained in the triplet excitons cannot be directed to luminescence, the efficiency of
`an OLED is reduced by 75%.
`
`Fortunately, although the decay of a triplet state is disallowed by the conservation of spin symmetry, it
`is occasionally observed if the triplet state is perturbed such that the transition becomes weakly allowed.
`
`*Lecture presented at the 4th International Symposium on Functional Dyes—Science and Technology of
`Functional p-Electron Systems, Osaka, Japan, 31 May–4 June 1999, pp. 2009–2160.
`†Corresponding author: E-mail: forrest@EE.Princeton.EDU
`
`2095
`
`RHEMK-1011.001
`
`

`
`2096
`
`M. A. BALDO et al.
`
`In this case, the decay of the triplet state may still be very slow, but phosphorescence is generated. But
`efficient phosphorescence is rare at room temperature, and attempting to find a material that also readily
`transports charge is a challenging task. Moreover, very few materials luminesce efficiently in
`homogenous films due to the quenching of emission by surrounding molecules. The solution to these
`demands on OLED materials is found by doping the luminescent material into a charge transport host
`material [7]. Emission then occurs in one of two ways: either by direct carrier trapping and exciton
`formation on the luminescent dye, or by exciton formation in the host and energy transfer to the
`luminescent guest. Hence, it is not sufficient merely for the guest to be phosphorescent from its triplet
`states; it must also be able to gather the triplets formed by electrical excitation.
`
`ENERGY TRANSFER
`
`There are two mechanisms for the transfer of energy in these materials: Fo¨rster and Dexter transfer [8].
`Fo¨rster transfer [9] is a long range (< 40 ˚A–100 ˚A), non-radiative, dipole-dipole coupling of donor (D)
`and acceptor (A) molecules. Since it requires that the transitions from the ground to the excited states be
`allowed for both D and A species, this mechanism only transfers energy to the singlet state of the acceptor
`molecule via:
`1D(cid:3) (cid:135) 1A ! 1D (cid:135) 1 A(cid:3)
`or,
`3D(cid:3) (cid:135) 1A ! 1D (cid:135) 1A(cid:3)
`Frequently, phosphorescent dyes possess strong intersystem crossing (ISC) from the singlet to the
`triplet excited state. For these materials, little or no fluorescence is observed from the singlet state of the
`acceptor (1A*), and all excited states in the donor are ultimately transferred to the triplet acceptor state
`(3A*). Typically the donor exciton must also be a singlet to participate, however, there is an exception for
`donor materials where a triplet-to-ground state transition is weakly allowed (see Eqn 2) [9,10]. For
`example, energy transfer from a triplet to a singlet state has been demonstrated when the donor is an
`efficient phosphor [11]. Here, the slower rate of energy transfer from a weakly allowed transition is
`compensated by the long lifetime of the donor exciton.
`
`(cid:133)1(cid:134)
`
`(cid:133)2(cid:134)
`
`Dexter transfer [12] is a short-range process where excitons diffuse from D to A sites via
`intermolecular electron exchange. In contrast to Fo¨rster transfer, Dexter processes require only that the
`total spin of the D-A pair be conserved under the Wigner–Witmer selection rules [8], via:
`1D(cid:3) (cid:135) 1A ! 1D (cid:135) 1A(cid:3)
`or,
`3D(cid:3) (cid:135) 1A ! 1D (cid:135) 3A(cid:3)
`Thus, Dexter transfer permits both singlet-singlet and triplet-triplet transfers. However, Fo¨rster
`transfer dominates singlet-singlet transfer at low acceptor concentrations because it is faster over long
`distances. Energy transfer to phosphorescent dyes then proceeds by Fo¨rster transfer of singlets and Dexter
`transfer of triplets. Dexter transfer processes are slow for all
`interactions except those between
`neighboring molecules; thus it is possible that direct charge trapping and exciton formation on a
`phosphorescent dye may be an equally effective method for generating efficient phosphorescence.
`
`(cid:133)3(cid:134)
`
`(cid:133)4(cid:134)
`
`Lanthanide complexes
`
`The previous discussion of excitons applies to organic materials, but as discussed in the introduction
`phosphorescence is also possible in ‘disallowed’ atomic transitions. The most efficient examples of
`phosphorescence from an atomic species are the lanthanide complexes [13]. Since these transitions
`possess very sharp spectral bands, near monochromatic or ‘saturated’ luminescence results. This is
`obviously also desirable in OLEDs, and examples are red-emitting complexes [14,15] of Eu3(cid:135), green-
`emitting complexes [5] of Tb3(cid:135), and more recently 1.54 mm electroluminescence has been generated [16]
`from Er3(cid:135), although the efficiencies in all cases have been low (typically < 1%). It is illustrative to
`examine the most successful application [15] yet of a lanthanide complex to OLEDs. Tris(1,3-diphenyl-
`1,3-propanediono)(monophenanthroline)Eu(III) (Eu(DBM)3(Phen)) exhibits red-orange phosphorescence
`
`q 1999 IUPAC, Pure Appl. Chem. 71, 2095–2106
`
`RHEMK-1011.002
`
`

`
`Organic light emitting devices
`
`2097
`
`at 614 nm and has been used to generate electroluminesce in organic devices. The emission mechanism is
`
`the ‘forbidden’ 5D0! 7F2 transition of Eu3(cid:135) which has a radiative lifetime < 100 ms. In OLEDs,
`Eu(DBM)3(Phen) is preferably codeposited with a charge transport material such as [17] biphenyl-p-(t-
`butyl)phenyl-1,3,4-oxadiazole (PBD), thereby reducing self-quenching and improving charge carrier
`mobilities.
`Approximate energy levels of the Eu(DBM)3(Phen):PBD system are shown [18] in Fig. 1. Excitation
`of the Eu3(cid:135) ion occurs via the triplet energy level of the Eu3(cid:135) complex ligand [18]. As discussed in
`Section I, for high efficiency, it is necessary that both host singlets and triplets be transferred to the Eu3(cid:135)
`ion. Fo¨rster transfer of singlets is possible [15], as demonstrated by the observation that photoexcitation
`of PBD results in Eu3(cid:135) phosphorescence. However, since the relevant triplet energy levels in the host and
`the guest triplet state are frequently unknown and, moreover, are difficult to quantify due to their small
`oscillator strengths, optimizing guest-host systems for resonant triplet transfer is problematic. In the case
`of Eu(DBM)3(Phen), the energy of ligand phosphorescence [18] is large (< 2.6 eV) and the triplet
`absorption energy is expected to be even higher. Hence triplet transfer from the host to the ligand requires
`a host triplet of sufficiently high energy to be in resonance with the guest, and a device structure that
`minimizes nonradiative losses of host triplets. Analysis of Eu(DBM)3(Phen):PBD is difficult because the
`triplet energy in PBD is unknown. However,
`the maximum quantum efficiency [15] of
`Eu(DBM)3(Phen):PBD is < 1%. Thus, it is suspected that in this material system, energy transfer and
`particularly triplet-triplet transfer is poor.
`
`Fig. 1 The chemical structures of the lanthanide complex Eu(DBM)3(Phen) and a host material PBD. Also shown
`is a proposed energy level diagram (adapted from [13]) indicating the energy transfer process. Note that the triplet
`energy level of PBD is unknown, complicating analysis of this system.
`
`Despite the fact that lanthanide materials should be relatively insensitive to triplet-triplet annihilation,
`quantum efficiency is nevertheless found to decrease as current density increases, and consequently the
`maximum luminance [15] of the Eu3(cid:135) complex is only 460 cd/m2. Notwithstanding these deficiencies, the
`saturated emission of lanthanide-based complexes remains extremely attractive for many luminescent
`
`q 1999 IUPAC, Pure Appl. Chem. 71, 2095–2106
`
`RHEMK-1011.003
`
`

`
`2098
`
`M. A. BALDO et al.
`
`applications. In the case of vacuum deposited OLEDs, molecular design is required to ensure that the
`complexes can withstand vacuum sublimation. It is also necessary to minimize the distance between the
`ligand and the central lanthanide atom to maximize energy transfer within the complex [19]. Then if a
`host material is chosen for effective energy transfer to a lanthanide complex, or efficient charge trapping
`on the dye, the uniquely saturated emission of lanthanide complexes will be coupled with high
`electroluminescent efficiencies.
`
`Purely organic phosphors
`
`Another example of a phosphorescent material applied to OLEDs is benzophenone [4]. This material has
`phosphorescent emission at l < 450 nm and is frequently employed in the study of organic triplet
`excitons [8]. Benzophenone emits from a triplet state; thus the additional process in the lanthanide
`complexes of energy transfer from the ligand triplet to the ionic excited state is not required. In general,
`removing a step from the energy transfer process reduces losses and lowers the inital energy required for
`excitons to propagate from the host to emissive state of the dye.
`
`Phosphorescence in benzophenone arises from a forbidden transition from excited (p*) to ground (n)
`states. Spin-orbit coupling is enhanced [8] since both states are localized on the carbonyl group, hence the
`p*ˆ n transition becomes weakly allowed. However, the rate of phosphorescence is slow and must
`compete with non-radiative transitions that acquire intensity by coupling to the vibrations of the molecule
`[20]. Consequently, phosphorescence from benzophenone is strongly temperature dependent. When
`employed in OLEDs (see Fig. 2), benzophenone was spun cast in poly(methylmethacrylate) (PMMA). It
`exhibited negligible phosphorescence at room temperature, however a < 900-fold increase in quantum
`efficiency was observed [4] by reducing the temperature to 100 K. A lifetime of < 5.3 ms at 100 K has
`been observed for 1.7% benzophenone doped into PMMA. Evidently, at room temperature the rate of
`phosphorescence in benzophenone is too slow to compete with thermally activated non-radiative modes.
`
`Fig. 2 The chemical structure of the triplet emitter benzophenone and the device structure of the OLED used to
`obtain phosphorescence at 77 K. Poly(methylmethacrylate) (PMMA) serves as the host for benzophenone; the
`hole transporting layer is poly(methylphenylsilane) (PMPS) and the electron transport and hole blocking layer is
`PBD. From [4].
`
`Organometallic phosphors
`
`The example of benzophenone demonstrates that for efficient room temperature phosphorescence from
`organic ligands, short triplet lifetimes are required. This may be achieved by spin-orbit (L-S) coupling
`which mixes singlet and triplet excited states [20]. Spin-orbit coupling is significantly enhanced by the
`presence of a heavy atom in an organometallic complex. In contrast to the lanthanide complexes
`discussed earlier, these materials, which typically are complexes of Os, Ru, Pd, Pt, Ir or Au, do not
`luminesce from an atomic transition. Rather, the lowest energy excited state is frequently a metal-ligand
`charge-transfer triplet state, mixed with the excited singlet state by L-S coupling. Consequently,
`
`q 1999 IUPAC, Pure Appl. Chem. 71, 2095–2106
`
`RHEMK-1011.004
`
`

`
`Organic light emitting devices
`
`2099
`
`phosphorescent lifetimes are short (< 100 ms) and high photoluminescent efficiencies are possible. The
`mixing between singlet and triplet excited states is also responsible for very high probabilities (> 99%) of
`intersystem crossing. Thus, both singlet and triplet excitations of these complexes can result in
`phosphorescent emission.
`
`Because these materials emit from a triplet state, and because the ground state of molecular oxygen is
`also a triplet, oxygen reversibly quenches phosphorescence by triplet-triplet annihilation. Some
`complexes may also be quenched by electron transfer and other processes, hence many suitable
`organometallic complexes have been studied for use as oxygen sensors [21,22]. Application of
`organometallic complexes in OLEDs was demonstrated [2] by Ma et al., doping several osmium
`complexes in poly(N-vinyl carbazole). Although very low electroluminescent quantum efficiencies
`(< 0.1%) were reported, it was shown that this promising class of materials can ultimately yield efficiency
`improvements in OLEDs. Recently [23] these workers also demonstrated two new organometallic
`phosphors based on either gold(I) or copper(I). Both phosphors exhibited strong intersystem crossing and
`high photoluminescent efficiencies of 23% and 42%, respectively. But once again, owing to poor energy
`collection by the phosphor, only low quantum efficiencies (< 0.1%) were obtained.
`
`High efficiency electroluminescence from an organometallic complex triplet state was ultimately
`demonstrated [1] in OLEDs using the phosphorescent dye 2,3,7,8,12,13,17,18-octaethyl-21H,23H-
`porphine platinum (II) (PtOEP). Porphyrin complexes are known to possess long-lived triplet states useful
`in oxygen detection [24]. The addition of platinum to the porphine ring reduces the phosphorescence
`lifetime by increasing L-S coupling; the triplet states gain additional singlet character and vice versa. This
`also enhances the efficiency of intersystem crossing from the first singlet excited state to the triplet
`excited state. Transient absorption spectroscopy gives a singlet lifetime in PtOEP of < 1 ps, and the
`fluorescence efficiency is extremely weak [25]. In contrast, the room temperature phosphorescence
`efficiency of PtOEP in a polystyrene matrix is [21] 50% with an observed lifetime of 91 ms. Thus, both
`singlet and triplet excitations in PtOEP yield efficient phosphorescence. Consequently, no significant
`emission is found for the previously identified singlet state, expected at approximately 580 nm [25], but as
`shown in Fig. 3, strong emission is observed from the triplet excited state of 650 nm, with weaker
`emission at the vibronic harmonic overtones at 623 nm, 687 nm and 720 nm. The emission of PtOEP at
`650 nm is almost as sharp and saturated as the atomic transition of Eu3(cid:135) at 614 nm. However, PtOEP is a
`much deeper red and possesses Commission Internationale de L’E´ clairage (CIE) chromaticity co-
`ordinates of (x,y) (cid:136) (0.72, 0.29).
`
`Fig. 3 Spectra of phosphorescent OLEDs with different molar concentrations of PtOEP in Alq3 at different
`current densities. (a) 1%, 6% and 20% PtOEP in Alq3 OLEDs at 25 mA/cm2 (b) 1%, 6% and 20% PtOEP in Alq3
`OLEDs at 250 mA/cm2. Note the increased Alq3 emission at 530 nm in the 1% PtOEP OLED due to saturation of
`PtOEP sites and poor energy transfer. From [1].
`
`q 1999 IUPAC, Pure Appl. Chem. 71, 2095–2106
`
`RHEMK-1011.005
`
`

`
`2100
`
`M. A. BALDO et al.
`
`Conclusive evidence for triplet transfer to PtOEP from the well-known electron transport material
`tris(8-hydroxyquinoline) aluminum (Alq3) was obtained by examining the unnormalized spectra of the
`two devices in Fig. 4. A 100 ˚A thick layer of Alq3 doped with < 1% of the fluorescent dye DCM2 is
`placed at the heterojunction between Alq3 and 4,40-bis[N-(1-napthyl)-N-phenyl-amino] biphenyl (a-
`NPD). Alq3 preferentially transports electrons, whereas a-NPD preferentially transports holes; thus
`exciton formation is localized to this interface. Since DCM2 exhibits efficient energy transfer from Alq3
`[26], the 1% DCM2 in Alq3 layer effectively removes singlet excitons. Remaining singlets eventually
`recombine in Alq3, yielding the small shoulder in the spectra at < 530 nm. However, in device 2, an
`additional layer of < 10% PtOEP in Alq3 is introduced 200 ˚A away from the exciton formation zone. In
`this device, emission is seen from PtOEP without any change in the intensity of emission from either
`DCM2 or Alq3. Hence, PtOEP cannot be an efficient electron trap, since carriers removed by PtOEP in
`device 2 would result in a decrease in the DCM2 and Alq3 emission: an effect clearly not observed.
`Since the DCM2 acts as a ‘filter’ that removes singlet Alq3 excitons, the only possible origin of the
`PtOEP luminescence is Alq3 triplet states that have diffused through the DCM2 and intervening Alq3
`layers.
`
`Fig. 4 Two electroluminescent devices demonstrating that Alq3 triplets are transferred to PtOEP. Each device
`contains a 100 ˚A thick layer of < 1% DCM2 in Alq3 at the recombination zone. This layer acts to remove singlet
`states. Remaining singlets recombine in Alq3, yielding the shoulder apparent in the spectra at 530 nm. Device 2
`contains an additional layer of < 10% PtOEP in Alq3 positioned 200 ˚A away from the Alq3/a-NPD interface.
`Strong emission is seen from the PtOEP without a corresponding decrease in emission from DCM2 or Alq3. The
`spectra were measured at a current density of 6 mA/cm2. From [1].
`
`An alternative demonstration of triplet transfer to PtOEP was described by Cleave et al. [3]. Here,
`PtOEP was doped into the polymer host poly[4-(N-4-vinylbenzyloxyethyl,N-methylamino)-N-(2,5-di-
`tert-butylphenylnapthalimide)] (PNP) and the transient electroluminescence was examined in OLEDs
`with the general structure ITO/polyvinylcarbazole/PtOEP:PNP/calcium [3]. The observed decay rate is
`the convolution of the rate of energy transfer from PNP to PtOEP and the PtOEP triplet relaxation rate.
`Analysis of the electroluminescent decay determined [3] that a fraction of excitons participated in slow
`(< 10 ms) energy transfer. Moreover, such slow energy transfer from PNP to PtOEP was not found in the
`photoluminescent transient response of PtOEP:PNP. Thus, it was inferred that slow Dexter transfer of
`host
`triplet excitons was
`responsible for
`the difference between the photoluminescent and
`electroluminescent decay transients. By separating the electroluminescent decay into components
`exhibiting fast or slow energy transfer, it was determined [3] that in the 0.1% PtOEP:PNP films only 0.4
`triplets are transferred to PtOEP for every singlet exciton. Since the expected [6] singlet-to-triplet ratio in
`the host material is 1:3, triplet energy transfer is evidently inefficient in this system.
`
`q 1999 IUPAC, Pure Appl. Chem. 71, 2095–2106
`
`RHEMK-1011.006
`
`

`
`Organic light emitting devices
`
`2101
`
`Owing to long lifetimes, triplet diffusion lengths in organic materials may also be substantially longer
`than for singlets. Indeed, it has been found [6] that in Alq3, the triplet diffusion length is > 1400 ˚A, as
`compared to singlet diffusion lengths of < 100 ˚A. To optimize efficiency, phosphorescent OLEDs can be
`modified to trap triplets within the luminescent layer, thereby increasing the probability for energy
`transfer from the host to the phorphor. A material suitable for this purpose is 2,9-dimethyl-4,7 diphenyl-
`1,10-phenanthroline (bathocuproine, or BCP), which has previously been used as a hole blocking layer in
`OLEDs [27]. When placed between a doped HTL and an Alq3 ETL, it was found that light emission
`originated from the HTL. As confirmed by the proposed energy level diagram [28] of Fig. 5, BCP has a
`large ionization potential and blocks the passage of holes out of the HTL. These results also suggest that
`the lowest unoccupied molecular orbital (LUMO) level of BCP freely allows the transport of electrons
`resulting in exciton formation in the HTL. Furthermore, since the energy gap in BCP is < 3.5 eV, it
`should act as a barrier to exciton diffusion.
`
`Fig. 5 Proposed energy level diagram of the electroluminescent devices containing PtOEP. The luminescent
`region is sandwiched between electron blocking a-NPD and hole blocking BCP. Also shown are the chemical
`structures of (a) Alq3, (b) CBP, (c) PtOEP and (d) BCP.
`
`The electroluminescent efficiency of PtOEP can also be optimized by the selection of the host
`material. For example, the photoluminescent efficiency of PtOEP in a 4,40-N,N0-dicarbazole-biphenyl
`is approximately twice that of PtOEP in an Alq3 host [29]. The external quantum
`(CBP) host
`efficiencies of several OLEDs employing these hosts and a BCP barrier layer are shown in Fig. 6. As
`expected, the combination of a CBP host and a BCP barrier layer maximizes PtOEP emission. The
`benefits of the BCP barrier are most noticeable for the thinner (250 ˚A) luminescent layers, where
`energy collection by PtOEP is particularly inefficient. Here, the quantum efficiency in CBP doped
`devices is h (cid:136) (2.2 6 0.1)% at 100 cd/m2, which is nearly twice the reported best result of 1.3% in
`PtOEP:Alq3 devices [1]. The peak efficiency of h (cid:136) (5.6 6 0.1)% for the PtOEP:CBP device is
`equivalent to an internal quantum efficiency of < 32% [30], higher than is expected if only singlet
`excitons are transferred to PtOEP molecules [6]. BCP is less useful for devices employing PtOEP in
`
`q 1999 IUPAC, Pure Appl. Chem. 71, 2095–2106
`
`RHEMK-1011.007
`
`

`
`2102
`
`M. A. BALDO et al.
`
`an Alq3 host where previous studies [1] have shown that under optimum conditions, energy transfer is
`nearly complete with up to < 90% of excitons transferred to the dye within a luminescent layer with a
`thickness of 400 ˚A.
`
`Fig. 6 External quantum efficiencies of PtOEP:CBP and PtOEP:Alq3 devices as a function of current with and
`without a BCP blocking layer. The top axis shows the luminance of the device with a 80 ˚A thick BCP layer and a
`400 ˚A thick CBP luminescent layer doped with 6% PtOEP. Inset: Schematic cross section of the high efficiency
`OLED, consisting of a 450 ˚A thick 4,40-bis[N-(1-napthyl)-N-phenyl-amino] biphenyl (a-NPD) HTL, PtOEP doped
`electron transport layer co-deposited with either Alq3 or CBP acting as the host, an 80 ˚A thick layer of BCP and a
`further 200 ˚A thick cap layer of Alq3 to prevent non-radiative quenching of PtOEP excitons at the cathode.
`Finally, a shadow mask with 1 mm diameter openings was used to define the cathode consisting of a 1000 ˚A thick
`layer of 25:1 Mg:Ag, with a 500 ˚A thick Ag cap. From [29].
`
`Unlike the case of PtOEP in Alq3, the triplet transfer efficiency of PtOEP in CBP is unknown. Indeed,
`the larger offset in HOMO and LUMO levels between PtOEP and CBP may encourage charge trapping
`and direct exciton formation on PtOEP molecules. Further work is required to clarify this issue. However,
`similar to PtOEP in Alq3, the maximum electroluminescent quantum efficiency for PtOEP in CBP is
`obtained at relatively high doping densities (< 6–8% by mass).
`In Fig. 6, all the quantum efficiency curves are characterized by a rapid roll-off above a maximum of
`‘threshold’ current density. At very high current densities, it is known that long phosphorescent lifetimes
`cause saturation of the emissive sites. This is particularly significant at low concentrations of the
`phosphorescent dopant, and is marked by an increase in emission from the host material. However, recent
`work [31] has demonstrated that the roll-off, which occurs at a much lower current density than saturation
`is due to triplet-triplet annihilation. For example, the efficiency of the 250 ˚A thick CBP device begins to
`decrease at a lower current density than that of the 400 ˚A thick CBP device (Fig. 6) due to the higher
`concentration of triplets within the thinner luminescent layer. The roll-off in efficiency also limits the
`maximum brightness obtainable in a phosphorescent OLED, and typically PtOEP-based devices are
`limited to < 1000 cd/m2. This is a particular concern for passive matrix display applications, since these
`require OLEDs to be strongly excited by short electrical pulses. Unfortunately, triplet-triplet annihilation
`increases with the square of the triplet concentration [8]; thus unless the OLEDs can be pulsed faster than
`the phosphorescent decay rate, severe reductions in efficiency are likely.
`
`q 1999 IUPAC, Pure Appl. Chem. 71, 2095–2106
`
`RHEMK-1011.008
`
`

`
`Organic light emitting devices
`
`2103
`
`Triplet-triplet annihilation and saturation are minimized if the phosphorescent lifetime is short. This
`was demonstrated [32] using the green phosphorescent material fac tris(2-phenylpyridine) iridium
`(Ir(ppy)3) [22,33–35]. As in the case of PtOEP, Ir(ppy)3 was doped into a CBP host. The device structure
`and the proposed energy levels [28] of the charge transport materials are identical for Ir(ppy)3:CBP and
`PtOEP:CBP devices (see Fig. 5), although the ionization potentials of PtOEP and Ir(ppy)3 relative to their
`hosts are unknown. Once again, it was found that a thin (60 ˚A) BCP barrier layer was necessary to confine
`excitons within the luminescent zone and achieve high efficiencies.
`Figure 7 shows the external quantum efficiencies of several Ir(ppy)3-based OLEDs. In contrast to
`PtOEP-based devices, the Ir(ppy)3 doped devices exhibit a slow decrease in quantum efficiency with
`increasing current. In addition to the doped device, a heterostructure was fabricated where the
`luminescent region was a homogeneous film of Ir(ppy)3. The reduction in efficiency to (< 0.8%) of neat
`Ir(ppy)3 is reflected in the transient decay, which has a lifetime of only < 100 ns as compared with
`< 500 ns in the 6% Ir(ppy)3 in CBP devices.
`
`Fig. 7 The external quantum efficiency of OLEDs using Ir(ppy)3:CBP luminescent layers. Peak efficiencies are
`observed for mass ratio of 6% Ir(ppy)3:CBP. The 100% Ir(ppy)3 device has a slightly different structure than
`shown in Fig. 1: the Ir(ppy)3 layer is 300 ˚A thick and there is no BCP blocking layer. The efficiency of a 6%
`Ir(ppy)3:CBP device grown without a BCP layer is also shown. From [32].
`
`In Fig. 8, the luminance and power efficiencies are plotted as functions of voltage. The peak power
`efficiency is 31 lm/W with a quantum efficiency of 8%, (28 cd/A). At 100 cd/m2, a power efficiency of
`19 lm/W with a quantum efficiency of 7.5% (26 cd/A) is obtained at a voltage of 4.3 V. The transient
`response of Ir(ppy)3 in CBP is an approximately mono-exponential phosphorescent decay of < 500 ns,
`compared with a measured lifetime [22,33,34] of 2 ms in degassed toluene at room temperature. Slow
`triplet relaxation can form a bottleneck in electrophosphorescence, thereby encouraging triplet-triplet
`annihilation [31] and saturation. But these lifetimes are short and result in only a gradual decrease in
`efficiency with increasing current, leading to a maximum luminance of < 100 000 cd/m2.
`
`q 1999 IUPAC, Pure Appl. Chem. 71, 2095–2106
`
`RHEMK-1011.009
`
`

`
`2104
`
`M. A. BALDO et al.
`
`Fig. 8 The power efficiency and luminance of the 6% Ir(ppy)3:CBP device. At 100 cd/m2, the device requires
`4.3 V and its power efficiency is 19 lm/W. Inset: The chemical structure of Ir(ppy)3. From [32].
`
`In Fig. 9, the emission spectrum and CIE coordinates of Ir(ppy)3 are shown for the highest efficiency
`device. The peak wavelength is l (cid:136) 510 nm, and the full width at half maximum is 70 nm. The spectrum
`and CIE coordinates (x (cid:136) 0.27, y (cid:136) 0.63) are independent of current. Even at very high current densities
`(< 100 mA/cm2), blue emission from CBP is negligible – an indication of complete singlet energy
`transfer. As is the case of PtOEP in CBP, the triplet transfer efficiency is unknown and trapping and direct
`exciton formation on Ir(ppy)3 molecules cannot be ruled out.
`
`Fig. 9 The electroluminescent spectrum of 6% Ir(ppy)3:CBP. Inset: The Commission Internationale de
`L’Eclairage (CIE) chromaticity co-ordinates of Ir(ppy)3 in CBP are shown relative to the fluorescent green
`emitters Alq3 and poly(p-phenylenevinylene) (PPV). From [32].
`
`We note that the device structure has the potential for further optimization. For example, the use of LiF
`cathodes [36,37], shaped substrates [30], and novel hole transport materials [38] that result in a reduction
`in operating voltage or increased quantum efficiency are also applicable to this work. These methods have
`yielded power efficiencies of < 20 lm/W in fluorescent small molecule devices [38]. The quantum
`
`q 1999 IUPAC, Pure Appl. Chem. 71, 2095–2106
`
`RHEMK-1011.010
`
`

`
`Organic light emitting devices
`
`2105
`
`efficiencies in these devices [39] at 100 cd/m2 are typically > 5%, and hence green-emitting
`electrophosphorescent devices with power efficiencies of > 40 lm/W are anticipated.
`
`RELIABILITY OF PHOSPHORESCENT OLEDs
`
`The suitability of semiconducting organic thin films for practical use is ultimately determined by device
`reliability. Unless OLEDs can demonstrate thousands of hours of high performance then applications will
`be scarce. But fluorescent OLEDs have already achieved such standards [40] and there is no reason to
`believe that phosphorescent materials should be any less stable; they may in fact improve reliability. For
`example, there has been speculation that molecules in the triplet state may be particularly susceptible to
`degradation because of the long lifetime of the excitation [41]. Although the triplet lifetime of phosphors
`possessing significant ISC may approach < 100 ms, these lifetimes are still much shorter than those in
`fluorescent materials, where the triplet decay is strongly forbidden. Thus, it is possible that by acting as
`sinks for triplet excitations, phosphorescent materials may in fact ultimately improve device reliability.
`
`Owing to the novelty of phosphorescent dyes, there is little data to test such a contention. But
`preliminary reliability data is available for PtOEP-based OLEDs, and a lifetime of > 105 h at a luminance
`of 35 cd/m2 is observed for OLEDs incorporating PtOEP. This is at least as reliable as the longest lived
`fluorescent devices employing all the same materials except PtOEP. Another study [42] has identified a
`derivative of PtOEP for oxygen sensing applications, where long operational lifetimes are required in
`relatively uncontrolled environments.
`
`CONCLUSION
`
`Although this work has highlighted the performance advantages inherent to phosphorescence, these
`advantages are eliminated if the phosphor cannot efficiently gather triplet and singlet excitons within the
`device. For example, the lanthanide complexes exhibit losses in energy transfer within the complex, and
`a