Anal. Chem. 1980,
't
I
\
---._
__-350
450
550
(nmi
I
Flgure 3. Solid line = room-temperature fluorescence spectrum of fraction 15. Broken line = room-temperature fluorescence spectrum of benzo[ghi]perylene. Excitation wavelength in both cases was 300 nm.
I
1
I
I
i
52, 2095-2100
2095
bands appear which almost exclusively emanate from the emission bands of benzo[ghi]perylene. Only four peaks can clearly be pointed out as belonging to some other compound. This clearly illustrates the selectivity effect of the Shpol'skii fluorescence, which is observed in most cases of real sample analysis. The concentration of emission energies into small wavelength regions favors and facilitates the identification of certain compounds. This also leads to the fact t h a t many fractions have predominating Shpol'skii spectra originating from unidentified compounds which are only minor constituents of the PAH content of that particular fraction. The total result of analysis can be seen in Table I. T h e results of analysis by Shpol'skii fluorescence are presented and compared with the identifying and quantifying of the PAH content by glass capillary GC of each fraction. In conclusion, we have established that the identification of compounds in samples by aid of their quasilinear fluorescence can readily be made in those cases where a sample contains, among others, certain compounds t h a t fit the properties of strong quasilinear emission, even if the concentration of the compounds in question is relatively small.
ACKNOWLEDGMENT We thank Beryl Holm for reviewing the manuscript.
I
LITERATURE CITED
Flgure 4. Shpol'skii spectrum of fraction 15. Temperature = 63 K. Solvent = n-pentane. Arrows indicate emission bands not belonging to the spectrum of benzo[ghi]perylene. Nonselective excitation A,, I348 nm.
cence spectrum of the fraction, but identification is ambiguous a n d uncertain. If the same solution of fraction 15 is frozen t o 63 K, and the fluorescence spectrum is recorded (Figure 4) a considerable difference appears in the peak distribution and the relative intensities. About 40 well-defined emission
(1) "Proceedings of Third International Symposium on Chemistry and BiologyCarcinogenesis and Mutagenesis"; Jones, P., Leber, P., Eds.; Ann Arbor Science Publishers: Ann Arbor, MI, 1979; pp 83-452. (2) StruoDe. R. C.: Tokousbalides. P.: Dickinson. R. 6.: Wehrv. E. L. Anal. Chem. 1977, 49, 701. (3) Woo, C. S . ; D'Silva, A. P.; Fassel, V. A. Anal. Chem. 1980, 52, 159. (4) Colrnsjo, A. L.; Stenberg. U. Anal. Chem. 1979, 51, 145. (5) Colmsjo, A. L.; Stenberg, U. "Proceedings of Third International Symposium on Chemistry and Biology-Carcinogenesis and Mutagenesis"; Jones, P., Leber, P., Eds.; Ann Arbor Science Publishers: Ann Arbor, MI. 1979: D 121. (6) coimsio, .A.L.; Stenberg, U. m e m . scr. 1976, 9, 227 (7) Shpol'skii, E. V. Usp. Fir. Nauk. 1959, 68, 51.
RECEIVED for review March 12,1980. Accepted July 30,1980. This investigation was sponsored by the Swedish Natural Science Research Council (Contract No. 0369-100).
Comparison of Spectrofluorometry and Ion-Selective Electrode Potentiometry for Determination of Complexes between Fulvic Acid and Heavy-Metal Ions Robert A. Saar and James H. Weber' Department of Chemistry, Parsons Hall, University of N e w Hampshire, Durham, New Hampshire 03824
We analyzed the fluorescence properties of two fuivic acid samples, one derived from soil and the other from fresh water. A broad, featureless emission peak, which is unaffected by oxygen and KNO:, electrolyte, occurs at 445-450 nm upon excltation at 350 nm. Emission intensity reaches its maximum at pH 5 and drops off markedly toward low pH. Bound Cuz+ ion dramatically quenches the fulvic acid fluorescence; unbound metal ions (Cu2+, Pb2+, Co2+, Ni2+,and Mn2+) do not quench the fluorescence. We demonstrate a close relationship between fluorescence quenching by Cu2+ and fulvic acid complexation of Cuz+ as measured by ion-selective electrode potentlometry. With further development, complexation studies and stability-constant calculations for complexes containing fulvic acid and paramagnetic metal ions may be possible. 0003-2700/80/0352-2095$01 .OO/O
Complexation of heavy-metal ions by naturally occurring dissolved organic matter is widely studied because this complexation can change the transport pathways and toxicity of metal ions. I t is important t o know not only the types of complexes that form but also the strength of association between the metal ion and ligand. The association strength is commonly reported as a stability constant for the complex. Stability-constant calculations are frequently done by measuring the concentrations of uncomplexed metal ion in a solution containing known concentrations of total metal ion and total ligand (1-3). A stoichiometry or several stoichiometries may be assumed for the complexes. A second method of stability-constant calculation depends on measuring the amount of hydrogen ion released as complexes between metal 0 1980 American Chemical Society
2096
ANALYTICAL CHEMISTRY, VOL. 52, NO. 13, NOVEMBER 1980
ion and ligand form (4, 5 ) . The mathematics of stability constant calculations do not require the investigator to measure free metal ion concentration and then infer the concentration of complex. Many equations are symmetric, so that a measure of complexed or uncomplexed ligand would be equally suitable for determination of the stability constant (6). What is required is a way to distinguish complexed ligand from uncomplexed ligand. One spectroscopic property of naturally occurring organic matter t h a t may be measured is its fluorescence. Various researchers have demonstrated this fluorescence for organic matter derived from soils (7, 8) and waters (9-11). T h e fluorescence of organic molecules is often diminished (quenched) by heavy-metal ions, especially paramagnetic ones (12). The quenching of fluorescence from biological molecules is the subject of much research (13). However, relatively little work appears in the literature on the quenching of humic fluorescence by metal ions (14-16). We examined the fluorescence properties of two fulvic acid (FA) samples, one derived from a podzol soild (SFA) and the other derived from a freshwater river (WFA). Fulvic acid is the fraction of naturally occurring organic matter that is soluble in both acidic and basic solution. Work in this group during t h e last 7 years has provided us with detailed information about these fulvic acids: functional group and elemental analyses (I7), number-average molecular weights (18), the free radical nature (19),and the ability of these fulvic acids t o complex with Cu2+ (20), Cd2+ ( 3 ) , and Pb2+ (21, 22). Fluorescence measurements of uncomplexed ligand concentration in the presence of metal ion provides yet another perspective on fulvic acid systems.
EXPERIMENTAL SECTION Apparatus. We obtained fluorescence spectra with a Perkin-Elmer Model 204 fluorescence spectrophotometer. A Nucleopore polycarbonate filtration apparatus and 0.4-pm polycarbonate filters removed particulate matter that may have been present in the fulvic acid solutions. We measured pH with a Corning Model 476050 combination electrode which was attached to an Orion Model 701A pH/mV meter and performed ion-selective electrode titrations with an Orion Model 94-29 Cu2+ electrode. A Corning Model 476050 electrode served as the reference for the Cu2+electrode and also measured pH. Two Orion Model 701A meters allowed simultaneous recording of pH and metal ion concentration. A Shimadzu Spectronic 200UV spectrophotometer provided measurements of solution absorbance. Reagents. The characteristics of the two fulvic acids used in this study appear in earlier publications (17-19). Because we know their number-average molecular weights (644 for SFA and 626 for WFA), we can express SFA and WFA concentrations in moles per liter. The electrolyte was KNO,; we adjusted the pH of solutions with HNO, and KOH. The sources of metal ion were Fisher 1000 ppm atomic absorption standards. We prepared solutions of the model compound salicylic acid from Mallinckrodt crystals. Procedures. For nearly all experiments involving fulvic acid, we weighed out a sample of fulvic acid powder and dissolved it in 0.1 M aqueous KNOB. After adjusting the sample pH to the value of the coming experiment, we filtered it through the Nuclepore apparatus. Then, to 5.0-mL fulvic acid solution aliquots, we added various amounts of metal ion, depending on the mole ratio of total metal ion to total fulvic acid to be achieved. In all cases, the amount of metal ion added was less than the amount needed to cause precipitation of M2+-FA solids. We had determined the precipitation thresholds earlier (22). Any changes in fluorescence intensity were due to quenching not to the settling out of fluorescing material. We then adjusted the pH of the solution and added electrolyte until the total solution volume was 6.00 f 0.05 mL. The final FA concentration for all samples was 4.9-5.1 X M (approximately 32 ppm). The experimental temperature was 23 f 1 "C. Several types of experiments were necessary to characterize FA fluorescence.
100
I
AA PA
,a A
80
P
A
60
X
40
X X
20
X
0 0,o
2,o
400
6,O
0 8,O
PH Figure 1. Fluorescence of 5 X 10-5 M soilderived fuhic acid solutions with no metal ion (A) and with a threefold molar excess of Cu2+ (X).
(1) To determine the best wavelength for excitation and emission, we checked wavelengths from 320 to 500 nm. (2) The electrolyte might affect FA fluorescence. We checked M and FA fluorescence at two levels of electrolyte: 4.2 X 7.5 X M KNO,. (3) Oxygen is paramagnetic, and as such might quench FA fluorescence. We purged a pH 4.9 sample of SFA with nitrogen and compared its fluorescence intensity with an unpurged SFA sample. (4) To measure the effect of pH on FA fluorescence, we prepared series of SFA and WFA samples with pH values ranging from 1.5 to about 8 in steps of 0.3-0.5 pH units and measured the fluorescence for each. We corrected the values for variation in absorbance with pH changes. ( 5 ) We performed a series of titrations with various metal ions to see the effect each had on FA fluorescence. The ions included Cu2+,Pb2+,Cd2+,Mn2+,Co2+, and Ni2+. The goal of these titrations was to show whether the loss of fluorescence was specific to the amount of metal bound to FA. (6) Checking the correspondence between the quenching of FA fluorescence and the amount of bound metal ion required use of model compounds. Such a compound must fluoresce, and it must be possible to calculate the amount of complex formed at various pH values and metal ion concentrations. The most useful model compound was salicylic acid; we performed Cu2+,Co2+,and Ni2+ spectrofluorometric titrations with this ligand. We also performed ion-selective electrode titrations with Cu2+. The techniques for these titrations are reported elsewhere (21).
RESULTS AND DISCUSSION Movement of the emission peak to longer wavelengths as the excitation wavelength is lengthened and the presence of a broad, featureless emission peak indicate that FA is a mixture of fluorescing compounds (12). T h e maximum emission intensity occurred a t 445-450 nm upon excitation a t 350 nm. Neither oxygen nor electrolyte concentration affected the fluorescence emission of a n SFA solution. A change in electrolyte concentration, however, would affect the amount of metal ion complexation (5,23),which would alter the amount of fluorescence quenching in a solution containing fulvic acid and certain metal ions. The intensity of fluorescence emission from SFA and WFA varies with pH. T h e data (corrected for absorbance) represented by triangles in Figure 1 show the p H effect. Nearly identical results arise for WFA. T h e maximum emission
ANALYTICAL CHEMISTRY, VOL.
I
F
/ - t
80
I 1,5
metal ion
CMM/CSFA
Nil' Mn2
227 24 3
co2+
227
Pb2
100
V 40 0,5 20
0
0,o 0,O
1,0 2,O
3,O
4,O
5,O
+
65
CU2+
209
Cd2
119
pH
% quenchedb
1.45 1.45 1.41 1.44 1.47 1.42
FA
Flgure 2. Comparison of Cu2+ titrations into soiklerived futvic acid (5 X M) as performed by fluorescence spectrometry (A, left-hand scale) and by ion-selective electrode potentiomtry (X, righthand scale). 3 equals bound Cu2+ concentration divided by total SFA concentration.
intensity occurs a t about p H 5, with a fairly steep drop in intensity as the p H drops below 4. Above pH 5, there is a gentle decline in emission intensity. Williams noted that the degree of protonation of compounds with one or several functional groups (-OH, -COOH, -0Me) dramatically altered the fluorescence intensity (24). Phenolic and carboxylic acid groups are prevalent in our SFA and WFA (17). Futhermore, the various compounds in Williams' survey had maximum fluorescence intensity a t widely differing p H values. It appears that our fulvic acids are heavily weighted with substituted aromatic compounds t h a t do not fluoresce a t low pH, hence the major reduction in fluorescence intensity a t low pH. Multiple deprotonation of fulvic acid fluorophores also reduces the emission intensity. Of the metal ions examined, Cuz+ is most efficient a t quenching fulvic acid fluorescence. This observation agrees with eariler work (14, 16). We further observe that the quenching by Cu2+increases with increasing pH; data showing this effect are represented by crosses in Figure 1. Almost no quenching occurs a t low p H (pH 2.1 and below), and nearly quantitative quenching occurs a t p H 6.0. Earlier work in this laboratory (3,20,22) and elsewhere (2) shows that complexation of divalent metal ions increases with increasing pH. We wished, then, to see if the effect of pH was the same on Cu2+ quenching of FA fluorescence as it was on Cu2+complexation by FA. It became necessary to do fluorescence and ion-selective electrode titrations with Cu2+titrant at several pH values. The ionic strength (0.1 M), FA concentration (5 X 10" M), and temperature (23-25 "C) were the same for the two types of experiments. Figure 2 shows the results for pH 3.0, 4.0, 5.0, and 6.0 titrations. The x-axis variable is the mole ratio of total Cu2+ ion to total SFA (CCJCSFA). This variable is a measure of the progress of either type of titration. The y-axis variable for the fluorescence titrations (left) is the percentage of the original, metal-free fluorescence remaining a t specified p H and CcU/CsFA values. The y-axis variable for ion-selective electrode titrations (right) is D, which is the number of bound Cu2+ ions per fulvic acid molecule. i~also equals (Ccu [cu2+])/c,,A, where [cu2+]is the free c u 2 +concentration as measured by the ion-selective electrode.
0.9 -0.5'
1.8 3.8
3.5 -0.7 a Fulvic acid concentration was 5.1 X M. The experimeiits were performed in 0.1 M KNO, at 23 "C. Uncertainty in % quenched values is * 2%. Negative values indicate fluorescence intensity greater for SFA with metal ion than for the reference sample that had no metal ion. +
The fluorescence and ion-selective electrode results coincide closely: a t all pH and CcU/CsFA values, the proportionality between percentage quenched and 3 is 57. T h a t is
57n = percentage quenched CCU%
2097
Tabie I. Quenching of Soil-Derived Fulvic Acid Fluorescence by High Concentrations of Metal Ions at Low p H a
+
60
52, NO. 13, NOVEMBER 1980
(1)
The proportionality constant 57 has no physical meaning; it depends on the units used to express fluorescence quenching. The results in Figure 2 suggest two conclusions: that a metal ion must be bound to fulvic acid in order to cause quenching and possibly that a bound metal ion quantitatively quenches a nearby fluorophore. To verify the f r s t conclusion, one must calculate whether FA molecules and metal ions collide to a substantial degree during the lifetime of the fluorescence excited state. A double exponential corresponding t o lifetimes of 1.3 and 7.0 ns closely fits the excited-state decay pattern for our SFA (25). For Cu2+and SFA M a t 23 "C, little concentrations of approximately 1 X collisional quenching would occur (12). It appears that fluorescence quenching is selective for bound metal ion at the concentrations of these experiments. Further verification that unbound metal ions do not cause appreciable quenching of fulvic acid fluorescence comes from experiments done a t pH 1.4-1.5 with high mole ratios of metal ion to FA. The high hydrogen ion concentration of this experiment prevents divalent metal ions from binding to FA, so any quenching is due to free metal ion. The results are listed in Table I. Except for Pb2+ and Cu2+,there is no significant quenching of SFA fluorescence by metal ions. The small amount of quenching by Pb2+and Cu2+is not a problem for the usual titration study done at higher pH, because the amount of metal ion in the samples listed in Table I is 1 to 2 orders of magnitude greater than are used in titrations. Use of salicylic acid as a model for fulvic acid helped to verify the second possible conclusion-that a bound metal ion completely quenched a nearby fluorophore. We measured the fluorescence of a series of pH 6.0 solutions, all containing Cu2+ (except the metal-ion blank) and salicylic acid in 0.1 M KN03, but each sample had a different mole ratio of total Cu2+to total salicylic acid. Values in the literature (26) for the Cu2+-salicylic acid stability constant allowed calculation, for any concentration of Cu2+,of the fraction of total salicylic acid t h a t was attached to a Cu2+ ion. T h e calculated and fluorescence results appear in Figure 3. The correspondence of these results shows that a salicylic acid molecule attached t o a Cu2+ ion does not fluoresce. T h a t is, bound Cu2' quenching is complete. Cu2+is not the best ion with which to test the capability of fluorescence analysis. Its performance in fluorescence tests is unusually good: Cu2+ is paramagnetic and hence, i t quenches fluorescence efficiently. It also forms strong complexes with fulvic acid; the close association of metal ion and ligand is necessary for quenching. We tested other metal ions
2098
ANALYTICAL CHEMISTRY, VOL. 52, NO. 13. NOVEMBER 1980
loo
3
=3 z
w
Y
0
m
X
I
brp
A
l x
P
I w
0 0
I
I I
I
5
10
15
20
CCU'CSALI C Y L I C ACID Flgure 3. Comparison of the percentage of salicylic acid fluorescence quenched by Cu2+ (A) and the calculated percentage of salicylic acid complexed with Cu2+ (X) at pH 6.0. The salicylic acid concentration is 1.5 X lO-'M.
that were not paramagnetic or did not bind strongly to FA to gauge the sensitivity of fluorescence analysis. A comparison of fluorescence and ion-selective electrode results for Pb2+-SFA complexes appears in Figure 4. Comparably poor results arise for Pb2+-WFA complexes. There are several differences between the data in Figure 4 and the data for Cu2+-SFA complexes shown in Figure 2. First, the P values are scaled by a factor of 20 in order to have the ion-selective electrode results coincide roughly with the percentage-quenched numbers from the fluorescence experiments. Earlier work (21)showed that a t low mole ratios of total metal ion to total fulvic acid CMM/CFA, Cu2+ and Pb2+ bind with approximately the same strength to fulvic acids. The smaller scaling factor for Pb2+compared with that for Cu2+indicates that each bound Pb2+ ion does not quench as much fluorescence as does each bound Cu2+ion. The ratio of the D scaling factors for Cu2+ and Pb2+ (57/20) is the ratio of fluorescence-quenching efficiency for the two metal ions. The probable reason for the smaller Pb2+factor is that Pb2+ is diamagnetic; it does not quench fluorescence as effectively as does a paramagnetic metal ion. It is also conceivable that Pb2+binds less tightly t o the same FA sites than Cu2+does or binds to different sites-sites farther from fluorophores and thus less able to quench their fluorescence. There is a second major difference between the Pb2+-SFA data in Figure 4 and the Cu2+-SFA data in Figure 2. The maximum mole ratio of metal ion to FA for the Pb2+ experiments is 1.0 because precipitation of Pb-SFA solids occurs a t higher ratios (21,22).Mole ratios of Cu2+to FA can reach 4 or 5 before precipitation begins, so it is possible to obtain a much wider range of Cu2+data. The scattered appearance of the data for Pb2+plotted in Figure 4, in contrast to the orderly data for Cup+in Figure 2, results from the two differences discussed above: the quenching per bound Pb2+ion appears to be only one-third that per bound Cu2+ion, and the amount of Pb2+that can be added is only one-fifth of the amount of Cu2+that can be used. Considering the magnification of the axes, the scatter in the Pb2+figure is not surprising nor does it weaken the conclusions reached for Cuz+. Because fluorescence analysis does not work well with the Pb2+-fulvic acid system, complexes with other diamagnetic
ions and FA would probably not be analyzed easily by fluorescence. Pb2+,among diamagnetic metal ions, is probably the most amenable to fluorescence analysis because it is very heavy (high atomic weight, in addition to unpaired electrons, is a factor promoting quenching) and binds strongly to fulvic acid. The possibility of high C M / C F A without precipitation is one advantage that some diamagnetic ions might have over Pb2+,but this advantage may not be enough to overcome the problems such a diamagnetic ion might have-inefficient quenching and weak complexation. T o confirm our conclusions about diamagnetic ions, we performed a fluorescence titration with Cd2+and SFA. Any usable results from fluorescence experiments with Cd2+could possibly be compared to our Cd2+-FA titration data obtained with an ion-selective electrode ( 3 ) . Howver, a t pH 7.5, there is no measurable SFA fluorescence quenching by Cd2+,even up to C a / C S F A = 15. Ion-selective electrode experiments show that an easily measurable amount of SFA is bound to Cd2+ a t pH 7 or 8 (3),so the fluorescence experiment is not useful for analysis of complexes between Cd2+and fulvic acid. Complexation of paramagnetic ions by humic substances appears to be the chief potential application of this type of fluorescence analysis. Two such metal ions for which there are no ion-selectiveelectrodes are Co2+and Ni2+. On the basis of stability constants they form the model ligands such as salicylic acid (26), Co2+ and Ni2+ should form moderately strong complexes with fulvic acid. A comparison of SFA fluorescence without metal ion and SFA fluorescence with either Co2+or Ni2+as a function of pH appears in Figure 5. This figure shows results of experiments that are like earlier ones done for Cu2+and SFA (Figure 1): fluorescence diminishes in the presence of any of these metal ions as the pH is raised. Co2+and Ni2+ cause nearly the same amount of quenching and somewhat less than that caused by Cu2+. It is not surprising that Co2+and Ni2+cause the same response in the presence of SFA; their ionic radii and hydrolysis constants are very similar. If quenching of a fluorophore by a bound Co2+ or bound Ni2+ ion is complete, then the relatively small amount of quenching caused by Co2+and Ni2+ indicates that they bind to SFA less strongly than does Cu2+. Figure 6 shows pH 4.0 and pH 6.0 titrations of Co2+and SFA; similar results arise for Ni2+ and SFA. This figure, like Figure 5, shows the increased quenching and presumably increased complexation a t pH 6.0 compared t o that a t pH 4.0. It remained to be seen whether the bound Co2+ or Ni2+ caused complete quenching. Experiments with salicylic acid and either Co2+or Ni2+ appear to confirm this expectation, but the stability constants for Co2+and Ni2+ complexes with salicylic acid are low and the pH and metal ion concentrations had to be very high t o cause a measurable amount of complexation. Such high metal-ion concentrations required us to correct for collisonal quenching by the metal ions. We did this by measuring the quenching a t pH 5.0, where virtually no complexation occurs. We performed the complexation measurements near pH 8. Salicylic acid, therefore, does not provide a very sensitive experiment. A more rigorous test of Co2+ and Ni2+ effects on fulvic acid fluorescence awaits a model compound that has a strong fluorescence and a higher affinity for Co2+and Ni2+than salicylic acid has, a t least in relation to its proton affinity. Also, the complex formed should not have an ultraviolet or visible spectrum that intereferes with the fluorescence analysis. Such an interference arose for 5-nitrosalicylic acid, a model compound we tried. The data included in Figure 6 can be used for calculation of conditional stability constants for Co2+-SFA if a proportionality constant between P and the percentage quenched is determined. If we assume that the proportionality constant
ANALYTICAL CHEMISTRY, VOL. 52, NO. 13, NOVEMBER 1980
20
1,o
16
0,8
12
0,6
50
D
w
U x
v
c 3
8
b?
I
40t -
z w
0,4
2099
X
X
x
8 I V
z 3 w a be
P
4
A
082 A
080
0,3
0,6
1.2
0,9
0
CP~CSFA Figure 4. Comparison of Pb2+ titrations into soilderived fulvic acid (5 X 10" M) as performed by fluorescence spectrophotometry (A,left-hand scale) and by ion-selective electrode potentiometry (X, right-hand scale). P equals bound Pb2+ concentration divided by total SFA concentration.
*
A
A A A
80
5
B
W u
B V
w
60
01
0
-I 3 LL W
2
4
8 CCO'CS
12
16
FA
Figure 6. Fluorescence titrations at pH 4.0 (A) and pH 6.0 ( X ) for Co2+ and soilderived fulvic acid (5 X M).
as expressed by B. For CoZ+-SFAand Ni2+-SFA, fluorescence and ion-selective electrode analysis may not be measuring the same complexation phenomenon as these two techniques appear to do for Cu2+-SFA. Fluorescence, then, may provide a valuable different view of the interaction between metal ions and organic matter. However, the possibility that fluorescence quenching and B may be directly proportional for a variety of paramagnetic metal ions should not be discarded. The rewards for determining such proportionality constants are generous: fluorescence could allow complexation studies with paramagnetic ions for which no ion-selective electrodes exist. Also, the attempts to tailor a model compound to the acid-base, metal complexation, and fluorescent properties of fulvic acid may provide further understanding of the molecular structures in fulvic acid.
100
3
A
0,D
0
9
A
40
E w
20
LITERATURE CITED
n
U
0,o
2,o
4,O
6,O
8,O
PH Figure 5. Fluorescence of 5 X M soilderived fuhric acid solutions with no metal ion (A), with a threefold molar excess of Co2+ (X), and with a threefold molar excess of Ni2+ (0).
is 57 (as it is for Cu2+),and further assume simple 1:l complexation
w++ FA2-
K
MFA ( 2) the log K a t B = 0.2 is 2.8 a t pH 4.0 and 3.8 a t 6.0. The same values arise for Ni*+-SFA complexes a t the same two pH values. These conditional stability constants are smaller than those reported by Schnitzer and Hansen (27) for Ni2+-SFA complexes: log K = 3.1 a t pH 3.0 and log K = 4.2 a t p H 5.0. Our Co2+-SFA and Ni2+-SFA conditonal stability constants are also smaller than Cd2+-SFA conditional stability constants t h a t we determined by ion-selective electrode titration (3). On the basis of stability constants with simple organic ligands (26),we would expect Co2+and Ni2+to form stronger complexes with fulvic acid than Cd2+does. What appears likely is that the proportionality constant of 57 that applies to Cu2+ does not apply to Ni2+or Co2+. Indeed, for Ni2+or Co2+,there may be no proportional constant between percent of SFA fluorescence quenched and the amount of complex formed
(1) Buffle, J.; Greter, F.-L.; Haerdi, W. Anal. Chem. 1977, 4 9 , 216-222. (2) Takamatsu, T.; Yoshida. T. Soil Sci. 1978, 125, 377-386. (3) Saar. R . A.; Weber, J. H. Can. J. Chem. 1979, 5 7 , 1263-1268. (4) Stevenson, F. J. Soil Sci. SOC. Am. J . 1978, 4 0 , 665-672. (5) Stevenson, F. J. Soil Sci. 1977, 123, 10-17. (6) Rossotti, F. J. C.; Rossotti. H. "Determination of Stability Constants and
other Equilibrium Constants in Solutions"; McGraw-Hill: New York, 1961. (7) Seal, 8. K.; Roy, K. B.; Mukherjee. S.K. J . Indian Chem. SOC.1964, 4 1 , 212-214. (8) Mueller-Wegener, U. 2. Pflanrenernaehr. Bodenkd. 1977, 140, 563-570. (9) Black, A. P.; Christman, R . F. J.-Am. Water Works ASSOC.1963, 55. 753-770. (10) Ghassemi, M.; Christman, R. F. Limnol. Oceanogr. 1968, 13,583-597. (11) Smart, P. L.; Finlayson, B. L.; Rylands, W. D.; Ball, C. M. Water Res. 1976, IO, 805-811. (12) Parker, C. A. "Photoluminescence of Solutions with Applications to Photochemistry and Analytical Chemistry"; Elsevier: Amsterdam, 1968. (13) Chen, R. F. in "Biochemical Fluorescence: Concepts"; Chen, R. F., Edelhoch, H.. Eds.; Marcel Dekker: New York, 1976; Vol. 2, Chapter 13. (14) Banerjee, S. K.; Mukherjee, S. K. J . Indian SOC. Soil Sci. 1972, 2 0 , 13-18. (15) Lgvesque, M. Soil Sci. 1972 7 73, 346-353. (16) Cline, J. T.; Holland, J. F. ERDA Symp. Ser. 1977, 42, 264-279. (17) Weber, J. H.; Wilson, S.A. Water Res. 1975, 9 , 1079-1084. (18) Wilson, S. A.; Weber, J. H. Chem. Geo/. 1977, 9 , 285-293. (19) Wilson, S. A.; Weber, J. H. Anal. Lett. 19778 70, 75-84. (20) Bresnahan, W. T.; Grant, C. L.; Weber, J. H. Anal. Chem. 1978, 5 0 , 1675-1 679. (21) Saar. R. A.; Weber, J. H. Geochim. Cosmochim. Acta, in press. (22) Saar, R. A.; Weber, J. H. Environ. Sci. Techno/. 1980, 14, 877-880. (23) Gamble, D. S.;Can. J. Chem. 1973, 57,3217-3222. (24) Williams, R. T. J . R . Inst. Chem. 1959, 8 3 , 611-626. (25) Seitz, W. R., personal communication, University of New Hampshire, 1980.
2100
Anal. Chem. 1980, 52, 2100-2104
(26) Martell, A. E.: Smith, R. M. "Critical Stability Constants: Other Ligands"; Plenum: New York, 1977, Vol. 3, p 186. Schnitzer, M.; Hansen, E. H. Soil Sci. 1970, 709, 333-340.
Organic
(27)
RECEIVED for review May 14,1980. Accepted August 14,1980.
National Science Foundation Grant OCE 79-10571 partly supported this work, which was presented to the Division of Environmental Chemistry at the 180th National Meeting of t h e American Chemical Society, Las Vegas, NV, Aug 1980.
Flow-Through Electrode Unit for Liquid Membrane Electrodes D. S. Papastathopoulos, E. P. Diamandis, and T. P. Hadjiioannou' Laboratory of Analytical Chemistry, University of Athens, 104 Solonos Street, Athens (144), Greece
The construction of a flow-through electrode unit for the preparation of liquid membrane electrodes is described. The construction is based on the use of a triangular piece of glass frit embedded in a plastic block. The glass frit serves as an inert matrix holding the electroactive liquid phase. A hole drilled through the frit forms the sensing path for the test solution channeling. The unit is equipped with an internal reference solution-Ag/AgCI electrode system and a reservoir for ion exchanger storage. Flow-through liquid membrane electrodes for calcium, nitrate, perchlorate, and picrate Ions are prepared by using conventional Ion exchangers for the evaluation of the constructed unit. The electrodes are tested for their stability, reproducibility, response time, and worklng range under various continuous-flow conditions.
Ion-selective electrodes and gas-sensing probes have already been employed for continuous-flow measurements in industrial process control, biomedical analysis, and environmental or pollution research (14). Various types of automatic or semiautomatic sytems are now available, using such sensors under the basic configurations of the flow-through cells and the sensor end-cups. In the majority of these applications regular dip-type sensors are used, where the sample solution is flowing across the sensing membrane surface (7). Nevertheless, this design may cause a series of problems, such as, air bubble trapping, easy clogging from particulate matter, instability of t h e measured potential, and development of streaming potentials due to the improper geometry of the flow-through cell (8). Also, in the case of liquid membrane electrodes the potential response becomes erratic a t high flow rates, owing to distortion of the membrane by the high pressure of the stream ( 1 , 7). Most of these drawbacks can be obviated by using a design where the sample solution is channeled through t h e sensing membrane. Flow-through sensors based on this configuration have been developed recently, but so far these have only been of t h e solid-state or glass capillary type membrane electrodes (8-10). T h e present paper describes the construction of a new flow-through electrode unit for the preparation of liquid membrane electrodes. The design is based on the use of a piece of fine porosity glass frit embedded into a polymeric plastic. T h e glass frit serves as an inert matrix which holds the electroactive liquid phase. A hole drilled through the glass frit forms the sensing path for the sample stream channeling. T h e unit has been successfully used for the preparation of several flow-through ion exchange membrane electrodes.
EXPERIMENTAL SECTION Chemicals. All chemicals used were of reagent grade. Stock solutions of ions and ionic strength/pH adjustors were prepared 0003-2700/80/0352-2100$01.OO/O
by using deionized, distilled water. Liquid ion exchangers and the corresponding internal reference solutions for the nitrate, perchlorate, and calcium ion electrodes (6) were obtained from Orion Research Inc. The liquid ion exchanger for the picrate ion electrode consisted of tetrapentylammonium picrate in 2-nitrotoluene, and it was prepared as described elsewhere (11).A solution of 0.01 M sodium p i c r a M . 1 M sodium chloride was used as the internal reference solution. Apparatus. A schematic diagram of the automated analysis flow system employed for the evaluation of the constructed flow-through electrodes is shown in Figure 1. Sampler I1 and proportioning pump I11 were from the Technicon Auto-Analyzer I1 system. The timing functions of sampler I1 were controlled by a specially designed digital "sample-wash timer" connected to sampler I1 as described elsewhere (12). The sample-reagent metering system on pump 111, consisting of Solvaflex calibrated tubes, was the same for all of the evaluated ion selective electrodes except for the perchlorate electrode. The constructed flow-trhough electrodes were used in conjunction with an Orion 90-01 single-junction reference electrode. All potentiometric measurements were carried out with a Corning Model 12 Research pH/mV meter and recorded on a HeathSchlumberger SR-255 B potentiometric recorder. Flow-Through Electrode Unit Construction. The final configuration of the constructed flow-through electrode unit is shown in Figure 2. Glass frit disks of fine porosity (nominal maximum pore size of approximately 5 km and thickness of about 2.5-3 mm) were obtained from filtering glass crucibles. An equilateral triangular piece (side length approximately 1 cm) was cut from the disk with a fine hacksaw blade, and the cut sides were smoothed with fine emery paper. All sides of the entire triangular piece were covered with a thin layer of silicon rubber and embedded by using a circular mold in Serifix or Epofix resin, a casting plastic obtained from H. Struers, Scientific Instruments, Copenhagen. The silicon rubber layer serves as a protective film to prevent the absorption of the plastic into the pores of the glass frit. The embedding was made by initially preparing a supporting layer that filled half of the mold and allowing it to harden. Then, the triangular piece of the glass frit was placed on the supporting layer and the mold was filled to the top with a covering layer of plastic. The filled mold was allowed to harden completely for at least 24 h. The resultant Cylindrical cast block (3 cm in diameter and 2 cm high) was ground and polished by using various sizes of emery paper. Appropriate diameter holes were made on the cylindrical surface of the block, as shown in Figure 2, by relatively slow speed drilling with regular drill bits. A flow-through channel 1.6 mm in diameter was also drilled on the longitudinal axis of the block crossing the glass frit, in order to form the electrode sensing path. The side filling and deaeration holes were drilled in such a way as to form an angle of about 60". The upper section of the fiing hole was sufficiently enlarged and stuffed with cotton or glass wool, thus serving as a reservoir for the electroactiveliquid phase. The internal reference electrode compartment was made of a Plexiglas tube (1cm 0.d.) 5-6 cm long, which was mounted into a hole drilled between the side filling and deaeration holes and sealed with Araldite epoxy resin glue. The junction between the internal reference solution and the liquid electroactive phase, 0 1980 American Chemical Society
