Use of a Mono-Fiber Optrode in Remote and in Situ Measurements by the Raman/Laser/Fiber Optics (RLFO) Method

N G U Y E N Q U A N G HUY, M I C H E L J O U A N , and N G U Y E N QUY DAO* Laboratoire de Chimie et Physico-Chimie Molgculaires, URA D0441 CNRS, Ecole Centrale Paris, 92295 Chdtenay-Malabry Cedex, France

A new versatile mono-fiber optrode based on the combined use of a dual- fiber optrode and a short mono-fiber is proposed. The quantitative an- alytical possibility of this optrode configuration is tested on benzene/ heptane mixtures with the use of a PLS procedure, and good results are obtained. The new all-purpose mono-fiber optrode is extremely simple and user friendly and can be used for in situ and in vivo detection.

Index Headings: Mono-fiber optrode; Raman/laser/fiber optics; Quan- titative chemical analysis; PLS procedure.

INTRODUCTION

In recent years, the coupling of optical fibers to an optical spectrometer has been increasingly investigated due to its possibilities for remote and in situ multi-mea- surements? -s In terms of Raman spectroscopy, some Ra- man/laser/fiber optics (RLFO) systems are now proposed by manufacturers, 7,s marking a new step in development. Because of its selectivity and nondestructivity, this method offers tremendous potential for applications in different domains of human life including industry, med- icine, and the environment. Among its industrial appli- cations, process on-line control, the monitoring of radio- active or hostile sites, etc., can be found. In medicine, the potential applications of this method seem to be even more spectacular. By inserting flexible optical fibers through natural openings or small incisions in the human body, one can perform immediate in vivo chemical anal- yses. With the delivery of laser energy inside the body for surgery and therapy, this method may soon combine diagnosis and treatment. Many studies have been made in these directions. 9-17

The simplest and most successful configuration to be found for the RLFO method is a mono-fiber optrode. This geometry has the advantage of providing perfect overlap between the excitation and collection fields at the end of fiber, and thus neither adjustments nor ad- ditional components are necessary. Unfortunately, the exploitation of this optrode is not straightforward. The principal drawback is the presence of the very intense Raman spectrum of the fiber itself. This problem makes detection with mono-fiber optrodes possible only for very short fiber lengths. In approaches to overcome this dif- ficulty, multiple-fiber optrodes have been proposed. In these optrode configurations, the excitation and collec- tion beams are transported by different fibers. This ge- ometry allows the participation of the fiber spectrum in the observed RLFO spectra to be significantly reduced.

Received 5 April 1993; revision received 13 July 1993. * Author to whom correspondence should be sent.

Consequently, the length of the fiber used in remote analyses can be increased. However, this benefit does not mean that the multiple-fiber optrodes have no obstacles. The optical collection efficiency is markedly lower than that of mono-fiber optrodes, and the adjustment is more difficult. The Raman spectrum of optical fibers in these multiple-fiber optrodes now reaches the spectrometer because of reflections at the measurement sites. As a result of this factor, the contribution of the fiber Raman signal in each RLFO spectrum varies enormously and uncontrollably from one spectrum to another. This lim- itation may cause problems for the quantitative appli- cations of this method. Different optrode configurations using two or more fibers have been developed or sug- gested. Various studies, such as the optimization of ad- jacent-fiber optrode efficiency, is the utilization of a dou- ble-core fiber 9,19 or of a bundle of fibers where one excitation fiber is surrounded by six or more collection fibers, 17,2°,21 the introduction of GRIN lenses 22,23 and op- tical filters 24 to increase the optical collection efficiency and to remove the fiber spectrum, etc., have been carried out. Thanks to these developments, several important improvements have been obtained, and analyses can now be performed with long fibers. Nevertheless, the major disadvantage of these optrodes is that their size is sig- nificantly increased.

It is clear that multiple-fiber optrodes are useful for remote analysis while mono-fiber optrodes are more suit- able for in situ and in vivo applications.

In a preceding paper, 25 we announced the possibility of using mono-fiber optrodes for Raman measurements. This paper presents quantitative analytical results of such a system.

METHOD

The quantitative analysis was carried out with the use of the partial least-squares (PLS) regression. This meth- od was chosen for its performances and rapidityY G-33 Moreover, it has been shown that this method is able to determine both analyte concentrations and sample tem- perature at the same time, 34 which is useful for various applications. In addition, the capability of this method to determine the chemical composition in mixtures of very similar chemical formula has also been demonstrat- ed. 35,36 Our version of PLS was developed in order to take into account the contribution of the fiber Raman signal in each RLFO spectrum. 34 In this procedure, the fiber signal participation in each RLFO spectrum of the cal- ibration set is estimated and is entered into the model

Volume 47, Number 12, 1993 0003-7028/93/4712-201352.00/0 APPLIED SPECTROSCOPY 2013 © 1993 Society for Applied Spectroscopy

Intensity

121 ~ Mean intensity for a ~o ~ - - ~ . ~ 7 ~ "

30 ~ - - ~ x ~ / 7 ~ _ ~ -

Spectrum b

01 D Wavenamber (cm -1 )

Before mean intensity shifting

Intensity

Wavenumber (cm -1 )

After mean intensity shifting

FIG. 1. Mean intensity shifting.

calculations as another component. The so-called "fiber explicit modeling" makes it possible to use the "raw" RLFO spectra directly without subtraction of the fiber spectrum. This capability is very important in those cases where multiple-fiber optrodes are used and optical filters are not applicable because of optrode dimension require- ments. Since the PLS method is described in detail else- where, 26-2s it is not presented here.

Spectrum Treatments. Before any calculations are ini- tiated, it is usually necessary to pretreat the spectral information to remove the irrelevant or useless data. Two pretreatment techniques used in this work are pre- sented below.

Mean-Centering Procedure. In this procedure, the mean calibration spectrum should be subtracted from all spectra before their delivery for calibration and anal- ysis. Using this technique, one can make the spectrum-

Laser ~ Excitation fiber

Raman spectrometer

Collection fiber

Holographic beam splitter

Mono-fiber

MiNor DILOR Super Head

Sample

FIa. 2. Instrumental setup for mono-fiber op t rodeRLFOspec t rom- etry.

1.600 10 4

1.400 104

1.200 104

1.000 104

8000

6000

4000 .~ 2000

O5od

FIG. 3.

' ' ' ' 1 ' ' ' ' 1 . . . . I ' ' ' ' l ' ' ' ' l ' ' ' ' l ' ' ' '

region used for analysis

1000 1500 2000 2500 3000 3500 4000

wavenumber (era" ~)

RLFO spectrum recorded for 12% benzene/heptane sample.

to-spectrum variation predominant, as should be the case for the spectral information concerning the concentra- tion of the components under investigation. In particu- lar, this technique was found to be useful for the RLFO applications in reducing the influence of the fiber signal? 4

Mean Intensity Shifting Procedure. According to this procedure, the mean intensity of the useful part of the spectrum is subtracted from each intensity. This treat- ment is intended to solve the problem of intensity shift- ing, as illustrated in Fig. 1. It may also be useful in reducing the fiber spectrum effect in the RLFO quan- titative applications, as will be seen later.

EXPERIMENTAL

Apparatus. The experimental setting for mono-fiber optrode RLFO spectrometry is shown in Fig. 2. It couples the new commercial DILOR RLFO 260V system with a short fiber used as a mono-fiber optrode. The DILOR RLFO 260V system is an inexpensive instrument spe- cially constructed for remote and on-line applications. It is equipped with the DILOR Super Head, which is a retro-Raman dual-fiber optrode, s This head is linked to the laser and Raman spectrometer by two 10-m-long, 100-#m-core optical fibers. Longer lengths up to 100 m are also available from the catalog or on customer de- mand. The high performance of lenses, holographic beamsplitter, and optical filters ensures good collection efficiency and effective rejection of fiber Raman emission. At the end of the Super Head, the laser beam is focused into the other fiber. The mono-fiber optrode used is a 50-cm-long PCS 600-#m-core fiber with one end fixed on the Super Head and the other end dipped directly into the liquid sample contained in a small flask. The Raman spectrometer is a multi-channel type which detects the whole Raman spectrum (from 150 to 3500 cm -1) in one single scan by means of a 700-intensified photodiode array (the resolution is about 30 cm-0. The 514.5-nm radiation excitation is delivered by an air-cooled argon- ion laser, with 25 mW on the sample.

Quantitative Analytical Procedure. Quantitative anal- ysis was performed on benzene/heptane solutions. This mixture was chosen for illustration since quantitative analysis of hydrocarbon fuel mixtures is important for the petroleum industry. The molar fractions of benzene ranged from 1 to 13 %, calculated from the mass ratios.

2014 Volume 47, Number 12, 1993

1.300 104

1.o38 lO 4 i A ~ benzene

.~ 5125 (b) (c)

2500 (a) 800 850 900 950 1000 1050 1100 1150 1200

wavenumber (cm" x)

Flc. 4. RLFO spectra recorded for (a) 12% benzene, (b) 7.5% ben- zene, and (c) 2% benzene in heptane. Region used: 800-1200 cm-L The pseudo-noise observed is the photodiode array response.

600

4 0 0

200

0 .~. 0

-200

-400800' 850 900 950 1 0 0 0 1050 1100 1 1 5 0 1200

wavenumber (cm 1)

FIG. 6. RLFO spect rum of 12 % benzene/heptane, obtained after mean intensi ty shifting and mean centering.

Forty samples were prepared and recorded with an in- tegration time of one second and an accumulation num- ber of 240. These 40 samples were divided into two sets; 21 samples were gathered in the calibration set to build the model, and the 19 remaining samples formed the verification set to check the model validity. As it is not necessary to use the entire spectrum, but only the spec- tral ranges where characteristic bands exist for both com- pounds, a 400-cm-1 broad spectral region (800-1200 cm -1, or 58 points for each spectrum) was used, to reduce the calculation time cost. The spectra were mean centered and also mean intensity shifted before their introduction into the calculations. The precision of the results ob- tained can be evaluated by the Standard Error of Pre- diction:

S E P = ~ ~ (Cpred -- Ctrue) 2 .

An example of an RLFO spectrum recorded for a 12 % benzene sample with the working window is shown in Fig. 3.

RESULTS AND DISCUSSION

Figure 4 shows some of our RLFO spectra in the region used for analysis. In this figure, one can note a difference

7000 . . . . i . . . . i . . . . D . . . . I . . . . J . . . . i . . . . 4 . . . .

4750

"~ 2500

.2

250

- 2 0 0 0 . . . . t . . . . L . . . . L . . . . ~ . . . . n . . . . i . . . . i . . . .

---800 850 900 950 1 0 0 0 1 0 5 0 1 1 0 0 1 1 5 0 1200

wavenumber (¢m 1)

Fm. 5. The same spectra as in Fig. 4, after mean intensi ty shifting.

in the contribution of the fiber Raman signal in each of our RLFO spectra. However, it should be noted that, with the use of this mono-fiber optrode, the variations in the contribution of the fiber Raman signal in RLFO spectra should not be very great, compared with those of a multiple-fiber optrode. In this optrode configuration, the Raman signal of the fiber reaches the spectrometer mainly by retro-Raman diffusion, and the reflection in the measurement site has only a slight effect. This slight effect of the fiber spectrum may be seen as an intensity shift from one to another RLFO spectrum, and the mean intensity shifting should be exploited here. The RLFO spectra obtained after such a procedure are shown in the Fig. 5. In these spectra, no significant difference in con- tribution of the fiber Raman signal could be observed. Consequently, the mean-centering treatment enables the fiber spectrum effect to be reduced (Fig. 6). The explicit fiber modeling procedure is not necessary here.

Quantitative analysis has been carried out in two cases. The first one uses only mean centering; the other also includes the mean intensity shifting. The results ob- tained are compared in Table I. A small difference in these results shows that the fiber spectrum has only a slight effect on the quantitative analysis. Furthermore, this slight effect may be effectively eliminated by using the mean intensity shifting. The predicted values ob- tained for this case are plotted against the real values in Fig. 7. The regression equation calculated from these values is y = 0.05 + 0.998 x, which is very close to the ideal line y = x.

The good results obtained prove the reliability of this mono-fiber optrode configuration. They also show that the use of longer fibers is possible. Moreover, the focal point diameter of the Super Head is about 20 ~m and allows the injection of the laser beam into a 100-#m-core fiber. As the fiber spectrum depends only on the fiber

TABLE I. Results obtained for benzene/heptane mixtures, with and without mean intensity shifting.

Mean Correla- intensi ty tion coef- Regression

Case shifting SEP ficient equation

1 no 0.19 0.997 y = -0 .05 _+ 1.O07x 2 yes 0.13 0.999 y = 0.05 + 0.998x

APPLIED SPECTROSCOPY 2015

14

12

O 10

8

6

"~ 4

2

Ideal line

O 0 2 4 6 8 10 12 14

Real c o n c e n t r a t i o n s

FIG. 7. Prediction results obtained for verification set. Pretreatments used: mean intensity shifting, mean centering.

length, but not on the fiber diameter, the use of a 100- ~m-core fiber instead of the present 600-ttm one should enhance the signal-to-noise ratio by an order of magni- tude. The applications of this mono-fiber optrode to bio- medical analysis using 1-m-long, 100-ttm-core fiber are at present being investigated, this time with a near-in- frared excitation.

CONCLUSION

The possibilities of the new mono-fiber optrode for remote and in situ quantitative analysis have been dem- onstrated in the present study for a case of chemical analysis. With its simplicity and small dimensions, the mono-fiber optrode can easily be introduced into the human body as well as other sites of difficult access. Therefore, it is very useful not only for in situ but also for in vivo detection. In such applications, the Super Head can be kept away from the spectrometer, and thus remote analyses can be obtained. The same fiber can also be used for both the detection and the therapeutic laser beams. The coupling of this optrode with an inexpensive industrial and biomedical Raman spectrometer makes it possible to look forward to successful applications of the RLFO method in industry and medicine.

ACKNOWLEDGMENTS

The authors wish to thank Dr. E. Da Silva, Director of DILOR S.A., for helpful discussions and for the loan of an INDURAM RLFO Raman spectrometer.

1. R. L. McCreery, M. Fleischmann, and P. Hendra, Anal. Chem. 55, 146 (1983).

2. Nguyen Quy Dao, M. Jouan, P. Plaza, and J. Robieux, Brevet Fran~ais 84.15047 (1984).

3. P. J. Hendra and G. Ellis, J. Raman Spectrosc. 19, 413 (1988). 4. D. D. Archibald, L. T. Lin, and D. E. Honigs, Appl. Spectrosc. 42,

1558 (1988). 5. J. P. Alarie, D. L. Stokes, W. S. Sutherland, A. C. Edwards, and

T. Vo-Dinh, Appl. Spectrosc. 46, 1608 (1992). 6. Nguyen Quy Dao, M. Jouan, and P. Plaza, Les Cahiers de Spectra

2000 168, 8 (1992). 7. Ph. Roussel and Ph. Cleon, Analusis 20, 58 (1992). 8. DILOR RLFO 260V spectrometer technical data sheet (DILOR,

Lille, France, 1993). 9. H. Fevrier, H. Saisse, P. Plaza, and Nguyen Quy Dao, Brevet Fran-

~ais 85.09725 (1985). 10. Nguyen Quy Dao, M. Prod'homme, P. Plaza, and M. Joyeux, C. R.

Acad. Sci. Paris 302 II, 313 (1986). 11. Nguyen Quy Dao, P. Plaza, and M. Joyeux, Analusis 14, 334 (1986). 12. S. Hameau, M. Jouan, and Nguyen Quy Dao, Analusis 16, 173

(1988). 13. C. Gomy, M. Jouan, and Nguyen Quy Dao, Anal. Chim. Acta 215,

211 (1988). 14. D. Phat, P. Plaza, Nguyen Quy Dao, P. N. Vuong, and P. G. Steg,

in Proceedings of the 12th International Conference on Raman Spectroscopy (John Wiley & Sons, Chichester, 1990), p. 704.

15. D. Phat, P. Plaza, P. N. Vuong, F. Cheilan, and Nguyen Quy Dao, in Proceedings of the 64th Scientific Sessions of the American Heart Association, supplement to Circulation 84(4), II-725 (1991).

16. D. Phat, P. N. Vuong, P. Plaza, F. Cheilan, and Nguyen Quy Dao, in Proceedings of the Future Trends in Biochemical Applications of Lasers, SPIE Vol. 1525 (SPIE, Bellingham, Washington, 1991), p. 196.

17. T. D. Nguyen Hong, D. Phat, P. Plaza, M. Daudon, and Nguyen Quy Dao, Clin. Chem. 38, 292 (1992).

18. P. Plaza, Nguyen Quy Dao, M. Jouan, H. Fevrier, and H. Saisse, Appl. Opt. 25, 3448 (1986).

19. P. Plaza, Nguyen Quy Dao, M. Jouan, H. Fevrier, and H. Saisse, Analusis 15, 504 (1987).

20. K. P. J. Williams, J. Raman Spectrosc. 21, 147 (1990). 21. C. G. Zimba and J. F. Rabolt, Appl. Spectrosc. 45, 162 (1991). 22. M. Jiaying and L. Zhong, Appl. Spectrosc. 45, 1302 (1991). 23. N. V. Tien, H. Nguyen Quang, M. Jouan, Nguyen Quy Dao, Opto

92, 554 (1992). 24. M. L. Myrick and S. M. Angel, Appl. Spectrosc. 44, 565 (1990). 25. Nguyen Quy Dao, M. Jouan, H. Nguyen Quang, and E. Da Silva,

Analusis 21,219 (1993). 26. P. Geladi, B. R. Kowalski, Anal. Chim. Acta 185, 1 (1986). 27. A. Lorber, L. E. Wangen, and B. R. Kowalski, J. Chemometrics 1,

19 (1987). 28. K. R. Beebe and B. R. Kowalski, Anal. Chem. 59, 1007 {1987). 29. D. M. Haaland and E. V. Thomas, Anal. Chem. 60, 1193 (1988). 30. W. Lindberg, J. Persson, and S. Wold, Anal. Chem. 55, 643 (1983). 31. D. M. Haaland, R. G. Easterling, and D. A. Vopicka, Appl. Spec-

trosc. 39, 73 (1985). 32. M. P. Fuller, G. L. Ritter, and C. S. Draper, Appl. Spectrosc. 42,

228 (1988). 33. D. M. Haaland and E. V. Thomas, Anal. Chem. 60, 1202 (1988). 34. H. Nguyen Quang, B. Vu Thi, M. Jouan, and Nguyen Quy Dao,

Analusis 20, 141 (1992). 35. M. B. Seasholtz, D. D. Archibald, A. Lorber, and B. R. Kowalski,

Appl. Spectrosc. 43, 1067 (1989). 36. H. Nguyen Quang, N. V. Tien, M. Jouan, Nguyen Quy Dao, N.

Zanier, V. Ruffier-Meray, and E. Behar, in Proceedings of the 13th International Conference on Raman Spectroscopy (John Wiley & Sons, Chichester, 1992), p. 1030.

2016 Volume 47, Number 12, 1993