J. MATER. CHEM., 1991, 1(6), 1001-1005 1001 Luminescence Spectra of Pure and Doped GaBO, and LiGaO, Gradus J. Dirksen, Arthur N. J. M. Hoffman, Teus P. van de Bout, Maurice P. G. Laudy and George Blasse* Debye Research Institute, Utrecht University, Solid State Chemistry Department, P.O. Box 80 000, 3508 TA Utrecht, The Netherlands The luminescences of GaBO, and LiGaO, are reported. The former has Gal" in octahedral sites (calcite structure), the latter in tetrahedral sites (ordered wurtzite structure). The Gall' co-ordination number has the same influence on the luminescence as in the case of 0x0-do complexes: the lower the co-ordination number, the higher the energy of the first absorption band, the larger the Stokes shift, and the less mobile the excited state.The broad- band spectra with large Stokes shift indicate that the nature of the optical transition involved is complicated. Dopants in GaBO, are Eu"'and Cr"'. They are characterized by magnetic-dipole emission due to the inversion symmetry site in calcite. The Cr"' ion yields ,E emission at 4.2 K, in contrast to Cr"' in isomorphous ScBO, where it yields 4T, emission. The dopant in LiGaO, is Fell' which yields an intense deep-red emission with pronounced vibrational structure. Keywords: Luminescence; Gallium(Ir1) ; Gallium borate ; Lithium gallium dioxide 1. Introduction LiGaO, was prepared by dispersing P-Ga203 in a solution of LiOH with or without 0.5 atom.% iron. The dispersion In a recent paper, one of us has summarized evidence for the was dried and the residue firedin air twice after milling.The occurrence of luminescence from complexes consisting of a final firing temperature was 800 "C. central d" metal ion surrounded by oxygen ions in solids Samples were checked by X-ray powder diffraction using and molecules. Examples are Zn"-O" in Zn,O(a~etate),~ Cu-Kcr radiation. and ZII~O(BO~),,~ Cd"-O" in CaO,, and In"'-O" in Y203.5 Luminescence measurements were performed using aLater we extended this series with the luminescence of InB0,.6 Perkin-Elmer MPF-44B spectrofluorometer equipped with The relevant luminescence consists of a strongly Stokes- a liquid-helium cryostat. Excitation measurements in theshifted broad emission band, the nature of which is unknown, 190-250 nm region were performed with a Perkin-Elmerbut consists undoubtedly of a considerable amount of charge- MPF-3 spectrofluorometer equipped with a deuterium lamp.transfer character.' There is a similarity with the well known Diffuse reflection spectra were measured on a Perkin-Elmer luminescence of 0x0-complexes of the do metal ions (such as Lambda-7 spectrometer. vanadate and tungstate). In order to characterize this luminescence further, we investigated two simple gallium oxides, viz. GaB0, and LiGaO,. The luminescence of GaB0, was mentioned super- 3. Results ficially in the literat~re,~ but not much is known about this 3.1 GaB03 compound at all. It turns out to have the calcite crystal structure, i.e. the Gat" ions are six-co-ordinated by oxygen According to its X-ray diffractogram GaBO, has the calcite structure.The lattice parameters are a=4.57 A and c= 14.22ions which belong to triangular borate groups. It is isomorph- A. These values are comparable to those of CrBO, (4.57 and ous with the previously investigated JnBO,.The luminescence 14.23 A) and smaller than those of InB03 (4.82 and 15.45 of two dopant ions was also investigated in GaB03, viz. that A).'' All samples contain two second phases in small amounts, of Cr"' and Eu"'. The crystal structure of LiGa02 is an ordered variant of viz. P-Ga,O, and a borate glass phase (see below). In According to the diffuse reflection spectrum the optical wurtzite with Ga"' ions in tetrahedral co-~rdination.~,~ absorption edge of GaB0, is at 250 nm at 300 K.This implies this way the influence of the co-ordination number of Ga"' on the luminescence can be investigated. Since the Fe"' ion that upon host lattice excitation the small amount of P-Ga2O3 is known to be a very efficient activator in LiGa02,10 it was does not interfere, since its absorption edge is at the same Here already we note that the coincidence of also investigated in more detail. p~sition.'~*'~ It turns out that GaB0, as well as LiGaO, show efficient the optical absorption edge of GaB0, and P-Ga203 is rather luminescence at room temperature. It is surprising that the peculiar, since, generally, that of the borate is at higher energy efficient luminescence of such simple compounds has been than that of the corresponding oxide in view of the eiectro- overlooked for such a long time.negative character of the borate group. Upon excitation with 250nm radiation GaBO, shows an intense luminescence with high quantum efficiency (270%). 2. Experimental The emission consists of a broad band. At 300 K the emission maximum is at 460 nm. At 4.2 K, however, there is a different High-purity starting materials were used, among others emission band with a maximum at 375 nm (see Fig. 1). Both P-Ga203 (99.999%). Since this is not very reactive, it was emissions have an excitation spectrum which consists of a dissolved in concentrated KOH solution and precipitated as band coinciding with the absorption in the diffuse reflection the oxalate. This was fired in air with boric acid in order to spectrum.The excitation band maximum is at 250 nm. Upon obtain GaBO,. The final firing temperature was 825 "C.The increasing the temperature from 4.2 to 300 K, the ultraviolet dopants Cr,03 and Eu,03 were added in a concentration of emission quenches at ca. 150 K in favour of the blue emission. 0.5 atom.%. At 4.2 K the blue emission can be excited selectively by \ \ \ 300 500 wavelength/nm Fig. 1 Luminescence spectra of GaBO,. EX =Excitation spectrum of the emission of GaBO, at 4.2 K; qr gives the relative quantum output in arbitrary units. EM =Emission spectra of GaBO, at 4.2 K (LHT) and 290 K (RT) under 250 nm excitations; CD gives the relative spectral radiance in arbitrary units radiation with A I260 nm. The similarity between the lumi- nescence properties of GaB0, and P-Ga203 is ~triking.'~.' Therefore we follow the interpretation given for P-Ga203, viz.the ultraviolet emission is intrinsic and the blue emission is extrinsic. The Stokes shift of the intrinsic emission is ca. 13 500 cm-'. 3.2 GaB0, :Ed1' The luminescence properties of GaB0, :Eu"' are very similar to those of ScBO, :Eu"' which has the calcite structure also.I4 The Eu"' ion occupies a site with inversion symmetry so that 5Do-7F1 emission dominates. The other emission transitions occur as broad and weak features, since they can only take place as vibronic transitions. In GaBO, :Eu"' the europium emission can hardly be excited in the host lattice. Excitation with 250 nm yields mainly the blue/UV broad emission band on which the Eu"' lines are just observable.The excitation spectrum of the Eu"' emission does not show the host lattice excitation band. The study of GaB0,:Eu"' showed, however, that our sample contains a glassy second phase. This is most clear upon excitation into the 7Fo-5D2 transition of Eu"' which is forbidden in GaBO,. This excitation yields an Eu"' emission which is very similar to that in borate gla~ses:'~~'~ it consists of very broad lines and the 5D,-7F2 emission dominates. The broadness of the lines is due to inhomogeneous broadening in the glass phase. Whereas X-ray diffraction reveals one second phase, viz. P-Ga203, optical spectroscopy reveals another, viz. a borate glass. Fortunately, it is possible to separate the Eu"' emissions in the glass phase and the calcite phase, since the electric-dipole transitions dominate in the former and the magnetic-dipole transitions in the latter.Spectroscopically we did not observe any evidence for Eu"' in the B-Ga203 second phase. 3.3 GaBO, :Cr"' The luminescence properties of Cr"' in the calcite structure have been reported for ScB0, by several author^'^^'^ and the material has been proposed as a laser host. In ScB03 :Cr"' the chromium emission consists of the broad-band 4T2+4A2 emission, down to 4.2 K. 'This is due to a weak crystal field as is evident from the maximum of the 4A2-4T2 absorption band at 15 750 ~m-'.'~ J. MATER. CHEM., 1991, VOL. 1 In GaBO, :Cr"' the crystal field is stronger; the absorption maximum is now at 17 100 cm-' and at 4.2 K the emission consists of the 2E-4A2 transition (see Fig.2). The stronger crystal field is, of course, due to the smaller cation site in GaBO, (the ionic radii for six co-ordination are 0.615, 0.620 and 0.745 8, for Cr"', Ga"' and Sc"', respectively). At room temperature the chromium emission contains also the 4T2- 4A2 band emission, with a maximum at ca. 710 nm. Table 1 shows the spectral features observed for the emission of GaB0,:Cr"' at 4.2 K together with their assignment. The vibronic lines are relatively strong (see Fig. 2) which is due to the fact that the pure electronic line is electric-dipole forbidden 760 700 waveIe ng thin rn Fig. 2 Emission spectrum of GaBO, :Cr at 4.2 K Table 1 The vibronic side lines in the 2E-+4A2 emission of Cr"' in GaBO, at 4.2 K, and an assignment.Frequencies for ScB0,:Cr"' are given for c~mparison'~ GaBO, : Cr"' ScBO, :Cr"' position relative to vibronics in zero-phonon lineicm -assignment" 4~2-4~, emission 14 435 0-0 -65 V1 40 -180 '6 I70 -292 v4 320 -360 -430 v3 405 -662 V1 610 -710 v1 +Vl -850 pair line -960 pair line -1250 v3(B0,3 -1 1200 vl, lattice mode; v1 -6, vibrational modes of Cr06 octahedron; v,(BO,,-), asymmetrical borate stretching vibration J. MATER. CHEM., 1991, VOL. 1 due to the inversion centre at the cation site of the calcite structure. It can only occur as a magnetic-dipole transition. In Table 1 the vibronic lines are tabulated and tentatively assigned.The dominating vibronics are ascribed to coupling with the CrO, vibrational modes. It is interesting to note that in ScBO, :Cr these were also observed, although the emission transition is a different one, viz. 4T2-+4A2.However, in both cases the ungerade v3, v4 and v6 are necessary to break through the parity selection rule. Table 1 shows that in ScBO, these Cr06 vibrations are at lower frequency, which is to be expected in view of the larger Cr-0 distance. In addition we observed other side lines, uiz. one due to coupling with a lattice mode, vl, and one due to coupling with the asymmetric borate stretching vibration v,. The latter is very weak. Two lines are ascribed to pair lines since they are relatively sharp and their intensity varies upon changing the excitation wavelength.Actually, 1 -0.99512=6% of the Cr"' ions are expected to have a nearest Cr"' neighbour if we assume a statistical distribution of Cr"' on the gallium sites and 12 nearest cation neighbours. So the presence of pair lines is not unexpected. Fig. 3 shows the excitation spectrum of the Cr"' emission of GaBO, :Cr"' at room temperature. It shows the well known transitions 4A2-'4T2, 4T1 (F) and 4T1 (P) in sequence of increasing energy. The excitation band at the highest energy is the same as observed for the host-lattice emission and is ascribed to host-lattice excitation. At 4.2 K this host-lattice excitation band disappears. For comparison we investigated also P-Ga203 :Cr"'.Its luminescence characteristics are very similar to those of GaB0,:Cr"' with two exceptions: (i) the 4.2 K emission spectrum shows a much stronger zero-phonon line in the 2E-4Az emission for /I-Ga,O3:Cr1'' owing to the absence of inversion symmetry at the metal-ion sites of P-Ga203; (ii) host-lattice excitation is more effective in P-Ga203 :Cr"' than in GaB0, :Cr"'. At 4.2 K the excitation spectrum of the Cr"' emission of B-Ga2O3 :Cr"' contains the host-lattice excitation band, whereas that of GaBO, :Cr"' does not. Our samples of GaB0,:Cr"' contain, according to X-ray diffraction, a small amount of P-Ga203. In view of the results for GaBO, :Eu"' they may also contain a certain amount of the glass phase.The emission spectra at 4.2 K do not give any evidence for Cr"' in a glass phase, which is expected to give a broad-band 4T2-4A2 emission.16-18 This holds for every reasonable excitation wavelength. Therefore the Cr"' ion prefers the crystalline phases above the glass phase, as is well known (see e.g. ref. 18). Under a suitable excitation 1 300 600 wave lengthin m Fig. 3 Excitation spectrum of the Cr"' emission of GaBO,: Cr at 4.2 K. The excited levels of Cr"' have been indicated. H means host- lattice excitation wavelength it is possible to observe the Cr"' emission of 8-Ga203:Cr"'. It is easy, however, to account for this impurity. 3.4 LiGaO, The compound LiGaO, has an ordered wurtzite structure. Its diffuse reflection spectrum shows that the optical absorp- tion edge is at ca.215 nm at 300 K. The compound shows efficient luminescence at room temperature if excited with sufficiently short wavelengths. Fig. 4 shows the emission and excitation spectrum of LiGaO, at 290 K. The emission consists of a broad band with a maximum at 360 nm. The corresponding excitation band has a maximum at 220 nm which corresponds nicely with the diffuse reflection spectrum. From these values a Stokes shift of ca. 18 000 cm-' is derived. Note the higher energy positions of the spectral bands and the larger Stokes shift in comparison with the luminescence of GaBO, (section 3.1). 3.5 LiGaO, :Fe"' Rabatin" has claimed efficient and deep-red Fe"' emission from LiGa02:Fe"' and this was confirmed in the present study.Here we present only data which were not given by Rabatin, viz. the 4.2 K emission spectrum with a rich vibrational structure (see Table 2 and Fig. 5) and the Fe"' excitation spectrum. The latter spectrum consists of the well known crystal field transitions within the 3d5 configuration, the Fe"'-O" charge-transfer band at 260 nm, and the LiGaO, excitation band. The latter two have about the same intensity. 250 400 wave1 en gth in m Fig. 4 Emission and excitation spectra of the luminescence of LiGaO, at 290 K Table2 The vibronic side lines in the 4T1-6A1emission of Fe"' in LiGaO, at 4.2 K, and an assignment position relative to zero-phonon line/cm- assignment" 14 085 0-0 -1 10 Vl -225 v4 -325 -430 v3 -515 -635 -725 V1 1'1 +v, -900 '1 +'4 -1005 -1065 v1 +v3 -1120 a vl,lattice mode; v1 ,3 ,,, vibrational modes of FeO, tetrahedron I I I 4 760 700 wavelength/nm Fig.5 Emission spectrum of LiGaO, :Fe at 4.2 K In view of the Fe" concentration in LiGaO, (0.5 atom.%) this points to a restricted amount of energy transfer from host to activator. The 4.2 K emission spectrum consists of an intense zero- phonon line followed by a large number of side bands. These are tabulated in Table2 which contains also a possible assignment. The latter needs a more extended study for it to be confirmed. 4. Discussion Here we wish to concentrate on the gallate luminescence. The dopant emissions are essentially known and understood, and were treated above.It has been shown before that there exists an analogy between the luminescences of oxo-d10 and oxo- do complexes.' The present results underline this statement as will be shown now without entering into the problem of the nature of the optical transitions involved. We use the short-hand notation introduced above, uiz. oxo-d" and oxo- do for complexes such as Ga"' (d'") [O"In and Wv' (do) [O" I,, respectively. The luminescence properties of oxo-do complexes have been reviewed in ref, 20. Their emission and excitation spectra are characterized by very broad bands with Stokes shifts of 1-2 eV. Generally speaking the octahedral oxo-d" complexes have a smaller Stokes shift than the tetrahedral ones: for example, the WO: -octahedron in ordered perovskites shows a Stokes shift of 12 000 ern-', and the WOZ-tetrahedron in scheelites has 16 000 cm-'.' Finally, the optical absorption J. MATER.CHEM., 1991, VOL. 1 edge shifts to higher energies if the co-ordination number decreases. These properties are clearly evident in the present investi- gation. All gallate spectra observed are of the broad-band type (Fig. 1 and 4) and the Stokes shift of the emission is large (see Table 3). Since we now have available data for a compound with six-co-ordinated Ga"', uiz. GaBO,, and another with four-coordinated Ga"', uiz. LiGaO,, it is possible to evaluate the influence of the co-ordination number on the luminescence properties.Clearly the Stokes shift is larger for tetrahedral than for octahedral co-ordination (see Table 3). From a comparison of the excitation maxima for GaBO, and LiGa02 it is clear that the optical absorption edge is at lower energy for the case of six co-ordination (see Table 3). However, it should be realized that interaction between the optical centres involved may, in the case of a solid, lead to energy-band formation, so that the absorption edge shifts to lower energies. In case of the oxo-do complexes this is nicely illustrated by examples like TiO, and WO, where the edge shifts into the visible region. This effect plays most probably an important role in the case of In203 also (see Table 3). The comparison of InBO, and In203 leads immediately to the question why the optical absorption edges of GaBO, and /?-Ga203 are about the same.Above it was already noted that this coincidence is unexpected in view of the different nature of the borate and the oxygen ligands. However, in /?-Ga203 only 50% of the Ga'" ions are in six co-ordination, the others being in four co-ordination. In addition, both type of ions form layers. A nice presentation of the crystal structure has been given by Clark.21 In the octahedral layers each gallate octahedron has oxygen ions in common with four other octahedra. In GaB03 each Ga"' ion has 12 Ga"' nearest neighbours. This might bring the optical absorption edge to a lower value than expected at first sight. Unfortunately, we could not find reliable absorption data for a-Ga203 with all Ga"' ions in octahedral co-ordination. However, for SrGa12019 this edge is lower than for /?-Ga203.In SrGal2OI9 with magnetoplumbite structure there are spinel layers in which every octahedrally co-ordinated Ga"' ion has six Ga"' neighbours, a higher number than in P-Ga203. In view of our data we assume that the gallate tetrahedra in p-Ga203 show optical absorption at energies above the absorp- tion edge, so that it cannot be observed. In spite of the qualitative nature of the discussion it is clear that the analogy between the luminescence properties of oxo- d" and oxo-do complexes is striking. It is interesting to note that recently Nikol and Vogler2, observed for Sb"' and Bi"' chloro complexes the same spectral dependence on co- ordination number.In order to discuss the mobility of the excited state it is useful to consider how far the intrinsic excitation energy can be transferred to the impurity centres. Let us again first summarize the situation for the oxo-do complexes which has Table 3 Some data on the luminescence of compounds containing d" metal ions compounda excitation maximum/1O3 cm-' excitation maximum/103 cm- ' Stokes shift/103 cm-' ref. 40.0 26.5 13.5 this work 40.0 26.5 13.5 13 37.0 22.2 14.8 19 InBO, (6) ca. 45 33 ca. 12 6 InZ03 (6) 25' ca. 15 ca. 10 5 LiGaO, (4) Zn,O(acetate),c (4)Zn,O(B02)6 (4) 45.0 40.0 46.3 27.0 22.5 26.9 18.0 17.5 19.4 this work 3 2 a The co-ordination number of the d'O ion is given in parentheses; indirect band gap'; molecular complex in solution.J. MATER. CHEM., 1991, VOL. 1 been well studied.,, If the Stokes shift is large, the excitation is completely localized and the total amount of energy transfer restricted. However, if the Stokes shift is not too large, the excitation energy is mobile at room temperature and reaches the impurities from where efficient emission occurs. An example of the former is YNb04: Eu (niobate Stokes shift 16 000 cm-'), and of the latter YVO, :Eu (vanadate Stokes shift 10 000 cm-')',23 The same situation prevails for the 0x0-d" complexes. The compounds GaB03 and P-Ga203 are nice examples. At 4.2 K the gallate group shows intrinsic ultraviolet emission. At higher temperatures this emission disappears in favour of a blue emission.As has been shown for P-Ga203, this is due to energy migration to the blue-emitting centre.13 Unfortu- nately the nature of the centre is unknown. Harwig and Kellendonk have proposed that we are dealing with a gallate group with an oxygen vacancy.24 This is reminiscent of similar defect centres in 0x0-do complexes.' The presence of Eu"' or Cr"' in GaB0, can hardly compete with the blue centres for the excitation energy, the presence of Cr"' in P-Ga203 can at least trap part of the excitation energy. In contrast to these observations In,O,:Eu"' is able to show efficient luminescence upon host-lattice e~citation.~ This shows that the mobility of the excited state in In,O, is much higher, in agreement with the broader energy bands, which in turn are responsible for an absorption edge at low energy.In In203, with C-type lanthanide oxide structure, every In"' has 12 In"' neighbours which seem to be the basis for our observations. It is interesting that the case of Mn" in SrGa1201, shows an amount of energy transfer from host to activator" which is in between that in GaB0,:Eu"' and that in In,O,:Eu"'. This is to be expected from the position of the absorption edge. In ZII,O(BO~)~ with a larger Stokes shift, energy transfer to Mn" does not seem to take place., The same is expected to be the case for Fe"' in LiGaO,. Actually some transfer was observed in that system. However, it should be realized that each Ga"' ion in LiGaO, has six Ga"' neighbours and that the Fe"' concentration is 0.5 atom.%.Consequently 1 -0.9956=3% of the Ga"' ions has an Fe"' neighbour to which one-step energy transfer is possible. This accounts for the greater part of the transfer observed. All these results are additional evidence for the earlier statement' that 0x0-d" complexes have a localized excited state which is able to give efficient broad-band emission. At higher temperatures this excited state may become mobile depending on the amount of relaxation as measured by the Stokes shift. Therefore, this emission has also been indicated as self-trapped exciton emission (see e.g. ref. 13). These con- siderations, however, do not solve the problem of the nature of the optical transition.For that purpose, at least, a detailed energy-level calculation of an 0x0-d" complex has to be performed. References 1 G. Blasse, Chem. Phys. Lett., 1990, 175, 237. 2 H. Kunkely and A. Vogler, J. Chem. SOC., Chem. Commun., 1990, 1204. 3 A. Meijerink, G. Blasse and M. Glasbeek, J. Phys. Condensed Matter, 1990, 2, 6303. 4 H. Lange, Techn. Wiss. Abh. Osram Ges. (Munchen), 1969, 10, 87. 5 H. Yamamoto and K. Urabe, J. Electrochem. Soc., 1982, 129, 2069. 6 G. Blasse and L. H. Brixner, Muter. Chem. Phys., 1991, 28, 275. 7 G. Blasse, J. Inorg. Nucl. Chem., 1967, 29, 266. 8 R. Hoppe, Bull. SOC. Chim. Fr., 1965, 1115. 9 M. O'Keeffe and B. G. Hyde, Acta Crystallogr, Sect. B, 1978, 34, 3519. I0 J. G. Rabatin, J. Electrochem. SOC., 1978, 125, 920. 11 0. Muller and R. Roy, The Major Ternary Structural Families, Springer-Verlag, Berlin, 1974. 12 G. Blasse and A. Bril, J. Phys. Chem. Solids, 1970, 31, 707. 13 T. Harwig, F. Kellendonk and S. Slappendel, J. Phys. Chem. Solids, 1978, 39, 675. 14 G. Blasse and G. J. Dirksen, Inorg. Chim. Acta, 1988, 145, 303. 15 J. W. M. Verwey, G. J. Dirksen and G. Blasse, J. Non-cryst. Solids, 1988, 107, 49. 16 J. W. M. Verwey and G. Blasse, Muter. Chem. Phys., 1990, 25, 91. 17 S. T. Lai, B. H. T. Chai, M. Long and R. C. Morris, IEEE J. Quantum Electron., 1986, 22, 1931. 18 G. Boulon, Muter. Chem. Phys., 1987, 16, 301. 19 J. M. P. J. Verstegen, J. Solid State Chem., 1973, 7, 468. 20 G. Blasse, Structure Bonding, 1980, 42, 1. 21 G. M. Clark, The Structures of Non-molecular Solids, Applied Science, London, 1972. 22 H. Nikol and A. Vogler, J. Am. Chem. SOC., to be published. 23 G. Blasse, in Handbook on the Physics and Chemistry of the Rare Earths, ed. K. A. Gschneidner Jr. and L. Eyring, North Holland, 1979, ch. 34. 24 T. Harwig and F. Kellendonk, J. Solid State Chem., 1978, 24, 255. Paper 1/024816; Received 28th May, 1991