- Email: [email protected]
A microscope for fourier transform Raman spectroscopy
Specmchimica Acta, Vol. 46A, No. 2, pp. 153-159. Printed in Great Britain
1990 0
A microscope for Fourier Transform
0.564-8.539190 53.00+ 0.00 199u Pergamon Press plc
Raman spectroscopy
F. J. BERGIN Shell Research Ltd., Thornton Research Centre, P.O. Box 1, Chester CH13SH, U.K. (Received 1 September 1989; accepted 6 November 1989) Abstract-Vibrational microscopy, and in particular Raman microscopy, has been used for many years as a powerful spectroscopic tool. The recent developments in FT Raman spectroscopy have opened up new areas in which Raman spectroscopy can play a useful role. In light of this, a simple Raman microscope, based on refractive optics, has been coupled to an FT Raman system. This allows data to be collected from a wide range of materials including single fibres and crystals. The reduced performance of the microscope system, when compared to a conventional FT Raman system, is rationalized in terms of the losses in the optical components used.
1. INTRODUCTION
the first report of FT Raman spectroscopy using near infrared excitation [l], there has been a steady growth of interest in both the instrumentational aspects and the applications of the technique. However, little attention has been devoted to the development of FT Raman microscopy. MESSERSCHMIDT and CHASE [2] have examined the possibility of using a Spectra-Tech IR plan microscope and have achieved results that “argue for the development and commercialization of a combined FT IR/FT Raman microscopic instrument.” In an earlier report [3] we outlined some preliminary results obtained using a conventional glass-optics microscope. This paper describes some further developments we have undertaken in this area. The relative strengths and weaknesses of FT Raman spectroscopy have been dealt with elsewhere in this issue. The additional use of a microscope would permit direct analysis of any given area of interest in a sample with a spatial resolution of the order of several microns. These advantages provide sufficient incentive to further investigate this aspect of the technique. SINCE
2. EXPERIMENTAL The experimental aspects of FT Raman spectroscopy, using “macro optics”, have been dealt with in several publications [l, 3-51. The salient features of our macro system are as follows. The spectrometer used was a Perkin-Elmer model 1760 near infrared instrument equipped with a liquid nitrogen-cooled germanium detector (NEP = 2 x lo-l4 WHz-I’*). The Nd: YAG laser was supplied by Spectron Lasers and the Rayleigh rejection achieved by using two dielectric-coated filters from Glen Creston. The combination of detector and filters resulted in Raman data being collected in the range 400-31OOcm-‘. The Raman-scattered signal was collected in a 180” backscattered configuration using an ellipsoid mirror supplied by Perkin-Elmer. A more detailed description can be found in Ref. [3]. For our microscopy studies the 180” collection ellipsoid was replaced by a Nikon Optiphot metallurgical microscope.. In the original configuration of the microscope system the laser was brought to the sample via a 12.5% reflecting beamsplitter and a x40 lens, as shown schematically in Fig. l(a). This combination gave an overall throughput (laser head to sample) of approximately 4%. In an effort to improve the throughput, the microscope was modified by replacing the beamsplitter with a dielectric-coated Nd : YAG reflection filter. The reflection characteristics of this filter are illustrated in Fig. 2. By suitable adaptation of the beam-steering optics, this filter was incorporated into the microscope as shown in Fig. l(b). The overall throughput was thus increased to 30-40% allowing powers of up to 1 W to be focused onto the sample. Whilst this filter arrangement served the dual purpose of increasing the throughput and simultaneously incorporating some Rayleigh rejection, it suffers from one drawback. Although the filter is transparent to visible light, allowing the sample to be viewed through the microscope, the 153
154
F. J. BERGIN T.V.
(b)
T.V.
Spatial
7_
aperture ------_____ To spectrometer
Tospectrometer
Lens I
0
&Prism
/ / I
0
Filter
---IT
Mcroscope objective
Microscope
objective
+
A
-i-
Sample
Sample
Fig. 1. (a) Schematic illustration of original configuration
of FT Raman microscope. (b) Modified FT Raman microscope. The filter used was an ND : YAG Rayleigh rejection filter.
exact position of the focused laser spot could not be identified directly. However, its position could be located by using the HeNe laser which was set to run coaxial to the YAG laser beam. With the original beamsplitter configuration the diameter of the laser spot on a glass slide was about 5-10 microns. There should be no reason to assume that the spot size should change as a result of using the filter configuration. It is worth noting here the approximate dimensions of the optical imaging. A focused spot size of approximately 5-10 microns and a magnification of 40 produces a primary image that is, at most, only 400 microns in diameter. This first primary image, in our system, is formed approximately 200 mm from the objective lens-in essence, a very narrow, almost parallel beam is produced by the microscope objective. This type of optical system is far from ideal for our current FT spectrometer. A single lens was positioned approximately at the primary image formed by the microscope objective, which in turn produced a further image at the Jacquinot stop of the spectrometer. This lens was mounted on an X-Y-Z stage for ease of alignment. All the spectra presented here have been recorded at a nominal resolution of 4 cm-’ and have not been corrected for the instrument response function. The laser powers used ranged from 300 to 600 mW at the sample.
3. RESULTS
AND DISCUSSION
As an illustration of the performance of the microscope, Fig. 3(a) shows the FT Raman spectrum of anthracene recorded from a solid pellet mounted under the microscope objective. The scan time was 1.5 min and this shows that spectra with good
E
6 _
P state
u .4
-
<
s
filter E s state
Y J”1
‘2-
0
s
10
20
30
Angle of tilt
40 (degrees)
Fig. 2. Reflectivity curves for Nd: YAG Rayleigh rejection filter.
A microscope for IT Raman spectroscopy
155
signal-to-noise (S/N) ratios can readily be obtained for such highly scattering samples. However, this illustration is somewhat misleading as an example. If the primary image is stopped down such that light from a spot corresponding to 25 microns at the sample is allowed to reach the spectrometer, the S/N is greatly reduced. This is shown in Fig. 3(b). In this case the data collection time was 7.5 min. For completeness Fig. 3(c) shows the FT Raman spectrum of a single flake of anthracene, approximately 50 x 50 microns in size, again using 7.5 min collection time. The same laser power (550 mW) was used for all Anthracene
(a)
3200
2500
2000
1500
Raman
Shift
1000
500
0
(cm-‘)
Fig. 3. IT Raman spectrum of anthracene using (a) micro optics, full aperture, (b) aperture closed to 25 microns and (c) single flake. The spectra have not been corrected for instrument response and the data are only valid between the limits 400 and 3100 cm-‘.
three examples. The significant decrease in S/N ratio between these results is primarily due to the high degree of multiple scattering which takes place within the anthracene pellet. As noted in our earlier report, the difference in S/N between the macro and the micro optics was approximately 6.5 : 1. For comparison with the results described in Ref. [2], Fig. 4 shows the FT Raman spectrum of a single fibre of Kevlar. This was taken using 7.5 min collection time and with approximately 350 mW at the sample position. MESSERSCHMIDT and CHASE [2] have noted that at power levels above 125 mW their sample of Kevlar decomposed. This suggests that perhaps the laser spot size in our configuration is somewhat bigger than the 12 micron-diameter fibre. It is perhaps worth noting here that the “anomalous background”, that has been associated with FT Raman data and that is clearly evident in Fig. 5 of Ref. [2], is not apparent in our results. It is possible to collect data on a microscopic scale even from samples that have very weak Raman cross-sections. As an illustration of this, Fig. 5(a) shows the FT Raman spectrum of a highly-coloured polyurethane elastomer recorded using our micro optics system. This spectrum was obtained with a laser power of approximately 550 mW, a spot size of approximately 50 microns in diameter and a data collection time of 7.5 min. For comparison Fig. S(b) shows a spectrum of the same material recorded through the macro system with a considerably longer data collection time of 37.5 min.
F. J. BERGIN
156
Single
Kevlar
fibre
c 3200
2500
2000
1500
Raman
Shift
1000
500
0
(cm-l)
Fig. 4. Ff Raman spectrum of a 12 micron diameter Kevlar tibre. The spectrum has not been response and the data are only valid between the limits 400 and 3100 cm-‘.
corrected for instrument
Polyurethane
elastomer
Roman
(cm-‘)
:a)
Shift
Fig. 5. Ff Raman spectrum of a polyurethane elastomer recorded (a) using the microscope and (b) using conventional macro optics. The spectra have not been corrected for instrument response and the data are only valid between the limits 400 and 3100 cm-‘.
A microscope for lT Raman spectroscopy
157
For some strongly scattering materials, it is possible to collect data from relatively small single crystals or aggregates. To illustrate this, Fig. 6(a) shows the FI Raman spectrum of a sample of an antidepressant drug (flunitrazepam), recorded from a sample approximately 50 x 100 microns in size; scan time was 15 min and laser power 600 mW. * Figure 6(b) shows the spectrum of the same material recorded using the macro system. Flunitrazepam
2000
1600 Raman
1200 Shift
800
(cm-‘)
Fig. 6. m Raman spectrum of an antidepressant drug (a) using the microscope and (b) using conventional macro optics. The spectra have not been corrected for instrument response and the data are only valid between the limits 400 and 3100 cm-‘.
Finally Fig. 7 shows the spectrum of an aggregate of an explosive material, triaminotrinitrobenzene. * The total diameter of the aggregate was approximately 50 microns, again with a laser power of 600 mW and a data collection time of 15 min. In attempting to address the reason for the somewhat reduced performance of the microscope, compared to that of the macro system, there are several points that need to be considered. One such aspect is the throughput of each part of the system. The throughput or Ctendu of a system is defined as
where A is the area and B is the solid angle at the point in the system where this quantity is the smallest. For the Perkin-Elmer system the Ctendu is about 1.4 mm* sterads. For the microscope, again using a numerical aperture (N.A.) of 0.65 and a spot size of 50 microns, the throughput is approximately 2.9 x lo-’ mm2 sterads-not 1 x lo-* as suggested in Ref. [3]. Thus the microscope may determine the limiting throughput of the system. The microscope, however, only becomes a limiting factor in the case where one is limited by the maximum power density the sample can tolerate. Although the image area * Editor’s note: a more extensive coverage of drug and explosive spectra are included in this Special Edition: see Refs [6,7].
F. J.
1.58
BERLIN
in the microscope system is considerably lower than that of the macro system the power density is much higher. Therefore, with the same power at the sample in both the microscopic and macroscopic system, the S/N ratio should be the same in both cases. Since in comparing the S/N ratio in our macro and micro systems we have not been limited by power density, there must be other loss factors. One such loss factor arises from the problem of matching the instrument and microscopefnumbers. The Perkin-Elmer instrument accepts light between f/S and f/10, whereas the microscope, with its numerical aperture (N.A.) of 0.65, has an f number of 0.77. However, this mismatch off numbers is compensated for, in part, by a suitable choice of coupling optics. An alternative means of comparison is to look at the solid angle of collection. A comparison of our micro and macro systems reveals a reduction, by approximately a factor of 4, as a result of using the microscope optics instead of the macro optics. (The x40 objective lens of the microscope has a numerical aperture of 0.65 which results in a solid angle of collection of 1.51 sterads. In the macro system the sample lies within the cone of collection of the ellipsoid, giving a solid angle of collection of 2~ sterads.) Although the scattered Raman signal is not equally distributed into the entire solid angle of 2n, the reduced collection efficiency of the microscope will certainly account for some of its reduced performance. Another loss factor may be the difficulty of steering a narrow, almost parallel beam through the spectrometer. In the Perkin-Elmer system, there are two laser pick-off reflectors and a central l/8 II waveplate which are needed for the internal reference HeNe laser. These do not present a problem with the relatively large beam areas used in the conventional operation of the instrument, but they will obscure some of the light
I
2000
1800
1600
1400
1200
1000
I
800
I
600
RAMANSHIFT(cm-')
Fig. 7. FT Raman spectrum of an explosive, triaminotrinitrobenzene, recorded from a sample approximately 50 microns in diameter. The spectrum has not been corrected for instrument response and the data are only valid between the limits 400 and 3100 cm-‘.
450
A microscope for Ff Raman spectroscopy
1.59
when the microscope is used. To overcome this, the beam was actually skewed through the side of the Jacquinot stop to produce the optimal signal. This obstruction of the beam coupled with the smaller solid angle of collection and the losses due to the additional optical components in the micro system (beamsplitter, right angle prism and coupling lens) account for most of the loss in the S/N ratio when comparing the micro and macro systems. The anti-reflection properties of the objective lens have not been optimized for the near infrared. Further improvement may be obtained in this area.
4. CONCLUSIONS Despite the limiting aperture, or Ctendu, of a microscopic system, some modifications to a simple, refractive, metallurgical microscope have allowed us to record FT Raman data from a range of materials with a spatial resolution which is of the order of 20-50 microns. Since, in general, a microscopic objective is a compound lens system, some further improvements could be achieved by using specially designed refractive optics for the near infrared. As noted earlier, the image sizes in such a microscopic experiment are considerably smaller than those used in conventional FT Raman experiments and hence a reduction in the size of the detector could also prove useful.
REFERENCES [l] T. Hirschfeld and D. B. Chase, Appl. Spectrosc. 40, 133 (1986). [2] R. G. Messerschmidt and D. B. Chase, Appl. Spectrosc. 43, 11 (1989). [3] F. J. Bergin and H. F. Shurvell, Appl. Spccrrosc. 43, 516 (1989). [4] D. B. Chase, J. Am. Chem. Sot. 108, 7485 (1986). [5] C. G. Zimba, V. M. Hallmark, J. D. Swalen and J. D. Rabolt, Appl. Specrrosc. 41, 721 (1987). [6] M. C. Davies, J. S. Binns, C. D. Melia and D. Bourgeois, Spectrochim. Acta 46A,277 (1990). [7j C. M. Hodges and J. Akhavan, Specrrochim. Acta 46A, 303 (1990).
.
1990 0
A microscope for Fourier Transform
0.564-8.539190 53.00+ 0.00 199u Pergamon Press plc
Raman spectroscopy
F. J. BERGIN Shell Research Ltd., Thornton Research Centre, P.O. Box 1, Chester CH13SH, U.K. (Received 1 September 1989; accepted 6 November 1989) Abstract-Vibrational microscopy, and in particular Raman microscopy, has been used for many years as a powerful spectroscopic tool. The recent developments in FT Raman spectroscopy have opened up new areas in which Raman spectroscopy can play a useful role. In light of this, a simple Raman microscope, based on refractive optics, has been coupled to an FT Raman system. This allows data to be collected from a wide range of materials including single fibres and crystals. The reduced performance of the microscope system, when compared to a conventional FT Raman system, is rationalized in terms of the losses in the optical components used.
1. INTRODUCTION
the first report of FT Raman spectroscopy using near infrared excitation [l], there has been a steady growth of interest in both the instrumentational aspects and the applications of the technique. However, little attention has been devoted to the development of FT Raman microscopy. MESSERSCHMIDT and CHASE [2] have examined the possibility of using a Spectra-Tech IR plan microscope and have achieved results that “argue for the development and commercialization of a combined FT IR/FT Raman microscopic instrument.” In an earlier report [3] we outlined some preliminary results obtained using a conventional glass-optics microscope. This paper describes some further developments we have undertaken in this area. The relative strengths and weaknesses of FT Raman spectroscopy have been dealt with elsewhere in this issue. The additional use of a microscope would permit direct analysis of any given area of interest in a sample with a spatial resolution of the order of several microns. These advantages provide sufficient incentive to further investigate this aspect of the technique. SINCE
2. EXPERIMENTAL The experimental aspects of FT Raman spectroscopy, using “macro optics”, have been dealt with in several publications [l, 3-51. The salient features of our macro system are as follows. The spectrometer used was a Perkin-Elmer model 1760 near infrared instrument equipped with a liquid nitrogen-cooled germanium detector (NEP = 2 x lo-l4 WHz-I’*). The Nd: YAG laser was supplied by Spectron Lasers and the Rayleigh rejection achieved by using two dielectric-coated filters from Glen Creston. The combination of detector and filters resulted in Raman data being collected in the range 400-31OOcm-‘. The Raman-scattered signal was collected in a 180” backscattered configuration using an ellipsoid mirror supplied by Perkin-Elmer. A more detailed description can be found in Ref. [3]. For our microscopy studies the 180” collection ellipsoid was replaced by a Nikon Optiphot metallurgical microscope.. In the original configuration of the microscope system the laser was brought to the sample via a 12.5% reflecting beamsplitter and a x40 lens, as shown schematically in Fig. l(a). This combination gave an overall throughput (laser head to sample) of approximately 4%. In an effort to improve the throughput, the microscope was modified by replacing the beamsplitter with a dielectric-coated Nd : YAG reflection filter. The reflection characteristics of this filter are illustrated in Fig. 2. By suitable adaptation of the beam-steering optics, this filter was incorporated into the microscope as shown in Fig. l(b). The overall throughput was thus increased to 30-40% allowing powers of up to 1 W to be focused onto the sample. Whilst this filter arrangement served the dual purpose of increasing the throughput and simultaneously incorporating some Rayleigh rejection, it suffers from one drawback. Although the filter is transparent to visible light, allowing the sample to be viewed through the microscope, the 153
154
F. J. BERGIN T.V.
(b)
T.V.
Spatial
7_
aperture ------_____ To spectrometer
Tospectrometer
Lens I
0
&Prism
/ / I
0
Filter
---IT
Mcroscope objective
Microscope
objective
+
A
-i-
Sample
Sample
Fig. 1. (a) Schematic illustration of original configuration
of FT Raman microscope. (b) Modified FT Raman microscope. The filter used was an ND : YAG Rayleigh rejection filter.
exact position of the focused laser spot could not be identified directly. However, its position could be located by using the HeNe laser which was set to run coaxial to the YAG laser beam. With the original beamsplitter configuration the diameter of the laser spot on a glass slide was about 5-10 microns. There should be no reason to assume that the spot size should change as a result of using the filter configuration. It is worth noting here the approximate dimensions of the optical imaging. A focused spot size of approximately 5-10 microns and a magnification of 40 produces a primary image that is, at most, only 400 microns in diameter. This first primary image, in our system, is formed approximately 200 mm from the objective lens-in essence, a very narrow, almost parallel beam is produced by the microscope objective. This type of optical system is far from ideal for our current FT spectrometer. A single lens was positioned approximately at the primary image formed by the microscope objective, which in turn produced a further image at the Jacquinot stop of the spectrometer. This lens was mounted on an X-Y-Z stage for ease of alignment. All the spectra presented here have been recorded at a nominal resolution of 4 cm-’ and have not been corrected for the instrument response function. The laser powers used ranged from 300 to 600 mW at the sample.
3. RESULTS
AND DISCUSSION
As an illustration of the performance of the microscope, Fig. 3(a) shows the FT Raman spectrum of anthracene recorded from a solid pellet mounted under the microscope objective. The scan time was 1.5 min and this shows that spectra with good
E
6 _
P state
u .4
-
<
s
filter E s state
Y J”1
‘2-
0
s
10
20
30
Angle of tilt
40 (degrees)
Fig. 2. Reflectivity curves for Nd: YAG Rayleigh rejection filter.
A microscope for IT Raman spectroscopy
155
signal-to-noise (S/N) ratios can readily be obtained for such highly scattering samples. However, this illustration is somewhat misleading as an example. If the primary image is stopped down such that light from a spot corresponding to 25 microns at the sample is allowed to reach the spectrometer, the S/N is greatly reduced. This is shown in Fig. 3(b). In this case the data collection time was 7.5 min. For completeness Fig. 3(c) shows the FT Raman spectrum of a single flake of anthracene, approximately 50 x 50 microns in size, again using 7.5 min collection time. The same laser power (550 mW) was used for all Anthracene
(a)
3200
2500
2000
1500
Raman
Shift
1000
500
0
(cm-‘)
Fig. 3. IT Raman spectrum of anthracene using (a) micro optics, full aperture, (b) aperture closed to 25 microns and (c) single flake. The spectra have not been corrected for instrument response and the data are only valid between the limits 400 and 3100 cm-‘.
three examples. The significant decrease in S/N ratio between these results is primarily due to the high degree of multiple scattering which takes place within the anthracene pellet. As noted in our earlier report, the difference in S/N between the macro and the micro optics was approximately 6.5 : 1. For comparison with the results described in Ref. [2], Fig. 4 shows the FT Raman spectrum of a single fibre of Kevlar. This was taken using 7.5 min collection time and with approximately 350 mW at the sample position. MESSERSCHMIDT and CHASE [2] have noted that at power levels above 125 mW their sample of Kevlar decomposed. This suggests that perhaps the laser spot size in our configuration is somewhat bigger than the 12 micron-diameter fibre. It is perhaps worth noting here that the “anomalous background”, that has been associated with FT Raman data and that is clearly evident in Fig. 5 of Ref. [2], is not apparent in our results. It is possible to collect data on a microscopic scale even from samples that have very weak Raman cross-sections. As an illustration of this, Fig. 5(a) shows the FT Raman spectrum of a highly-coloured polyurethane elastomer recorded using our micro optics system. This spectrum was obtained with a laser power of approximately 550 mW, a spot size of approximately 50 microns in diameter and a data collection time of 7.5 min. For comparison Fig. S(b) shows a spectrum of the same material recorded through the macro system with a considerably longer data collection time of 37.5 min.
F. J. BERGIN
156
Single
Kevlar
fibre
c 3200
2500
2000
1500
Raman
Shift
1000
500
0
(cm-l)
Fig. 4. Ff Raman spectrum of a 12 micron diameter Kevlar tibre. The spectrum has not been response and the data are only valid between the limits 400 and 3100 cm-‘.
corrected for instrument
Polyurethane
elastomer
Roman
(cm-‘)
:a)
Shift
Fig. 5. Ff Raman spectrum of a polyurethane elastomer recorded (a) using the microscope and (b) using conventional macro optics. The spectra have not been corrected for instrument response and the data are only valid between the limits 400 and 3100 cm-‘.
A microscope for lT Raman spectroscopy
157
For some strongly scattering materials, it is possible to collect data from relatively small single crystals or aggregates. To illustrate this, Fig. 6(a) shows the FI Raman spectrum of a sample of an antidepressant drug (flunitrazepam), recorded from a sample approximately 50 x 100 microns in size; scan time was 15 min and laser power 600 mW. * Figure 6(b) shows the spectrum of the same material recorded using the macro system. Flunitrazepam
2000
1600 Raman
1200 Shift
800
(cm-‘)
Fig. 6. m Raman spectrum of an antidepressant drug (a) using the microscope and (b) using conventional macro optics. The spectra have not been corrected for instrument response and the data are only valid between the limits 400 and 3100 cm-‘.
Finally Fig. 7 shows the spectrum of an aggregate of an explosive material, triaminotrinitrobenzene. * The total diameter of the aggregate was approximately 50 microns, again with a laser power of 600 mW and a data collection time of 15 min. In attempting to address the reason for the somewhat reduced performance of the microscope, compared to that of the macro system, there are several points that need to be considered. One such aspect is the throughput of each part of the system. The throughput or Ctendu of a system is defined as
where A is the area and B is the solid angle at the point in the system where this quantity is the smallest. For the Perkin-Elmer system the Ctendu is about 1.4 mm* sterads. For the microscope, again using a numerical aperture (N.A.) of 0.65 and a spot size of 50 microns, the throughput is approximately 2.9 x lo-’ mm2 sterads-not 1 x lo-* as suggested in Ref. [3]. Thus the microscope may determine the limiting throughput of the system. The microscope, however, only becomes a limiting factor in the case where one is limited by the maximum power density the sample can tolerate. Although the image area * Editor’s note: a more extensive coverage of drug and explosive spectra are included in this Special Edition: see Refs [6,7].
F. J.
1.58
BERLIN
in the microscope system is considerably lower than that of the macro system the power density is much higher. Therefore, with the same power at the sample in both the microscopic and macroscopic system, the S/N ratio should be the same in both cases. Since in comparing the S/N ratio in our macro and micro systems we have not been limited by power density, there must be other loss factors. One such loss factor arises from the problem of matching the instrument and microscopefnumbers. The Perkin-Elmer instrument accepts light between f/S and f/10, whereas the microscope, with its numerical aperture (N.A.) of 0.65, has an f number of 0.77. However, this mismatch off numbers is compensated for, in part, by a suitable choice of coupling optics. An alternative means of comparison is to look at the solid angle of collection. A comparison of our micro and macro systems reveals a reduction, by approximately a factor of 4, as a result of using the microscope optics instead of the macro optics. (The x40 objective lens of the microscope has a numerical aperture of 0.65 which results in a solid angle of collection of 1.51 sterads. In the macro system the sample lies within the cone of collection of the ellipsoid, giving a solid angle of collection of 2~ sterads.) Although the scattered Raman signal is not equally distributed into the entire solid angle of 2n, the reduced collection efficiency of the microscope will certainly account for some of its reduced performance. Another loss factor may be the difficulty of steering a narrow, almost parallel beam through the spectrometer. In the Perkin-Elmer system, there are two laser pick-off reflectors and a central l/8 II waveplate which are needed for the internal reference HeNe laser. These do not present a problem with the relatively large beam areas used in the conventional operation of the instrument, but they will obscure some of the light
I
2000
1800
1600
1400
1200
1000
I
800
I
600
RAMANSHIFT(cm-')
Fig. 7. FT Raman spectrum of an explosive, triaminotrinitrobenzene, recorded from a sample approximately 50 microns in diameter. The spectrum has not been corrected for instrument response and the data are only valid between the limits 400 and 3100 cm-‘.
450
A microscope for Ff Raman spectroscopy
1.59
when the microscope is used. To overcome this, the beam was actually skewed through the side of the Jacquinot stop to produce the optimal signal. This obstruction of the beam coupled with the smaller solid angle of collection and the losses due to the additional optical components in the micro system (beamsplitter, right angle prism and coupling lens) account for most of the loss in the S/N ratio when comparing the micro and macro systems. The anti-reflection properties of the objective lens have not been optimized for the near infrared. Further improvement may be obtained in this area.
4. CONCLUSIONS Despite the limiting aperture, or Ctendu, of a microscopic system, some modifications to a simple, refractive, metallurgical microscope have allowed us to record FT Raman data from a range of materials with a spatial resolution which is of the order of 20-50 microns. Since, in general, a microscopic objective is a compound lens system, some further improvements could be achieved by using specially designed refractive optics for the near infrared. As noted earlier, the image sizes in such a microscopic experiment are considerably smaller than those used in conventional FT Raman experiments and hence a reduction in the size of the detector could also prove useful.
REFERENCES [l] T. Hirschfeld and D. B. Chase, Appl. Spectrosc. 40, 133 (1986). [2] R. G. Messerschmidt and D. B. Chase, Appl. Spectrosc. 43, 11 (1989). [3] F. J. Bergin and H. F. Shurvell, Appl. Spccrrosc. 43, 516 (1989). [4] D. B. Chase, J. Am. Chem. Sot. 108, 7485 (1986). [5] C. G. Zimba, V. M. Hallmark, J. D. Swalen and J. D. Rabolt, Appl. Specrrosc. 41, 721 (1987). [6] M. C. Davies, J. S. Binns, C. D. Melia and D. Bourgeois, Spectrochim. Acta 46A,277 (1990). [7j C. M. Hodges and J. Akhavan, Specrrochim. Acta 46A, 303 (1990).
.