0016-7037/88/$3.00 + MJ
Geochimica d Cosnwhimica Acta Vol. 52, pp. 1849-1857 Cowright 0 1988 Pergamon Press plc.Printedin U.S.A.
~eas~~ment
of copper(I) in surface waters of the su~~opi~al Atlantic and Gulf of Mexico JAMES W. MOFFBTT~ and ROD G. ZIKA
University of Miami, Rosenstiel School of Marine and Atmospheric Science, 4600 Rickenbacker Causeway, Miami, FL 33 149, U.S.A. (Received July 14, 1987; uccepted in revisedform April 6, 1988) Abstract-Cu(1) has been measured in the upper marine water column at various locations in the Atlantic Ocean and the Gulf of Mexico. Deptb profiles characteristically display surface maxima, where Cu(1) comprises 5 to 10% of the total copper. Concen~ations decline rapidly with depth to below the limit of detection (0.015 nM) at the base of the mixed layer and throughout the rest of the water column. Profiles show variability withpepth and time of day consistent with photochemically mediated Cu(I1) reduction. Cu(1) distribution appears to be controlled by a variety of factors including Cu(II) speciation, photochemically, or biologically produced Cu(I1) reductants and possibly Cu(1) chelators which are significant only at subnanomolar Cu(1) levels. Hydrogen peroxide, which occurs in the photic zone and reduces Cu(II) to Cu(1) in seawater, accounts for a signifi~nt fraction of the Cu(1) observed, but other processes must also be involved. Cu(1) is analyzed by solvent extraction of its complex with 2,9dimethyl 1,lO phenanthroline, with ethylenediamine added simultaneously to mask Cu(II) interference. Details of the procedure are presented, along with an investigation of potential interferences.
of day, along with potentially related parameters such as hydrogen peroxide concentrations and Cu(I1) speciation. The objectives were to establish if Cu(1) was detectable in the upper water column and to determine if any of the processes studied in the laboratory can account for its concentration and spatial and temporal variability. The procedure is an extension of a method recently developed for Cu(I) dete~ination at levels greater than IO-* M (MOFFETT et al., 1985). In that work, the formation of a Cu(I) complex with bathocuproine disulfonic acid (2,9-dimethyl-4,7-diphenyl- 1, lo-phenanthroline disulfonic acid) was measured s~c~ophotomet~~ly. In this work, a closely related compound, neocuproine (2,9-dimethyl-1, lo-phenanthroline (dmp)) was used and the complex preconcentrated by solvent extraction with methylene chloride and back extraction into dilute acid, followed by atomic absorption analysis for total extractable copper. This extended detection limits to subnanomolar levels. Preliminary results indicated that only a small fraction of the total copper is present as Cu(I). Therefore, evaluation of potential Cu(I1) interference was an important part of this study. In particular, since interferences in the spectrophotometric determination of Cu(1) were proportional to Cu2+ activity and reducing agent concentration, it was important to demonstrate that the assay does not merely respond to changes in these parameters in the water column.
INTRODUCITON
THERE IS CONSIDERABLE interest in dynamic chemical processes in the marine photic zone and their importance in biological productivity and geochemical cycling (BURTONet al., 1986). The importance of redox reactions in the upper water column and the role they play in the speciation, transport, and bioavailability of minor eiements is poorly understood. However, recent evidence suggests that photochemitally and biologically driven redox cycling may be important for elements such as iron (ANDERSONand MOREL, 1982; WAITE and MOREL, 1984a,b) and manganese (SUNDA et al., 1983). Furthermore, hydrogen peroxide, which may’fun~ion as an oxidant or a reductant of many metal complexes has been measured at levels exceeding 10m7M in surface seawater (ZIKA et al., 1985). Recent Iaboratory studies of the redox chemistry of copper in seawater (MOFFETT and ZIKA, 1983, 1987a,b) indicate that a number of potentially important reactions lead to redox cycling of copper between Cu(1) and Cu(II), many of which am phot~hemi~y induced. Cu(f) may be produced through the direct photoIysis of Cu(II) organic complexes or through the reduction of Cu(I1) by photochemically or biologically produced reductants. For instance, hydrogen peroxide, which is produced photochemi~ly in surface waters is an effective Cu(II) reductant in seawater (MOF?%TT and ZIKA, 1987b). It has been suggested that such reactions could lead to significant steady-state levels of Cu(1) because of its stabilization by chloride complexation (MOFFETT and ZIKA, 1983). Direct measurement of Cu(I) in the water column is necessary to evaluate this hypothesis. In this study, a procedure for the determination of Cu(1) at subnanomolar levels in seawater was developed and rigorously evaluated for potential artefacts. The procedure was used to measure Cufi) at several oligotrophic and coastal sites in the Gulf of Mexico and the western subtropical Altantic Ocean as a function of depth and time
SAMPLING
REGIONS
Samples were collected on cruises I-IV and VIII of the SOLARS (Studies of Light Activated Reactions in Seawater) program at a variety of locations off the Florida coast, in the Sargasso Sea, and the Gulf of Mexico. Station locations are shown in Fig. 1. Sampling dates, coordinates, and weather ~nditions are given in the data tables in the Appendix. Stations I-2 and IV- 1 in the Florida Current, II-6 in the Tongue of the Oceans, Bahamas, and VIII-D in the Sargasso Sea were in oligotrophic regions. In particular, the Sargasso Sea station was characterized by low nutrients, low light attenuation, and a chlorophyll a ~ximum at 120 m. The station was far enou~ offshore to be free of any in8uence from the Antilles Current system which runs along the eastern edge of the Bahamas and is probably fairly representative of the Southern Sargasso Sea.
t Present address:Woods Hole Oceanographic Institution, Woods Hole, MA 02543, U.S.A. 1849
J. W. Moffett and R. G. Zika
1850
added. This was shaked for 30 seconds and the phases allowed to separate (about 1 min). The solvent was transferred to a 250 ml separatory funnel and 2 ml of 5% nitric acid added. It was shaken vigorously and allowed to separate (about 10 min). Subsequent analysis was by graphite furnace atomic absorption spectroscopy using standard operating parameters.
3o”
VIII-D 0
25”
FIG. 1. Map of sampling region. Further details about sites given in the Appendix.
Sampling Seawater samples were collected using trace metal clean techniques. Samples from the University of Miami dock, Biscayne Bay, were collected using a submersible electric pump with a Teflon-coated interior and Teflon tubing. Samples for immediate measurement were collected in acid-washed 1-liter Teflon bottles. Samples reserved for subsequent experiments were collectedin acid-washedPyrex carbcys. At sea, samples were collected using Teflon-coated Go-Flo bottles (General Oceanics, Miami, FL) attached to polypropylene-jacketed hydrowire. Equipment
Station III-2 in the Gulf of Mexico was located in shelf waters off the Florida west coast. Levels of particulates and productivity were higher than at the oligotrophic stations. Station I-2, off the Shark River mouth, Everglades National Park, Florida, was characterized by extremely high levels of dissolved organic matter and particulates. Preliminary studies were carried out with samples collected at the University of Miami, RSMAS, dock, Biscayne Bay, Florida. EXPERIMENTAL The ligand, 2,9_dimethyl-1, IO-phenanthroline (dmp) was selected because of its similarity to bathocuproine, so that the spectrophotometric assay (MOFFETTet al., 1985) could serve as a guideline, and because the ligand and Cu(1) complex are water soluble yet readily extracted into certain solvents. Furthermore, a great deal of information is available about the chemistry of its copper complexes. Methylene chloride was selected as the solvent because the Cu(1) dmp complex is readily extractable into it and because naturally occurring Cu(I1) organic complexes, which would interfere with the determination, were not extracted. For the spectrophotometric assay, it was determined that there was a substantial Cu(I1) interference in the presence of reducing agents, but this could be effectively inhibited using ethylenediamine (en) as a masking l&and. This is evidently due to its rapid complexation with Cu(II), and lack of any significant interaction with calcium or magnesium. In evaluating the spectrophotometric assay, detailed kinetic studies of potential interference processes were made and the effectiveness of Cu(I1) masking by en confirmed. We used data from that study, along with literature data to develop a procedure which, based on the known thermodynamic and kinetic properties of the assay reagents and their copper complexes should be highly selective for Cu(1). As in the earlier work, we studied the rates of interference processes to determine if they were important on the time scale of the assay. Reagents A stock solution containing 2.5 X lo-” M dmp and 1.25 X IO-’ M en was prepared and adjusted to pH 8 using a high-purity ammonia solution. HPLC-grade methylene chloride was used without any pretreatment before use. For some studies, marine organic material isolated from Biscayne Bay, Florida, by uItra8ltration through an Amicon PM 10 filter was used. Artificial seawater was prepared for some studies and trace metal contaminants removed by passing through a Chelex 100 column before use. “Chloride free” artificial seawater was prepared by replacing NaCl and MgCl2 with MgS04 and enough NazS09 to maintain a constant ionic strength. Extractionprocedure 2 ml of dmp/en reagent were added to 1 liter of seawater in a separatory funnel, shaked briefly, and 80 ml methylene chloride
Teflon bottles and separatory funnels were used throughout all stages of the analysis. Reagents were also stored in Teflon bottles. Analyses were performed in a Class 100 vertical flow laminar flow hood equipped with a duct at the base to exhaust solvent vapors. At sea, samples were drawn from the Go-FIo bottles directly into the hood via Teflon tubing to avoid atmospheric contamination. The Go-FIo bottles were pressurized with filtered compressed air to facilitate transfer. Atomic absorption analyses were performed using a Perkin Elmer 403 spectrophotometer. Total copper Samples for total copper were collected in 500-ml acid-washed polyethylene bottles along with Cu(1) samples, and acidified to pH 2 with nitric acid. Samples were worked up ashore using the APDC/ DDDC procedure described by DANIELLSON et al. (1978). For SOLARS I and II, Cu samples were not made on all casts because of a shortage of sample bottles. For SOLARS III and VIII, total Cu samples were taken with every Cu(1) sample. RESULTS Evaluation of procedure Thermodynamic considerations. Thermodynamic data for the Cu complexes of dmp and en are shown in Table 1. Calculations based on these data indicate that dmp and en chelate virtually all of the Cu(I) and Cu(II), respectively at the concentrations used. The resultant Cudmpf and Cue@ complexes are very stable, dissociate slowly, and are unreactive with any oxidants or reductants likely to be present in seawater during sample workup. Therefore, thermodynamic considerations indicate the technique is selective for Cu(1) and should produce a 100% yield. However, selectivity and yield will depend on the rates of formation Cudmpi and Table 1. Stability constants for lx(I) and CU(I1) complexes with &IQ and en. dmpa CU(I) cu(II) H+
log 81 log 82 log 131 loI3 82 pKa
-19.1 5.2 11.7 5.9
a. 25' I = 0.1 n b. 25' I = 0.5 I From SMITH and -TELL
enb -11.2 10.7 20.0 10.0
(1975).
1851
Copper in surface seawaters Cuer$ compared with other processes induced by the reagent addition which change the initial Cu(I)/Cu(II) ratio. Yield. In the spectrophotometric study (MOFFETT et al., 1985) it was observed that quantitative chelation of Cu(I) by bathocuproine in oxygenated seawater solution depended on the ratio of bathocuproine to en. It was desirable that the concentration of en be high, to minimize Cu(I1) interference. However, if en was present at greater than a 5-fold excess over bathocuproine, a substantial fraction of Cu(1) was chelated by en and rapidly oxidized, resulting in lower yields. Similar results were found in preliminary studies using dmp, by measuring the formation of Cudmp; spectrophotometrically. When elevated levels of Cu(I) (>20 nM) were added to oxygenated seawater containing dmp and en at the concentrations used in this work, greater than 91% was bound by dmp. The yield was measured by adding aliquots of Cu(I) stock solution to seawater from Biscayne Bay containing the reagents. At standard additions below 1 nM, the use of dilute Cu(I) standard solutions, coupled with small aliquot sizes, led to erratic results, so Cu(I) was added as the neocuproine complex in this region. The yield was 80 + 5% over the concentration range 0.2 to 10 nM. The yield was not pH dependent over the range 1.9 to 8.3, which includes most marine waters. Data obtained using the procedure were adjusted using this value for the yield and the adjusted values reported in this study. The yield in “chloride-free” seawater was 70 k 5%, only slightly less. This indicates that the counterion which facilitates extraction of the charged complex is not chloride, but another anion present in both solutions such as hydroxide. It is not clear why the yield for the extracted samples is less than that observed spectrophotometrically. The yield was also 80% in samples with 50 nM- 100 nM standard additions, where a much higher yield in solution was measured spectrophotometrically. This indicated that the extraction process was leading to a lower yield. However, extracting a second time or using more solvent did not increase the yield. One possibility is that the neutral dmp ligand is extracted more rapidly than the charged Cu(I) complex, leading to a reequilibration in the presence of en. However, since the yield was constant over the range studied, the procedure was used without further modification. Data obtained were corrected for this yield and the corrected values reported. Cu(ZZ)interference. Cu(I1) interference leading to an artefact in Cu(I) determination may arise in two general ways: 1) Extraction of naturally occurring Cu complexes, Cueny or Cudmp:’ into the solvent phase. 2) Changes in the initial Cu(I)/Cu(II) ratio in favor of more Cu(I) induced by reagent addition. Interference due to extraction of other copper complexes was easily tested. In control experiments where no reagents were added, no detectable copper was extracted in any of the waters where Cu(I) was measured. Therefore, naturally occurring Cu species were not contributing to the signal. Neither Cudmp:+ nor Cuenp were extractable either. Therefore, the extraction process was selective for Cudmpt . A shift in the Cu(I)/Cu(II) ratio leading to an increase in Cu(I) induced by the reagents can occur through two processes summarized schematically in Fig. 2.
cu (II)
en;
+
en t
+reductnnt
I
cu (II)
(1)
l
4
cu (1)
+ oxidant (-1)
r---------i Process (1)
Cu (II) dmp;
-
I
Cu (1) dmp; + rcductant
FIG. 2. Interference processesassociated with the assay.
In process (l), chelation of Cu(I) by dmp induces a shift in the prevailing redox equilibrium in favor of more Cu(1). In process (2), Cu(I1) is chelated by dmp, and Cudmpp is reduced irreversibly to the Cu(I) complex. Chelation of Cu(I1) by a masking ligand such as en effectively inhibits either process. At the en concentration used in this work, [Cu&$]/[Cu(II)],,,, is three to four orders magnitude lower than in the original seawater. In order to validate the assay, it is necessary to demonstrate that neither artefact processes are competitive with Cu(I1) complexation by en. This requires that estimates be made for rate constants for the reactions in Fig. 2. Estimates can be made with reasonable confidence based on kinetic data in the literature for Cu complexation by en and 1, lo-phenanthroline and for copper redox reactions in seawater. For a process ( 1)-type artefact to occur, copper must exhibit redox cycling through reactions 1 and -1 which must be fairly rapid, i.e. on the time scale of the assay. There is now considerable evidence that this cycling occurs (MOFFETT and ZIKA, 1987a), although all the oxidants and reductants participating in the cycle have not been fully characterized. However, it is highly probable that the dominant pathways for Cu(I) oxidation are through reaction with 02 and H202, for which rate data are available (MOFFETT and ZIKA, 1983, 1987b). These data yield a pseudo first-order rate constant for Cu(I) oxidation of - 10-3 s-l. Preliminary data indicate that Cu(I) comprises at most 10-l 5% of the total Cu (MOFFETT and ZIKA, 1987a), so a maximum pseudo first-order rate constant for Cu(I1) reduction derived under steady state conditions is - 10m4s-l. Studies of Cu(I1) reduction in seawater indicate that only minor species such as CuClOH, CuClz and possibly CL& are likely to be reactive, and they constitute about 10m3of the total copper in surface waters. Therefore, a pseudo first-order rate constant for reduction of these complexes required to maintain the steady state = 10m4 s-‘/1()-3
=
10-1
s-1.
Rate constants for the complexation of Cu& by en are shown in Table 2. At pH 8, 1% of en is present as the kinetically reactive neutral species. Therefore, a pseudo first-order rate for complexation at 2.5 X 10e5 Men = 2.5 X low5 X lo-* X 4 X 10’ = lo3 SK’.Data are not available for the reaction with CuC12 or CuCIOH, but they are probably within an order of magnitude of the Cu& rate. Therefore, the rate of chelation of kinetically reactive species by en exceeds the rate
1852
J. W. Moffett and R. G. Zika Table 2.
Rata constants for the reaction of Cu(II) with en and phen
2.5~
Reaction CU m~2:ye;_m;;; 2+ Cuphen2+ ClJ + phen 2+ + phen cuphen; cuphen
-1 k (ma1 L s-l) 3.8 K 109 a
2.0.S f
1.5.-
S~
1.0..
1.9 x log a 6.4 K 10 1.5
x 1,:
E
6T
0.5-
KIRSCHEmALm end KUSTIti(1970). b FABIAN ana DIEBLER (1987).
Q 0.0 0
= 6.4 X 10’ mol-’ L s-’ X 5 X lo+ mol L-’ = 3 x 102 s-1 which is only a factor of 3 lower than the constant for en. Chelation of the second phen is even faster, although this may not be the case for dmp because of steric hindrance of the methyl groups due to the square planar geometry of the Cudmp:+ complex. These data indicate that the formation of Cudmpp cannot be easily ruled out by kinetic considerations. Once formed, Cudmp$+ can be reduced rapidly by a variety of reducing agents. However, steady state levels of Cudmp$+ should be low, especially in the presence of en. Stopped flow kinetic studies of the reduction of the Cu(I1) bathocuproine complex, which is also rapidly reduced, showed that the rate of reduction was slowed down dramatically in the presence of en, even with strong reductants like hydroquinone (MOFFETT et al., 1985). It was shown that at 10e6 M hydroquinone, only a 2% interference would occur on the time scale of a spectrophotometric measurement. And en completely inhibited Cu(II) interference in solutions of humic acid, despite rapid Cu(II) reduction in the same solutions containing bathocuproine alone. Further evidence that a process (2)-type artefact was not important was obtained under conditions closer to those encountered in this work by studying the influence of chloride on the assay. Cu(I) oxidation in seawater is orders of magnitude slower than in chloride-free solutions (MOFFETT and ZIKA, 1983). Therefore, steady-state levels of Cu(I) in oxygenated chloride-free solution should be undetectable, or at least very much lower than in an analogous solution with seawater chloride levels. However, the interference resulting from process 2 depends only on the formation of Cudmp:+ and its resultant reduction to Cudmpf and extraction, which is not chloride dependent. Solutions containing ultrafiltrate isolated from Biscayne Bay in artificial seawater and “chloride-free” artificial seawater were exposed to sunlight in lliter round-bottomed quartz flasks. In natural seawater such treatment leads to a significant increase in Cu(I) (MOFFETT
+
+
Y
a
of reduction by three to four orders of magnitude. Therefore, a process (I)-type artefact is unlikely. A process (2)-type artefact depends on the relative rates of Cu(I1) chelation by dmp and en, and the rate of reduction of Cudmpp to Cudmpt . No data are available for dmp complexation rates, but a recent study shows that the parent compound, 1, lo-phenanthroline (phen), reacts at rates considerably lower than en (Table 2). However, 100% of the dmp is present as the neutral form. A pseudo first-order rate constant calculated from these data
+"
A 30
AA
AA
60
90
A 120
A
A
150
I 180
TIME (minutes)
PIG. 3. Chloride dependence of Cu(1) signal in sunlight irradiated solutions containing UF, Artificial seawater (00); “Chloride-free” artificial seawater (AA). Experiments performed in duplicate (open and closed symbols). Total Cu lo-12 nM. and ZIKA, 1987a). In artificial seawater an increase in the Cu(1) signal is also observed as indicated in Fig. 3. However, in the chloride-free media a very low signal (approximately 2% of total) was observed, which did not increase with irradiation. This pronounced chloride dependence is strong evidence that a process (2)~type artefact is not occurring. In summary, Cu(I1) interference through process (I) can be ruled out because data on Cu(1) oxidation set upper limits on how fast Cu(II) reduction can be, which is slow compared with en complexation. Process (2) was shown to be unlikely to be important in stopped flow studies with bathocuproine, and this was confirmed here by studying the chloride dependence of the Cu(1) signal. Time dependence. The time dependence of the signal for each stage of the assay was studied to establish a procedure for routine determinations. The results in Fig. 4 show the dependence of the Cu(1) signal on 1) time between seawater sampling and sample workup, 2) time between reagent addition and extraction, and 3) duration of extraction. Cu(1) signal does not decay detectably for at least an hour between sampling and analysis. However, after the reagents are added, the signal increases slowly with time before extraction. This may represent a slow interference process or slow exchange of Cu(1) between a very strong complex and dmp. 0.6-r 0.5
4
zz 5
-
,A/
0.3
I
,' _/-_-----
s
:
/.
0.4
-0
,_&--1
--
a
-a---&__-__i
1
1
0.2
0.1 0.0
4 0
I
15
30
45
60
TIME (minutes)
FIG. 4. Change in Cu(1)signal as a function of time between addition of reagents and extraction (A); Change in Cu(1) signal as a function of shaking time (0); Change in Cu(I) signal over the time of the other experiments with immediate extraction and separation after reagent addition (0).
1853
Copper in surface seawaters Currently it is not possible to distinguish between these two processes. Exchange with a strong Cu(1) chelator seems more plausible, since the kinetic considerations outlined above indicate that interference, even at this slow rate, is unlikely. However, on this time scale other processes such as surface reactions on container walls cannot be ruled out. Therefore, to consider a slowly exchanging fraction in the assay makes data interpretation much more difficult. Consequently in this work extraction was always carried out immediately after reagent addition, recognizing that a significant, relatively nonlabile pool of Cu(I) may be neglected. In Fig. 4 the influence of shaking time on the signal was studied to see if reactions at the interface affected the signal. An increase in signal occurred during continuous shaking, but was much less than in the sample containing reagents only. The most likely explanation is that free dmp is extracted into the solvent, reducing the rate of interference and/or exchange reactions in solution. Reproducibility and limit of detection in seawater. Measurements of Cu(1) in seawater were performed to assess precision and limit of detection of the technique at natural levels in an oligotrophic station in the Florida Current and a low productivity coastal site off the Dry Tortugas, Florida. All samples were worked up within one hour of collection through the solvent extraction stage and showed no variability with time over this period. The organic phases were back extracted within three hours of collection and showed no variability over this time. The results are shown in Table 3. Cu(I) in surface samples from several sites ranged from 5 to 10% of the total copper present. Replicates from each sample had a relative standard deviation of about 20%. Measurements of a sample collected at 225 m in the Florida Current are similar to the reagent blank in milliQ water. A theoretical limit of detection based on 2.5 times the standard deviation of this sample is 0.015 nM. These data suggest that the estimated error associated with a given measurement should be equal to 20% of the total or 0.0 15 nM, whichever is greater.
0 [CuO)l 0.03
0.00
0
3 = 15m*
Concentrations in nII d = lmb d = mmb
d = 225ma
0.07 0.045 0.05 0.07 0.045
0.11 0.075 0.11 0.082 0.085
0.11 0.13 0.082 0.11
0.015 0.01 0.005 0.005 0.005
m = 0.11 0 - 0.02 R.S.D. = 18%
ID. = 0.01 a = 0.006 R.S.D. = 60x
a. Florida Current. total Cu - 1.20 nn. b. Off LaKKerheod Key, Dry Tortugas. 15 m total depth. Total Cu = 1.20 nlI.
0.09
0.12
0.15
20
25
---o-
eo50.. 2 E
to+ 0
-o-
o-olOO--
CL E 15o'.pZOO? 0
5
TEM&N"R:;OC)
FIG. 5. Depth profile of Cu(1) in the Florida current, April 1985 (Station I-2).
and a decline in concentration with depth, with a rapid decline at the base of the mixed layer. Concentrations decline more rapidly with depth at Station III-2 (Fig. 8) than at the more oligotrophic Station VIII-D (Figs. 6 and 7). There is a general correlation with Hz02 , indicated in Fig. 5 and in the Appendix. In Fig. 6 measurements were made down to 900 m in the Sargasso Sea. Cu(1) was below the limit of detection at all depths below the photic zone. In Figs. 7 and 8 night and day profiles for Cu(I) are shown for the Florida west coast and Sargasso Sea stations. Levels are higher during the day but still detectable at night, consistent with decay studies of photo-chemically produced Cu(1) at subnanomolar levels which indicate a half life on the order of 12 hours (MOFFETTand ZIKA, 1987a). The large difference between day and night at the surface in Fig. 7 may be due in part to the higher surface total copper concentration measured at this site at 1400 hr. Despite these data, Cu(I) is by no means ubiquitous in surface waters. On SOLARS IV in
0
Depth profiles of Cu(I) compiled on SOLARS I, III, and VIII are shown in Figs. 5-8 and are tabulated in the Appendix, along with total copper, hydrogen peroxide, and hydrographic data. Each data point represents the mean of two replicates. Error bars are calculated from the considerations discussed above and were generally greater than or equal to the range between replicates. Profiles show surface maxima in Cu(I)
15 0.055 m = 0.092 0 - 0.013 0 - 0.017 B.S.D. I 22,. R.S.D. = 18%
0.06
PWI
Cll(I)measurements in near surface and deep waters
h.021 ( JJM)
l
01
Water column data
Table 3.
(nM)
0.04
(nMvl> 0.06
0.12
0.16
600
POTENTIAL
TEMPERATURE
(OC)
FIG. 6. Depth profile of Cu(1)in the Sargasso Sea, June 1986,160O hr (16 m to 160 m); 1900 hr (200 m to 950 m), (Station VIII-D).
J. W. Moffett and R. G. Zika
1854
2oot 0.00
0.04
0.08 [Cd01
0.12
0.16
(nM)
FIG. 7. Depth profiles of Cu(1) in the SargassoSea, June 1986. (a) 1600 h (0); (b) 0600 (0). (Station VIII-D)
the Florida Current in December 1985 during bad weather (cloudy, with a mixed layer depth SO-90 m), Cu(I) at the surface was below the limit of detection. Measurements of &(I) at the mouth of the Shark River, Everglades National Park, were below the limit of detection despite high peroxide and sunny calm conditions, which was attributed to strong Cu(I1) binding by terrestriaUyderived humics (MOPPP~ and ZIKA, 1987a). Data from these stations are shown in the Ap pendix. Data for total copper shown in the Appendix are in agreement with other studies in these regions (SPENCERet al., 1982; WINDOM and SMITH, 1979; BRULANDand FRANKS, 1983), indicating no serious contamination problems for total cop per. The most likely contamination source which would lead directly to Cu(1) is particulate Cu metal, which might be derived from shipboard electric motors. However, this would have led to erratic profiles so is unlikely to be a serious problem. DISCUSSION Depth profiles of Cu(1) at all three stations are characterized by surface maxima and a decrease with depth. The profiles are consistent with a photochemical mechanism for Cu(1) formation, in agreement with an earlier study in which the formation of Cu(1) in seawater during sunlight irradiation using this technique was demonstrated (MOFFETT and ZIKA, 1987a). In that work, a variety of sunlight-driven reactions were identified which lead to Cu(1) formation. However, it is still not clear which processes are directly involved. In laboratory studies of Cu(1) oxidation in seawater in the range [Cu(l)] (t = 0) = 0.1 to 5 PM, Cu(1) has a half life of only 46 minutes (MOFFETTand ZIKA, 1983). Clearly, the slow decay of Cu(1) at natural levels observed in laboratory studies and at night are due to either a long-lived reductant produced in the photic zone, or to further stabilization of Cu(1) speciation at the subnanomolar level by natural chelators, rendering the oxidation kinetic data inapplicable because it is based solely on chloride complexation. Primary photoprocesses or secondary processes involving highly reactive, short-lived species cannot by themselves account for the slow decay. Hydrogen peroxide was originally proposed to be important in this system based on thermodynamic considerations. However, calculations based on kinetic data for the reduction of Cu(I1) by Hz02 (MOFFE~ and ZIKA, 1987b) and new Cu(I1) speciation data for these regions (M0PPEl-r and ZIKA, 1987~;
MOFFETT, 1986) indicate that only 10% to 15% of the measured Cu(1) can be accounted for by this pathway (MOFFETT and ZIKA, 1987a). While errors associated with speciation measurements may contribute to this discrepancy, it seems that other reductants are also involved. Such reductants may be humic compounds or compounds of more recent biological origin. There is indirect evidence for the production of reducing agents in humic materials in studies of the reduction of Mn oxides (SUNDA et al., 1983), Fe(III) (WAITE and MoREL, 1984b), and F%(V) (CHOPPINet al., 1985). And the experiments with ultra&ate in this study suggest that macromolecular organic species are involved directly or as precursors. Cu(1) at subnanomolar levels decays slowly even in filter sterilized seawater (MOFFETT and ZIKA, 1987a) so reactions involving cell or detrital surfaces or short-lived metabolites are not implicated, although they could play an important role in the water column. Alternatively Cu(1) may be stabilized by additional complex formation. Extrapolation of laboratory data for Cu(1) oxidation rates to natural levels is based on the assumption that Cu(1) speciation at all concentrations is dominated by its strong chloride complexes. This assumption is reasonable given the high stability constants of these complexes and the fact that most functional groups on humic materials probably interact weakly with class B metals such as Cu(1). However, reduced sulfur compounds form strong Cu(1) complexes. For example, glutathione could dominate Cu(1) speciation in seawater at concentrations as low as 5 nM. Recent evidence indicates that SH- may be present in surface seawater at subnanomolar levels (CUTTER and OATTS, 1987), making it potentially important in Cu(1) complexation. ELLIOT et al. (1987) have proposed that the hydrolysis of atmospherically derived carbonyl sulfide to HS could account for its accumulation at subnanomolar levels, and a surface source is consistent with Cu(1) maxima. However, photochemical formation and slow decay of Cu(1) in filter sterilized, aged seawater (MOFFETT and ZIKA, 1987a), where significant levels of reduced sulfur compounds are unlikely, indicates that such species are not required, though they may be significant in the water column. It was pointed out previously that the slow increase in signal observed in Fig. 3 may be the result of slow exchange of Cu(1) between a natural chelation and dmp instead of an artefact. However, at present there is no way to distinguish between these possibilities. O-
--c
+O-
8OT 0.00
-04 :
: 0.10
:
I': 0.20
0.30
: 0.40
:
c 0.50
[NOI (nW FIG. 8. Depth profiles of Cu(I) off the Florida west coast. (a) 1400 h (0); (b) 0300 h (0). (Station 111-2)
1855
Copper in surface seawaters
Studies of Cu(I1) speciation have been made at these sites using a ligand exchange, liquid liquid partition procedure. Details of the procedure and results are described elsewhere (MOFFETT and ZIKA, 1987a,c; MOFFETT, 1986). These data yield some information on the parameters which influence Cu(1) distribution. In general, Cu(I1) is bound most tightly in the chlorophyll maximum, where the ratio Cu&/Cu$~~, is less than 10e4, presumably because of strong chelation by biologically-produced chelators. Between this region and the increases by over an order of magnsurface, Cu&/Cu&~ itude to 10V3,which may be due to the photodegradation of organic chelators. There is a general correlation between increasing CU$&/C~$&I, and Cu(I) in this region. This is consistent with kinetic data, which indicate that certain minor species-notably CuClOH and CuC12-are more reducible than Cu(I1) organic complexes (MOFFETT and ZIKA, 1987a). These minor species are proportional to Cu&. However, ratios of CU$@&I, also increase with depth below the chlorophyll maximum to values at 900 m which are comparable with surface values, yet no Cu(1) was detected at any depth below the photic zone. Cu(I) distribution is probably determined by a variety of factors, including Cu(I1) speciation and the distribution of reduced species-the latter evidently confined to the photic zone at these sites. Negligible Cu(1) levels at the productivity maxima at all stations are probably due to strong binding of Cu(I1) and do not necessarily mean that biologically mediated reduction processes are not significant in this region. Cu(I) formation is unlikely to have dramatic implications for copper transport by adsorptive processes as both redox states form strong, highly water-soluble complexes. However, the results are important because they indicate the potential for minor elements to exhibit a dynamic redox chemistry in the upper water column involving kinetically facile one electron transfer steps, in addition to multielectron, enzymatically mediated transformations involving such couples as I-/IO;. They suggest the presence of a significant class of redox reactive compounds which function as reducing agents in the presence of Cu(I1). If so, Cu effectively acts as a catalyst in the oxidative decomposition of these species, which may IX a significant contribution to the overall influence of photochemistry on carbon and oxygen cycling in the upper water column. Similar processes may be important in the reduction of other metals. For instance, SUNDAet al. (1983) have demonstrated the photoreductive dissolution of naturally occuring Mn oxides in seawater in the presence of humic materials which could contribute to a longer residence time and consequent near surface maxima which are observed in dissolved Mn profiles. Acknowledgements-The assistance of the captains and crews of the research vessels Calanus, Cape Florida, and Columbus I&in of the University of Miami is gratefully acknowledged. Supported by the Office of Naval Research, Contract number NOOOl4-85c-0020.
Editorial handling: K. W. Bruland
REFERENCES ANDERSONM. A. and MOREL F. M. M. (1982) The influence of aqueous iron chemistry on the uptake of iron by the coastal diatom Thalassiosira weissflogii. Limnol. Oceanogr. 27, 789-8 13. BRULANDK. W. and FRANKSR. P. (1983) Mn, Ni, Cu, Zn and Cd in the western North Atlantic. In Trace Metals in Seawater (ed. C. S. WONG), pp. 395-414. Plenum, New York. BURTONJ. D., BREWERP. G. and CHESSELETR. (eds.) (1986) Dynamic Processes in the Chemistry of the Upper Ocean. Plenum, New York, 246~. CHOPPING. R., ROBERTSR. A. and MORSEJ. W. (1985) Effects of humics on plutonium speciation in marine systems. ACS Symposium Series 305, pp. 382-388. CUTTERG. A. and OATTS T. J. (1987) Determination of dissolved sulfide and sedimentary sulfur speciation using gas chromatogmphy-photoionization detection, Anal. Chem. 59, 7 17-72 1. DANIELSSON L. G., MAGNUSSONB. and WESTERLUNDS. (1978) An improved method for the determination of trace metals in seawater by atomic absorption spectroscopy with electrothermal atomization. Anal. Chim. Acta 98, 45-57. ELLIOTS., Lu E. and ROWLANDF. S. (1987) Carbonyl sulfide hydrolysis as a source of hydrogen sulfide in open ocean seawater. Geophys.Res. Lett. 14, 131-134. FABIANI. and DIEBLERH. (1987) Kinetics of the consecutive binding of bipyridyl ligands and of phenanthroline ligands to copper (II). Inorg. Chem. 26,925-928. KIRSCHENBAUM L. J. and KUSTINK. ( 1970) Kinetics of copper (II)ethylenedianine complex formation. J. Chem. Sot. (A). 684-688. MOF’FETT J. W. (1986) The photochemistry of copper complexes in seawater. Ph.D. dissertation, Univ. of Miami. MOFFETTJ. W. and ZIKA R..G. (1983) The oxidation kinetics of Cu(I) in seawater. Implications for its existence in the marine environment. Mar. Chem. 13,239-25 1. MOFFETTJ. W. and ZIKA R. G. (1987a) The photochemistry of copper complexes in seawater. In Photochemistryof Environmental AquaticSystems(eds. R. G. ZIKA and W. J. COOPER);Adv. Chem. Ser. 327. DD. . . 116-l 30. Amer. Chem. Sot., Washington. MOFFETTJ. W. and ZIKA R. G. (1987b) Reaction kinetics of copper and iron with hydrogen peroxide in seawater. Environ. Sci. Tech. 21, 804-8 10. MOF’FETT J. W. and ZIKA R. G. (1987~) Solvent extraction of copper acetylacetonate in studies of Cu(II) speciation in seawater. _._-. ,+ffir Cheh. 21,301-313. MO~TT J. W. and ZIKAR. G. and F%TASNE R. G. (1985) Evaluatinn . of bathocuproine for the spectrophotometric hetermination of copper(I) in copper redox studies with applications in studies of natural waters. Anal. Chim. Acta 175, 17 l-179. SMITHR. M. and MARTELLA. E. (1975) CriticalStabilityConstants, vol. 2. Plenum, New York, 418~. SPENCERM. J., BETZERP. S. and PIOTROWICZS. R. (1982) Concentrations of cadmium, copper, lead and zinc in surface waters of the Northwest Atlantic Ocean. A comparison of Go-Flo and Teflon water samplers. Marine Chem. l&403-410. SUNDAW. G., HUNTSMANS. A. and HARVEYG. R. (1983) Photoreduction of manganese oxides and the supply of manganese to marine plants. Nature 301,234-236. WAITET. D. and MORELF. M. M. (1984a)Photorcductive dissolution of colloidal iron oxides in natural waters. Environ. Sci. Tech. 18, 860. WAITET. D. and MOREL F. M. M. (1984b) Ligand exchange and fluorescence quenching studies of the fulvic acid iron inter&t;; Effects of DH and liaht. Anal. Chim. Acta 162.263-274. WINDOMH.-L. and SMITH R. G. (1979) Coppe; concentrations in surface waters ofthe Southeast Atlantic Coast, U.S.A. Mar. Chem. 7, 157-163. ZIKA R. G., MOFFETTJ. W., PETASNER. G., COOPERW. J_. 2nd -___ SAL~ZMA~E. S. (1985) Spatial and temporal variations ofhydrogen peroxide in Gulf of Mexico waters. Geochim. Cosmochim. Acta 49, 1173-l 184. .._“..
1856
J. W. Moffett and R. G. Zika Table 4-A. SOLAILSIII, Sta. X11-2, Gulf of Mexico, g/13/85, 1400 hr., g/13/85, Calm, Sunny (27*40'11,84*30'u)
APPEUDIX All concentrations in Ml. 0 = Contamination suspected for data. UD = Uo data - sample not collected, or lost due to procedural error. Table 1-A. SOL.ARSI, Sta. I-l, Shark River Wouth, 4/4/85, Die1 Study, Sunny. Calm (2S020'U, 81°15'U)
Iy9
Surface
I
Samrdes
0600 0600 0800 1000 1000 1200 1200 1400 1400 1600 1600 1800 1800 2000 2000 2300 2300 0200 0500 0500
0.03 0.01 0.02 0.02 0.02 0.02 0.02 (0.31) 0.02 0.025 0.015 0.01 0.01 0.01 0.01 0.01 0.015 0.015 0.02
134
S("/o.)
2 2
0.42 0.34
2.70
28.68
35.35
25 25 45 45 70 70
0.13 0.08 0.08 0.06 0.05 0.03
1.80
26.46
35.69
1.80
24.69
35.93
1.70
20.81
36.24
Table 5-A. SOLARS III, Sta. 111-2, Gulf of Mexico. 9114185, 0300 hr., Calm, Dark (27*40'8, 84'3O'U)
139 97 Depth
44 33 26
w2021
10 10 25 25 50 50 70 70 90 90 140 140 180 180
0.081 0.125 0.055 0.070 0.044 0.075 0.028 0.048 0.020 0.030 0.00 0.006 0.00 0.00
1.13
65
1.30
55
ND
47
UD
8
1.20
4
0.93
0
WD
0
Table 3-A. SOL&US II, Sta. 11-6, Tongue of the Ocean, Bahamas. 6/16/85. 1740 hr (2S016.6'N, 77*42'uJ
[Cultotal
T(OC)
25.76
5 5 15 15 25 25 45 45 70 70
[Cultotal
T(*C)
S('I..)
1.90
28.70
35.33
1.85
27.31
35.56
UD
27.01
35.73
UD
24.71
35.86
1.75
20.80
36.25
0.17 0.13 0.08 0.06 0.07 0.05 0.01 0.03 0.01 0.01
Table 6A. SOLASS IV, Sta. IV-1 12/17/85, 1550 hr.. cloudy, Rough (25'6'U. 80*12'U)
Each of the following tables (3-A to 9-A) comprises data from a single Go-P10 cast. Hydrographic data for each table was collected from CTD casts prior to, or after the Go-Flo casts - generally within one hour.
1.25
[CU(I)l
85
[Cultotal
0.054 0.059 0.069 0.069 0.044 0.046 0.031 0.028 0.013 0.01 0.01 0.01
T(*C)
88
Icu(I)l
10 10 25 25 50 50 80 80 120 120 180 180
tCultotal
32 68
Depth
ICU(l)l
Icu(I)l
24
0.01
Table 2-A. SOLARS I Sta. I-2, Florida Current, 4/6/85, 1300 hr, Sunny, Calm (24.46'11,80'21'U)
Depth
Depth
S(O/..)
37.96
ND
25.53
37.96
UD
26.28
36.38
UD
25.39
36.45
1.45
24.03
36.83
ml
22.92
36.93
Depth
Icu(I)l
lCultotal
5m 5m 5m 5m
0.02 0.01 0.03 0.01
1.38 1.45
Table 7-A. SOLARS VIII, June 1986, Sta. D, Sargaoso Sea, 6/22/86, 1600 hr, Calm, Sunny (26'46'u, 7Sa26*U)
Depth
10 16 16 20 25 25 40 50 50 80 80 120 120 160 160
[Co(I)1
lCultotal
T(Y)
S(O/..)
"202
105.2 0.11 0.16
1.62
0.09 0.11
1.45
0.09 0.08 0.01 0.01 0.04 0.02 0.00 0.01
1.35
26.63
36.47
61.6
25.20
36.59
41.2
1.31
23.80
36.67
20.2
1.45
22.60
36.68
6.0
1.30
20.88
36.72
0.1
Copper in surface seawaters Table 8-A. SOLAM VIII, 0630 hr. Calm, Dam
Depth
10 10 20 25 25 50 50 80 80
[Cu(I)l
Table 9-A. SOL&S VIII, Sta. D, SargaSsO Sea, 6/24/N, 1900 hr, Calm, Gusk (26'46'1, 75*26'U)
Sta. D, SarSasso Sea, 6123187. (26*46'S, 75'26'W)
Icultotal
T('C)
sce/..j
1857
Depth
[cu(I)l
[mltotal
T('C)
S('I..)
200 200 250 250 300 300 400 400 500 600 600 750 950 950 1000
.Ol .Ol .Ol .Ol .Ol .Ol .Ol .Ol
1.66
20.29
36.70
(2.54)
19.11
36.59
(3.00)
18.45
36.53
16.93
35.53
11.71
35.35
6.99
35.10
H202
0.10 --
1.37
27.17
36.48
0.07 0.13 0.04 0.03 0.02 0.02
1.44
26.14
36.49
1.44
24.82
36.62
46.5
-_
23.33
36.66
27.6
82.7
1.47
.Ol .Ol
1.18
.02 .Ol
1.65