Time-resolved spectroscopy of long-path fluorescence and scattering for measuring total absorption of fluids

ANALYTICA

CHIMICA ACTA ELSEVIER

Analytica Chimica Acta 330 (1996) 183-I 87

Time-resolved spectroscopy of long-path fluorescence and scattering for measuring total absorption of fluids K.-H. Mittenzwey”?*,

G. Sinnb

“Gesellschaft fiir Me- und Systemtechnik mbH, Rudower Chaussee 5, 12489 Berlin, Germany bGeographisches Institut, Humboldt-Universitiit Berlin, Chausseestrasse 86, 10115 Berlin, Germany

Received 8 January 1996; revised 1 April 1996; accepted 6 April 1996

Abstract Optical methods are appropriate tools to detect organic micro-pollutants in fluids. A new technique is introduced which uses the decay of interaction processes like fluorescence and elastically scattered radiation by a fluid. Principally two different parameters are determined: (i) the decay-time of the conventional interaction ro, which occurs at relatively short path-lengths of the incidence beam in the fluid, and (ii) the decay-time rMp of the multi-path-saturation interaction originating at long pathlengths, e.g. in multi-path-reflection cuvettes, where the incidence beam is fully absorbed by the fluid. A relation between the decay-time and the absorption coefficient of a fluid is theoretically derived. A simple preliminary experiment is performed considering distilled water polluted with non-fluorescent azobenzene and fluorescent quinine-sulphate. A nitrogen laser has been used to generate the fluorescence and scattering signals. The reciprocal value of the difference between the decay-time of the multi-path and conventional signals, l/(7 MP-rc), yields the total absorption coefficient directly. In comparison to the conventional absorption technique the decay-time method is characterized by a higher sensitivity. Keywords: Fluorimetry; Long-path fluorescence;

Scattering;

Organic

substances

1. Introduction Optical techniques are often used for analytical purposes. Methods of high sensitivity are needed to detect organic micro-pollutants in fluids, e.g. fluorescence techniques [l-S]. Fluorescence seems, in principle, to be appropriate for monitoring of waters in the environment and industrial processes. Unfortunally, conventional fluorescence methods can only

* Corresponding author. Present 1 l/II, 10439 Berlin, Germany.

address:

Finnlkdische

Strae

0003-2670/96/$15.00 0 1996 Elsevier Science B.V. All rights reserved PII SOOO3-2670(96)00168-7

be applied to fluorescent substances, and are influenced by varying fluorescence quantum yields. Recently, a new optical method has been introduced considering the fluorescence and elastical scattering [9-111. According to this the interaction photons of fluorescence and scattering are measured at two different path-lengths of the incident radiation: (i) The intensity of the conventional interaction (CI) which occurs at relatively short path-lengths of the incidence beam in the fluid, and (ii) the multi-pathsaturation interaction (MPSI) originating at long path-lengths, e.g. in multi-path-reflection cuvettes, where the incidence beam is fully absorbed by the

184

K.-H. Mittenzwey,

G. SinnLAnalytica Chimica Acta 330 (1996) 183-187

fluid. The ratio CIZMPSI yields the total absorption of the fluid directly. In contrast to the conventional absorption spectrometry, this ratio is characterized by a higher sensitivity. To get a higher degree of selectivity time-resolved measurements are often used. Therefore, the timeresolved behaviour of the MPSI signals is investigated. The correlation between the decay-time of the MPSI and the absorption are examined. Besides theoretical considerations, preliminary experiments with waters samples polluted by aromatic substances were performed.

2. Theory

CI = la * H * m * a.

We consider a fluid with the total absorption coefficient EC. The terms E and C denote the specific absorption coefficient and the substance concentration, respectively. The fluid contains dissolved organics and is located in a cuvette. An incident photon beam with the intensity Zo, at the wavelength X, penetrates the fluid in the x-direction (Fig. 1). The interaction of the incidence beam with the fluid causes fluorescence and elastical scattering. These interaction photons are measured perpendicularly to Ia in the y-direction. It is assumed that the attenuation of the interaction photons in the y-direction is of minor influence, which can be achieved by short path-lengths in the y-direction (below a few mm). Then, the intensity Zr of interaction photons results in: ZI = IO *H

The term a stands for the cuvette length. The operational parameter H includes the diameter of the incident beam, the filter transmission and the aperture of the optical system. In the case of fluorescence, the term M is characterized by the fluorescence quantum yield QF and the absorption coefficient E&F of the fluorescent substances (m=Qr&nCr). In the case of scattering, m stands for the scattering coefficient /I of the fluid (m=P). Now the exponential term in Eq. (1) is considered. At values of d*a lower than 0.2, e.g. at short pathlengths (a=l-2cm) of the incident radiation, Eq. (1) can be transformed into a linear one yielding the well-known linear or conventional interaction CI:

*$

* (1 -

exp(-ecu)).

IWtMlectlon

(1)

cell with

pulwd N,-laser ( 500 ps, 350 kW )

low cut-on

Fig. 1. Experimental setup.

glass

UG5,

395nm

+

(2)

In contrast to this, the exponential function reaches a saturation level at values K*a higher than around 2.5. Assuming low substance concentrations (BeerLambert!), a saturation can only be achieved at long path-lengths, e.g. a=0.3-2.5 m, which can be realized using multi-path-cuvettes with high reflecting mirrors. Then, Eq. (1) results in the multi-path-saturation interaction MPSI: MPSI=Z,,*H*s.

(3)

In case of fluorescence, the interaction photons CI results in the conventional fluorescence (CF), and MPSI in the multi-path-saturation fluorescence (MPSF). We get analogous results in the case of scattering CI=CS (conventional scattering) and MPSI=MPSS (multi-path-saturation scattering). The CI signals (Eq. (2)) occur at short path-lengths of the incident beam in the fluid, e.g. l-2cm. The incident beam needs around 50-100 ps to pass such lengths when water is used. Because the fluorescence decay of many aromatic molecules falls into the range of greater than 1 ns, the measured timeresolved conventional fluorescence CF is, besides the apparatus function, dominantly influenced by the decay of aromatic molecules. In contrast to this the elastical scattering is a very fast process. Thus, the measured time-resolved conventional scattering CS is dominated by the apparatus function of electronical device used. The MPSI signal (Eq. (3)) occurs at long pathlengths of the incident beam, at which it is fully

K.-H. Minenzwey, G. SintdAnalyrica Chimica Acra 330 (1996) 183-187

absorbed by the water. This is the case when the intensity I of the incident beam is decreased to at least 10% of its initial intensity. Because of the correlation between the path-length xgo% and time t90% needed by the incident beam, this approximation results in: X90%

=

c * t9()s

=

2,3 EC

The term c means the velocity of the incident beam in water. For example, for absorption coefficients of typically l-20 m- ‘, a full absorption of the incident beam can be achieved at path-lengths of 0.115-2.3 m. For such path-lengths, the incident beam takes 0.510ns. Thus, t90s influences the time-resolved behaviour of the multi-path-saturation signals MPSI essentially. Usually the band-width of the measured MPSI time-function is expressed by the measured decaytime rMF But besides t90s MPSI is also influenced by the apparatus function of the electronical device used, and in case of fluorescence, further by the fluorescence decay of aromatic molecules. To get a decaytime I- only influenced by tws, the measured decaytime rMp of MPSI has to be corrected by subtraction of the decay-time rc of the conventional CI signals discussed above. Considering c=230.000 km/s, we get a linear correlation between the inverse decay-time l/ r and the absorption coefficient EC: 1 r[ns]

1 (TMP -

= k * &[m-‘1.

TC)[ns]

Thus measuring T the absorption coefficient EC can be determined directly. This is in contrast to the conventional absorption spectrometry, which is expressed by a logarithmic correlation between the transmission and absorption causing a relatively small sensitivity.

3. Experimental The distilled water (high purity) polluted by nonfluorescent azobenzene (puriss.) was first investigated. For this purpose azobenzene was dissolved in ethanol (spectran.). Then a small volume of the ethanol/azobenzene solution was added to the

185

distilled water at concentrations of O-1 mg/l. Secondly, distilled water with a constant amount of 100 pg/l quinine sulphate (99%) was investigated, which was also polluted by azobenzene. The samples prepared did not shown any turbidities. A nitrogen laser was used in the laboratory generating short pulses of 0.5 ns at a wavelength of 337nm (Fig. 1). The pulse power amounted to 350 kW. The repetition rate was 20 Hz. A multi-path cell was used to create long path-lengths of the incident beam, necessary to generate the MPSI signal. A 2cm standard quartz-cuvette was located in the center of this unit. The incident beam entered the cuvette through a small hole in the first cylindrical mirror. The second mirror was a concave one. The multi-path cell was of a Herriott type [12]. The mirrors were coated with dielectric layers yielding a reflectivity of more than 0.995 at 337nm. With a motorized translation stage, the mirrors were moved into the incident beam. The conventional signals CI were generated without mirrors. A combination of a cut-off and a blue band-pass filter (390-460nm) was used for fluorescence measurements (CF and MPSF). A band-pass filter UG5 was applied for measurements of the scattering signals (CS and MPSS). Both fluorescence and scattering were measured perpendicularly to the incident beam, using a photomultiplier with a rise time of 0.6ns. The measurements were controlled with a boxcar averager at a sampling gatewidth of 220~s. Synchronously, the conventional absorption by means of a fibre-optic polychromator was measured using a 2cm quartz cuvette.

4. Results and discussion Fig. 2 shows measured time-resolved spectra of MPSS of distilled water plus quinine sulphate (DW+QS) at two very different concentrations of non-fluorescent azobenzene (AB at 0 and 2.5 mg/l). The scattering is characterized by a large band-width at Omg/l and small band-width at 2.5 mg/l. Low concentrations lead to long path-length x9o.x, of the incident beam, causing large values of the measured decay-time rMF At high concentrations, the incident beam is already fully absorbed after a short path xgo%, leading to a small decay-time. This is in accordance with the theory (Eqs. (4) and (5)).

186

K.-H. Mittenzwey, G. SindAnalytica

Chimica Acta 330 (19%)

0

183-187

0.2

0.4

0.6

AZOBENZENE CONCENTRATION

Fig. 2. Measured time-resolved multi-path-saturation scattering (MPSS) of distilled water plus quinine-sulphate at two different concentrations of non-fluorescent azobenzene (quinine sulphate concentration: 100 ug/l, excitation source: nitrogen laser at 337 nm, emission filter: UG5).

The measured time-resolved spectra, especially at 0 mg/l, show two local maxima. One reason could be the following. The very first reflections occur in the vicinity of the entrance hole of the cylindrical mirror. Thus a part of marginal photons of the incident beam could leave the multi-path cell through this hole. This causes a loss of its intensity at short times, and thus leads to small values of MPSS. After the first reflections, the distance between the hole and the incident beam increases. Thus photons of the incident beam can no longer go through the hole leading to MPSS intensities without losses at longer times. The time-resolved behaviour of the fluorescence MPSF is similar to MPSS in principle. But the decaytime rMp of the measured scattering MPSS (10.8 and 3 ns for 0 and 2.5 mg/l, resp.) is lower than that of the measured fluorescence MPSF (13 and 5.9ns, resp.). Clearly, this is due to the fluorescence decay of the aromatic quinine sulphate, which is slower than the fast scattering decay. Fig. 3 shows the inverse decay-time l/7=1/ (rMp--rc) of elastical scattering of distilled water polluted by non-fluorescent AB in the concentration range of O-l mg/l. Principally, l/r increases with rising AB concentration. A linear regression analysis yielded a squared correlation coefficient of ?=0.98, indicating a good correlation between the inverse decay-time and AB concentration. But the correlation is worse in the concentration range below 5Opg/l (?=OSl). Figs. 4 and 5 show the inverse decay-time l/r of scattering and fluorescence of DW+QS polluted by non-fluorescent AB in the concentration range of

Fig. 3. Inverse decay-time l/r water as a function of the azobenzene ($=0.98 for O-l excitation source: nitrogen laser

0.8

of elastical scattering of distilled concentration of non-fluorescent mg/l; ?=0.51 for O-O.O5mg/l; at 337 nm, emission filter: UG5).

1.4 ,

I

I

01

0

1

(mg/ll

0.2

0.4

0.6

AZOBENZENE CONCENTRATION

0.8

1

(mg/l)

Fig. 4. Inverse decay-time l/r of elastical scattering of distilled water plus quinine sulphate as a function of the concentration of non-fluorescent azobenzene (?=0.89 for O-l mg/l, ?=0.90 ‘for O0.05 mg/l, quinine-sulphate concentration: 100 pg/l, excitation source: nitrogen laser at 337 nm, emission filter: UG5).

. .I . 1

0

0.2

0.4

0.6

AZOBENZENE CONCENTRATION

0.8

1

(mgll)

Fig. 5. Inverse decay-time l/r of fluorescence of distilled water plus quinine sulphate as a function of the concentration of nonfluorescent azobenzene (AO.91 for 0-lmgfl, ?=0.90 for O0.05 mg/l, quinine sulphate concentration: 100 pg!l, excitation source: nitrogen laser at 337nm, emission filter: cut-off and band-pass filter at 39OA60 nm).

K.-H. Mittenzwey,

G. SinnLAnalytica Chimica Acta 330 (1996) 183-187

O-l mg/l. l/r increases with rising AB concentration. A linear regression analysis yielded a squared correlation coefficient of ?=0.89 for scattering and 0.91 for fluorescence. At concentrations below 50 ug/l, the rz values amount to 0.90 for both scattering and fluorescence, showing a much better correlation as in case of AB in DW. Further l/r is somewhat higher for AI3 in DW than for AB in DW+QS at concentration below 50 ug/l although the absorption of DW+QS is higher than that of DW. Both phenomena indicate that in the case of AB in DW the path-length x9o% necessary to generate a correct MPSI signal cannot be achieved at such low concentrations. This causes to small MPSI values resulting in to large CVMPSI ratios. Obviously, the multi-path cell used is not appropriate in this case. In contrast to the decay-time measurements, the conventional absorption is characterized by ?=0.24 for AB in DW and ?=0.76 for AB in DW+QS at Al3 concentrations below 50 ug/l. These lower correlations are likely to be caused by its lower sensitivity which is in accordance to the theory. Principally, the experiments confirm the theory. The total absorption can be measured by UT directly. In comparison with the conventional absorption spectrometry, the decay-time method is characterized by a higher sensitivity. This could lead to lower detection limits in the analysis of aromatic substances in fluids. The method introduced is independent of the fluorescence quantum yield. Clearly, this is advantageous in case of fluids with unknown and varying quantum yields. The decay-time method works even in the case of non-fluorescent fluids using the elastical scattering. The decay-time method introduced is not disturbed by varying and different device parameters such as initial intensity, filter transmission, and aperture of the optical system described by the operational parameter H. The method offers the possibility of measuring the absorption coefficient by setting two different timegates. The one time-gate is set at short times characterizing the conventional short-time signals CF and CS. The other one is a wide gate to measure the long-time signals MPSF and MPSS.

187

5. Conclusions The method introduced may be used for monitoring of waters in the environment and industrial processes. For example, dangerous and harmless samples could be discriminated in-situ. Early-waming systems seem to be possible. But as only preliminary experiments have been performed, further investigations are necessary to verify the suitability of the method for the applications mentioned. Especially, mixtures have to be investigated. Clearly, selectivity and sensitivity depend on the application and are influenced by the number of excitation wavelengths and the possibility of sample treatment. Furthermore, multi-path cells also appropriate for very low absorptions should be designed and possible influences by both the raman and cuvette scattering have to be investigated.

References 111 S.A. Tucker, J.S. Griffin, W.E. Acme, Jr., M. Zander and R.H. Mitchell, Appl. Spectrosc.,

48 (1994) 458.

121 S.A. Tucker, H.C. Bates, V.L. Amszi and W.E. Acree, Jr., Anal. Chim. Acta, 278 (1993) 269. 131 I. Makinen and E.L. Poutanen, Sci. Total Environ., 81/82 (1989) 32. Proc. Environment [41 N.H. Eisium and A. Lyngaard-Jensen, and Pollution Measurement Sensors and Systems. Society of Photo Optical Instrumentation Engineers, Bellingham Washington, 1269, 1990, p.167. PI K.M. Bark and R.K. Force, Talanta, 38 (1993) 181. KY U. Panne, F. Lewitzka and R. Niessner, Analusis, 20 (1992) 533. Fluorimetrische Analyse, Verlag Chemie, [71 G. Schwedt, Weinheim, 1981. Verbindungen, VandenPI T. Forster, Fluoreszenz organ&her hoeck und Ruprecht, Giittingen, 1982. r91 K.-H. Mittenzwey and G. Sinn, EARSeL Adv. Remote Sensing, 4 (1995) 87. [101 K.-H. Mittenzwey, J. Rauchfu, G. Sinn and H.-D. Kronfeldt, Fresenius’ J. Anal. Chem., 354 (1996) 159. K.-H. Mittenzwey and H.-D. Kronfeldt, Offenlegungsschrift, [Ill DE 43 37 227 Al (1995). 1121 H.-D. Kronfeldt and J. Berger, SPIE, Lens and Optical Systems Design, 1780 (1992) 650.