You are on page 1of 3

Excitation Frequency Dependence and Fluorescence in the

Raman Spectra of Al2O3


A. AMINZADEH*
Centre for Instrumental and Development Chemistry, QUT, Brisbane, Australia

Raman spectra of aluminu m oxide, Al2O3, have been a subject of Al2 O3 in order to characterize its ¯ uorescence back-
interest over the last two decades. The origin of the ¯ uorescence ground.
background in the Raman spectra of Al2O3 is still a subject of dis-
cussion. It has been believed to be due to hydrocarbon, iron ion EXPERIMENTAL
impurities, or hydroxy groups on the surface of Al2O3. In this paper,
Raman spectra of Al2O3 are reported that were acquired with dif- Materials. g -Aluminum oxide was purchased from
feren t excitation frequencies as well as differen t spectrom eters and commercial sources (Merck) and was used without fur-
differ ent temper atures. It is shown that, whereas at an excitation ther puri® cation. An atomic absorption measurement
frequency of 1064 nm a strong ¯ uorescence background is observed, showed its iron content to be about 300 ppm. Heat-treated
a sample of Al2O3 heated at 12508 C when excited with a 632.8-nm
frequency exhibits strong and remarkable ¯ uorescence ® ne struc-
Al2 O3 was obtained by heating g -aluminum oxide for 2
ture. It is concluded that the origin of the ¯ uorescen ce background h at 1250 8 C in an electric furnace.
in the Raman spectra of Al2O3 is iron, and Fe impur ity, and the Instrumentation. Raman spectra were recorded with
intensity of this ¯ uorescen ce depends upon the structure of Al2O3 the use of three different excitation frequencies and three
and the excitation frequen cy used. kinds of instruments, as follows:
Index Headings: Aluminum oxide; Raman spectroscopy; Fluores- 1. A Renishaw Raman microprobe Model 1000 equipped
cence.
with a charge-compled device (CCD) detector. Exci-
tation at 632.8 nm was obtained from a HeNe laser,
Spectra Physics Model 127.
INTRODUCTION 2. A Perkin± Elmer FT-Raman System 2000 equipped
with a GaInAs detector. Excitation at 1064 nm was
Aluminum oxide, Al2 O3, sometimes referred to as alu- obtained from a Nd:YAG laser, I.E. Optomech Model
mina, is a very important mineral with wide applications 385.
as a catalyst support,1 as a ceramic material,2 and as a 3. A Cary Model 82 equipped with a P.M. R943-02 pho-
dielectric in electrolytic capacitors.3 tomultipler. Excitation at 514.5 nm was obtained from
Al2O3 has several different phases. The modi® cations a Coherent Radiation Ar1 laser.
of Al2 O3 and the nature of the phases depend upon the
method and the temperature of production. With boehm- RESULTS AND DISCUSSION
ite as a starting material at 500 to 850 8 C, g -alumina is
formed; from 850 to about 1100 8 C, u -alumina is ob- Figure 1 shows Raman spectra of g -aluminum oxide
tained; and above 1100 8 C, the a -phase is obtained. Al- (g -Al2O3) with excitation wavelengths of (a) 1064, (b)
though with different starting materials the nature of 514.5, and (c) 632.8 nm, respectively. Figure 2 shows
phases at different temperatures is different, above 1100 Raman spectra of a -aluminum oxide (heated at 1250 8 C),
8 CÐ with any starting materialÐ only the a -phase is ob- at excitation frequencies of (a) 1064, (b) 514.5, and (c)
tained. The naturally occurring forms of Al2 O3 Ð namely, 632.8 nm, respectively. These spectra have been recorded
corundum, emery, sapphire, and rubyÐ are more or less under the following conditions: For 632.8-nm excitation:
pure forms of a -Al2 O3. 8 mW power of a HeNe laser, 10 scans, and the resolution
The ® rst Raman spectrum of corundum was reported of the spectrometer estimated to be 4 cm2 1 . For 1064-nm
by Krishnan,4 who identi® ed seven allowed vibrations. In excitation: 50 mW power of a Nd:YAG laser, 5 scans,
1967 Porto and Krishnan5 described the Raman spectrum and resolution of either 4 or 8 cm2 1 . For 514.5-nm ex-
of a single crystal of corundum. In their Raman spectrum, citation: 400 mW of an Ar1 laser, 5 scans, and resolution
Porto and Krishnan reported seven Raman bands at 378, of about 4 cm2 1.
418, 432, 451, 578, 645, and 751 cm2 1 . Since then, sev- It can be seen from Fig. 1 that, apart from a broad,
eral additional reports have appeared in the literature on weak band at about 1300 cm2 1 in Raman spectrum c,
the Raman spectrum of Al2O3 , all of which show a ¯ u- which was recorded with 632.8-nm excitation, all spectra
orescence background and deal with the origin of this show a typical ¯ uorescence background. However, it can
background.6± 12 be seen from Fig. 2 that, although at 1064-nm excitation
The origin of the ¯ uorescence background in all these a strong and at 514.5-nm a rather weak ¯ uorescence
reports has been a subject of debate.6± 12 We therefore un- background (spectra a and b) are observed, in Raman
dertook a careful investigation of the Raman spectra of spectrum c, which was recorded with excitation at 632.8
nm, a well-resolved spectrum with ® ne structure is ob-
Received 29 May 1996; accepted 5 December 1996. tained. In this spectrum the bands are located at 622, 856,
* Permanent address: College of Chemistry, Isfahan University of Tech- 1000, 1153, 1356, 1400, 1660, 1792, and 2034 cm2 1.
nology, Isfahan, IR, Iran. None of these bands appear at the anti-Stokes side of the
0003-7028 / 97 / 5106-0817$2.00 / 0
Volume 51, Number 6, 1997 q 1997 Society for Applied Spectroscopy APPLIED SPECTROSCOPY 817
FIG. 1. Raman spectra of g -Al 2O3 with the use of excitation frequen-
cies at (a) 1064, (b) 514.5, and (c) 632.8 nm, in the wavenumber shift
range of 200± 2000 cm2 1 .

Raman spectrum. We therefore designate these bands as


¯ uorescence bands. The only Raman bands which can be
observed in this spectrum might be at 410 and 622 cm2 1 , FIG. 2. Raman spectra of a -Al2O3 with the use of excitation frequen-
and they are very weak.13 cies at (a) 1064, (b) 514.5, and (c) 632.8 nm, in the wavenumber shift
range of 200± 2000 cm2 1 .
As early as 1945 and 1948, DeMent14 and Kroger,15
respectively, pointed out that pure Al2O3 is neither ¯ uo-
rescent nor phosphorescent. However, a very small por-
tion of chromic oxide (about 10 ppm) will produce ruby At this temperature, most normal and light hydrocarbons
with an intense ¯ uorescence. Egerton et al.6 reported that are burned, and other heavy hydrocarbons are either de-
the main source of ¯ uorescence background in Al2 O3 is composed or vaporized.16
due to hydrocarbon impurities, with some due to Fe31 . In There remains our proposal that Fe31 is the source of
a similar interpretation, Jeziorowski and Knozinger ex- ¯ uorescence. This conclusion is supported with the fact
plained the ¯ uorescence background in terms of hydro- that Fe31 has a strong emission band centered at 700 nm
carbon impurities. Fe31 , and excitations from structural (about 14,300 cm2 1)17 when located in a spinel environ-
defects.7 Two recent reports of Mortensen et al.9,10 explain ment. As has been pointed out, it is well known that, at
the ¯ uorescence background in the Raman spectra of high temperatures, Al2O3 is converted to a -Al2 O3 , which
Al2O3 in terms of Fe impurity and surface hydroxyl has a spinel structure. The excitation frequency at 632.8
groups. Another very recent report of Chen et al.12 has nm is close to the maxima of the ¯ uorescence band
appeared, with the conclusion that the background is due (14,300 cm2 1), and the position of the two strongest
only to surface hydroxyl groups. bands (spectrum c, Fig. 2) at Raman shifts of 1350 and
On the basis of our results presented in Figs. 1 and 2, 1400 cm2 1 will, on the absolute wavenumber scale, be
we propose that the origin of ¯ uorescence in the Raman close to the maxima of the emission of the Fe31 ¯ uores-
spectra of Al2 O3 is due solely to iron impurity and that cence band. In agreement with our ® ndings, Mortenson
its intensity depends upon the structural environment of et al., using two samples of Al2O3 with 193 ppm and less
the host. In our Raman spectra, three excitation frequen- than 20 ppm of Fe, report a higher background ¯ uores-
ciesÐ namely, 1064, 632.8, and 514.5 nmÐ were used. If cence with 193-ppm Fe impurity.9
one regards the surface hydroxyl groups as the main or-
igin of ¯ uorescence background in the Raman spectra of CONCLUSION
Al2O3, then all three spectra should show a rather similar
background. Moreover, one should also observe Raman We conclude that, in agreement with Mortenson et al.,9
bands due to hydroxyl groups as well. No such bands are the origin of ¯ uorescence in the Raman spectra of Al2 O3
observed in this respect. If one regards hydrocarbon im- is due to Fe impurity. We assign the strong bands at spec-
purity as the source of ¯ uorescence background, the Ra- trum c in Fig. 2 to the laser-induced ¯ uorescence bands
man spectrum taken with 632.8-nm excitation could be of Fe31 , which has replaced the Al31 ion in a spinel struc-
interpreted only as a rigorous resonace Raman spectrum ture.
originating from an absorption band close to 632.8 nm. Finally, we should point out that a recent work on the
This possibility is rejected since there has been no such Raman spectra of Al2O3 reported by Prassad et al.,18 in
reference either to the nature of this absorption band or which more than 16 Raman bands have been assigned to
to the concentration of hydrocarbon. Moreover, at tem- Al2O3 and two Raman bands to NiO as well, might be
peratures as high as 1250 8 C, no hydrocarbon can exist. wrong and should be revised.

818 Volume 51, Number 6, 1997


ACKNOWLEDGMENTS 8. H. Jeziorowski and H. Knozinger, Chem. Phys. Lett. 51, 519
(1977).
I would like to thank CIDC, Queensland University of Technology,
9. A. Mortensen, D. H. Christensen, and O. Fauurskov Nielsen, J.
for the use of Raman equipment. I am grateful to Professor P. Fredericks
for helpful advice and to Dr. L. Rintoul for his excellent technical as- Raman. Spectrosc. 22, 47 (1991).
sistance. I am also grateful to Dr. M. Riley from the University of 10. A. Mortensen, D. H. Christensen, and O. Fraurskov Nielsen, J. Ra-
Queensland for the use of the Cary Model 82 Raman instrument. Fi- man. Spectrosc 24, 667 (1993).
nally, I would like to thank the Isfahan University of Technology (IR, 11. C. Dyer, P. J. Hendra, W. Forsling, and M. Ranheimer, Spectrochim.
Iran) for ® nancial support. Acta 49A, 691 (1993).
12. Y. Chen, J. Hyldtoft, C. J. H. Jacobson, and O. Faurskov Nielsen,
Spectrochim. Acta 51A, 216 (1995).
13. A. Aminzadeh and H. Sarikhani-Fard, Aus. J. Chem., paper sub-
1. C. L. Thomas, Catalytic Processes and Proven Catalysis (Academ-
ic Press, New York, 1970). mitted (1996).
2. E. Dorr and H. Hubner, Alumina (Springer-Verlag, Berlin/Heidel- 14. J. DeMent, Fluorochemistry (Chemical Publishing Company, New
berg, 1994). York, 1945).
3. K. S. Chari and B. Mathur, Thin Solid Films 891, 271 (1981). 15. F. A. Kroger, Some Aspects of the Luminescence of Solids, (Elsevier,
4. R. S. Krishnan, Proc. Indian. Acad. Sci. 26A, 450 (1947). New York, 1948).
5. S. P. S. Porto and R. S. Krishnan, J. Chem. Phys. 47, 1009 (1967). 16. J. B. Maxwell, Data Book on Hydrocarbones (Krieger, Melbourne,
6. T. A. Egerton, A. H. Hardin, Y. Kozirovski, and N. Sheppard, J. Florida 1975).
Catal. 32, 343 (1974). 17. G. T. Pott and B. D. McNicol, J. Chem. Phys. 56, 5246 (1972).
7. H. Jeziorowski and H. Knozinger, Chem. Phys. Lett. 42, 162 18. R. Prasad, A. Garg, S. Mathews, and A. Bhardwaj, Orlent. J. Chem.
(1976). 10, 231 (1994).

APPLIED SPECTROSCOPY 819