Environ. Sci. Technol. 1980, 20, 1047-1050
A New Model Describing the Adsorption of Copper on MnO, Simcha Stroes-Gascoyne,*iLQJames R. Kramer,$ and William J. SnodgrassQ
Departments of Geology and Civil Engineering, McMaster University, Hamilton, Ontario L8S 4M 1, Canada The use of 6-Mn02in determining conditional stability constants (CSC) for trace metal-organic complexes depends upon the precise modeling of trace metal uptake by 6-Mn02. Cu adsorption deviates from Langmuir linearity at low surface coverages due to a nonconstant binding energy. An implicit Langmuir expression, based upon surface complexation theory, has the form A
acts
from which CC (L,) is obtained from the slope and CSC (K'Cu-L) from the intercept. The Langmuir adsorption isotherm
r, = rm
[CU2+]
1/B
+ [Cu2+]
or its linearized form
ru+in
Adsorption isotherms for various 6-Mn02batches, obtained in the pH range 6-8.5, fit the implicit model well. CSC values obtained for Cu-NTA by using this model agree with literature data. Introduction The 6-Mn02method (1-3) was developed to determine complexing capacities (CC) and conditional stability constants (CSC) for trace metal-organic complexes in the pH range 6-8.5. This method has been shown to be more applicable in this pH range compared to other methods ( 4 ) . The 6-Mn02 method, however, has shortcomings, which, if not corrected for, will produce false CSCs. This in turn will lead to incorrect trace metal speciation calculations. A modified model for Cu adsorption on 6-Mn02 is developed and tested in this work. The 6-Mn02method, as originally proposed (I), consists of adding a small amount of synthetic 6-Mn02to a water sample. (The 6-Mn02has previously been calibrated for Cu uptake at constant pH and ionic strength.) Cu is then added to the sample, and there is a partitioning between solution and substrate. After a suitable equilibration time, a sample is taken, filtered, and acidified, and total dissolved Cu (cud) is measured by anodic stripping voltammetry. The titration is continued by adding more copper. CC and CSC are then obtained by manipulation of mass balance and equilibrium expressions: [Cud] = [Cu2+]+ [Cu-L]
+ [Cu-hydroxides] + ...
[cut] = [cud] + [cual [L,] = [CU-L] + [L] Kbu-L = [Cu-L]/([Cu2+][L])
(1) (2)
(3)
(4)
where cud = total dissolved c u , cu2+= free c u , cu-L = Cu-ligand complex, Cut = total copper added, Cu, = Cu adsorbed on MnO,, L, = total ligand, and L = free ligand. Other complexes may be eliminated by sample preparation (degassing of COz)or may be calculated. It is assumed that there is one ligand, L, only. Combining the above equations and rearranging give [CU2+] [CU2+] 1 -(5) [CU-LI Khu-L[Ltl [Ltl
+-
*Address correspondence to this author at the Atomic Energy of Canada Ltd., Whiteshell Nuclear Research Establishment, Pinawa, Manitoba ROE 1L0, Canada. *Department of Geology. f Department of Civil Engineering. 0013-936X/86/0920-1047$01.50/0
in which Fa, rm, and B are the amount of Cu adsorbed, maximum value for Cu adsorption, and binding constant, respectively, was used to calibrate the uptake of Cu2+by 6-Mn02 ( I ) , even though a Langmuir expression should be applied only for constant charge surfaces. In the procedure, [Cud] is measured, and from eq 2, [Cu,](I',) is determined; eq 6a is then used to calculate [Cu2+],and values of [Cu-L] are determined from eq 1; eq 5 is then used to determine CC and CSC. Equation 5 shows that even if values for [Cu2+]are incorrect, the value for [L,] can be determined accurately, provided that [Cu-L] values are correct. [Cu-L] values are calculated from eq 1,and at pH values >6, Cu2+concentrations are usually much less than cud and cu-L concentrations. Thus, a test of [L,] is no guarantee that the correct CSC has been determined. The 6-Mn02method, therefore, produces valid CSCs only if the Langmuir equation correctly reflects the adsorption process. Another requirement is that the organic ligand does not adsorb on the 6-Mn02 surface. Measurements with 14Clabeled NTA, glycine, and aspartic acid did not indicate substantial adsorption of these ligands on 6-Mn02 during Cu titration ( 4 ) , and van den Berg (5) has also argued against the adsorption of humics on 6-MnOz. Various batches of 6-Mn02were prepared following the prescribed recipe (1). Adsorption experiments were carried out with emphasis placed on low concentrations of Cu. The adsorption parameters, rmand B, determined for these surfaces did not agree with early values. The differences in results are illustrated in Figure 1 for rm and in Figure 2 for B. During this study it was also noted that the linearized plot (eq 6b) displayed a clear and consistent deviation from linearity at low Cu concentrations. An example is shown in Figure 3. This deviation implies a nonconstant value for B throughout the isotherm. During titrations of natural waters, Cu concentrations need to be kept small in order to avoid precipitation and to be affected by ligands. Coverage of the 6-Mn02 is therefore low, and it is in this region of coverage that nonlinearity of eq 6b occurs. Ignoring this effect leads to overestimation of [ Cu2+]and hence underestimation of
csc.
CSCs determined for Cu-NTA in the previous study (1) were an order of magnitude below literature data. This is likely due to the nonlinearity of eq 6b. In addition, other Cu-organic CSC values reported in ref 1 could also be affected by the nonlinearity. The difference in values of the adsorption parameters of this study and previous work (Figures 1 and 2) is probably due to precipitation of Cu in the previous study
0 1986 American Chemical Society
Environ. Sci. Technol., Vol. 20, No. 10, 1986
1047
g -
1.o
1
van den Berg (1979) Stroet-Gascoyne (1983)
0.8-
0 0
LE$
0
060.4 .
0.2~~
6.5
6
7
7.5
8
8.5
9
PH
Figure 1. Comparison of the adsorption capacity rm for Cu adsorption on 6-Mn0, as obtained by van den Berg ( 7 ) and in this work.
1
/
-e- van den Berg (1979) Stroes
- Gascoyne(l9831
9.5
reported deviation from linearity for low coverages of Co and Zn and of Zn, respectively. Loganathan and Burau ascribed the deviation to a second adsorption site on the Mn02 surface. Attempts were made to fit the adsorption data obtained in this study to a double Langmuir expression ( 4 ) . This model accommodates two classes of sites with different binding energies and predicts deviation from linearity, if the binding ehergies are substantially different. However, fitting of numerous adsorption data to a double Langmuir model by using nonlinear regression did not result in a particular pattern for the model parameters. Moreover, model fit was hard to obtain in a considerable amount of cases, because the data basically described different functions (4). It was therefore concluded that a second adsorption site does not adequately explain the data. Points pertinent to the development of the modified model are that (1)the adsorption process is not unique, Le., various adsorption mechanisms are possible, and (2) as adsorption progresses, the surface charge and hence B change. Adsorption of Cu on 6-Mn02can be represented by any or all of the following expressions (9, IO): (=MnO(H),) + [Cu2+]8 (=MnO CU@-~)+) + x[H+]
(7)
( ~ M n 0 - 1+ [Cu2+]+ H 2 0 + (=MnO CuOH) + [H+] (8) (=MnO-) 6
6.5
7
7.5
8
8.5
9
+ [Cu2+]
(=MnO Cu+]
(9)
Adding these equations yields
PH Ftgure 2. Comparison of the binding constant, 6,for Cu adsorption on 6-Mn0, as obtained by van den Berg ( 7 ) and in this work.
+
+
+ +
Ka
(=MnO(H),] 2(=Mn0-] 3[Cu2+] H,O r (=MnOCu(2-r)+) (=MnO CuOH) (EMnO Cu+) (1 + Z)[H+I (10)
+
+
Assuming that rm - ra= free sites = (=MnO(H),) + 2(= MnO-1, that ra= the occupied sites = (=MnO C U ( ~ - ~+) + ) (=MnO CuOH) (=MnOCu+),that [Cu2+]is the solution concentration of Cu in equilibrium with the occupied sites, that [H+]”(n = 1 x) represents the [H+]ions involved, and that [H20] = 1, eq 10 can be rearranged to
+
+
OF-
I 10
[cU2+]
20
rm[Cu2+] ru+in
30
lo9 M
Flgure 3. Linearized Langmuir isotherms for 6-Mn02, batch 1, at a pH of 7.5, showing the nonlinearity at low Cu concentrations. Different symbols represent duplicate isotherms.
because Cu additions between 10 and 100 pM were used (1). Calculations performed by using REDEQLZ (6) on the previous data suggest that this is the case, and the amount of precipitate increases with increasing pH. The precipitated Cu would be measured as adsorption, thus increasing rmwith pH and influencing B. It is also inherent to the form of the adsorption isotherm that the lower adsorption values more strongly influence the estimation of B, while the higher values affect rm. Equal distribution of data points along the isotherm is therefore necessary. In addition, modifying the isotherm to a linear form tends to overestimate the value for rmand underestimate the value for B. Modified Model Deviations from Langmuir linearity, as shown in Figure 3, imply that the binding energy is not constant. Loganathan and Burau (7) and Gabano et al. (8) previously 1048
ra=
Environ. Sci. Technol., Vol. 20, No. 10, 1986
Schindler et al. (9) derived an equation that corrects for the effect of changing surface charge upon the activity of surface groups on an oxide. Applying this correction to eq 11 yields
in which rl, = difference between the surface potential and the bulk solution potential, K’, = “surface equilibrium” constant, related to the maximum number of available surface sites, which is here assumed to be constant and independent of pH, and x = assumed constant and independent of pH (i.e., the reactions shown in eq 7 have a constant ratio). Surface potential, +, and surface charge, u, are related by capacitance (C) which is assumed constant (11): lJ
while
u
= u/C
(13)
depends on the relative amount of specific ad-
Table I. Model Parameters for Adsorption of Cu on 6-Mn02 Using the Implicit Langmuir Expression sample 6-Mn02 size used 1 2
95
3
112
97
b (=log B )
rm f lo
n
f
lo
*Model predictions showing deviation from linearity
f lo
0.24 f 0.005 1.85 f 0.05 -4.78 h 0.43 0.25 f 0.006 1.87 f 0.06 -4.83 f 0.42 0.20 f 0.008 1.66 f 0.06 -3.12 f 0.45
LA 10
OO
- Model
.
n 0.34
30
M
Flgure 5. Linearized implicit Langmuir model predictions for 6-Mn02, batch 1 at a pH of 7.5.
I
A
20
[ c U 2 + i x 10'
A experimental data
11-
5
10
[cu2+i '01
15
-;
M
10-
Figure 4. Verification of the implicit Langmuir model for 6-Mn02, batch 1, with independent data, at a pH of 7.5.
3
u_
k
-8 9 -
sorption on the surface, Fa/Frn, For this simple model it is not important exactly where the plane of adsorption is located. Keeping this in mind, it is proposed here that the term K\e-(2-x)+/kTin eq 12 be replaced by the term Be(l-ra/rm), which yields
6
rrn[cu2+] (14)
IH+1"
0-
+ [CU2+]
Be(l-ra/rm)
in which [H+]"/ [Be(l-ra/rm)] represents the binding energy that varies with pH and degree of coverage of the surface. It is noted that to make such a substitution it is argued ( 4 ) that the effect of pH on the surface charge at constant ionic strength is linear at pH values a few units away from the zero-point-of-charge (11,121. In addition, if ion pair formation occurs between electrolyte ions and surface groups as proposed by Davis (111, it is assumed to have a linear effect on the surface charge. These two linear effects are combined in B. Equation 14 shows that if adsorption of [Cu2+]is minimal, the overall binding energy term is minimal (strongest adsorption). At almost complete surface coverage, the binding energy term is maximum and approaches a constant value. The modified expression (eq 14) is called the "implicit Langmuir model" because the adsorption parameter, rrn, and the pH are now an implicit part of the binding energy term.
Parameter Estimation and Testing of Model Various batches of 6-Mn02were prepared at different times during the study. The batches were kept separate because they showed small differences in Cu adsorption (4).For each of the batches of 6-Mn02, a number of isotherms were obtained in the pH range 6 I pH I 8.5 and fixed ionic strength (0.01 M KNOB). Seven to ten data were obtained for each isotherm. All adsorption data for one batch of 6-Mn02were used to obtain the parameters in eq 14 by using nonlinear regression. Table I summarizes the model parameters n, b (=log B ) , and rmobtained for the various batches of 6-Mn0,. In order to verify the model, different adsorption isotherms were obtained for
7
8
PH Flgure 8. Summary of conditional stability constants for Cu-NTA. (0) Data from this study, obtained with the implicit Langmuir function. (I) Literature data range (13, 14). (A)Original study (I), using the simple Langmuir function. (M)This study using the simple Langmuir function.
batch 1 and compared with model predictions using eq 14. Figure 4 shows that there is an excellent coincidence of model predictions and observations, and Figure 5 illustrates how eq 14 predicts the deviation from linearity. The 6-Mn02method using the implicit Langmuir model was used to determine CSCs for the Cu-NTA complex in the pH range 6-8. Figure 6 compares the CSCs for the Cu-NTA complex obtain%dby using the implicit Langmuir equation with CSCs obtained in the previous study (1) using the simple expression. In addition the range of literature values obtained for Cu-NTA by other methods is shown (13,14). It is clear from Figure 6 that the CSCs obtained by use of the implicit equation fall within the (rather large) range of literature values, whereas values of CSCs obtained by using the simple expression are much less. This suggests that the MnO, method using the implicit Langmuir model is capable of estimating CSCs for metal-organic complexes in natural waters.
Conclusions This study has derived the implicit Langmuir model for use with the Mn02 adsorption method, to determine CCs and CSCs for trace metal-organic complexes in natural waters. This model appears quite satisfactory because it accurately predicts the observed deviation at the lower end of the isotherm; the model incorporates pH in the binding term expression, making the determinations of CSCs possible at different pHs without MnO, calibration at the specific pH, and the model yields values for the Cu-NTA complex that fall within the range of literature values. Testing of the modified model for other (organic) complexes will critically evaluate the general applicability of the method. Environ. Sci. Technoi., Vol. 20, No. 10, 1986
1049
Environ. Sci. Technol. lQ86, 20, 1050-1055
Registry No. Cu, 7440-50-8; MnOz, 1313-13-9.
Literature Cited
van den Berg, C. M. G. Ph.D. Thesis, McMaster University, Hamilton, Ontario 1979. van den Berg, C. M. G.; Kramer, J. R. ACS Symp. Ser. 1979, 115-132.
van den Berg, C. M. G.; Kramer, J. R. Anal. Chim. Acta 1979,106, 113-120.
Stroes-Gascoyne, S. Ph.D. Thesis, McMaster University, Hamilton, Ontario, 1983. van den Berg, C. M. G. Mar. Chem. 1982, 11, 307-322, 323-342.
McDuff, R. E.; Morel, F. M. M. California Institute of Technology Technical Report EQ-73-02,1973,1974,1975; W. M. Keck Laboratory of Environmental Engineering Science. Loganathan, P.; Burau, R. G. Geochim. Cosmochim.Acta 1973, 37, 1277-1293.
(8) Gabano, J. P.; Etienne, P.; Laurent, J. F. Electrochim.Acta 1965,10,947-963. (9) Schindler, P. W.; Ftirst, B.; Dick, R.; Wolf, P. U. J. Colloid Interface Sci. 1976, 55, 469-475. (10) McKenzie, R. M. Geochim. Cosmochim. Acta 1980, 43, 1855-1857. (11) Davis, J. A. PbD. Thesis, Stanford University, Stanford, CA, 1978. (12) Breeuwsma, A,; Lyklema, J. J.Colloid Interface Sci. 1973, 43,437-448. (13) Sillen, L. G.; Martell, A. E. Spec. Pub1.-Chem. SOC.1961, No. 17. (14) Sillen, L. G.; Martell, A. E. Spec. Pub1.-Chem. SOC.1971, No. 25.
Received for review October 17,1985. Accepted May 29,1986. This research was supported by research grants from the Natural Sciences and Engineering Research Council of Canada and from Environment Canada.
Enhancement of N-Nitrosamine Formation on Granular-Activated Carbon from N-Methylaniline and Nitrite Andrea M. Dietrich,* s t Danlel L. Galiagher,t Patricia M. DeRosa,' David S. Mililngton,5and Francis A. DlGlanot
Department of Environmental Sciences and Engineering, University of North Carolina, Chapel Hill, North Carolina 275 14, State of North Carolina, Division of Health Services, Raleigh, North Carolina 27602-2091, and Department of Pediatrics, Division of Genetics and Metabolism, Duke University, Durham, North Carolina 27706 Sterile aqueous N-methylaniline solutions were allowed to equilibrate at various nitrite, F-400 granular-activated carbon, and pH levels for 1 week. The aqueous and activated carbon phases were extracted and analyzed for nitrosamines relative to an added internal standard. Selected ion monitoring GC/MS, utilizing continuous monitoring of the NO+ ion (m/z 29.9980) characteristic of nitrosamines, at medium resolution ( R = 2500-3000) was applied to quantitatively measure nitrosamines at picograms per microliter concentrations. This method selected for nitrosamine products only and eliminated interferences from non-nitrosamine reaction products. Results indicate that the presence of granular-activated carbon significantly enhanced the formation of nitrosamine from N-methylaniline (F = 145, p < 0.0001). The amount of N-nitrosomethylaniline formed in the presence of activated carbon was 75 times more than that formed in the absence of activated carbon under the same nitrite, pH, and precursor amine conditions. High nitrite concentrations and low pH values significantly increased the conversion of secondary amine to nitrosamine. Introduction
Granular-activated carbon (GAC) accumulates natural and anthropogenic organic material in addition to inorganic salts, minerals, and microorganisms during water treatment. Chemical and biochemical activity may produce new compounds from those originally present, with possible human health consequences ( 1 , 2 ) . This research investigated the role of activated carbon in mediating the formation of N-nitrosamines from nitrite and secondary amine precursors. N-Nitrosamines are of particular environmental and health importance because, as a class, these compounds possess potent mutagenic and carcino'University of North Carolina. Carolina, Division of Health Services. *Duke University. *State of North
1050
Environ. Sci. Technol., Vol. 20, No. IO, 1986
genic activity (3). Although secondary amines and nitrite are id in only very low concentrations in water supplies (4, a), GAC may still mediate the formation of N-nitrosamines. Secondary amines can accumulate on GAC over a long period of operation, and nitrite is an intermediate in nitrification, a microbial process that occurs in GAC filters (6). Nitrosamine formation depends on the direct nitrosation of secondary amines (7), although primary and tertiary amines are also potential precursors since they may be converted to secondary amines. The overall reaction is R2NH
+ NOp- + H+
-
RzN-N=O
T H2O
The rate of reaction is first order with respect to the secondary amine and second order with respect to nitrite (8):
rate of formation = k,[R2NH][HN02]2 Direct nitrosation occurs optimally at pH 3.4, the pKa of nitrous acid (9),but another reaction pathway catalyzed by certain carbonyl compounds (including formaldehyde and trichloroacetaldehyde) can produce nitrosamines under neutral and even basic conditions (10). The nitrosating potential of tapwater has been demonstrated when secondary amines are added (11). Whether activated carbon affects nitrosation remains unclear. However, activated carbon has been implicated in the formation of nitrosamines from airborne precursors (12). The occurrence of nitrosamines in a variety of environmental samples has been cited. N-Nitrosamines have been detected in drinking water ( I I ) , industrial wastes (13-15), and water passed through deionizing resins (16-18). This study was designed to investigate the effects of pH and nitrite on the abiotic F400-GAC-mediated nitrosation of N-methylaniline. N-Methylaniline was selected because of its high solubility, low volatility, low pKa, strong adsorption on GAC, and known occurrence in certain drinking water supplies (19). The experimental conditions
0013-936X/86/0920-1050$01.50/0
0 1986 American Chemical Society