650
Anal. Chem. 1980, 52, 650-653
(3) Rosencwaig, A.; Hall, S. S. Anal. Chem. 1975, 47, 548. (4) Bennett, H. S.;Forman, R. A. Appl. Opt.. 1977, 76, 2834. (5) Hordvik, A.; Skolnik, L. Appl. Opt. 1977, 16, 2919. (6) Farrow, M. M.; Auzannea, M.; Burnham, R. K.; Olsen, S. L.; Purdie, N.; Eyring, E. M. Appl. Opt. 1978, 77, 1093. (7) Adams, M. J.; Beadle, B. C.; Kirkbright, G. F.; Menon, K. R. Appl. Spectrosc. 1978, 3 2 , 430. (8) Nordal, P. E.; Kanstad, S. 0.Opt. Commun. 1978, 24, 95. (9) Wong, Y. H.; Hawkins, G. F.; Thomas, R. L. Appl. Phys. Lett. 1978, 32, 538. (10) Fujihira, M.; Osa, T.; Hursh, D.; Kuwana, T. J . Nectroanal. Chem. 1978, 88, 285. (11) McClelland, J. F.; Knisely. R. N. Appl. Opt., 1976, 75, 2658. (12) Lahman, W.; Ludewig, H. J.; Welling, H. Anal. Chem. 1977, 49, 549. (13) Gray, R. C . ; Fishman, V. A.; Bard, A. J. Anal. Chem. 1977, 49, 697. (14) Adams. M. J.; Highfield, J. G.; Kirkbright, G. F. Anal. Chem. 1977. 49, 1850. (15) Rockley, M. G.; Waugh, K. M. Chem. Phys. Lett. 1978, 5 4 , 597. (16) Shtrikma, S.; Slatkine, M. Appl. Phys. Lett. 1977, 3 7 , 830. (17) Gerlach, R.; Amer, N. M. Appl. Phys. Lett. 1978, 3 2 , 228. (18) Koch, K. P.; Lahman, W. Appl. Phys. Lett. 1978, 32, 289. (19) Krltchma, E.; Slatkine, M.; Shtrikrna, S.J . Opt. SOC.A m . 11978, 68, 1257. (20) Gray, R. C.;Bard, A. J. Anal. Chem. 1978, 5 0 , 1262. (21) Rosencwaig, A.; Ginsberg, A. P.; Koepke, J. W. Inorg. Chem. 1976, 75, 2540.
(22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) (34) (35) (36) (37) (38)
Fernelis, N. C. Appl. Spectrosc. 1978, 32, 554. Wong, K. Y . J . Appl. Phys. 1, 1978, 49, 3033. Nordal, P. E.; Kanstad, S. 0. Opt. Commun. 1977, 2 2 , 185. Monahan. E. M.; Nolle, A. W. J . Appl. Phys. 1, 1977, 48. 3519. Beadle, B. C.; King, A. A,; Kirkbright, G. F.; Adams, M. J. Analyst (London) 1976, 101, 553. Ortner, P. B.; Rosencwaig, A. Hydrobiologia 1977, 56, 3. Rosencwaig. A.; Pines, E. Biochim. Biophys. Acta 1977, 493, 10. Cahen, D.; Caplan, S. R.; Garty, H., FEBS Lett. 1978, 91, 131. Cahen, D.; Lerner, E. I.; Malkin, S. FEBS Lett. 1978, 91, 339. Lerman. S.; Borkman, R.; Palmer, R . A,; Roark, J. C. Opthalmic Res. 1978, 70, 168. Rosencwaig, A.; Gersho, A. J . Appl. Phys. 1976, 47, 64. Murphy, J. C.; Aamodt, L. C. Appl. Phys. Lett. 1977, 37, 728. VonBenken, W.; Kuwana, T. Anal. Chem. 1970, 42, 11 14. Szentirmay, R.; Kuwana. T., Anal. Chem. 1977, 49, 1348. Instruction Manual Lockin Amplifier Model 126, Princeton ADDlied Research Corp., 1971. Beranek, L. L. "Acoustics"; McGraw-Hill: New York, 1954. Armstrong, N. R., private communication.
RECEIVED for review June 11, 1979. Accepted January 10, 1980. This work was supported by grants from NSF (Grant No. CHE76-81591) and PHS-NIH (Grant No. GM19181).
Analysis of Turbid Solutions by Laser-Induced Photoacoustic Spectroscopy Shohei Oda, * Tsuguo Sawada, Toyohiko Moriguchi, and Hitoshi Kamada Department of Industrial Chemistry. Faculty of Engineering, The University of Tokyo, 7-3- 1, Hongo, Bunkyo-ku, Tokyo, Japan
The application of laser-induced photoacoustic spectroscopy (LIPAS) to the determination of ultra trace particles in suspension was attempted. Using BaSO, particles produced chemically in solution, the effect of particle size distribution upon photoacoustic and turbidimetric signal intensity was investigated. I n comparison with a turbidimetric analysis, photoacoustic signal intensity was less affected by the particle sire distribution. The detection limit was about two orders of magnitude lower than that obtained by turbidimetry. A linear concentration range of three orders of magnitude was obtained. Therefore, LIPAS has been found to be well suited for quantitative determination of particles in suspension in terms of sensitivity and accuracy.
Quantitative measurement of turbid particles in liquid has been carried out with a turbidimeter, which is based on the measurement of the attenuation of transmitted light by a solution containing a finely divided precipitate. There is a direct relationship between an optical density log Zo/Z (Ioand I are t h e intensities of the incident and transmitted beams respectively), and the amount of material in suspension ( I ) . T h a t is,
1 log & / I = ~KNdl 2.303 where N is number of spherical particles per unit volume, r is the radius of the particle, 1 is the thickness of a suspension, and K is sometimes called the total scattering coefficient and was deduced from electromagnetic theory by Mie assuming uniform spherical particles. K depends upon (1)the particle size, (2) the wavelength, (3) the refractive index, and (4) the complex refractive index in the case of absorbing particles (2). Turbidimetry is often not very precise since it is difficult to 0003-2700/80/0352-0650$01 .OO/O
prepare a stable and reproducible suspension of the precipitate. A precision and accuracy of *5-10% of the amount present is usually obtained. On the contrary, as photoacoustic spectroscopy provides only information concerning light intensity absorbed by particles, the particle size distribution would have less effect on the photoacoustic signal intensity. Recently, laser-induced photoacoustic spectroscopy (LIPAS) has been applied to the measurement of absorption of light by aerosols ( 3 , 4 )and dust (5)in the atmosphere. Terhune et al. ( 3 ) have reported that very weak absorption down to a level of m-' in the measurement of aerosols in the atmosphere could be detected by comparing photoacoustic signals with and without aerosols. Absorption measurements of ultra trace turbid or colloidal particles in liquid are important, particularly in the fields of biochemistry and environmental chemistry. However, few studies have been carried out owing to the lack of sensitivity and accuracy in their measurements. In the present investigation, an application of LIPAS to a trace determination of material in suspension was attempted using BaS04 particles produced chemically in water. Turbidimetric analysis of BaS04 in suspension for the determination of S042-in water has been carried out frequently as a well-known method. In comparison with this ordinary turbidimetry, LIP AS has several advantages, particularly in respect to sensitivity and accuracy even with suspensions which have extremely small absorption coefficients in the visible region such as the B a S 0 4 suspension.
EXPERIMENTAL Apparatus. The apparatus used in the present investigation was almost the same as the one in our previous papers (6-9).
However, some improvements were made for the measurement of material in suspension. Laser light was expanded t o about 7-mm diameter with a concave lens and was incident on a sample cell in order t o lower the effect of the particle size distribution
in the vertical direction upon the photoacoustic signal. A thin platinum foil (0.1 mm thickness) was attached inside the pie1980 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 52, NO. 4, APRIL 1980
zoelectric transducer (NPM, N-21, Tohoku Kinzoku Co. Ltd.), photoacoustic sensor of our cell, in order to prevent the light scattered by particles from being absorbed by the piezoelectric transducer (PZT). The detection sensitivity of PZT was scarcely lowered by these modifications. An argon ion laser (Spectra Physics Model 164-03) was modulated at a given frequency in the range from a few Hz to 100 kHz with an acousto-optic light modulator (Intra-Action Corp. Model AOM-40). The photoacoustic spectrum of BaS04 powder was measured with a commercially available double-beam photoacoustic spectrometer (Gilford Instrument Lab. Inc. Model R1500). A spectrophotometer (Hitachi Model 101) was used as a turbidimeter. The beam diameter of the incident light was about 10 mm. Reagents and Procedure. All reagents were of ultrapure grades. Twice-distilled deionized water was used. Two kinds of turbid BaSO, solutions were prepared as follows. Method 1. Ten mL of glycerine solution (1:l) and 5 mL of NaCl solution ( 2 5 % , HCl solution) were added to 50 mL of a known concentration of solution. While the solution was stirred, BaCI, powder (0.3 g) was added. The solution was stirred for 1 min, left for 4 min, and stirred again for 15 s just before measurement. Method 2. The same glycerine solution (1:l)and NaCl solution ( 2 5 % , HCl solution) were added to 40 mL of a known concentration of S042-solution as was done in method 1. While the solution was stirred, 10 mL of BaClz solution (3%) were added. The same procedure was carried out just before measurement. BaS04 powder was prepared as described below. BaS0, precipitate, formed by adding BaCl, solution (3%) to SO4'- solution (1 mg/mL was filtered off, washed with distilled water until the excess C1- could not be detected, and dried at 100 O C in a drying oven.
651
Wavelength ( n m )
Flgure 1. Photoacoustic spectrum of BaS0, powder. Modulation frequency: 80 Hz; time constant: 3.0 s; scan rate: 20 nmlmin; band pass: 13.2 nm
RESULTS AND DISCUSSION In order to investigate the effect of a thin platinum foil attached inside the P Z T upon the photoacoustic signal intensity, the variation of photoacoustic signal intensity with changing laser beam diameter was measured using BaS0, turbid solution (50 pg/mL). In the case of the cell in which a platinum foil was attached inside the PZT, the photoacoustic signal intensity held almost constant in spite of a laser beam diameter change. On the other hand, for the case of the cell without the platinum foil, the increase of absorption of scattered light by the P Z T caused the intensity to change greatly. Moreover, when a small portion of an argon ion laser (488 nm, 1mW) directly irradiated the PZT, the photoacoustic signal intensity obtained for the PZT with a platinum foil was in comparison with that for the PZT reduced to about 1/2w without the platinum foil. The above two experiments show that the light scattered by suspended particles has less effect on the measured photoacoustic signal, when the PZT is covered by platinum foil. T h e effect of the scattered light will be discussed again later. Figure 1 shows the photoacoustic spectrum of BaS0, powder. As BaSO, powder is nonfluorescent, this spectrum can be considered to coincide with the absorption spectrum. As shown in Figure 1, the absorption spectrum of B a S 0 4 powder has a maximum a t about 330 nm. At present, an assignment of this absorption band is not clear. Its trail is extended to the visible region. Though the color of BaS04 powder is white, it absorbs some of the visible light. The photoacoustic signal intensities of the BaS0, suspension were also measured using five different lasing lines. T h e results are shown in Figure 2. The intensities increased when the lasing line was shifted to a shorter wavelength. A spectrum similar to Figure 1 was obtained, although it was only a small portion of the whole spectrum. Curve 1 in Figure 2 was corrected by subtracting the photoacoustic signal of the solvent from the measured signal. Curve 2 shows the dependence of the photoacoustic signal of the solvent upon the wavelength. The power of each lasing line was monitored by a calibrated photocell. Though the wavelengths of the lasing lines used
2 ... JI 3 0
zi E
r
25-
-----
--_I___
48:
500 iCa\elength
---52C
nv)
Figure 2. Photoacoustic spectrum of BaSO, turbid solution. (1) BaSO, turbid solution. Concentration. 15 Fg/mL BaSO,. (2) Solvent. Laser power: 200 mW
for the measurement of the BaSO, suspension were far from its absorption maximum, a comparatively large photoacoustic signal was obtained. This shows that LIPAS is highly sensitive, and also that a fine particle in suspension xts somewhat as a black body and actual light-path lengths are enlarged by multiple reflection in a particle. As shown in Figure 2, it is favorable to use a lasing line of shorter wavelength in order to obtain a larger signal. However, considering that the photoacoustic signal is proportional to the intensity of a lasing line (6, 7 ) ,a line of 488 nm lasing with high power was chosen in the present experiment. The photoacoustic signal dependency upon the modulation frequency was investigated in the range from 10 Hz to 100 kHz (Figure 3). The photoacoustic signal showed its maximum at 33 Hz, and after that decreased monotonously with increasing frequency. The photoacoustic signal followed neither 0 - l nor w-3/2 ( w = frequency) below 400 I-Iz. However, beyond 400 Hz, it showed w-l dependency, as was indicated for solid samples by Rosencwaig's theory (10, 11). Several sharp peaks observed a t over 10 kHz were probably due to the resonance frequency of our cell. These tendencies have not been clarified yet. In the case of absorbing particles in suspension, pressure fluctuation generated in the photoa-
652
ANALYTICAL CHEMISTRY, VOL. 52, NO. 4, APRIL 1980
Choppng
Frequency
:% 1
Flgure 3. Relation between photoacoustic signal and modulation fre-
quency. Concentration: 25 pg/mL BaSO,; laser power: 400 m W Variation of BaSO, turbid solution with time. Concentration: (1) 150 pg/mL BaSO,, (2) 50 pg/mL BaSO,, (3)25 pg/mL BaSO,; laser power: 400 mW; (-) photoacoustic signal: (-- -) turbidimetric signal Figure 4.
coustic cell is limited by the rate a t which the heat is transferred to the liquid from the particles (12). Therefore, by varying the light modulation frequency, the particle size distribution could be measured. However, the theoretical calculation for the present experiment showed that a modulation frequency around 1 MHz would permit the determination of particle size. This was impractical owing to the lack of sensitivity. Therefore, subsequent experiments were carried out a t 33 Hz. The modulated heating of particles a t 33 Hz was efficient in generating pressure waves for every BaS04 particle in suspension. The diameter of the BaS0, particles is small when they are produced. They grow up gradually and finally sediment. Though a small amount of glycerine is added to the solution in order to prevent particles from precipitating rapidly, measurement in an ordinary turbidimeter is recommended within a few minutes after the particles are produced. Figure 4 shows the photoacoustic and turbidimetric signal variation with time when BaS04 particles are produced by adding BaClz powder to SO:- solution (method 1). The photoacoustic signal intensity held almost constant within 1 h after the particles were produced, although a slight decrease was observed. The size diameter of chemically produced BaSO., particles is considered to be in the range from 0.5 to 6 pm ( 1 ) . Under our experimental conditions (viscosity of the solvent, q = 1.3 g cm-' s-l), a particle 6 pm in diameter falls a t approximately 0.1 mm/min (13). I t takes about 1 h for particles over 6 gm to precipitate out of the expanded laser beam. The slight decrease in the photoacoustic signal after 1h would be caused by the precipitation of BaS04 particles. On the other hand, in a turbidimeter, the apparent absorptivity of the BaSO, turbid solution decreased with time, and this tendency became remarkable when the BaS0, concentration increased. The scattered light intensity is significantly dependent upon the particle size distribution, as mentioned previously. According to Mie's theory, the intensity of the forward scattered light increases when the particle size diameter increases. Therefore, the decrease in apparent absorptivity arising comparatively soon after the particles were produced would be due to the growth of the particles. The effect of the particle size diameter upon photoacoustic and turbidimetric signal intensity was investigated by using another method. According to Turnbull ( 1 4 ) , the size distributions of BaS04 particles produced chemically in solution are dependent upon the way BaC1, is added to S042-solution. In order to make the particle size smaller, the formation of local supersaturation is required during the mixing process. T h a t is, the addition of BaClz powder makes the particle size smaller than that of BaClz solution. For the case of the addition of BaC12 solution, the particle size distribution ranges
from 2 to 20 pm (14). Measurements of calibration graphs were carried out by photoacoustic and turbidimetric methods in various concentrations of BaS04 particles produced chemically in solution by using BaC1, powder (method 1) and solution (method 2 ) respectively. In a turbidimeter, two different lines were obtained, while two calibration curves coincided well with each other in LIPAS. The turbidimetric measurement clearly reflects the difference in the size distribution from the two different ways of adding BaClZ. On the other hand, in LIPAS the photoacoustic signal intensity was not affected by the difference of the size distribution. However, the possibility that absorption of the scattered light by the P2T with a platinum foil is contained to some extent in the photoacoustic measurement is still not ruled out. The amount of contribution of the scattered light by particles to the photoacoustic signal could not be estimated. However, if absorption of the scattered light by the PZT greatly affects the photoacoustic signal, the photoacoustic signal should exhibit the same tendency as the turbidimetric signal. From the results obtained with LIPAS and turbidimetry, as described above, the scattered light would be regarded to have less effect on the present photoacoustic measurement. A t 400 mW of laser power, the relationship between the photoacoustic signal intensity and the concentration of BaS04 was linear over a range of three orders of magnitude, that is, from 0.05 pg/mL to 100 pg/mL BaS04. In comparison with ordinary turbidimetry, the linear concentration range offered by LIPAS is more than one order of magnitude greater. The detection limit was calculated to be 0.03 pg/mL BaS04, based on a limiting signal-to-noise ratio of 2:l. As the result of calculation with extinction coefficient K [defined by I = Io exp(-4.lrKx/X0), x = effective thickness, Xo = wavelength in vacuo] of BaS04 from the literature (I), the value of the detection limit was found to correspond to an optical density of 1.6 x 10-j. This value was about two orders of magnitude lower than that obtained with a turbidimeter with a 50-mm length cell. In addition, the variation coefficients measured with 24 pg/mL and 0.5 pg/mL BaS04 turbid solutions were 4.4% and 5.4%, respectively (10 repeated runs). These values could be regarded as fairly good in comparison with those by turbidimetry. The application of LIPAS to colored particles such as AszS3 in suspension will be more interesting practically and is now in progress.
LITERATURE CITED ( 1 ) Lewis, P. C : Lothian. G. F. t3r. J . Appl. Phys., Suppl. 1953, 71-75. (2) Van d e Hulst, H. C. "Light Scattering by Small Particles"; John Wiley 8 Sons: New York, 1957; pp 114-130.
Anal. Chem.
1980, 52, 653-656
(3) Terhune, R. W.; Anderson, J. E. Opt. Lett. 1977, 7 , 70-72. (4) Bruce, C. W.; Pinnick, R. G. Appl. Opt. 1977, 76, 1762-1765. (5) Schleuserner, S.A,; Lindberg, J. E.; White, K. 0.; Johnson, R. L. Appl. Opt. 1976, 75, 2546-2550. (6) Oda, S.;Sawada, T.; Karnada, H. 6unseki Kagaku 1978, 27, 269-273. (7) Sawada, T.; Oda, S.; Karnada, H. Proc. Jpn. Acad. Ser. 6 1978, 54, 189- 193. (8) Oda, S.;Sawada, T.; Karnada, H. Anal. Chem. 1978, 50, 865-867. (9) Oda, S.;Sawada, T.; Nomura, M.; Karnada, H. Anal. Chem. 1979, 57, 686-688.
653
Rosencwaig, A.; Gersho, A. J. Appl. Phys. 1976, 47, 64-69. Rosencwaig, A. J. Appl. Phys. 1978, 49, 2905-2910. Chan, C. H. Appl. Phys. Lett. 1975, 26, 628-631. . . Daniels, F.; Alberty. R. A. “physical Chemistry”; John Wiley & Sons: New York, 1955; pp 491-535. (14) Turnbull, D. Acta Mettalurg. 1953, I , 684-691.
RECEIVED
for review June 29, 1979. Accepted January 18,
1980.
Halide Ion Enhancement of Chromium(III), Iron(II), and Cobalt(I1) Catalysis of Luminol Chemiluminescence Cheng A. Chang and Howard H. Patterson” Department of Chemistry, University of Maine, Orono, Maine 04469
The bromide and chloride ion enhancement of chemiiuminescence (CL) of trace metal ion catalyzed iumlnoi oxidation by hydrogen peroxide was examined for chromium( 111), iron(II), and cobalt(I1) ions. The enhancement remains nearly constant at different metal ion concentrations but it varies with the pH of the iuminol-H,O, solution. I n the presence of 0.3 M Br-, an increase of 6.2-, 3.5-, and 1.4-fold of CL signal was observed for Cr( III),Fe( II), and Co( II), respectively, at the optimized pH of analysis. The feasibility of analytical applications for the halide effect is discussed. A general mechanism for the halide effect is proposed.
Applications of the chemiluminescence (CL) technique in analytical chemistry have provided many fruitful results in recent years, especially in trace element analysis ( 1 , 2 ) . For certain analyses of trace metal ions, the application is based on the fact that metal ions can catalyze the luminol oxidation reaction with hydrogen peroxide (or molecular oxygen) in basic aqueous solution ( 3 , 4 ) . Since the intensity of light emission is usually proportional to the “free” metal ion concentration, t h e technique is particularly useful in the determination of the speciation of many metal ion systems in the environment. Very recently, as a result of a determination of chromium speciation in marine and fresh-water environments, Bause and Patterson have found that concentrated inorganic salt solutions, particularly those of halide ions, can cause a n enhancement of the CL intensity (5). A bromide concentration of 0.5 M yields an eightfold increase in the CL signal intensity for chromium(III), relative to the signal with no bromide present, lowering the detection limit to 1.3 X lo-’’ M (7 parts-per-trillion) for fresh-water systems. I n order to explore the “halide effect” further in the field of analytical chemistry, we report here some detailed studies of halide ion enhancement of chromium(III), iron(II), and cobalt(I1) catalysis of luminol chemiluminescence. T h e objectives of the present study are twofold: first, to explore the analytical applications of the “halide effect” for the determination of metal ions other than chromium(II1); second, to understand the nature as well as possible mechanisms of the halide effect.
EXPERIMENTAL Apparatus. The chemiluminescence was measured using a flow system similar to that reported by Seitz et al. ( 3 ) . However,
for most of the measurements, only two 20-mL plastic syringes were used. One syringe contains the metal ion sample to be tested with or without halide ions; the second syringe contains a mixed solution of luminol and hydrogen peroxide at the desired pH. The syringes were driven by a Sage Model 351 syringe pump. The metal ion solution was mixed with the luminol-H202 solution before entering a quartz flow cell. The cell was contained in a Perkin-Elmer MPF-44A Fluorescence Spectrometer by which the chemiluminescence was measured. Light emission was monitored at 430 nm. Reagents. Luminol was purchased from Aldrich Chemical Company. All other reagents were purchased from Fisher Scientific Company and were certified grade. They were used as received. Solutions were prepared using doubly purified deionized water by a Bion-Exchanger Water Purification System. The stock luminol solution was prepared using a doubly recrystallized sodium salt of luminol which was mixed with sodium bicarbonate and disodium salt of ethylenediaminetetraacetate (EDTA). The final pH was adjusted by adding 2.0 M NaOH solution. The stock hydrogen peroxide solution was freshly prepared each day by diluting the appropriate amount of 30% H202in the presence of EDTA. The working luminol-H202 mixture was made using stock solutions immediately before measurement. For all measurements, luminol = 2.5 X M, EDTA = 2.5 X M, and H202= 6 X M for Cr(II1) and Fe(II), 1.2 X M for Co(I1). A different H,Op concentration for Co(I1) analysis was chosen because the sensitivity was better. Metal ion solutions were prepared using stock solutions and standard dilution techniques with or without halide ions. Procedures. Standard procedures for routine CL analysis are published elsewhere ( 3 ) . For the analysis of Cr(III), EDTA was added to complex those possible kinetic-labile interfering metal ions. For Co(I1) and Fe(1I) analysis,no EDTA was added because EDTA reduces the “free” ion concentration due to complexation. However, the EDTA present in the luminol-H202 solution apparently does not interfere in the analysis for both ions. This may be due to the fact that the catalytic reactions are faster than the complexation reactions. O2was carefully excluded for the Fe(I1) CL analysis. We define the halide ion enhancement factor as the ratio of the enhanced CL signal t o that of the unenhanced signal under exactly the same conditions with and without halide ion, respectively. All data obtained have been corrected for background blank solution signals. The impurities in the commercially available metal halide salts do not interfere with the analyses as long as the trace metal ion concentration being determined is greater than lo-@M. However, when the metal ion concentration is down to sub-part-per-billionlevels, the impurities may be a serious problem for Co(I1) and Fe(I1). Separate experiments were done with the halide ion in the luminol-H202 solutions so that EDTA could
0003-2700/80/0352-0653$01.00/0 0 1980 American Chemical Society
