Particle-size and chemical control of As, Cd, Cu, Fe, Mn, Ni, Pb, and Zn in bed sediment from the Clark Fork River, Montana (U.S.A.)

The Science of the Total Environment, 76 (1988) 247-266 Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

247

PARTICLE-SIZE AND CHEMICAL CONTROL OF As, Cd, Cu, Fe, Mn, Ni, Pb, and Zn IN BED SEDIMENT FROM THE CLARK FORK RIVER, MONTANA (U.S.A.)

EDWARD J. BROOK* and JOHNNIE N. MOORE Department of Geology, University of Montana, Missoula, M T 59812 (U.S.A.) (Received April 5th, 1986; accepted May 20th, 1988)

ABSTRACT Mining and smelting in the headwaters of the Clark Fork River have significantly enriched Clark Fork River bed sediment in As, Cd, Cu, Mn, Pb, and Zn. These enriched elements are present predominantly in (operationally defined) "reducible" and "oxidizable" chemical phases, with small contributions from "residual" phases. In size fractionated samples (> 300, 300-63, 6~38, 38-17, and < 17#m) metals and arsenic concentrations generally increase with decreasing particle size, with the greatest contribution to this increase from the "reducible" phase. Copper is an exception, with the strongest contribution to this increase from the "oxidizable" phase. Anomalously high concentrations in the coarsest ( > 300 #m) fraction of some of the samples described in this study are due to preferential concentration of iron and manganese oxides on coarse particles, and entrapment of coarse organic material on 300#m sieves. Clark Fork bed sediment is, in general, quite coarse; the relatively high concentrations of metals in the coarse fractions of these sediments are therefore important to the bulk metal and As content of this system.

INTRODUCTION Although time consuming, studies of metal speciation within different particle-size classes can provide relatively detailed information about processes i n f l u e n c i n g t h e m e t a l c o n t e n t o f s e d i m e n t s (see f o r e x a m p l e : W h i t n e y , 1975; G i b b s , 1977; F i l i p e k a n d O w e n , 1979; F o r s t n e r a n d P a t c h i n e e l a m , 1980; T e s s i e r e t al., 1982). W e h a v e a p p l i e d t h i s m e t h o d o f i n v e s t i g a t i o n t o b e d s e d i m e n t f r o m the Clark Fork River, in western Montana, which has recently received attention due to elevated concentrations of arsenic, cadmium, copper, m a n g a n e s e , l e a d , a n d z i n c f o u n d t h r o u g h o u t t h e d r a i n a g e ( A n d r e w s , 1987; A x t e m a n n a n d L u o m a , 1987; M o o r e , 1987; M o o r e e t al., 1988a, b). T h e r i v e r begins in the Butte~Anaconda mining district, a region of large-scale base and precious metal mining of a large sulfide ore body in the Boulder batholith (Fig. 1). F r o m 1884 t o t h e e a r l y 1970s, l a r g e q u a n t i t i e s o f c o p p e r , zinc, m a n g a n e s e , lead, silver, gold, cadmium, and a number of other elements were produced in * Present address: Chemistry Department, Woods Hole Oceanographic Institute, Woods Hole, MA 02543, U.S.A. 0048-9697/88/$03.50

© 1988 Elsevier Science Publishers B.V.

248

"~.

O0t

;

%

r

"'',..,

?~,,,,;o.~, . . . . .

i

.....

:

)

..

1:42

~"~Drumrnond

.... "-

/ --

t

....

i

! \~-r-.... ~ \

I

\~

v._

~ 2

~

o

'..~

'"

\

~

I~

i

_.....:

/-.~.

o

. . . .

:'

v

"~x..f Fig. 1. Study area and sample locations. (z~) 1986 sample sites, (A) 1987 sample sites, ( 0 ) 1986 and 1987 sites. Clark Fork River samples identified by distance (km) downstream from Warm Springs Creek. Tributary samples designated as follows: (1) Little Blackfoot River, (2) Flint Creek, (3) Blackfoot River.

this district (Miller, 1973). These ores also contained a significant amount of arsenic (Meyer et al., 1968). Until the late 1950s, when containment ponds were built on lower Silver Bow Creek (see Fig. 1), much of the waste from ore processing was dumped locally. Floodplain and reservoir sediments t h r o u g h o u t the river now contain much of this material (Moore, 1987; Moore et al., 1988b). In this study we concentrated on the 200 km region between Warm Springs, MT, and Missoula, MT, (see Fig. 1), where the highest metal concentrations appear to exist (Axtmann and Luoma, 1987; Moore, 1987). In this area the Clark Fork is a relatively shallow stream with a gradient of -~ 2.5 m km -1. In 1985-86, instantaneous streamflow varied from ~ 0.85-28.3 m 3 s - ' at Deer Lodge, MT, to 9.5 254.7m 3 s 1 above the Milltown Reservoir (Lambing, 1987) (see Fig. 1). Most of the sediment deposited in the river bed is quite coarse; previous sampling has rarely recovered samples with less than 40-50% sand ( > 63 #m). Some of the work in the drainage concentrated on the role of particle size as a control of metal and arsenic concentrations in bed and floodplain sediment (Brook and Moore, 1987; Moore et al., 1988a). In these sediments the expected trend o f increasing metal concentrations with decreasing particle size are subdued (in the case of bed sediment) or predominantly absent (in the case of floodplain sediment). The work presented here was u n d e r t a k e n to obtain more information about the role of particle size as a control of metal and arsenic concentrations in river-bed sediment, and the chemical speciation of the large excess concentrations of metals and arsenic in this system.

249 METHODS

Sample collection, separation, and size analysis Bed sediment was collected in the summers of 1986 and 1987 (see Fig. 1 for sample locations). At all sample sites surface sediment was collected in areas of low flow velocity. Collection and treatment of the two groups of samples are described separately below. In 1986, seven river-bed samples, and one sample from each of three Clark Fork River tributaries (Little Blackfoot River, Flint Creek, and Blackfoot River) were collected (see Fig. 1). These samples were transported on ice to the University of Montana in Missoula. Samples were homogenized, subsamples were withdrawn for bulk chemical and particle-size analysis, and the remaining portions were frozen. Within -~ 1 month of collection these samples were thawed, and separated into the following six size fractions: > 300, 300-63, 63-38, 38-17, 17-4, and 4~.1 #m. Separation was accomplished by wet sieving with filtered (0.45 #m) river water (300, 63, 38, and 17 ~m Gilson polyester mesh screens), centrifugation with an IEC (International Equipment Company) UV centrifuge (17-4 ttm fraction), and centrifugation with a Sharples super-centrifuge (4-0.1 #m fraction). The size fractions were dried at 70°C and stored for later analysis. In late August-early September of 1987, six river-bed sediment samples were collected for size separation and subsequent sequential chemical extraction (see Fig. 1 for locations). These samples were returned to the laboratory on ice and immediately separated into five size fractions ( > 300, 300-63, 63-38, 38-17, and 17-0.1#m) using procedures identical to those described above, omitting separation of the 17-4 ~m fraction. The 17-0.1 #m fraction was not subdivided because the previous summer's work had shown that it was difficult to obtain enough material in the 17-4 and 4-0.1 #m size classes for sequential extraction procedures ( ~ 1 g). Individual size fractions were frozen (overnight), and then freeze dried. Unfiltered river water, collected separately at each site, was used for wet sieving. Filtered (0.45 t~m) and unfiltered aliquots of this water, and the water passing through the Sharples super-centrifuge at the end of the separation procedures, were collected and preserved with 3 ml l-1 nitric acid (Baker Intra-Analyzed). Because fluid passing through the super-centrifuge comes into contact with stainless steel parts of the equipment, deionized water passing through the centrifuge was also collected and preserved (as above) to assess possible contamination. Particle-size analysis of all samples was performed by standard sieving and pipette methods on separate sample splits. Triplicate analyses of splits of four samples gave results that differed from the mean of the triplicate analysis by no more than 1% of the total sample weight. Precision was significantly better for size fractions containing relatively large amounts of material ( > approximately 5% by weight). For size fractions that often made up less than 1% of the

25O samples (38-17, 17-4, 4-0.1pm) absolute precision was poor, however this does not affect conclusions derived from these data.

Analytical methods Size fractions of samples collected in 1986 were digested with a microwave technique modified from Nadkarni (1984). Approximately 0.2g of powdered, dried sample was placed in a 50 ml Teflon, screw-top, reaction vessel (Savillex Corp.). Five millilitres of aqua regia and 2ml of HF (Baker Instra-Analyzed acids) were added, and the vessel was tightly sealed. Vessels were placed, six at a time, in a resealable plastic bowl on a rotating carousel in a JET209D General Electric microwave oven. The oven was operated at high power for 5 min, then vessels were removed and allowed to cool. When cool, 40 ml of 2.5% (w/v) boric acid (Baker Analyzed) were added. The resulting solutions were allowed to stand for at least 0.5 h, filtered (0.45 pm cellulose triacetate filters), and diluted with deionized water to 100 ml in glass volumetric flasks. Size fractions collected in 1987 went through a three-step extraction scheme designed to remove metals in operationally defined "reducible", "oxidizable", and "residual" phases. This procedure is a simplification of a method used with other Clark Fork River sediments (Moore et al., 1988 b). To remove arsenic and metals bound in the "reducible' phase, ~ 0.5 g of sediment was placed in a 50 ml polyethylene centrifuge tube and shaken for 12 h with 10ml of 0.25 M hydroxylamine hydrochloride in 25% (v/v) acetic acid. These slurries were centrifuged for 20 min at 4000 r.p.m, and the supernatant was decanted and diluted with deionized water to 50ml in glass volumetric flasks. The acetic acidhydroxylamine extract is designed to liberate metals bound in iron and manganese oxide and hydrous oxide compounds (Chester and Hughes, 1967; Agemian and Chau, 1976; Sondag, 1981) and may also provide good anomaly/ background ratios for stream sediment geochemical exploration (Filipek et al., 1982; Tessier et al., 1982). Exchangeable and carbonate-bound metals should also be released (Chester and Hughes, 1967; Tessier et al., 1979). The residual material from the first extraction step was rinsed with 10 ml of deionized water (rinsing involved adding 10 ml deionized water, shaking, and centrifugation at 4000 r.p.m, for 20min). The supernatant from the rinsing step was discarded and the oxidizable phase of the sample was dissolved with a double potassium chlorate-hydrochloric acid extraction (Chao and Sanzolone, 1977). Potassium chlorate (0.5g, ACS grade, MCB Reagents) and 10ml of hydrochloric acid (Baker Instra-Analyzed) were added to the sediment in the centrifuge tubes and allowed to stand for 0.5 h. Then, 10 ml of deionized water were added to each, the tubes were shaken, and centrifuged for 20min at 4000 r.p.m. This procedure was repeated with fresh reagents and the solutions from the two "oxidizing" steps were combined and diluted with deionized water to 50 ml in glass volumetric flasks. The double potassium chlorate extraction, which was designed to dissolve crystalline sulfide minerals, dissolved 86-100% if galena (PbS), chalcopyrite (CuFeS2), cinnabar (HgS), orpiment (As2S3),

251 TABLE 1 Total-metal analysis of NBS reference materials 1645 (River Sediment) and 1646 (Estuarine Sediment) (Fe in wt. %, o t h e r values in gg g-1)a 1645 (n = 15) NBS Cd Cu Fe Mn Ni Zn

10.2 109 11.3 785 45.8 1720

(1.5) (19) (1.2) (97) (2.9) (170)

1646 (n = 12) This study

NBS

8.6 112 10.4 752 50 1670

0.36 18 3.35 375 32 138

(1.3) (12) (1.1) (90) (6.7) (191)

This study (0.07) (3) (0.1) (20) (3) (6)

BD 16 3.45 363 35 126

(8)b (0.3) (43) (6) (37)

a Values reported are m e a n s and ( + ) 2s (in parentheses). BD = below detection. b Analysis of t h e last eight s t a n d a r d s of this group gave _+ 1.4.

stibnite (Sb2S3), sphalerite (ZnS), and tetrahedrite (Cu12Sb4S13), and 71-86% of pyrite (FeS2) present in artificial samples (Chao and Sanzolone, 1977). This procedure was designed for use after extraction steps removing metals bound in iron and manganese oxide compounds and organic material; it is used here as an estimate of metals and arsenic bound in the "oxidizable" phase of the sediment (presumably mostly sulfides and organic material). The residual fraction from the "oxidizable" extraction step was rinsed as above, dried overnight at 70°C, ground in a corundum mortar and pestle, and digested with the total digestion procedure described above. All water samples and sediment extracts were analyzed for As, Cd, Cu, Fe, Mn, Ni, Pb, and Zn with a Jarrel-Ash Model 800 Atom Comp ICAPES (Induction Coupled Argon Plasma Emission Spectrometer) with matrix matched standards. Data for all procedures were adjusted for procedural blank concentrations, which were relatively low. Accuracy and precision for the total digestion procedure used for total and "residual" metals were estimated through repeated analysis of NBS 1645 (River Sediment) and 1646 (Estuarine Sediment) (Table 1). During development of the extraction procedures, bulk (not size-fractionated) samples from each of the 1987 sample sites went through the chemical extraction procedure, and splits of those samples were digested totally with the microwave technique. The sum of the concentrations in the individual extraction steps was generally within 10% of the measured totalmetal concentration for all elements analyzed. Duplicate acetic acidhydroxylamine extracts of separate splits of bulk samples and size fractions gave values within 6% of their means for Cu, Fe, and Zn, and 10% for Mn, Pb, and Ni. Duplicate potassium chlorate-hydrochloric acid extracts gave values within 6% of their means for Cu, and Mn, and 10% for Pb, Zn, and Fe. Duplicate Cd and Ni analyses were generally within 10% of the mean, but concentrations of these metals were near detection limits in some cases, where precision significantly worsened. Results of repeated analysis of laboratory prepared

252 TABLE 2

Detection limits for water (ugm1-1) and sediment(ttgg-l) analytical proceduresa Water

Sediment extractions "Reducible.

As Cd Cu Fe Mn Ni Pb Zn

0.019 0.004 0.002 0.005 0.001 0.004 0.012 0.001

.

.

.

Oxidizable

(NH2 H C L H O A c )

(KC103-HC1)

6.4 0.38 0.11 0.32 0.05 0.72 2.67 0.51

1.76 0.36 0.15 1.15 0.08 0.75 3.18 0.09

"Residual"-total

(HF-HC1-HNO3) 1.18 1.20 0.001 b 0.22 3.3 0.46

a D e t e c t i o n l i m i t = 3 SD of r e a g e n t b l a n k .

bWt. %. standards in the acetic acid-hydroxylamine and potassium chlorate-hydrochloric acid matrices gave accuracy within 4% and percent relative standard deviation of better than 4%. Precision and accuracy of water analysis were monitored through repeated analysis of USGS (United States Geological Survey) water standards T-91 and T-95, and both were generally within 10-15%. Concentrations in the river water used for sieving and in USGS water standards were often near ICAPES detection limits, however. Water data are used only to estimate the amount of metals mobilized during sieving, and the results of those calculations may be subject to a significant amount of error. As with the particle-size data, conclusions relying on this water data are not significantly affected by analytical error (see below). Detection limits for water and sediment analysis (defined as three standard deviations about the mean of the reagent blank concentration) are shown in Table 2. Total organic carbon (TOC) was determined for subsamples of all size fractions of 1987 bed-sediment samples at Rinehart Labs (Arvada, CO). Laboratory reported recovery was 98.5-100.4%, with measured concentrations in duplicate samples differing less than 5% from the mean. Due to small sample sizes, precision may be worse for samples containing less than ~ 0.2% TOC.

Metals loss during sieving Filtered aliquots of river water used for sieving 1987 samples and the water passing through the super centrifuge at the end of size-separation procedures were analyzed to investigate the possibility of metal and arsenic mobilization during size separation. These data, with the initial dry weight of the sample (measured on a separate split of wet sample), allow rough calculation of the amount of metals mobilized during sieving. As mentioned above, these data

253

4000I 3500

Cd 1

Cu

3000 2500 ~g/g

Fg/g

2000 1500 1000 500 0

0

" 2

'11"22'59'75

"162'

1 " 2

" 3 "

0

2

11

22

59

75

162

1

2

3

12000,

Mn

8000,

wt. ~,

~g/g

6000,

4000

2000

0

2

11

22

59

75

162

1

2

3

0

2

11

22

59

75

162

1

2

3

0

2

11

22

59

75

162

1

2

3

0

2

11

22

59

75

162

1

2

3

~g/g

Fig. 2. Total metals in six size fractions of 1986 bed-sediment samples. Each histogram gives concentrations (from left to right) in > 300, 30(P63, 6~38, 38-17, 17~4, and 4-0.1 #m size fractions. Samples labeled 1, 2 and 3 at the right end of each plot are in Clark Fork River tributaries. See Fig. 1 for sample locations. 1Missing values for Cd were below detection.

254 TABLE 3 Particle-size d i s t r i b u t i o n s for 1986 and 1987 bed-sediment samples 1986 samples Location a

> 300 #m

30(~63

63-38

38-17

17-4

<4

0 km 2 11 22 59 75 162 1 2 3

55.63% 55.03 42.35 61.14 20.68 64.44 73.31 42.91 28.61 32.26

40.67% 41.24 50.15 31.08 67.42 21.75 22.40 47.96 49.50 49.25

2.19% 2.91 5.14 4.56 9.44 6.36 2.88 5.66 16.87 15.01

0.55% 0.37 1.19 1.72 1.47 3.78 0.64 1.40 3.18 1.95

0.21% 0.15 0.50 0.57 0.36 1.35 0.23 0.52 1.01 0.78

0.04% 0.15 0.29 0.34 0.27 0.42 0.16 0.26 0.55 0.22

Location a

> 300 #m

300-63

63-38

38-17

< 17

0 km 22 35 59 91 188

14.59% 4.95 6.27 3.77 2.15 36.02

78.63% 89.47 77.60 83.08 15.62 54.06

5.87% 5.39 13.29 11.36 32.84 7.61

0.59% 0.77 1.80 1.02 28.89 1.42

0.25% 0.64 0.99 0.50 20.80 0.57

1987 samples

See Fig. 1 for sample locations. 1 = Little Blackfoot R., 2 = Flint Creek, 3 = Blackfoot R.

contain significant ambiguity, however sieving loss generally appeared to be minimal. For most of the samples, these calculations indicate that not more than 1% (Cu and Fe), 3% (Mn) and 7% (Zn) of the lowest concentration (of the five size fractions) in the acetic acid-hydroxylamine extract were mobilized during sieving. Other workers have reported minimal metals loss during sieving procedures (Forstner and Patchineelum, 1980; Jenne et al., 1980). Deionized water blanks passed through the Sharples centrifuge often contained significant amounts of zinc, so some zinc attributed to mobilization from the sediment may come from metal parts of the centrifuge. Zinc contamination from filters or filtration equipment has also been noticed during other procedures in our laboratory. Concentrations of As, Ni, Cd, and Pb in the supernatant water were generally below detection limits (see Table 2), however some mobilization of these elements during sieving may have occurred. In addition, in two cases, metal mobilization by sieving procedures did appear significant. A large amount of Mn ( ~ 20% of the total bulk concentration) was liberated during separation of the sample at 59 km, and a large amount of Ni (25% of the total bulk concentration) was liberated during separation of the sample at 91 km. The amount of mobilization of other metals during separation of these samples was not significant, however. These large values for metal removal during sieving are unusual, and hard to explain because they are

255 I •

Residual

[ ] Oxidizable

[ ] Reducible

8o, As'

7o, 6o,

50

Fg/g

~g/g 40 3o 20 lO 0

22

55

59

91

188

0

22

35

5g

gt

1B5

0

22

55

59

91

188

0

22

35

59

gl

188

0

22

55

59

91

188

Pg/g

~g/g

Pg/g

0

22

35

59

91

188 1400T 1

~g/g

0

22

35

59

91

188

0

22

35

59

91

188

Fig. 3. Sequential extractions of bulk 1987 bed-sediment samples. See Fig. 1 for sample locations. ~Residual concentrations of Cd were below detection; residual concentrations of Pb and As are not

reported (see text).

Fg/g C u

x

10 3

t~g/g

Cd 1

~

Fg/g

As1

o

Illlllll]lll~l

~-

~,!lllll]lll~

~ '~iiiiiiiiiii

~ ~]]]]]]]][~

~ ~]]llllllLlllllllllllillllllllllllllllllllllllllllll]llll~

,.,,~+~]~

,~ Jlllllllllllllllllllllllll[ll[llllllllllll[llllllllll]ll~l~

o

.

i

a

~ i

~ 1

~

~,,

IIIIlll[llLIIIlll~

|

o

o

i

i

o

o

+

~ IIIIIIIml~

~,~iiiiiiiiiii o b,

•~

o

1

UIIIIIIIl~!t

- IIIlltlIIIIIIIIIIMILH~

.--,.

o

~

o

~,

~

~

0 "++

IIIII[IIIIIH Illlllllli~.i~:~:~:~:i ::..'

m ?

~

-

~

~

~~ ,

-IIIII1~ ~1151111[11[1111[1~

:[

.

.

.

.

.

÷

====================================================

Z~ ....

~~

~IIIIIIIII[H ~lllll[ll[lil[l[ ~1111[11111111111[1111111115

+

~lllll]]ll]llllll~ o

o°

~

~o

o

~

,~

~

o

''

+

+

+

o

~

~

"

2~ . . . . o

?~

o ~

o

~,

~

~,

~

~

o=

o°

"1111111111 ÷

m IIIIm$111LII$11HHIM[IlmulmI~t

::::::::::::::::::::::::

~ Jllllllllllll[ll$$11lli:,:,i :,i :,i :,:,:,i

9g~

t~g/g Ni |

o

~

r~

Fg/g

Mn

~

4~ U l m

x

wt.

10 3 -~

00 ~

o

o

-

~

c,4

~Fo .,,..

~.

c..,

m

-~

'

?

o

~lllEiiiiiiii,iiiiiiiiiiiiiiiiiiiiiii n

iE ~TT~T

n

i[. "°* o

o

o

~II!!!!!!![!!!!! i

u

r

L~Z

n

n

"L

o

n

-,

i

-

~

~

-

~,.

. , 0~"

wt. ~ Size Fm~-~i~n

•

~'~

'

'

'

~.

Ftg/g Zn x 10 3

/Zg/9 Pbl

'

i *~ llllJIIIIIIti~ [ ;:i ::i ::;:i ;:::] ii c.,,~

~_ ~.~

o, ~- ,~

,~ HIIIIIIIIIII~ o,~+llfllllllllllllI.-'i

~

,~ ,~

o~~,~

,.o !I:....

~

I

~o

o,!111111"!!!!!!!!!!!

|

|

|

?~ll~ "-

i

o, "+IIHIE

~

.~

,~-

,~

'~ imi]l~l

'..,|i~FH

0

ii ~' ~' ~ ~'i ~

"~iifllilf[- :#:::"

O~lUfllllllJi'~iiiii !iii iiii iii

o, Illllllllllflii~ ! ! !! ! ] ~

.

~,pua~

i

i

o~ ~ ~o ~

I I

~. o,~lllllllllllllHfllE- i i i i i i i ! ! ! ! ! ! i! ! !:

259

restricted to two metals in two separate samples. The data are not corrected for this apparent loss. RESULTS AND DISCUSSION Metal enrichment

Total-metal concentrations in size-fractionated 1986 bed sediment samples (Fig. 2) display trends similar to those found in other studies of Clark Fork River sediments (Axtmann and Luoma, 1987; Moore, 1987; Moore et al., 1988a). Cadmium, copper, manganese, and zinc are enriched over tributary values. Relatively higher concentrations of cadmium, copper, manganese, and zinc in Flint Creek (see Fig. 2) have been noted by other workers (Moore et al., 1988a; Axtmann and Luoma, 1987) and are presumably due to mineralization and small mining operations in that drainage (Ingman and Bahls, 1979). Arsenic and lead data for these samples are not available. Particle-size analysis (Table 3) shows that most of these samples are composed of a large amount of sand ( > 63/~m material). These particle-size data, combined with the data in Fig. 2, suggest that although the highest metal concentrations occur in the 4~.1 pm fraction, the sand fraction contributes a large percentage to the total metal concentrations of these samples. Sequential extractions of bulk 1987 bed-sediment samples (Fig. 3) show that enriched cadmium, copper, manganese, and zinc are present in predominantly "reducible" and "oxidizable" phases, with relatively small concentrations in "residual" phases, while iron and nickel have larger contributions from '~residual" phases. The "reducible" phase dominates for manganese, and to a lesser extent for zinc, and copper is concentrated approximately equally (samples at 0, 22, 35, and 59 km) in the two non-"residual '' phases, or mostly in the '~oxidizable" phase (samples at 91 and 188 km). Concentrations of arsenic, lead, and cadmium in "reducible" and "oxidizable" phases only are plotted in Fig. 3; residual cadmium concentrations were below detection limits, and lead and arsenic analyses in the "residual phase" suffer from poor precision (lead), and significant interference in the ICP from aluminium (arsenic). Non~'residual" arsenic is fractionated approximately equally in "reducible" and '~oxidizable" phases, with the exception of higher "oxidizable" concentrations in the sample at 0km. Concentrations of lead in the "reducible" phase are generally higher than concentrations in the "oxidizable" phases, again with the sample at 0km as an exception. Although "residual" lead data are not presented, lead has a relatively significant "residual' phase ( ~ 30-60 #g g-l), and Andrews (1987) indicates that "residual" arsenic concentrations in Clark Fork River sediments are relatively low ( ~ 1-30 pg g-l). Andrews (1987) also describes a pattern of chemical fractionation between "residual" and non"residual" phases similar to that shown in Fig. 3, however in that study residual concentrations were determined by differences between "total" and extractable concentrations, using less severe chemical techniques to determine "total" concentrations.

260

0,4

T

0.3

SA/V (/Jm-1 ) 0.2

0.1 ~ 1 2 3 4 5 Fig. 5. Theoretical surface area/volume trend for spherical particles in the size ranges considered in this study. Surface area was calculated using the mid-point of each size class (500 ym was used for the > 300 ttm size class).

Chemical speciation and particle size effects Sequential extraction of size-fractionated bed sediments (Fig. 4) shows chemical speciation trends consistent with the bulk sample analysis (see Fig. 3). The significantly higher concentration of nickel in the size-fractionated 91km sample than in the bulk 91 km sample is presumably a result of sample inhomogeneity. Results of particle size-analysis (see Table 3 and Fig. 4) show that these samples are also composed predominantly of sand-sized material, with the exception of the sample at 91km. In addition, the sequential extraction of individual size fractions provides more information about particle-size control of metal concentrations in these sediments. Metal concentrations in the "reducible" phase are strongly dependent on particle size, with significantly higher concentrations of arsenic, cadmium, copper, iron, manganese, nickel, lead, and zinc in most of the 17-0.1pm fractions than in all coarser size fractions. The acetic acid-hydroxylamine extraction should liberate metals bound in non- or partially-crystalline iron and manganese oxide or hydrous oxide phases [highly crystalline oxides may not totally dissolve (Luoma and Davis, 1983)], carbonate associated metals, and exchangeable metals (Chester and Hughes, 1967; Tessier et al., 1979, 1982). Of these three, the oxide coatings are probably the most important scavengers of trace metals and arsenic in sediments (see: Chao and Theobold, 1976; Jenne, 1976; Lion et al., 1982; Luoma and Davis, 1983; Laxen, 1985; Takamatsu et al., 1985; Peterson and Carpenter, 1986) and should be present predominantly as coatings on particle surfaces or as small particulates (Chao and Theobold, 1976; Jenne, 1976; Salomons and Forstner, 1984). Predicted surface area/volume trends for the size classes of particles considered here suggest a significantly higher concentration of metals bound in surface area controlled chemical phases in the 17-0.1#m size fraction (Fig. 5). This increase is present in much of our data (see Figs 4 and 5). "Reducible" metal concentrations in coarser fractions are often higher than might be predicted (see Figs 4 and 5), however, suggesting enrichment of iron

261 and manganese coatings in these size fractions. The presence of orange-brown and black coatings on many coarse quartz and feldspar grains in these samples' supports this conclusion. Enrichment of this phase in the coarse fraction is also consistent with observations of coarse-fraction enrichment in other river sediments (Whitney, 1975; Gibbs, 1977; Wilber and Hunter, 1979; Tessier et al., 1982). This effect is probably due to an increase in iron and manganese coating thickness with increasing particle size, either due to long residence times of coarse particles in oxygenated, relatively high velocity currents (Whitney, 1975; Tessier et al., 1982), or the formation of iron and manganese oxide coatings in the weathering environment (e.g. floodplain or other sediments and soils), where coarser, more permeable sediments receive a greater supply of precipitating ions than finer, less permeable material (Gibbs, 1977). In our samples at 22, 35, and 59 km, higher concentrations of arsenic, cadmium (22 and 59km only), copper, nickel, lead, and zinc in the "reducible" phase of the >300pm fraction are mirrored by higher concentrations of iron and manganese in the ~'reducible" phase of this size fraction, lending support to the hypothesis that iron and manganese oxide coatings are concentrated in coarser fractions. These data support the suggestion that particle size plays a dual role in controlling the concentrations of metals in the "reducible" phase of fluvial sediments. Increases in surface area/volume tend to increase concentrations, however, because coarser particles may spend more time in chemical environments conducive to the formation of iron and manganese oxide coatings, concentrations in coarser sediment fractions are higher than predicted. This conclusion is also supported by previous work with size fractions of Clark Fork River sediments (Wilhelm, 1986, see Andrews, 1987). The exact mechanism of this coarse-fraction enrichment is unclear; Moore et al. (1988a), however, have shown that metals in Clark Fork River floodplain sediments do not show expected relationships between metal concentrations and particle-size distributions. This conclusion at least suggests that the growth of oxide coatings on coarse particles in floodplain sediments may explain some of the coarse-fraction enrichment in river-bed sediments. Metals in "oxidizable" phases should be present predominantly as diagenetic or detrital sulfides, and chemically bound with organic material. Sampling was intended to collect only oxidized, surface sediment, but the oxidized layer in these sediments is often extremely thin, and the possibility that some sediments containing diagenetic sulfides may have been collected cannot be ruled out. Metal concentrations in this chemical phase generally increase with decreasing particle size, however there are important exceptions to this trend (see Fig. 4), most notably nickel, zinc, iron, copper, and lead at 22 and 59km, and arsenic at 22, 35, and 59km. In addition, increases with decreasing particle size are generally not of the magnitude of increases in the "reducible" phase, except in the case of copper (and arsenic at 0km), where significant concentrations of metals and arsenic in the "oxidizable" phase occur in the 17-0.1 pm fraction. Copper also has the highest percentage of metal in this phase. Many workers have observed high concentrations of copper in

262 TABLE 4 Total o r g a n i c c a r b o n c o n t e n t (wt. %) of size.fractionated 1987 bed-sediment samples Location

> 300 pm

300~3

63-38

38-17

< 17

0 km 22 35 59 91 188

0.50 8.56 0.49 1.14

0.28 0.30 0.25 0.21 2.95 0.33

0.73 1.79 0.40 0.87 2.68 1.03

2.77 4.52 1.26 2.41 2.52 1.13

5.22 7.31 2.94 5.14 5.13 4.99

0.75

organic material in sediments Mantoura et al., 1978; Nriagu and Coker, 1980; Tessier et al., 1982), probably due to high stability constants for copper-organic complexes (Davis, 1984; Salomons and Forstner, 1984). Arsenic is often scavenged by iron and manganese oxide compounds in oxidizing sediments (Takamatsu et al., 1985; Peterson and Carpenter, 1986); the high "oxidizable" concentrations of arsenic at 0 and 188 km could be a result of the presence of a minor amount of arsenic-containing detrital sulfide minerals that were present in Butte ores [e.g. enargite (Cu3AsS4) or arsenopyrite (FeAsS)]. Manganese has relatively small "oxidizable" concentrations, while arsenic, cadmium, iron, lead, nickel, and zinc have significant concentrations in "oxidizable" phases (see Figs 3 and 4). The causes of trends in "oxidizable" metal concentrations related to particle size are difficult to assess because the two important phases, sulfides and organic material, are not discriminated. Total organic carbon data (Table 4) show trends related to particle size that are quite similar to trends of metal concentrations in the "oxidizable" phase (see Fig. 4). Organic carbon should be concentrated in smaller size fractions (Brownlow, 1972); the elevated concentrations of organic carbon in the > 300 ttm fraction of some of these samples (see Table 4) are due to entrapment of coarse plant material in the > 300ttm sieve. The presence of this coarse plant debris satisfactorily explains the anomalous concentrations of "oxidizable metals" in coarse size fractions (see Fig. 4). The relatively high organic carbon content of the 91km sample (see Tables 3 and 4) explains the large concentration of copper in the "oxidizable" phase of the 91km sample (see Fig. 3). The relative proportions of metals in the "oxidizable" fraction of the bulk samples (Fig. 3) support the conclusion that a significant fraction of the metals in "oxidizable" phases is bound by organic material. Whgn considered as a fraction of the non-residual phase, this proportion generally decreases in the order Fe >~ Cu /> As ~> Pb >/ Ni > Zn > Cd > Mn. This sequence partially resembles experimental and field observations of trends to the strength of metal binding by natural organic material (Mantoura et al., 1978; Kerndoff and Schnitzer, 1980; Nriagu and Coker, 1980). Given the large concentrations of metals in "oxidizable" phases in these samples, however, the possible contribution from metals bound in detrital or diagenetic sulfides cannot be discounted. Optical examination did not reveal the presence of any detrital sulfide

263 TABLE 5 Metal concentrations in shales, background Clark Fork River tributaries, and Clark Fork River "residual" fractions (all values in #g g 1) Shales a

Clark Fork Tributaries b

C l a r k F o r k " r e s i d u a l ''c

As Cd Cu Mn

0.2 45 600

0.40 b < 1.2 14 (4) 354 (78)

< 1.2 14 (4) 275 (65)

Ni Pb Zn

68 20 95

15 (2) 20b 33 (20)

11 (3) 57 (21)

a D a t a from T u r e k i a n a n d W e d e p o h l (1961). Compiled in S a l o m o n s a n d F o r s t n e r (1984). bE. Brook, u n p u b l i s h e d data. M e a n s of t o t a l c o n c e n t r a t i o n s of 12 s a m p l e s collected in Little B l a c k f o o t a n d B l a c k f o o t R i v e r s (see Fig. 1). V a l u e s in p a r e n t h e s e s = ls. Pb a n d As d a t a a r e from A n d r e w s (1987) for two s a m p l e s (Blackfoot a n d Little B l a c k f o o t Rivers). c M e a n s of r e s i d u a l c o n c e n t r a t i o n s in six b u l k (1987) s a m p l e s described in t h i s s t u d y (see Fig. 3). V a l u e s in p a r e n t h e s e s = ls.

minerals, however these predominantly soft compounds would be expected in finer particle-size fractions, where they are difficult to identify. Metals in the "residual" fraction should come primarily from silicate and other resistant minerals. Residual concentrations are similar to concentrations measured in Clark Fork River t r i b u t a r y sediments (Andrews, 1987; Axtmann and Luoma, 1987; Moore et al., 1988a; E. Brook, unpublished data). These background concentrations are in t urn similar to, but slightly lower than, global background concentrations in shales (Table 5). Salomons and F o r s tn er (1984) suggest that the shale values are, in general, slightly higher th an appropriate background concentrations for river sediments, and this seems to be the case in the Clark Fork River as well. As might be expected if metals in the "residual" phase are present mostly in lattice positions in minerals, there is little particle size control of "residual" metal concentrations (see Fig. 4). CONCLUSIONS Results of sequential extraction show t hat metal enrichment in Clark Fork River sediments is present predominantly in "reducible" and "oxidizable" phases. This observation is consistent with the source of metal enrichment; most of the excess concentrations are probably the result of weathering of unstable metal sulfide minerals extracted from the B u t t e - A n a c o n d a mining district, or the presence or detrital, crystalline sulfide minerals in river sediments, neither of which should contribute to "residual" phases. For the enriched metals cadmium, manganese, and zinc, the strong trend of increasing c o n c e n t r a t i o n with decreasing particle size has the greatest contribution from the "reducible" phase, presumably mostly iron and manganese

264 oxide compounds. These should be present as coatings on particles or as very small particulates, and therefore concentrations of metals in this phase should show strong increases with decreasing particle size. Lead, arsenic, and copper have significant contributions to this increase from the "oxidizable" phase as well. Iron and nickel are not significantly enriched, and show weak particle size trends; increases in c onc e nt r at i on with decreasing particle size are relatively minor (see Figs 2 and 4). The lack of large, non-reducible concentrations of nickel and iron, and therefore a lack of a significant surface-area controlled phase C'reducible" or ~'oxidizable" phases) is a probable explanation for the lack of a significant particle-size effect on nickel and iron concentrations. The relationships between particle size and metal concentrations in these sediments (Figs 2 and 4) are notable because of enrichment of metal concentrations in coarse size fractions. Metal enrichment in the coarse fractions of these sediments is emphasized here because of the coarse n a t u r e of Clark Fork River sediment; even though sampling in this study was targeted at the finest sediment in the river bed that was available in significant quantities, these samples are predominantly > 63 gm material. Comparison of the particle-size distributions of these samples (Table 3 and Fig. 4) with the corresponding extractable metal concentrations, indicates that, as in the case of the 1986 samples, most of the metals in these sediments are present in the three coarsest size fractions (with the sample at 91km as an exception). A large number of researchers have recommended t ha t sampling and analysis of sediment for metal contaminants be confined to a specific fine-size fraction (Oliver, 1973; de Groot et al., 1982; Salomons and Forstner, 1984), typically < 63 or 20 ttm. While the necessity of reducing particle-size effects on vertical or lateral trends of sediment-associated metals is apparent, studies of transport or ecological impact of metal contaminants in coarse-grained river systems should not overlook the potentially significant contribution to the total metal content of the system from the dominant particle-size fraction. ACKNOWLEDGEMENTS This research was funded through grants from Sigma Xi, The Geological Society of America, and The M o n t a n a Water Resources Research Center. Acquisition of the J a r r e l - A s h ICAPES was partially supported by NSF Instrumentation Grant No. EAR-8606061, MONTS, University of Montana, and ARCO Foundation funds. J o h n Cuplin assisted in the preparation of some of the figures. REFERENCES Agemian, H. and A.S.Y. Chau, 1976. Evaluation of extraction techniques for the determination of metals in aquatic sediments. Analyst, 101 (1207) 761-767. Andrews, E.D., 1987.Longitudinal dispersion of metals in the Clark Fork River, Montana. In: R.C. Averett and D.M. Mcknight (Eds), The Chemical Quality of Water and the HydrologicCycle. Lewis Publishers, Ann Arbor, Michigan.

265 Axtmann, E.V. and S.N. Luoma, 1987. Trace metal distributions in floodplain and fine grained bed sediments of the Clark Fork River, Montana. Proceedings, 6th Annual Conference on Heavy Metals in the Environment, New Orleans, September, 1987. CEP Consultants, Edinburgh, pp. 494-496. Brook, E.J. and J.N. Moore, 1987. Relationships between particle size and metal contamination in contaminated bed sediment of the Clark Fork River, western M o n t a n a (Abstr.). Geological Society of America Abstracts with Programs, 19(5), March, 1987. Brownlow, A.H., 1979. Geochemistry. Prentice-Hall, Inc., NJ. Chao, T.T. and R. Sanzolone, 1977. Chemical dissolution of sulfide minerals. J. Res. U.S. Geol. Sur. 5(4): 409412. Chao, T.T. and P.K. Theobald, 1976. The significance of secondary iron and manganese oxides is geochemical exploration. Econ. Geol. 71: 1560-1569. Chester R. and M.J. Hughes, 1967. A chemical technique for the separation of ferro-manganese minerals, carbonates, and adsorbed trace metals from pelagic sediments. Chem. Geol. 2: 249-302. Davis, J., 1984. Complexation of trace metals by adsorbed n a t u r a l organic matter. Geochim. Cosmochim. Acta, 48: 679-691. Filipek, L. and D. Owen, 1979. Geochemical associations and grain size partitioning of heavy metals in sediments. Chem. Geol. 26: 10~117. Filipek, L., T.T. Chau and P.K. Theolbald, 1982. Comparison of hot hydroxylamine hydrochloride and oxalic acid leaching of stream sediment and coated rock samples as anomaly e n h a n c e m e n t techniques. J. Geochem. Explor. 17: 3547. Forstner, U. and S.R. Patchineelum, 1980. Chemical associations of heavy metals in polluted sediments from the lower Rhine River. In: M.C. K a v a n a u g h and J.O. Leckie (Eds), Particulates in Water. American Chemical Society Advances in Chemistry Series, 189: 177-193. Gibbs, R.J., 1977. Transport phases of transition metals in the Amazon and Yukon rivers. GSA Bull. 88: 829-843. Groot, de A.J., K.H. Zchuppe and W. Salomons, 1982. Standardization of methods of analysis for heavy metals in sediments. Hydrobiologia, 93: 689-695. Ingman, G. and L. Bahls, 1979. An assessment of mining impacts on water quality in the Flint Creek Range, Montana. Unpublished Report, Water Quality Bureau, M o n t a n a Department of Health and Environmental Sciences, Helena, Montana. Jenne, E.A., 1976. Trace element sorption by sediments and soils sites and processes. In: R. Chappel and K. Peterson (Eds), Molybdenum in the Environment. Marcell Dekker, New York, Vol. 2, pp. 42~553. Jenne, E.A., V.C. Kennedy, J.M. Burchard and J.W. Ball, 1980. Sediment collection and processing for selective extraction and for total trace element analysis. In: R.A. Baker (Ed.), Contaminants and Sediments. Ann Arbor Science Publishers, Ann Arbor, Michigan, Vol. 2, pp. 169-191. Kerndorf, H. and M. Schnitzer, 1980. Sorption of metals on humic acid. Geochim. Cosmochim. Acta, 44: 1701-1708. Lambing, J., 1987. Water quality data for the Clark Fork River and selected tributaries from Deer Lodge to Milltown, Montana, M a r c h 1985 through June, 1986. U.S. Geological Survery OpenFile Report 87-110, M a r c h 1987. Laxen, D.H., 1985. Trace metal adsorption/coprecipitation on hydrous ferric oxide under realistic conditions: the role of humic substances. Water Res. 19(10): 1229-1236. Lion, L.W., R.S. Altman and J.O. Leckie, 1982. Trace metal sorption characteristics of estuarine particulate matter: evaluation of contributions from Fe]Mn oxide and organic surface coatings. Environ. Sci. Technol. 16: 660-666. Luoma, S.N. and J. Davis, 1983. Requirements for modeling trace metal partitioning in oxidized estuarine sediments. Mar. Chem. 12: 159-181. Mantoura, R.F.C., A. Dickson and J.P. Riley, 1978. The complexation of metals with humic substances in n a t u r a l waters. Estuarine Coastal Mar. Sci. 6: 387-408. Meyer, C., E.P. Shea and C.G. Goddard, 1968. Ore deposits at Butte, Montana. In: J. Ridge (Ed.), Ore Deposits of the United States, 1937-1967. The G r a n t o n Sales Volume, American Institute of Mining, Metallurgical, and Petroleum Engineers Inc., New York, pp. 137~1415.

266 Miller, R., 1973. Production history of the Butte district and geological function, past and present. Society of Economic Geologists Guidebook for the Butte field meetings. Economic Geology Publishing Company, Lancaster, PA, pp. F 1 - F l l . Moore, J.N., 1987. Processes of metal deposition and remobilization in a contaminated, high gradient river system: The Clark Fork River, Montana, USA. Proceedings, 6th Annual Conference on Heavy Metals in the Environment, New Orleans, September, 1987. CEP Consultants, Edinburgh, Vol. 2, pp. 175-177. Moore, J.N., E.J. Brook and C. Johns, 1988a. Grain size partitioning of metals in contaminated, coarse-grained floodplain sediment: Clark Fork River, Montana. Environ. Geol. Water Sci. (in press). Moore, J.N., W. Ficklin and C. J o h n s , 1988b. The role of sulfides in the partitioning of arsenic and heavy metals in reducing sediments. Environ. Sci. Technol., 22: 432~437. Nadkarni, R.A., 1984. Applications of microwave oven sample digestion in analysis. Anal. Chem. 56: 2233-2237. Nriagu, J.O. and R.D. Coker, 1980. Trace metals in humic and fulvic acids from Lake Ontario sediments. Environ. Sci. Technol., 14: 443~446. Oliver, B.G., 1973. Heavy metal levels of Ottowa and Rideau River sediments. Environ. Sci. Technol. 7: 135-137. Peterson, M.L. and R. Carpenter, 1986. Arsenic distributions in pore waters and sediments of Puget Sound, the Washington Coast and Saanich Inlet, B.C. Geochim. Cosmochim. Acta, 50: 353-369. Salomons, W. and U. Forstner, 1984. Metals in the Hydrocycle. Springer-Verlag, New York. Sondag, F., 1981. Selective extraction procedures applied to geochemical prospecting in an area contaminated by old mine workings. J. Geochem. Explor. 15: 645~52. Takamatsu, T., M. Kawashima and M. Koyama, 1985. The role of M n 2+ -rich hydrous manganese oxide in the accumulation of arsenic in lake sediments. Water Res., 19: 1029-1032. Tessier, A., P.G.C. Campbell and M. Bison, 1979. Sequential extraction procedure for the speciation of particulate trace metals. Anal. Chem. 51(7): 844~845. Tessier, A., P.G.C. Campbell and M. Bisson, 1982. Particulate trace metal speciation in stream sediments and relationships with grain size: implications for geochemical exploration. J. Geochem. Explor., 16: 77-104. Turekian, K.K. and K.T. Wedepohl, 1961. Distribution of the elements in some major units of the earth's crust. Geol. Soc. Am. Bull. 72: 175-192. Whilhelm, S.R., 1986. Smelter contaminated river sediments: Effect of particle size on heavy metal distribution, Clark Fork River, Anaconda to Missoula, MT. Unpublished senior thesis, Department of Geology, Princeton University, Princeton, New Jersey. Whitney, P.R., 1975. Relationships of manganese-iron oxides and associated heavy metals to grain size in stream sediments. J. Geochem. Explor., 4, 251 263. Wilber, W.G. and J.V. Hunter, 1979. Impact of urbanization on the distribution of heavy metals in bottom sediments of the Saddle River. Water Resour. Bull., 15(3): 790-800.