ANALYTICAL CHEMISTRY, VOL. 51, NO. 11, SEPTEMBER 1979
Table 111. Silver Results Obtained from Analyzing Fixer Solutions Using the HCDA and Atomic Absorption Methods" fixer, type HCDA AA A B C D E (bleach-fix) a
10.2 0.02
11.8
1.6
1.2 0.8
0.03 1.5 1.1 0.8
how efficiently a silver recovery unit attached to a fixer bath operates. In determining the quantities of silver in fixer solutions, all the steps in the General Procedure need not be followed. Usually there is adequate thiosulfate already present to maintain the integrity of the Ag(S2O3)?-complex. However, the final liquid volume in the beaker should be maintained as well as the pH control.
CONCLUSION
Silver data reported in g/L.
the ~ ~ ( ~ anionic ~ o ~complex. ) 2 - H ~ negative ~ ~ interference was evidenced when KBr and KI exceeded 500 mg and 25 mg, respectively, in a sample. Table 111 shows the silver data obtained by HCDA and atomic absorption on various types of "seasoned" photographic fixer solutions. There were no significant differences between t,he methods (correlation coefficient, r5 = 0.9995). Some chemicals commonly found in fixer baths are: sodium sulfite, aluminum sulfate, sodium acetate, sodium borate, and ammonium thiosulfate. In the case of a bleach-fix solution, iron(II1) may be present, complexed with an organic sequestering agent. The proposed method has been used to monitor silver in fixer solutions to determine when a bath needs replenishing. Also, the levels of silver found in solution help to determine
1865
Hexaamminecobalt(II1) chloride has been shown to be a useful reagent for, the gravimetric determination ~ precipitating ~ ~ of silver in Photographic and nonphotographic systems. EDTA can be added to prevent aluminum and iron from hydrolyzing and elements such as lead and cadmium from interfering. The method has the advantage of not having to destroy the silver thiosulfate complex or wet ashing silverbromide-chloride-thiocyanate matrices for subsequent
LITERATURE CITED (1) Yoshirnura, Jun; Takashirna, Yoshirnasa; Murakami, Yoshio; Kusaba, Toyoaki Bull. Chem. SOC.Jpn. 1962, 35, 1435.
RECEIVED for review February 28,1979. Accepted May 9, 1979.
Sample Cells for Photoacoustic Measurements David Cahen" and Haim Garty The Weizrnann Institute of Science, Rehovot, Israel
Photoacoustic (PA) measurements have been reintroduced in the past few years as an alternative (and at times the only) way to investigate spectroscopic properties of solid, semisolid, and liquid samples in the ultraviolet, visible, and infrared (1-4). The electronic equipment needed for PA measurements does not, in general, pose special problems, except perhaps for the choice of a suitable microphone. On the other hand, sample cells, and especially the combination of these with the microphone, tend to be a limiting factor in determining the success of the experiment. Although there are various descriptions of PA cells available in the literature, most of these need expensive microphones and/or sample cells that require considerable skill for their construction (2,5-7). We have been developing the PA method mainly for the investigation of energy conversion processes (by photocalorimetry), which puts high demands on the sensitivity of the method (8-11). The cells now in use in our laboratory, which are a further development of one described earlier ( 4 ) , were found to be very suitable for normal spectroscopic measurements as well, notwithstanding their simplicity of construction. Figure 1 shows two of these cells, whose main features are: (1)Short effective length, i.e., low gas volume to sample surface area (in contact with gas) ratio (VIA). All other factors being equal, a smaller ratio will give a better signal to background (= cell noise) ratio, SIB. (2) Use of commercially available rectangular glass or quartz tubing. (3) Versatility through the use of different, interchangeable, sample holders improving reproducibility and S I B especially for liquids and flat samples. While the round quartz tubes used in the earlier described cell gave reasonable results, experiments on less ideal samples and difficulties with reproducibility because of ill-defied light 0003-2700/79/0351-1865$01 .OO/O
intensity suggested the use of sample cells with lower V I A . The rectangular tubing (Vitro Dynamics, Rockaway, N.J.) we use, was chosen from tubing with several dimensions as the one giving the best results. Especially a minimal wall thickness (21mm) was found to be necessary. Although initially the sample was placed directly on the cell bottom, this method was found to introduce much noise because of such external factors as small temperature changes, air flow around the cell, and ambient acoustic noise. Closing the cell with a cover to which the sample holder is attached directly decreases the cell volume, while hardly affecting exposed sample surface area and improves reproducibility because the sample can be taken out, treated, and returned to the same position. Moreover the sample is now relatively insulated from outside interference. Silicone vacuum grease is used to seal the sample holders to the cell ( 4 ) . The two kinds of sample holder shown in Figure 1 each have their special advantages. Component 4 is a bath constructed from Perspex with a transparent bottom. Such a holder is particularly useful for liquids and powders. The small V I A of part 4 is important for liquids, which, because of their small surface area among other things, tend to give low PA signals. Figure 2 illustrates this for an aqueous suspension of bacteriorhodopsin-containing purple membranes of Hulobucterium halobium, a t several frequencies ( I O ) . Here light absorption initiates a cyclic photochemical process which drives the translocation of protons from one side of the purple membrane to the other. The occurrence of a photochemical process will cause a decrease in PA signal because less of the absorbed energy is converted into heat. At higher modulation frequencies, earlier intermediates of the photocycle are sensed, 0 1979 American Chemical Societv
1866
ANALYTICAL CHEMISTRY, VOL. 51,
6.LIGHT
/
4/
NO. 11, SEPTEMBER 1979 {SOURCE
/
1
5
Figure 1. Schematic drawing of PA cells. (1) Microphones (BT 1753, Knowles Electronics Inc.) (2) Support plate (brass, stainless steel, nickel, etc.) with grooves for rectangular cells and holes for contact with microphones. (3) Quartz or glass rectangular tubing, 7 X 6 X 3 mm, inside dimensions, 1.2-1.4 mm wall thickness (Vitro Dynamics). (4) Bath sample holder on cell cover (Perspex); 5.7 X 4.2 X 1.3 mm inside dimensions, bath wall 0.5 mm; 25-pL samples of liquid are sufficient to fill the cell. (5) Grid sample holder on cell cover (Perspex); 6.8 X 4.5 X 1 mm. (6) Light guide@) for modulated- and/or continuous side-illumination
I
I
I /--\
PM
/
‘
04
60-
I I
v
n
/I
1
W
a z
Hz
a
\
a
\
;20
‘.-
100 Hz
J
a
LL
ar
=
0
0-,
1
1
450
550
650
X (nm) Flgure 2. PA spectra of purple membrane (PM) suspensions (25 pL, 2 mM bacteriorhodopsin)at various modulation frequencies, compared with the normal absorption spectrum. Note the drastic increase in PA signal upon frequency decrease (e.g., 565 nm1660 nm), indicating increasing energy storage in the system (8, 70). Spectra obtained using Brookdeal/Ortec 9502 lock-in amplifier; BT 1753 microphones, 450-W Xe arc lamp
which store a larger fraction of the absorbed energy, than a t lower frequencies, and the PA signal will thus be smaller (8). The data in Figure 2 are influenced furthermore by the different frequency dependence for the absorbing membrane sample and the opaque carbon black powder, used for light intensity correction (1). This will tend to exaggerate the decrease of the PA signal a t higher frequencies. Because of the relatively high illumination intensities used in PA experiments and the necessary presence of a layer of gas between the sample and window (unless piezoelectric detectors are used, which introduce altogether different
problems, especially in the study of energetics), condensation on the window tends to occur if aqueous samples are used. In flat cells like component 4 we find this problem to be a minor one as, in a matter of minutes a practically stationary situation is obtained (asjudged by the constancy of the signal). Possibly the small volume of air between sample and window aids the rapid establishment of an equilibrium, thus allowing PA measurements to start. We have found no decrease in signal before and after condensation forms on the window’s inner side. The transparency of the cell bottom further decreases the cell’s background signal (12). [Recently S. Gasner of this Institute has constructed sample holders like no. 4 by injection molding. Such holders, which now can be manufactured in large enough quantities to be used as “throw away” ones if necessary, have superior clearness, better transparency and some two times higher S I B than those prepared normally (and used for the data presented here).] The second kind of holder shown, no. 5, is suitable for flat samples, especially those that should be in contact with the gas on both sides. If this last condition is not critical, the grid can be filled. (This may increase the background signal, though.) This grid is used for, among other things, pieces of leaf, and for suspensions absorbed on filter paper (e.g., biological samples on Millipore), in cases where the suspension alone yields too low a P A signal. This is a convenient way of increasing the sample concentration in a very thin layer, thus working with a highly absorbing sample without entering into the PA saturation range (1, 3). Because some space can still be available below the grid, depending on the grid thickness, drying out in wet samples can be avoided by placing some wet cotton wool under the grid. In this way long-term experiments (several hours) on the same, wet sample have proved to be easily possible. A different kind of experiment that can be done on holders like no. 5 is the investigation of thin surface layers on electrodes. We investigated thin film (-3 pm) polycrystalline CdSe photoelectrodes at high (-700 Hz) and lower (-35 Hz) modulation frequency after use in polysulfide electrolyte and after subsequent surface etching. Such electrodes undergo S/Se substitution in the top few Angstroms (13). The PA spectra before etching show indeed a small ( - 5 % ) increase in signal around 530 nm, corresponding roughly to the CdSe bandgap, on top of the CdSe spectrum. This increase is very small at 35 Hz (thicker layer sampled) and upon etching in HC1, which dissolves CdS, the increase disappears a t both frequencies. The broadened spectrum of polycrystalline semiconductors, the relatively low surface area of these samples, as well as the facts that one deals here with thin layers, much thinner than the thermal diffusion length, and that only a small portion of the light is absorbed in the thin CdS layer, make the performance of such experiments in cells with higher V I A and lower S I B impossible. Experiments with single crystals gave sharper, comparable results. In Table I we compare the various cells. The background signal given should be considered a maximal one only, especially for the sample holders described above, because if these are completely covered by the sample, the bottom of the holder does not contribute to the signal. The remaining or less than the one given. contribution from the cell is Minimal signal-to-noise ratios for carbon black (opaque) samples are: 80 (in grid) and 150 (in bath), and for the empty cells (-transparent sample): 20 (grid) and 50 (bath), all a t 1-s time constant. The transparency of not only the top, but also the side walls of the cell, makes side-illumination very easy, especially if light guides are used. Because strong side-illumination can cause considerable sample heating, heat-induced gas absorption and desorption can be followed in this way (cf. Ref. 14). Studies
ANALYTICAL CHEMISTRY, VOL. 51, NO. 11, SEPTEMBER 1979
1867
Table I. Comparison of Carbon Black Signal, Background, and Volume to Area Ratio ( V / A )of Various PA Cells (40-Hz Data; 500-nm Irradiation, BT 1753 Microphone) bkdl signal, VIA, signalb bkdb %' mm empty cell grid (dry) grid (wet) bath (dry) bath (wet) round cell" single cell"
1 3.5 (0.5)d 5.5 (1)
___
1 2.5 1.5 4 2.5
.__ __.
.__
1.5 1.1 0.7 1.1 0.6 2.1 1.5
3
2.4
__-
3
__-
4.4 4.4
" From
Ref 4 ;for comparison only ; 7 2-Hz data. Signal is Relative values with respect to empty cell. that obtained with dry carbon black. Same carbon black used for all measureineni-s. As certain powders can give signals higher than this carbon black, the values are for Values for wetted carbon black; comparison only. these are generally < of dry values.
I
I
1 C-BLACK
J
IO
50 FREQUENCY
100
500
(Hz)
Figure 3. Frequency dependence of non-activated carbon black (Canadian 1001) in cell of Figure 1 (no. 4). The dotted and dashed lines indicate the frequency dependence of the signal when a flat-response microphone is used. Rosencwaig: from Ref 14, Figure 4; MW: from Fig. 6 in McDonald and Wetsel, Ref 16
on a variety of porous clay samples show large differences in their behavior upon side-illumination, differences that correlate roughly with their active surface area as measured by BET. Although the sample holder material (Perspex) is not transparent below -400 nm, its useful wavelength range will not be limited by this, as most samples are thermally thick (1) down to at least 10 Hz (solid powders of >1mm thickness, solutions with depths 10.1 mm). This means that radiation which reaches the sample holder (its bottom, in the case of component 4) will not participate in the generation of the PA signal. Only the transmission characteristics of the quartz or glass cell will limit the wavelengths available for study. We have performed successfully several UV studies in holders like no. 4, e.g. the triplet state of anthracene in hexane solution, and solid state photodimerization of 9-cyano anthracene. The frequency dependence of carbon black in these cells is illustrated in Figure 3. For comparison, data on two other opaque samples are given, namely GaP (from Rosencwaig, Ref. 15) and a phenol red solution (from McDonald and Wetsel, MW, Ref. 16). The differences between our data and those from Refs. 15 and 16, which are especially pronounced below 100 Hz, are caused not only by the different samples and PA cells used (giving rise to different fractional light losses on, and contributions from the cell walls and sample holder), but mainly because of the different microphones used. In our case
Figure 4. Schematic drawing of flat cell as used for measurements on photovoltaic cells. (1) Support plate for PA cell (part of box with battery etc. for microphone). (2) Cell body (Perspex): sample chamber dotted; dimensions 11.5 X 9 X 2 mm. (3) Cell window (quartz). (4) BL 1685 microphone. (5) Cover with electrical connections for (7). (6) Cover. (7) photovoltaic cell
the microphone sensitivity starts t o decrease around 100 Hz and drops sharply below 50 Hz. Roth the E T (Condenserelectret) and BL (piezoelectric-ceramic) series of Knowles microphones show this drop which, however, makes measurements down to at least 3 Hz still possible. For measurements at lower frequencies, much more expensive microphones, Bruehl and Kjaer or General Radio for example, should be used. The 1-kHz sensitivity of the Knowles microphones (which is slightly less than that at 50 Hz) varies between 2 and 10 mV per N/m2. The B T 1753 microphones used by us in most cases, have the maximal 10-mV sensitivity at 1.3-V polarization, while the BL 1685 microphones are about 3 times less sensitive at this voltage. By increasing the polarization voltage for these latter ones, 4-mV sensitivity at 2.6 V can be reached. Because of their small entrance port, these microphones are useful as their incorporation in a cell is rather simple. Figure 4 illustrates this for a PA cell, we use inter alia for measurements on photovoltaic cells. The small V I A of this cell, 2.2 mm, makes it very suitable for such experiments (11). It is not included in Table I because of the different microphone used in it. By employing the newer, smaller EA 1841 microphone, which has 5-mV sensitivity a t 1 kHz and 1.3 V, the performance of such cells may be improved further.
ACKNOWLEDGMENT We thank A. Auerbach for skillful assistance and Knowles Electronics, Inc., for generously providing the microphones.
LITERATURE CITED (1) A. Rosencwaig in "Opto-acoustic Spectroscopy and Detection", Y . H. Pao. Ed., Acadamic Press, New York. 1977, Chapter 8, pp 192 ff. (2) M. J. Adams, B. C. Beadle, and G. F. Kirkbright, Anaiyst(London), 102,
567-575 (1977). (3) R. Somoano, Angew. Chem., Intl, E d Engl., 17, 238-245 (1978). (4) D. Cahen, E. 1. Lerner, and S. Malkin, Rev. Sci. Instrum.. 49, 1206-1209 (1978). (5) J. C. Murphy and L. C. Aarnodt, J . Appl Phys., 48,3502-3509 (1977). (6) D. Betteridge, H. E. Hallarn, and P. J. Meyler, Fresenlus Z . Anal. Chem., 290, 353-356 (1978). (7) R . C. Gray, V. A. Fishrnan, and A . J. Bard, Anal. Chem., 49,697-700 (1978). (8) S.Malkin and D. Cahen, Photochem. Photobiol., 29 803-813 (1979). (9) D. Cahen, S.Malkin, and E. I. Lerner, FEBS Lett., 91, 339-343 (1978). (10) D. Cahen, H. Garty, and S.R . CaDlan, FEBSLett., 91, 131-134 (1978). . . (11) D. Cahen, Appl. Phys. Lett.. 33, 810-811 (1978). (12) J. F. McClellandand R. N. Kniseley, Appl. Opt., 15, 2967-2968 (1978). (13) . , D. Cahen. G. Hodes. and J. Manassen. J . Electrochem. Soc.., 125.. 1623-1628 (1978). (14) R. C. Gray and A. J. Bard, Anal. Chem., 5 0 , 1262-1265 (1978). (15) A . Rosencwaig, J . Appl. Phys., 49, 2905-2910 (1978). (16) F. A. McDonald and G. C. Wetssl, Jr., J . Appl. Phys.. 49, 2313-2322 (1978).
RECEIVEDfor review January 22,1979. Accepted May 17,1979.