- Email: [email protected]
Field assessment of metal and sulfate fluxes during flooding of pre-oxidized mine tailings
Advances in Environmental Research 4 Ž2000. 295᎐306
Field assessment of metal and sulfate fluxes during flooding of pre-oxidized mine tailings U
Lionel J.J. Catalana, , Ernest K. Yanfulb, Luc St-Arnauda a
b
Noranda Inc., Technology Center, 240 Hymus Bl¨ d., Pointe Claire, Quebec, Canada H9R 1G5 Department of Ci¨ il and En¨ ironmental Engineering, The Uni¨ ersity of Western Ontario, London, Ontario, Canada N6A 5B9 Accepted 1 June 2000
Abstract This paper presents a field experiment performed at the Mattabi mine tailings site near Ignace, Ontario, Canada, to assess the evolution of water chemistry over flooded sulfide mine tailings. The tailings had oxidized for approximately 10 years prior to flooding. The experiment involved the construction of a 70 m2 test cell and a monitoring program that extended over more than 2 years. Metal and sulfate fluxes towards the water cover were calculated using measured solute concentrations, water cover volumes, and seepage rates. Results indicate that directly flooding pre-oxidized tailings can initially lead to the release of metals and sulfate to the water cover. However, dilution of the water cover by rain and snowmelt, flushing of the oxidation products by infiltration into the tailings, and removal of some metals by precipitation and sorption progressively reduced concentrations in the water cover below regulatory discharge limits. Because the tailings pore water was rich in ferrous iron, a thin layer of hydrous ferric oxide precipitated on the surface of the tailings, following the establishment of the water cover. The precipitate is believed to have removed zinc from the water cover through sorption or co-precipitation reactions. 䊚 2000 Elsevier Science Ltd. All rights reserved. Keywords: Oxidized tailings; Field experiment; Water cover; Flooding; Test cell; Flux; Metal; Sulfate
1. Introduction Underwater disposal is a common and effective method for reducing sulfide mineral oxidation and acid generation in mine tailings ŽRescan, 1989; Pedersen et al., 1997; Robertson et al., 1997.. Since the diffusion
U
Corresponding author. Department of Chemical Engineering, Lakehead University, 955 Oliver Rd., Thunder Bay, Ontario P7B 5E1 Canada. Fax: q1-807-343-8928. E-mail address: [email protected] ŽL.J. Catalan..
coefficient and solubility of oxygen in water are very low compared to their values in air, oxygen influx to the tailings is effectively reduced ŽDave ´ et al., 1997.. For weathered pre-oxidized tailings, the method usually involves the construction of containment structures such as dams and dykes around the tailings deposit to hold water. Flooding pre-oxidized tailings can enhance the release of metals and acidity from the tailings to the water cover, at least in the short term. Under such circumstances, the concentrations and loads of the released metals could be significant enough to warrant treatment of runoff and seepage leaving the tailings
1093-0191r00r$ - see front matter 䊚 2000 Elsevier Science Ltd. All rights reserved. PII: S 1 0 9 3 - 0 1 9 1 Ž 0 0 . 0 0 0 2 9 - 0
296
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
area for some time after the placement of the water cover. The purpose of a water cover over pre-oxidized tailings is threefold: Ž1. to reduce further oxidation of the tailings; Ž2. to provide sufficient improvement of the water cover quality to permit discharge of runoff to the surrounding environment without the need for secondary treatment after the transition period, that is, the period when previous oxidation products are being flushed out; and Ž3. to reduce over time metal and sulfate loads and treatment costs associated with seepage from the tailings. According to a recent review of water cover sites and research projects ŽYanful and Simms, 1997., flooding of oxidized tailings has already been implemented at a number of sites including three closed operations: the Solbec tailings site near Stratford, Quebec; the Quirke Lake tailings site in Elliot Lake, Ontario; and the Stekenjokk site in northern Sweden. Prior to flooding at the Solbec site, hydrated lime ŽCaŽOH.2 . was dumped in the tailings pond water, and calcite dust and granules ŽCaCO3 . were incorporated into the tailings by plowing to a depth of up to 30 cm in the area exposed to the atmosphere. On average, 230 t of alkaline material per hectare were applied to limit the solubility of metals ŽAmyot and Vezina, 1997.. The ´ water cover was implemented in 1994. Since the summer of 1995 all water quality parameters in the water cover have conformed to the applicable Quebec regulation. However, in the fall of 1996 the oxidized tailings pore water still contained iron and zinc concentrations of 150 mg ly1 and 9.3 mg ly1 , respectively. Further details on studies pertaining to the Solbec site are available in Karam and Guay Ž1994. and MEND Ž1994.. The Quirke Lake tailings site was flooded between 1992 and 1995. The water cover was designed as a series of five terraced cells separated by internal dykes. To neutralize the acidity generated from previously oxidized uranium tailings, lime was added to all cells, except to Cell 14 which is located upstream of the other cells. Within 2 years after flooding, the iron and zinc concentrations in the water cover of Cell 14 stabilized below 0.6 mg ly1 and 0.01 mg ly1 , respectively. The water cover pH varied between 6.5 and 7.1 during the same time period, and sulfate levels have shown a continuous decline since flooding was initiated. The mean water retention time in Cell 14 is estimated at approximately 3 months. Significant downward seepage in this cell, and the consequent flushing of oxidation products down into the pore water, are important in explaining the rapid attainment of discharge compliance standards in the water cover without lime addition. The high seepage loss is offset by diverting fresh water from a nearby lake and can be maintained in the long term. The quality of shallow pore water in the upper tailings also improved signifi-
cantly after flooding. The acidity declined from peak levels of 12 500 mg CaCO3 ly1 prior to flooding to a maximum of 240 mg CaCO 3 ly1 in 1996. Treatment requirements have increased in the short term due to increased seepage caused by flooding ŽKam et al., 1997.. At the Stekenjokk site, Zn᎐Cu tailings rich in pyrite and other sulfides were flooded in 1991. A small area of the tailings deposit Žapprox. 5% of the bottom of the pond. consists of material that has oxidized and weathered before the flooding. In the summer of 1995, average concentrations of Ca Ž20.0 mg ly1 ., S Ž11.1 mg ly1 ., Cd Ž0.69 mg ly1 ., Zn Ž139 mg ly1 ., and Ni Ž1.3 mg ly1 . in the water cover were higher than local background concentrations in surface water ŽLjungberg et al., 1997.. According to these authors, the most important source of Zn in the water cover is probably diffusion from the oxidized tailings pore water, where Zn concentrations are much higher than in the water cover. In addition, dissolution of gypsum in the oxidized tailings most likely accounts for the elevated concentrations of Ca and sulfate, both in the pore water and in the water cover. There are some indications that sulfide oxidation continues under the water cover Žalthough at a much smaller rate than in tailings exposed to the atmosphere., and also contributes to the pore water composition. The present paper describes a field experiment performed at the Mattabi mine tailings site located approximately 73 km north-east of Ignace, Ontario, Canada, to examine the impact of flooding previouslyoxidized sulfide mine tailings on the chemistry of the water cover without alkalinity addition to the water or to the tailings. The Mattabi tailings impoundment was active from 1972 to April 1991. During this period, approximately 15 500 000 t of tailings from three different ore bodies were end-spilled over an area of 125 ha. The experiment involved the construction and flooding of a test cell in a part of the tailings area that had been oxidizing for approximately 10 years. Metal and sulfate fluxes were calculated using water balance data and measured concentrations of metals and sulfate at different times during flooding of the test cell.
2. Materials and methods 2.1. Test cell construction and materials The field test cell was constructed in October 1991. The cell was 70 m= 70 m in plan, as measured between the centerlines of the perimeter dykes ŽFig. 1.. The dykes were constructed with glacial till compacted at the natural moisture content of 7᎐10%, slightly higher than the optimum water content of 7.5% based on the laboratory standard Proctor compaction test. The mea-
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
297
Fig. 1. Overall view of test cell. Photograph taken on 20 May 1992.
sured compacted dry density ranged from 2.0 to 2.2 Mg my3. Two 30-m long wooden rafts were placed at a 90⬚ angle, and fixed to the south and east embankments to provide access to the center of the test cell. The tailings in the cell measured approximately 54 m= 54 m in plan. Details of the construction can be found in Geocon Ž1993.. The tailings beneath the test cell had an average depth of approximately 14 m. Dry densities ranged from 1.5 to 2.1 Mg my3; specific gravity and porosity averaged 3.6 and 51%, respectively. The tailings were poorly graded and consisted, on average, of fine sand Ž60%., coarse sand Ž20%., and silt-sized Ž20%. particles. The average d50 was approximately 0.15 mm. The average chemical composition of oxidized Mattabi tailings samples determined by ICP is shown in Table 1. Combination of X-ray diffraction and chemical analysis results showed that sulfides accounted for 37% of the tailings mass. The dominant silicate minerals were quartz Ž21%., chlorite Ž11%., muscovite Ž10%., and albite Ž2.5%.. Gypsum Ž0.6%. and goethite Ž2.7%. were also identified. All percentages refer to the total tailings mass. Petrographic examination of the tailings ŽFreymond, 1994. revealed that pyrite was the domiTable 1 Average chemical composition of oxidized Mattabi tailings Element
Content Žgrg.
Al As Ca Cu Fe K Mg Mn Na Pb S Si Ti Zn
33 252 1785 2186 886 230 968 4701 6380 1076 2198 1444 199 288 141 156 1135 4686
nant sulfide mineral, followed by pyrrhotite, which was extensively oxidized near the surface of the tailings area. Trace amounts of sphalerite, arsenopyrite, and chalcopyrite were also present. Acid᎐base accounting ŽABA. tests were conducted on Mattabi tailings samples using the Sobek procedure ŽSobek et al., 1978.. These tests measure the overall balance between acidity potential ŽAP. and the neutralization potential ŽNP.. In a standard ABA test, the total sulfur is measured, and the AP is calculated based on the assumption that all the sulfur exists as pyrite and that the pyrite will oxidize completely to produce sulfuric acid. The NP is determined from the amount of acid that the sample neutralizes in a standardized digestion᎐back titration procedure. A sample is considered acid generating if the ratio NPrAP- 1.0 and AP ) 3.1 kg CaCO3 ty1 . If NPrAP) 3.0 or AP - 3.1 kg CaCO3 ty1 , the material is considered non-acid generating. Materials with 1.0- NPrAP- 3.0 and AP ) 3.1 kg CaCO3 ty1 are uncertain from an ABA point of view. With an average AP of 624 kg CaCO3 ty1 and an average NP of y0.2 kg CaCO3 ty1 , the oxidized Mattabi tailings were strongly acid generating. Their average paste pH was 3.6. The oxidized Mattabi tailings exhibited mineral oxidation and precipitation patterns similar to other tailings impoundments Že.g. Blowes and Jambor, 1990., where a sulfide-depletion zone is underlain by an active oxidation zone, and is further underlain by a zone affected by secondary mineral precipitation. The secondary mineral was typically amorphous ferric hydroxide, or a more crystalline ferric oxyhydroxide or hydroxysulfate ŽFreymond, 1994..
2.2. Test cell operation and monitoring During the winter following construction of the cell, snow and ice accumulated in the test cell. Snowmelt and rain during the spring elevated the water table from below to above the tailings surface. By July 1992, the depth of water covering the tailings had reached approximately 1 m in the deepest parts of the cell. On 7 July, surface water samples were collected in the
298
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
water cover. In this sampling event, as in the ones following, surface water samples were collected by lowering silicone tubing a few centimeters below the surface and pumping 500 ml of water using a portable peristaltic pump. Water samples were collected in duplicate or triplicate. The samples were filtered through 0.45-m filters, acidified, and then analyzed for Al, As, Ca, Cd, Cu, Fe, K, Mg, Mn, Na, Pb, S, Se, and Zn by inductively-coupled plasma atomic emission spectroscopy ŽICP-AES.. Concentration differences between replicate samples were generally small and within the limits of analytical error. Field measurements of solute concentration variations with depth in other water cover studies Že.g. Ljungberg et al., 1997. have demonstrated that the water column is well mixed during the ice-free season, due to wind-induced waves and natural convection. Moreover, because of the relatively small dimensions of the cell and the homogeneity of the tailings, it was assumed that the water chemistry would be independent of location in the cell. Since a closure option for the Mattabi site involved flooding the tailings with water from nearby Sturgeon Lake, rain and snowmelt water was pumped out of the cell between 7 and 9 July, 1992. Starting on 11 July, Sturgeon Lake water was pumped into the cell at a rate of 3500 l miny1 through a pipe located on the north side ŽFig. 2.. The total volume of water in the test cell after flooding was 4200 m3. The water depth was 1.0 m on the north side of the cell and 1.7 m on the south side, as the surface of the tailings in the test cell sloped at approximately 0.7⬚ downward from north to south. The surface water was sampled again on 14 July. Regular monitoring of water level in the cell began on 14 July and continued until 3 October, 1992. During this interval, precipitation, class A pan evaporation,
temperature, and humidity were also measured. There was no monitoring between October 1992 and July 1993. In July 1993, the test cell was instrumented with a pressure transducer to automate monitoring of the surface water elevation. Monitoring of the water level resumed on 22 July 1993 and continued until December 1993. The surface water was sampled on 25 July 1993. Starting on 26 July 1993, Sturgeon Lake water was again added to the cell. When the flooding was complete Ž3 August., the cell contained 5200 m3 of water. Additional surface water samples were collected on 1 August, 13 September, 12 October, and 3 November 1993. In 1994, no additional water was pumped to the cell. The last surface water samples were collected in July 1994. In addition to surface water samples, core samples of tailings were obtained before adding Sturgeon Lake water to the test cell in July 1992 and in July 1993. Cores were also obtained in July 1994. On each occasion, cores were collected at two locations within the test cell to approximately 1-m depth using thin-walled aluminum tubes. Slices of the cores were squeezed to recover pore water samples representative of various depth intervals. These samples were analyzed in the same manner as the surface water samples.
2.3. Seepage loss analysis Seepage loss from the test cell was established by carrying out a water balance analysis using the monitoring data collected during the test ŽGeocon, 1993..
Fig. 2. View of test cell with pipeline in place to supply fresh water from Sturgeon Lake. Photograph taken on 20 May 1992 before the water was pumped out Ž7᎐9 July, 1992..
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
The water balance was evaluated using the following equation: SL s Py Ey H
Ž1.
where SL is the seepage loss, P the factored precipitation, E the factored evaporation, and H the change in storage or cell water level Žnegative for drop.. The factored precipitation included an allowance for runoff from the interior slopes of the test cell dykes. This was achieved by increasing the measured precipitation by 10% Žthat is, using a factor of 1.1.. The resulting factored precipitation was then assumed to represent the rise in cell water level due to a precipitation event. Since the vertically projected area of the interior slope of the dykes represented approximately 20% of the water surface area in the cell, the factoring procedure translated into a runoff coefficient of 0.5 for the compacted till slopes of the dykes ŽGeocon, 1993.. This procedure assumed that there was negligible runoff from the crest of the dykes, a reasonable assumption since the crest was flat. To obtain the factored Žestimated. evaporation from the test cell, the measured class A evaporation was reduced by a factor of 0.8, which was slightly higher than the typical value of 0.7 used for lakes ŽFisheries and Environment Canada, 1978.. The value of 0.8 was used to allow for the much shallower depth of water in the test cell ŽGeocon, 1993.. As an example, for the period 14᎐17 July, 1992, the test cell water level dropped 33.8 mm and 21.4 mm of rainfall was recorded, while pan evaporation was measured to be 13 mm. Applying factors of 1.1 and 0.8 to precipitation and evaporation, respectively, and noting that the water level change is y33.8 mm, the seepage
299
loss is calculated to be wŽ21.4= 1.1. y Ž13 = 0.8. y Žy33.8.x mm, or ; 47 mm. The water surface area in the cell on 15 July Žmiddle of period. was estimated to be 3427 m2, and the elapsed time between measurements on 14 July and 17 July was 2.71 days. The rate of total seepage from the test cell for the period 14᎐17 327.42= 0.0470 July is then m3 dayy1 , or 59.4 m3 2.71 dayy1. Seepage rates were calculated for other days in a similar fashion.
3. Results 3.1. Water co¨ er chemistry The variation in the water cover volume with time for the monitoring periods of 1992 and 1993 is shown in Fig. 3. The elapsed time on the horizontal axis is calculated from 14 July, 1992. The volume reached a maximum after each flooding event, and then decreased during the following summer and fall, as evaporation and seepage exceeded precipitation. Average dissolved concentrations of selected elements in the water cover for each sampling event are shown in Table 2. Sturgeon Lake water-chemistry is also included for comparison. All solute concentrations decreased during the first flooding of the cell with Sturgeon Lake water in 1992, presumably as a result of dilution, but then increased between the end of flooding and 25 July, 1993. On that date, Fe and Zn reached elevated concentrations of 466 mg ly1 and 3.79 mg ly1 , respectively. During the second flooding Žbetween 26 July and 3 August, 1993., solute concentrations decreased, except
Fig. 3. Water cover volume in test cell vs. time.
300
Element
7 Jul 1992 Žbefore 1st flooding.
14 Jul 1992 Žafter 1st flooding.
25 Jul 1993 Žbefore 2nd flooding.
1 Aug 1993 Žduring 2nd flooding.
13 Sep 1993
12 Oct 1993
3 Nov 1993
Jul 1994
Sturgeon Lake water
Al As Ca Cd Cu Fe K Mg Mn Na Pb S Se Zn pH
9.08 - 0.3 257 - 0.02 0.12 974 6.7 233 29 11 - 0.15 1152 0.15 11.7 3.2
0.47 - 0.3 71 - 0.01 - 0.02 58 2.3 19 2.2 3.3 - 0.05 118 0.05 0.39
21.8 - 0.3 302 - 0.02 0.1 466 6 182 24 17 0.4 951 - 0.50 3.79
4.86 - 0.3 532 - 0.02 - 0.02 54 -5 194 6.1 25 - 0.25 806 - 0.5 0.69
3.03 - 0.3 627 - 0.02 - 0.02 0.58 5.8 196 3.2 24 - 0.25 791 0.5 0.74
0.56 - 0.3 596 - 0.02 - 0.02 0.12 -5 190 3 22 - 0.25 746 - 0.50 0.66 5.7
0.44 - 0.3 648 - 0.02 0.025 0.31 5.8 189 3.1 23 - 0.25 787 0.5 0.69
0.71 - 0.3 461 - 0.02 - 0.02 2.6 8 85 1.1 16 - 0.25 625 - 0.50 0.34 4.2
- 0.05 - 0.05 12.5 - 0.005 - 0.005 - 0.005 -1 1.8 0.001 3.9 - 0.05 6.22 - 0.1 - 0.005
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
Table 2 Average dissolved concentrations Žmg ly1 . of selected elements and pH in the water cover at various sampling dates and comparison with Sturgeon Lake water
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
for Ca and Mg. The evolution of solute concentrations after the second flooding differed markedly from that after the first flooding. Between 1 August, 1993 and July 1994, most solute concentrations decreased or remained stable. A slight increase in Al and Fe concentrations Žfrom 0.44 to 0.71 mg ly1 and from 0.31 to 2.6 mg ly1, respectively. occurred between November 1993 and July 1994. However, the concentration of the other solutes decreased or remained stable. The pH of the water cover increased as a result of flooding from 3.2 on 7 July, 1992 to 5.7 on 12 October, 1993, but then decreased to 4.2 in July 1994. Table 3 shows the evolution of dissolved masses of Al, Ca, Fe, Mg, Mn, S Žmostly as sulfate ., and Zn in the water cover on sampling events. These masses were determined as the product of solute concentrations and the volume of water in the cover at the various sampling dates. Note that solute masses could not be calculated for 7 July 1992 because the exact volume of the water cover was not measured on that day. All solute masses increased in the year that followed the first flooding of the cell. During the second flooding Žbetween 26 July and 3 August, 1993., the masses of Ca, Mg, Na and S increased dramatically, the masses of Al and Mn remained relatively stable, and the masses of Fe and Zn decreased. After the second flooding was complete, the masses of all solutes decreased with time. Iron differed from other solutes in that its mass began to increase again after 12 October, 1993.
3.2. Formation of hydrous ferric oxide precipitate layer On July 1992, before Sturgeon Lake water was pumped to the cell, a 3᎐4-mm thick iron hydroxide precipitate was observed coating the surface of the tailings below the water cover. This precipitate had formed during the previous spring when snowmelt and
301
rainwater accumulated in the cell. Chemical analysis of the precipitate revealed the presence of 2200 mg kgy1 of zinc. To determine the mobility of Zn, the sample was also leached using the Toxicity Characteristic Leaching Procedure ŽUS EPA, 1986., which involves shaking the sample in an acetic acid buffer solution at pH 4.9 for a period of 18 h, and then analyzing the leachate. The Zn concentration in the final leachate was 4.5 mg ly1 , which accounted for approximately 2% of the mass of zinc present in the precipitate. These results indicate that a fraction of the Zn initially released to the water cover was able to adsorb or co-precipitate within the hydrous ferric oxide precipitate.
3.3. Shallow pore water chemistry Table 4 shows the average concentrations of various elements and pH in shallow pore water extracted from cores collected on different dates during the experiment. The depth intervals for each set of concentration data are also indicated. Although significant differences exist between cores collected at two different locations in the cell on the same date, the overall trend is as follows: between 7 July, 1992 and July 1994 the concentrations of all measured solutes decreased, except for Ca and K which increased. This is consistent with the dissolution of oxidation products because of flooding and flushing due to infiltration from the water cover. The increase in pore water pH may have also led to the precipitation of some metals. The first flooding had a greater impact on shallow pore water concentrations than the second. As expected, most of the iron was present as ferrous iron in the pore water.
3.4. Metal and sulfate fluxes to the water co¨ er This section deals with estimates of the average flux of solutes to the water cover between sampling events. By definition, positive fluxes indicate a movement of
Table 3 Dissolved masses Žkg. of selected elements in the water cover at various sampling dates Element
14 Jul 1992 Žafter 1st flooding.
25 Jul 1993 Žbefore 2nd flooding.
1 Aug 1993 Žduring 2nd flooding.
13 Sep 1993
12 Oct 1993
3 Nov 1993
Jul 1994
Al Ca Fe Mg Mn Na S Zn
1.9 294 240 81 9.0 14 489 1.6
24.7 342 527 206 26.7 19 1076 4.3
24.1 2632 265 960 30.1 123 3989 3.4
9.1 1877 1.7 587 9.5 73 2367 2.2
1.2 1249 0.2 398 6.2 46 1563 1.4
0.9 1267 0.6 369 6.0 44 1540 1.3
1.7 1107 6.3 205 2.6 37 1500 0.8
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
302
Table 4 Dissolved concentrations Žmg ly1 . of selected elements and pH in shallow pore water at various sampling dates Element
Al As Ca Cd Cu Fe Fe 3q K Mg Mn Na Pb S Zn pH
7 Jul 1992 Žbefore 1st flooding.
25 Jul 1993 Žbefore 2nd flooding.
Jul 94
Depth 0᎐5 cm
Depth 0᎐20 cm
Depth 0᎐15 cm
15
1.4
413 0.56 0.90 2990
363 0.03 0.25 149
7.2 1080 151 110 0.98 3290 82 3.3
0.27 0.50 543 0.21 - 0.025 159 8.8 17 85 11 26 2.2 677 12 5.6
20 435 52 39 2.9 1093 20 3.7
solutes towards the water cover. The difference between the masses Mw Ž t2 . and Mw Ž t1 . of a given solute in the water cover determined at times t1 and t2 is related to the total solute flux FT to the water cover and the area of flooded tailings, ⍀, in the test cell as follows: Mw Ž t2 . y Mw Ž t1 . s FT ⍀ Ž t2 y t1 .
Ž2.
The total solute flux can be written as the sum of the advective flux of solute carried by the seepage FA, and the net flux of solute to the water cover FN : FT s FA q FN
Ž3.
The advective flux to the water cover is negative since seepage removes solutes from the water cover. It is related to the seepage rate S and the solute concentration in the water cover, Cw , between times t1 and t2 by: FA s y
1 ⍀ Ž t2 y t1 .
t2
Ht SC
w ⭈ dt
Ž4.
1
Combining Eqs. 2᎐4 yields the following expression for the net solute flux: FN s
1 ⍀ Ž t2 y t1 .
ž
Mw Ž t2 . y Mw Ž t1 . q
t2
Ht SC 1
w ⭈ dt
/
Ž5.
The net flux, if positive, represents the rate of solute release to the water cover due to dissolution or desorption from mineral phases present in the tailings or in the ferric hydroxide precipitate layer. Diffusion from pore water to the cover is not believed to have played a major role, given the significant downward advective
0.30 0.35 595 0.05 0.04 379 11.4 21.1 147 21.5 27.7 - 0.25 880 4.6 5.8
0.30 - 0.25 634 - 0.025 - 0.025 51 0.61 19 160 6.6 38 - 0.25 756 1.8
0.32 0.64 520 - 0.025 - 0.025 583 3.6 32 197 14.4 49 0.43 1056 4.8
fluxes due to seepage losses. A negative net solute flux indicates the rate of solute removal from the water cover by precipitation or adsorption on the ferric hydroxide precipitate layer. The integral in Eq. 5 can be estimated numerically from the plot of concentrations versus time and estimates of seepage rates between sampling events. Seepage losses were determined almost daily. Average seepage rates calculated over each week of monitoring are shown in Fig. 4. Calculation of net solute fluxes used average seepage rates for the periods between surface water sampling events in 1993 ŽTable 5.. As an illustration of the procedure for calculating the net fluxes, consider the estimated test cell volume and measured zinc concentrations in the water cover on 13 September and 12 October, 1993, shown in Table 2. From the measured zinc concentration of 0.74 mg ly1 Ži.e. 0.74 g my3 . and the test cell water volume of 2994 m3, for 13 Sept, 1993, the mass of zinc is 0.74= 2994, or 2215.6 g. Similarly, the mass of zinc in the water cover on 12 October is 1382.7 g. Using the estimated average seepage rate of 29.7 m3 dayy1 ŽTable 5. and an average zinc concentration of Ž0.74q 0.66.r2 mg ly1, or 0.70 mg ly1 , the net flux of zinc to
Table 5 Average seepage rates used in the calculation of solute fluxes Period
Seepage rate, S Žm3 dayy1 .
July 25᎐August 1, 1993 August 1᎐September 13, 1993 September 14᎐October 12, 1993 October 13᎐November 3, 1993
46.1 63.5 29.7 7.7
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
303
Fig. 4. Weekly average seepage rates vs. time.
the water cover during the 28 days between 13 Sept and 12 Oct, 1993, is calculated to be y0.0031 g my2 dayy1 as shown in Table 6. The negative sign means that zinc was actually removed from the water cover. Table 7 presents net fluxes for the various solutes between sampling events in 1993, calculated using the above procedure. Net solute fluxes were not calculated for 1992 due to lack of sufficient concentration data. A sensitivity analysis on net solute fluxes with respect to the seepage rates was performed, since seepage rates carry the most uncertainty among the parameters required to calculate net solute fluxes. Table 7 also shows net solute fluxes obtained assuming both an underestimation and an overestimation of the seepage rates by 10%. The results indicate that the order of magnitude and the sign of the net solute fluxes would generally not be affected by such an error in the seepage rate. The only exception is for Mg between 12 October and 3 November, 1993. The flooding of 1993 resulted in a net release Žby dissolution or desorption. of Al, Ca, Mg, Mn, Na and S Žas sulfate . to the water cover, as shown by the positive Table 6 Sample calculation of net solute fluxes a Date
Test cell water volume Žm3 .
Zn Žmg l -1 .
Zn Žg.
September 13, 1993 October 12, 1993
2994 2095
0.74 0.66
2215.6 1382.7
a Average seepage rate s 29.7 m3 dayy1 ŽTable 1.; Area of tailings surface s 54 m= 54 m. Net flux of zinc to water cover during 13 Sep᎐12 Oct, 1993 Ž28 days. using Eq. 5: wŽ138.27y 2215.6.grŽ28 days. q Ž29.7 m3 dayy1 . Ž0.74 q 0.66.r2 g my3 4rŽ54 m.rŽ54 m.x, or y0.0031 g my2 dayy1.
value of the corresponding net fluxes during the period 25 July to 1 August, 1993. In contrast, Fe and Zn displayed negative fluxes of y8.8 g my2 dayy1 and y0.007 g my2 dayy1, respectively, and were likely removed from the water cover by precipitation or sorption. As time progressed, however, the majority of fluxes decreased steadily. In the 2᎐3 months that followed, precipitation andror sorption reactions began removing the solutes that had previously been released to the cover during flooding. This is demonstrated by the fact that all the net fluxes were negative between 13 September and 12 October. Iron continued to be precipitated, but at a smaller rate than during flooding. Zinc appears to have been dissolved or desorbed at a small rate of 0.006 g my2 dayy1 just after flooding Ž1 August᎐13 September period., but the Zn net flux became negative again during the next period. During the last period Ž12 October᎐3 November., all solutes except aluminum displayed a positive, albeit small, net flux. The reason for this is not obvious. Table 8 compares the net solute fluxes to the water cover during the 1993 flooding Ž25 July to 1 August. with the solute concentrations both in the water cover and in the shallow pore water on the eve of flooding Ž25 July.. The data suggest that the net fluxes of some solutes to the water cover were not primarily controlled by diffusion to or from the pore water. For example, although the concentrations of Al, Mg, Mn and S were higher in the water cover than in the shallow pore water, the net flux of these solutes was towards the water cover Ži.e. positive.. In the case of Zn, the net flux was negative, although Zn concentrations were higher in the shallow pore water than in the water cover. These observations are consistent with solute
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
304
Table 7 Calculated net solute fluxes Žin g my2 dayy1 . for 1993 Solute
July 25᎐August 1, 1993
August 1᎐Sept 13, 1993
Sept 13᎐October 12, 1993
Oct 12᎐Nov 3, 1993
Al
0.18 a Ž0.16 b , 0.20 c . 118.8 Ž118.1, 119.4. y8.8 Žy9.2, y8.3. 39.9 Ž39.6, 40.2. 0.40 Ž0.38, 0.42. 5.40 Ž5.36, 5.43. 157 Ž155, 158. y0.007 Žy0.010, y0.003.
y0.033 Žy0.042, y0.024. 6.6 Ž5.3, 7.9. y1.51 Žy1.57, y1.45. 1.3 Ž0.9, 1.7. y0.063 Žy0.073, y0.052. 0.13 Ž0.08, 0.19. 4.3 Ž2.7, 6.3. 0.0060 Ž0.0045, 0.0076.
y0.078 Žy0.080, y0.077. y1.5 Žy2.1, y0.8. y0.015 Žy0.015, y0.014. y0.35 Žy0.55, y0.15. y0.009 Žy0.012, y0.006. y0.09 Žy0.11, y0.06. y2.0 Žy2.8, y1.2. y0.0031 Žy0.0038, y0.0023.
y0.0039 Žy0.040, y0.037. 1.9 Ž1.8, 2.1. 0.0065 Ž0.0065, 0.0066. 0.03 Žy0.02, 0.08. 0.0046 Ž0.0038, 0.0054. 0.029 Ž0.022, 0.034. 1.7 Ž1.5, 1.9. 0.0011 Ž0.0009, 0.0012.
Ca Fe Mg Mn Na S Zn a
Flux corresponding to the estimated seepage rate. Flux corresponding to 90% of the estimated seepage rate. c Flux corresponding to 110% of the estimated seepage rate. b
concentrations in the water cover being strongly affected by precipitation᎐adsorption᎐dissolution reactions within the hydrous ferric oxide precipitate covering the tailings surface. The pH and solute concentration data measured in the water cover on 12 October 1993 were used in geochemical equilibrium modeling of the water cover to calculate the saturation indices ŽSI. of various minerals. Minerals with saturation indices close to zero are possible candidates for controlling the concentration of the associated ions in solution. A large negative value indicates that the water is undersaturated with respect to the mineral and therefore the mineral will likely not precipitate. A high positive value is usually an indication that the mineral could theoretically form, but the
kinetics of the reactions controlling its formation are not favorable. Saturation indices for key minerals computed using the computer program MINTEQA2 ŽAllison et al., 1990. are presented in Table 9. SI values for zinc minerals were not tabulated, as they all indicated undersaturation ŽSI values of y6 or more negative.. The modeling results suggest that aluminum concentrations in the water cover were possibly controlled by AlŽOH.3Ža. or jurbanite ŽAlOHSO4 .. Saturation indices for gypsum ŽCaSO4 ⭈ 2H2 O. and ferrihydrite ŽFeŽOH.3 ⭈ 3H2 O. are positive and low, indicating that these minerals are likely present and could be controlling the concentrations of Ca, SO42y and Fe3q in the water cover. Zinc did not form any secondary minerals, and negative net
Table 8 Comparison of 1993 net solute fluxes to the water cover with solute concentrations in the shallow pore water and in the water cover Ž25 July. just before flooding
Al Ca Fe Mg Mn Na S Zn a
Pore water in top 20 cm of tailings on July 25, 1993 Žmg ly1 . a
Water cover on July 25, 1993 Žmg ly1 .
Average flux for July 25᎐Aug 1, 1993 Žg my2 dayy1 .
0.27, 0.30 543, 595 159, 379 85, 147 11, 21 26, 28 677, 879 11.7, 4.6
21.8 302 466 182 24 17 951 3.8
0.18 118.8 y8.8 39.9 0.40 5.40 157 y0.007
Values represent concentrations measured in pore water extracted from two core samples
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306 Table 9 Calculated mineral saturation indices Mineral
Saturation index ŽSI.
AlŽOH. 3 Ža. Jurbanite ŽAlOHSO4 . AlŽOH.10 SO4 Boehmite ŽAlŽOH. 3 . Ferrihydrite Gypsum Gibbsite Goethite Na-jarosite Diaspore
y0.783 0.309 6.32 0.978 0.591 0.313 0.992 4.420 1.679 2.818
fluxes of Zn are probably explained by adsorption onto hydrous ferric oxide precipitate.
4. Discussion and conclusions The field test cell results suggest that directly flooding previously oxidized sulfide mine tailings can lead to the release of metals and sulfate to the water cover. Iron and zinc concentrations in the water cover reached 466 and 3.8 mg ly1 , respectively, one year after the flooding of July 1992. After the second flooding, however, dilution of the water cover by rain and snowmelt, flushing of solutes by water infiltration from the cover to the tailings, and removal of some metals by precipitation and sorption progressively reduced the concentrations in the water cover. In July 1994, i.e. 2 years after the first flooding, metal concentrations in the water cover met regulatory discharge limits: As - 0.5 mg ly1 ; Cu - 0.3 mg ly1 ; Fe - 3 mg ly1 ; Pb - 0.2 mg ly1 ; and Zn - 0.5 g ly1. However, the water cover pH was lower than the minimum of 6.0. Because the pore water of the Mattabi tailings was rich in ferrous iron, the establishment of a water cover was accompanied by the precipitation of a thin layer of hydrous ferric oxide precipitate on the surface of the tailings. This was due to the oxidation of ferrous iron by dissolved oxygen, and subsequent hydrolysis of ferric iron. The precipitate contributed to the removal of zinc and other metals from the cover by sorption reactions. This reasoning is supported by the steady decline in the calculated net solute fluxes after the flooding of 1993. In the long term, as organic-rich sediment accumulates at the bottom of the water cover and uses up oxygen, conditions in the interface might turn from oxic to anoxic. Pedersen et al. Ž1997. describe a situation occurring in Buttle Lake, British Columbia, where a layer of organic-rich natural sediment covers pyrite-
305
rich tailings. The first 1᎐2 cm of the natural sediments remain aerobic, but anoxic conditions immediately below cause the dissolution of Fe and Mn oxides and the release of adsorbed or co-precipitated metals such as Zn to the pore solution. Below a 5-cm depth, Zn is consumed, presumably by sulfide precipitation. Upward diffusion of metals in the pore water leads to reprecipitation of oxyhydroxides near the surface water᎐ sediment interface, and possibly to release of Zn in the lake water ŽPedersen et al., 1997.. Hence at sites with little or no seepage, organic sediment deposition could affect the water cover quality in the long term. However, at sites with significant seepage as in the Mattabi test cell, upward diffusion of metals from the pore water to the water cover is unlikely. The results of the study suggest that some amount of water treatment will be required following flooding of pre-oxidized tailings. The extent of the treatment will depend on the accumulation of soluble oxidation products near the surface of the tailings prior to flooding and the seepage rate. Once oxidation products are flushed out of the water cover, surface water treatment costs will quickly decrease. If the seepage rate is high, as was the case in the present study, large volumes of tailings pore water will need to be collected for treatment. Treatment costs for the seepage will initially increase after flooding, but can be expected to eventually decrease as the contaminated pore water is replaced by cleaner water infiltrating from the water cover. Economic considerations suggest that at sites with high seepage rates, a water cover may be suitable only if it can be supplemented with gravity-fed fresh water from a nearby source to the cover, and thus compensate for seepage losses.
5. Nomenclature Cw : E: FA: FN : FT : H: Mw (t): P: S: SL: ⍀:
Solute concentration in the water cover Žg my3 . Factored evaporation Žmm. Advective flux to the water cover Žg my2 dayy1 . Net flux to the water cover Žg my2 dayy1 . Total flux to the water cover Žg my2 dayy1 . Change in storage or cell water level Žmm. Mass of a given solute Žg. in the water cover determined at time t Ždays. Factored precipitation Žmm. Seepage rate Žm3 dayy1 . Seepage loss Žmm. Area of flooded tailings in the test cell Žm2 .
306
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
Acknowledgements The test cell was constructed by Noranda Inc and monitored by personnel from Mattabi Mines and Noranda Technology Center. Roger Freymond, Phil Tibble, and Mark Woyshner were also involved in data collection and processing. The authors wish to thank Noranda Inc. for permission to publish this paper. References Allison, J.D., Brown, D.S., Novo-Gradac, K.J., 1990. MINTEQA2_ PRODEFA2. A geochemical model for environmental systems: Version 3.0 user manual. US EPA, Environmental Research Laboratory, Athens, GA. Amyot, G., Vezina, S., 1997. Flooding as a reclamation solu´ tion to an acidic tailings pond ᎏ the Solbec case. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, II, 681᎐696. Blowes, D.W., Jambor, J.L., 1990. The pore water geochemistry and the mineralogy of the vadose zone of sulfide tailings, Waite Amulet, Quebec. Appl. Geochem. 5, 327᎐346. Dave, ´ N.K., Lim, T.P., Horne, D., Boucher, Y., Stuparyk, R., 1997. Water cover on reactive tailings and waste rock: laboratory studies of oxidation and metal release characteristics. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, II, 779᎐794. Fisheries and Environment Canada, 1978. Hydrological Atlas of Canada. Surveys and Mapping Branch, Department of Energy, Mines and Resources, Ottawa, Ontario. Freymond, R., 1994. The effect of an overlying water body on sulfidic tailings pore-water ŽMattabi mine case study., B.A.Sc. Thesis, Department of Geological Engineering, University of Waterloo, Waterloo, Ontario. Geocon, 1993. Construction, evaluation and seepage test analysis of test cell on tailings beach Mattabi Mine Site, Ontario, Geocon Report T11498r15821, Montreal, Quebec, Canada. Kam, S.N., Knapp, R.A., Balins, J.K., Payne, R.A., 1997. Interim assessment of flooded tailings performance ᎏ
Quirke mine waste management area. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, II, 853᎐870. Karam, A., Guay, R., 1994. Inondation artificielle du parc ` a ´ residus miniers Solbec Cupra: Etudes microbiologiques et ´ chimiques, MEND Report 2.13.2c, CANMET, Natural Resources Canada, Ottawa, Ontario. ¨ Ljungberg, J., Lindvall, M., Holmstrom, B., ¨ H., Ohlander, 1997. Geochemical field study of flooded mine tailings at Stekenjokk, northern Sweden. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, III, 1401᎐1417. MEND, 1994. Flooding of a mine tailings site ᎏ Solbec Cupra. Suspension of Solids: Impact and Prevention, Report 2.13.2a, CANMET, Natural Resources Canada, Ottawa, Ontario. Pedersen, T.F., McNee, J.J., Flather, D., Mueller, B., Sahami, A., Pelletier, C.A., 1997. Geochemistry of submerged tailings in Buttle lake and the Equity Silver tailing pond, British Columbia, and Anderson Lake, Manitoba: what have we learned? Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, III, 989᎐1005. Rescan Environmental Services Ltd., 1989. Subaqueous disposal of waste rock and tailings, MEND Report 2.11.1a, CANMET, Natural Resources Canada, Ottawa, Ontario. Robertson, J.D., Tremblay, G.A., Fraser, W.W., 1997. Subaqueous tailings disposal: a sound solution for reactive tailing. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, III, 1029᎐1044. Sobek, A.A., Schuller, W.A., Freeman, J.R., Smith, R.M., 1978. Field and Laboratory Methods Applicable to Overburdens and Minesoils, EPA Report EPA-600r2-78-054, US EPA, Cincinati. US EPA, 1986. Hazardous Waste Management System; Identification and Listing of Hazardous Waste; Notification Requirements; Reportable Quantity Adjustments; Proposed Rule. Federal Register Volume 51 Ž 114 . , 21648᎐21693. Yanful, E.K., Simms, P., 1997. Review of water cover sites and research projects, MEND Report 2.18.1, CANMET, Natural Resources Canada, Ottawa, Ontario.
Field assessment of metal and sulfate fluxes during flooding of pre-oxidized mine tailings U
Lionel J.J. Catalana, , Ernest K. Yanfulb, Luc St-Arnauda a
b
Noranda Inc., Technology Center, 240 Hymus Bl¨ d., Pointe Claire, Quebec, Canada H9R 1G5 Department of Ci¨ il and En¨ ironmental Engineering, The Uni¨ ersity of Western Ontario, London, Ontario, Canada N6A 5B9 Accepted 1 June 2000
Abstract This paper presents a field experiment performed at the Mattabi mine tailings site near Ignace, Ontario, Canada, to assess the evolution of water chemistry over flooded sulfide mine tailings. The tailings had oxidized for approximately 10 years prior to flooding. The experiment involved the construction of a 70 m2 test cell and a monitoring program that extended over more than 2 years. Metal and sulfate fluxes towards the water cover were calculated using measured solute concentrations, water cover volumes, and seepage rates. Results indicate that directly flooding pre-oxidized tailings can initially lead to the release of metals and sulfate to the water cover. However, dilution of the water cover by rain and snowmelt, flushing of the oxidation products by infiltration into the tailings, and removal of some metals by precipitation and sorption progressively reduced concentrations in the water cover below regulatory discharge limits. Because the tailings pore water was rich in ferrous iron, a thin layer of hydrous ferric oxide precipitated on the surface of the tailings, following the establishment of the water cover. The precipitate is believed to have removed zinc from the water cover through sorption or co-precipitation reactions. 䊚 2000 Elsevier Science Ltd. All rights reserved. Keywords: Oxidized tailings; Field experiment; Water cover; Flooding; Test cell; Flux; Metal; Sulfate
1. Introduction Underwater disposal is a common and effective method for reducing sulfide mineral oxidation and acid generation in mine tailings ŽRescan, 1989; Pedersen et al., 1997; Robertson et al., 1997.. Since the diffusion
U
Corresponding author. Department of Chemical Engineering, Lakehead University, 955 Oliver Rd., Thunder Bay, Ontario P7B 5E1 Canada. Fax: q1-807-343-8928. E-mail address: [email protected] ŽL.J. Catalan..
coefficient and solubility of oxygen in water are very low compared to their values in air, oxygen influx to the tailings is effectively reduced ŽDave ´ et al., 1997.. For weathered pre-oxidized tailings, the method usually involves the construction of containment structures such as dams and dykes around the tailings deposit to hold water. Flooding pre-oxidized tailings can enhance the release of metals and acidity from the tailings to the water cover, at least in the short term. Under such circumstances, the concentrations and loads of the released metals could be significant enough to warrant treatment of runoff and seepage leaving the tailings
1093-0191r00r$ - see front matter 䊚 2000 Elsevier Science Ltd. All rights reserved. PII: S 1 0 9 3 - 0 1 9 1 Ž 0 0 . 0 0 0 2 9 - 0
296
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
area for some time after the placement of the water cover. The purpose of a water cover over pre-oxidized tailings is threefold: Ž1. to reduce further oxidation of the tailings; Ž2. to provide sufficient improvement of the water cover quality to permit discharge of runoff to the surrounding environment without the need for secondary treatment after the transition period, that is, the period when previous oxidation products are being flushed out; and Ž3. to reduce over time metal and sulfate loads and treatment costs associated with seepage from the tailings. According to a recent review of water cover sites and research projects ŽYanful and Simms, 1997., flooding of oxidized tailings has already been implemented at a number of sites including three closed operations: the Solbec tailings site near Stratford, Quebec; the Quirke Lake tailings site in Elliot Lake, Ontario; and the Stekenjokk site in northern Sweden. Prior to flooding at the Solbec site, hydrated lime ŽCaŽOH.2 . was dumped in the tailings pond water, and calcite dust and granules ŽCaCO3 . were incorporated into the tailings by plowing to a depth of up to 30 cm in the area exposed to the atmosphere. On average, 230 t of alkaline material per hectare were applied to limit the solubility of metals ŽAmyot and Vezina, 1997.. The ´ water cover was implemented in 1994. Since the summer of 1995 all water quality parameters in the water cover have conformed to the applicable Quebec regulation. However, in the fall of 1996 the oxidized tailings pore water still contained iron and zinc concentrations of 150 mg ly1 and 9.3 mg ly1 , respectively. Further details on studies pertaining to the Solbec site are available in Karam and Guay Ž1994. and MEND Ž1994.. The Quirke Lake tailings site was flooded between 1992 and 1995. The water cover was designed as a series of five terraced cells separated by internal dykes. To neutralize the acidity generated from previously oxidized uranium tailings, lime was added to all cells, except to Cell 14 which is located upstream of the other cells. Within 2 years after flooding, the iron and zinc concentrations in the water cover of Cell 14 stabilized below 0.6 mg ly1 and 0.01 mg ly1 , respectively. The water cover pH varied between 6.5 and 7.1 during the same time period, and sulfate levels have shown a continuous decline since flooding was initiated. The mean water retention time in Cell 14 is estimated at approximately 3 months. Significant downward seepage in this cell, and the consequent flushing of oxidation products down into the pore water, are important in explaining the rapid attainment of discharge compliance standards in the water cover without lime addition. The high seepage loss is offset by diverting fresh water from a nearby lake and can be maintained in the long term. The quality of shallow pore water in the upper tailings also improved signifi-
cantly after flooding. The acidity declined from peak levels of 12 500 mg CaCO3 ly1 prior to flooding to a maximum of 240 mg CaCO 3 ly1 in 1996. Treatment requirements have increased in the short term due to increased seepage caused by flooding ŽKam et al., 1997.. At the Stekenjokk site, Zn᎐Cu tailings rich in pyrite and other sulfides were flooded in 1991. A small area of the tailings deposit Žapprox. 5% of the bottom of the pond. consists of material that has oxidized and weathered before the flooding. In the summer of 1995, average concentrations of Ca Ž20.0 mg ly1 ., S Ž11.1 mg ly1 ., Cd Ž0.69 mg ly1 ., Zn Ž139 mg ly1 ., and Ni Ž1.3 mg ly1 . in the water cover were higher than local background concentrations in surface water ŽLjungberg et al., 1997.. According to these authors, the most important source of Zn in the water cover is probably diffusion from the oxidized tailings pore water, where Zn concentrations are much higher than in the water cover. In addition, dissolution of gypsum in the oxidized tailings most likely accounts for the elevated concentrations of Ca and sulfate, both in the pore water and in the water cover. There are some indications that sulfide oxidation continues under the water cover Žalthough at a much smaller rate than in tailings exposed to the atmosphere., and also contributes to the pore water composition. The present paper describes a field experiment performed at the Mattabi mine tailings site located approximately 73 km north-east of Ignace, Ontario, Canada, to examine the impact of flooding previouslyoxidized sulfide mine tailings on the chemistry of the water cover without alkalinity addition to the water or to the tailings. The Mattabi tailings impoundment was active from 1972 to April 1991. During this period, approximately 15 500 000 t of tailings from three different ore bodies were end-spilled over an area of 125 ha. The experiment involved the construction and flooding of a test cell in a part of the tailings area that had been oxidizing for approximately 10 years. Metal and sulfate fluxes were calculated using water balance data and measured concentrations of metals and sulfate at different times during flooding of the test cell.
2. Materials and methods 2.1. Test cell construction and materials The field test cell was constructed in October 1991. The cell was 70 m= 70 m in plan, as measured between the centerlines of the perimeter dykes ŽFig. 1.. The dykes were constructed with glacial till compacted at the natural moisture content of 7᎐10%, slightly higher than the optimum water content of 7.5% based on the laboratory standard Proctor compaction test. The mea-
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
297
Fig. 1. Overall view of test cell. Photograph taken on 20 May 1992.
sured compacted dry density ranged from 2.0 to 2.2 Mg my3. Two 30-m long wooden rafts were placed at a 90⬚ angle, and fixed to the south and east embankments to provide access to the center of the test cell. The tailings in the cell measured approximately 54 m= 54 m in plan. Details of the construction can be found in Geocon Ž1993.. The tailings beneath the test cell had an average depth of approximately 14 m. Dry densities ranged from 1.5 to 2.1 Mg my3; specific gravity and porosity averaged 3.6 and 51%, respectively. The tailings were poorly graded and consisted, on average, of fine sand Ž60%., coarse sand Ž20%., and silt-sized Ž20%. particles. The average d50 was approximately 0.15 mm. The average chemical composition of oxidized Mattabi tailings samples determined by ICP is shown in Table 1. Combination of X-ray diffraction and chemical analysis results showed that sulfides accounted for 37% of the tailings mass. The dominant silicate minerals were quartz Ž21%., chlorite Ž11%., muscovite Ž10%., and albite Ž2.5%.. Gypsum Ž0.6%. and goethite Ž2.7%. were also identified. All percentages refer to the total tailings mass. Petrographic examination of the tailings ŽFreymond, 1994. revealed that pyrite was the domiTable 1 Average chemical composition of oxidized Mattabi tailings Element
Content Žgrg.
Al As Ca Cu Fe K Mg Mn Na Pb S Si Ti Zn
33 252 1785 2186 886 230 968 4701 6380 1076 2198 1444 199 288 141 156 1135 4686
nant sulfide mineral, followed by pyrrhotite, which was extensively oxidized near the surface of the tailings area. Trace amounts of sphalerite, arsenopyrite, and chalcopyrite were also present. Acid᎐base accounting ŽABA. tests were conducted on Mattabi tailings samples using the Sobek procedure ŽSobek et al., 1978.. These tests measure the overall balance between acidity potential ŽAP. and the neutralization potential ŽNP.. In a standard ABA test, the total sulfur is measured, and the AP is calculated based on the assumption that all the sulfur exists as pyrite and that the pyrite will oxidize completely to produce sulfuric acid. The NP is determined from the amount of acid that the sample neutralizes in a standardized digestion᎐back titration procedure. A sample is considered acid generating if the ratio NPrAP- 1.0 and AP ) 3.1 kg CaCO3 ty1 . If NPrAP) 3.0 or AP - 3.1 kg CaCO3 ty1 , the material is considered non-acid generating. Materials with 1.0- NPrAP- 3.0 and AP ) 3.1 kg CaCO3 ty1 are uncertain from an ABA point of view. With an average AP of 624 kg CaCO3 ty1 and an average NP of y0.2 kg CaCO3 ty1 , the oxidized Mattabi tailings were strongly acid generating. Their average paste pH was 3.6. The oxidized Mattabi tailings exhibited mineral oxidation and precipitation patterns similar to other tailings impoundments Že.g. Blowes and Jambor, 1990., where a sulfide-depletion zone is underlain by an active oxidation zone, and is further underlain by a zone affected by secondary mineral precipitation. The secondary mineral was typically amorphous ferric hydroxide, or a more crystalline ferric oxyhydroxide or hydroxysulfate ŽFreymond, 1994..
2.2. Test cell operation and monitoring During the winter following construction of the cell, snow and ice accumulated in the test cell. Snowmelt and rain during the spring elevated the water table from below to above the tailings surface. By July 1992, the depth of water covering the tailings had reached approximately 1 m in the deepest parts of the cell. On 7 July, surface water samples were collected in the
298
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
water cover. In this sampling event, as in the ones following, surface water samples were collected by lowering silicone tubing a few centimeters below the surface and pumping 500 ml of water using a portable peristaltic pump. Water samples were collected in duplicate or triplicate. The samples were filtered through 0.45-m filters, acidified, and then analyzed for Al, As, Ca, Cd, Cu, Fe, K, Mg, Mn, Na, Pb, S, Se, and Zn by inductively-coupled plasma atomic emission spectroscopy ŽICP-AES.. Concentration differences between replicate samples were generally small and within the limits of analytical error. Field measurements of solute concentration variations with depth in other water cover studies Že.g. Ljungberg et al., 1997. have demonstrated that the water column is well mixed during the ice-free season, due to wind-induced waves and natural convection. Moreover, because of the relatively small dimensions of the cell and the homogeneity of the tailings, it was assumed that the water chemistry would be independent of location in the cell. Since a closure option for the Mattabi site involved flooding the tailings with water from nearby Sturgeon Lake, rain and snowmelt water was pumped out of the cell between 7 and 9 July, 1992. Starting on 11 July, Sturgeon Lake water was pumped into the cell at a rate of 3500 l miny1 through a pipe located on the north side ŽFig. 2.. The total volume of water in the test cell after flooding was 4200 m3. The water depth was 1.0 m on the north side of the cell and 1.7 m on the south side, as the surface of the tailings in the test cell sloped at approximately 0.7⬚ downward from north to south. The surface water was sampled again on 14 July. Regular monitoring of water level in the cell began on 14 July and continued until 3 October, 1992. During this interval, precipitation, class A pan evaporation,
temperature, and humidity were also measured. There was no monitoring between October 1992 and July 1993. In July 1993, the test cell was instrumented with a pressure transducer to automate monitoring of the surface water elevation. Monitoring of the water level resumed on 22 July 1993 and continued until December 1993. The surface water was sampled on 25 July 1993. Starting on 26 July 1993, Sturgeon Lake water was again added to the cell. When the flooding was complete Ž3 August., the cell contained 5200 m3 of water. Additional surface water samples were collected on 1 August, 13 September, 12 October, and 3 November 1993. In 1994, no additional water was pumped to the cell. The last surface water samples were collected in July 1994. In addition to surface water samples, core samples of tailings were obtained before adding Sturgeon Lake water to the test cell in July 1992 and in July 1993. Cores were also obtained in July 1994. On each occasion, cores were collected at two locations within the test cell to approximately 1-m depth using thin-walled aluminum tubes. Slices of the cores were squeezed to recover pore water samples representative of various depth intervals. These samples were analyzed in the same manner as the surface water samples.
2.3. Seepage loss analysis Seepage loss from the test cell was established by carrying out a water balance analysis using the monitoring data collected during the test ŽGeocon, 1993..
Fig. 2. View of test cell with pipeline in place to supply fresh water from Sturgeon Lake. Photograph taken on 20 May 1992 before the water was pumped out Ž7᎐9 July, 1992..
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
The water balance was evaluated using the following equation: SL s Py Ey H
Ž1.
where SL is the seepage loss, P the factored precipitation, E the factored evaporation, and H the change in storage or cell water level Žnegative for drop.. The factored precipitation included an allowance for runoff from the interior slopes of the test cell dykes. This was achieved by increasing the measured precipitation by 10% Žthat is, using a factor of 1.1.. The resulting factored precipitation was then assumed to represent the rise in cell water level due to a precipitation event. Since the vertically projected area of the interior slope of the dykes represented approximately 20% of the water surface area in the cell, the factoring procedure translated into a runoff coefficient of 0.5 for the compacted till slopes of the dykes ŽGeocon, 1993.. This procedure assumed that there was negligible runoff from the crest of the dykes, a reasonable assumption since the crest was flat. To obtain the factored Žestimated. evaporation from the test cell, the measured class A evaporation was reduced by a factor of 0.8, which was slightly higher than the typical value of 0.7 used for lakes ŽFisheries and Environment Canada, 1978.. The value of 0.8 was used to allow for the much shallower depth of water in the test cell ŽGeocon, 1993.. As an example, for the period 14᎐17 July, 1992, the test cell water level dropped 33.8 mm and 21.4 mm of rainfall was recorded, while pan evaporation was measured to be 13 mm. Applying factors of 1.1 and 0.8 to precipitation and evaporation, respectively, and noting that the water level change is y33.8 mm, the seepage
299
loss is calculated to be wŽ21.4= 1.1. y Ž13 = 0.8. y Žy33.8.x mm, or ; 47 mm. The water surface area in the cell on 15 July Žmiddle of period. was estimated to be 3427 m2, and the elapsed time between measurements on 14 July and 17 July was 2.71 days. The rate of total seepage from the test cell for the period 14᎐17 327.42= 0.0470 July is then m3 dayy1 , or 59.4 m3 2.71 dayy1. Seepage rates were calculated for other days in a similar fashion.
3. Results 3.1. Water co¨ er chemistry The variation in the water cover volume with time for the monitoring periods of 1992 and 1993 is shown in Fig. 3. The elapsed time on the horizontal axis is calculated from 14 July, 1992. The volume reached a maximum after each flooding event, and then decreased during the following summer and fall, as evaporation and seepage exceeded precipitation. Average dissolved concentrations of selected elements in the water cover for each sampling event are shown in Table 2. Sturgeon Lake water-chemistry is also included for comparison. All solute concentrations decreased during the first flooding of the cell with Sturgeon Lake water in 1992, presumably as a result of dilution, but then increased between the end of flooding and 25 July, 1993. On that date, Fe and Zn reached elevated concentrations of 466 mg ly1 and 3.79 mg ly1 , respectively. During the second flooding Žbetween 26 July and 3 August, 1993., solute concentrations decreased, except
Fig. 3. Water cover volume in test cell vs. time.
300
Element
7 Jul 1992 Žbefore 1st flooding.
14 Jul 1992 Žafter 1st flooding.
25 Jul 1993 Žbefore 2nd flooding.
1 Aug 1993 Žduring 2nd flooding.
13 Sep 1993
12 Oct 1993
3 Nov 1993
Jul 1994
Sturgeon Lake water
Al As Ca Cd Cu Fe K Mg Mn Na Pb S Se Zn pH
9.08 - 0.3 257 - 0.02 0.12 974 6.7 233 29 11 - 0.15 1152 0.15 11.7 3.2
0.47 - 0.3 71 - 0.01 - 0.02 58 2.3 19 2.2 3.3 - 0.05 118 0.05 0.39
21.8 - 0.3 302 - 0.02 0.1 466 6 182 24 17 0.4 951 - 0.50 3.79
4.86 - 0.3 532 - 0.02 - 0.02 54 -5 194 6.1 25 - 0.25 806 - 0.5 0.69
3.03 - 0.3 627 - 0.02 - 0.02 0.58 5.8 196 3.2 24 - 0.25 791 0.5 0.74
0.56 - 0.3 596 - 0.02 - 0.02 0.12 -5 190 3 22 - 0.25 746 - 0.50 0.66 5.7
0.44 - 0.3 648 - 0.02 0.025 0.31 5.8 189 3.1 23 - 0.25 787 0.5 0.69
0.71 - 0.3 461 - 0.02 - 0.02 2.6 8 85 1.1 16 - 0.25 625 - 0.50 0.34 4.2
- 0.05 - 0.05 12.5 - 0.005 - 0.005 - 0.005 -1 1.8 0.001 3.9 - 0.05 6.22 - 0.1 - 0.005
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
Table 2 Average dissolved concentrations Žmg ly1 . of selected elements and pH in the water cover at various sampling dates and comparison with Sturgeon Lake water
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
for Ca and Mg. The evolution of solute concentrations after the second flooding differed markedly from that after the first flooding. Between 1 August, 1993 and July 1994, most solute concentrations decreased or remained stable. A slight increase in Al and Fe concentrations Žfrom 0.44 to 0.71 mg ly1 and from 0.31 to 2.6 mg ly1, respectively. occurred between November 1993 and July 1994. However, the concentration of the other solutes decreased or remained stable. The pH of the water cover increased as a result of flooding from 3.2 on 7 July, 1992 to 5.7 on 12 October, 1993, but then decreased to 4.2 in July 1994. Table 3 shows the evolution of dissolved masses of Al, Ca, Fe, Mg, Mn, S Žmostly as sulfate ., and Zn in the water cover on sampling events. These masses were determined as the product of solute concentrations and the volume of water in the cover at the various sampling dates. Note that solute masses could not be calculated for 7 July 1992 because the exact volume of the water cover was not measured on that day. All solute masses increased in the year that followed the first flooding of the cell. During the second flooding Žbetween 26 July and 3 August, 1993., the masses of Ca, Mg, Na and S increased dramatically, the masses of Al and Mn remained relatively stable, and the masses of Fe and Zn decreased. After the second flooding was complete, the masses of all solutes decreased with time. Iron differed from other solutes in that its mass began to increase again after 12 October, 1993.
3.2. Formation of hydrous ferric oxide precipitate layer On July 1992, before Sturgeon Lake water was pumped to the cell, a 3᎐4-mm thick iron hydroxide precipitate was observed coating the surface of the tailings below the water cover. This precipitate had formed during the previous spring when snowmelt and
301
rainwater accumulated in the cell. Chemical analysis of the precipitate revealed the presence of 2200 mg kgy1 of zinc. To determine the mobility of Zn, the sample was also leached using the Toxicity Characteristic Leaching Procedure ŽUS EPA, 1986., which involves shaking the sample in an acetic acid buffer solution at pH 4.9 for a period of 18 h, and then analyzing the leachate. The Zn concentration in the final leachate was 4.5 mg ly1 , which accounted for approximately 2% of the mass of zinc present in the precipitate. These results indicate that a fraction of the Zn initially released to the water cover was able to adsorb or co-precipitate within the hydrous ferric oxide precipitate.
3.3. Shallow pore water chemistry Table 4 shows the average concentrations of various elements and pH in shallow pore water extracted from cores collected on different dates during the experiment. The depth intervals for each set of concentration data are also indicated. Although significant differences exist between cores collected at two different locations in the cell on the same date, the overall trend is as follows: between 7 July, 1992 and July 1994 the concentrations of all measured solutes decreased, except for Ca and K which increased. This is consistent with the dissolution of oxidation products because of flooding and flushing due to infiltration from the water cover. The increase in pore water pH may have also led to the precipitation of some metals. The first flooding had a greater impact on shallow pore water concentrations than the second. As expected, most of the iron was present as ferrous iron in the pore water.
3.4. Metal and sulfate fluxes to the water co¨ er This section deals with estimates of the average flux of solutes to the water cover between sampling events. By definition, positive fluxes indicate a movement of
Table 3 Dissolved masses Žkg. of selected elements in the water cover at various sampling dates Element
14 Jul 1992 Žafter 1st flooding.
25 Jul 1993 Žbefore 2nd flooding.
1 Aug 1993 Žduring 2nd flooding.
13 Sep 1993
12 Oct 1993
3 Nov 1993
Jul 1994
Al Ca Fe Mg Mn Na S Zn
1.9 294 240 81 9.0 14 489 1.6
24.7 342 527 206 26.7 19 1076 4.3
24.1 2632 265 960 30.1 123 3989 3.4
9.1 1877 1.7 587 9.5 73 2367 2.2
1.2 1249 0.2 398 6.2 46 1563 1.4
0.9 1267 0.6 369 6.0 44 1540 1.3
1.7 1107 6.3 205 2.6 37 1500 0.8
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
302
Table 4 Dissolved concentrations Žmg ly1 . of selected elements and pH in shallow pore water at various sampling dates Element
Al As Ca Cd Cu Fe Fe 3q K Mg Mn Na Pb S Zn pH
7 Jul 1992 Žbefore 1st flooding.
25 Jul 1993 Žbefore 2nd flooding.
Jul 94
Depth 0᎐5 cm
Depth 0᎐20 cm
Depth 0᎐15 cm
15
1.4
413 0.56 0.90 2990
363 0.03 0.25 149
7.2 1080 151 110 0.98 3290 82 3.3
0.27 0.50 543 0.21 - 0.025 159 8.8 17 85 11 26 2.2 677 12 5.6
20 435 52 39 2.9 1093 20 3.7
solutes towards the water cover. The difference between the masses Mw Ž t2 . and Mw Ž t1 . of a given solute in the water cover determined at times t1 and t2 is related to the total solute flux FT to the water cover and the area of flooded tailings, ⍀, in the test cell as follows: Mw Ž t2 . y Mw Ž t1 . s FT ⍀ Ž t2 y t1 .
Ž2.
The total solute flux can be written as the sum of the advective flux of solute carried by the seepage FA, and the net flux of solute to the water cover FN : FT s FA q FN
Ž3.
The advective flux to the water cover is negative since seepage removes solutes from the water cover. It is related to the seepage rate S and the solute concentration in the water cover, Cw , between times t1 and t2 by: FA s y
1 ⍀ Ž t2 y t1 .
t2
Ht SC
w ⭈ dt
Ž4.
1
Combining Eqs. 2᎐4 yields the following expression for the net solute flux: FN s
1 ⍀ Ž t2 y t1 .
ž
Mw Ž t2 . y Mw Ž t1 . q
t2
Ht SC 1
w ⭈ dt
/
Ž5.
The net flux, if positive, represents the rate of solute release to the water cover due to dissolution or desorption from mineral phases present in the tailings or in the ferric hydroxide precipitate layer. Diffusion from pore water to the cover is not believed to have played a major role, given the significant downward advective
0.30 0.35 595 0.05 0.04 379 11.4 21.1 147 21.5 27.7 - 0.25 880 4.6 5.8
0.30 - 0.25 634 - 0.025 - 0.025 51 0.61 19 160 6.6 38 - 0.25 756 1.8
0.32 0.64 520 - 0.025 - 0.025 583 3.6 32 197 14.4 49 0.43 1056 4.8
fluxes due to seepage losses. A negative net solute flux indicates the rate of solute removal from the water cover by precipitation or adsorption on the ferric hydroxide precipitate layer. The integral in Eq. 5 can be estimated numerically from the plot of concentrations versus time and estimates of seepage rates between sampling events. Seepage losses were determined almost daily. Average seepage rates calculated over each week of monitoring are shown in Fig. 4. Calculation of net solute fluxes used average seepage rates for the periods between surface water sampling events in 1993 ŽTable 5.. As an illustration of the procedure for calculating the net fluxes, consider the estimated test cell volume and measured zinc concentrations in the water cover on 13 September and 12 October, 1993, shown in Table 2. From the measured zinc concentration of 0.74 mg ly1 Ži.e. 0.74 g my3 . and the test cell water volume of 2994 m3, for 13 Sept, 1993, the mass of zinc is 0.74= 2994, or 2215.6 g. Similarly, the mass of zinc in the water cover on 12 October is 1382.7 g. Using the estimated average seepage rate of 29.7 m3 dayy1 ŽTable 5. and an average zinc concentration of Ž0.74q 0.66.r2 mg ly1, or 0.70 mg ly1 , the net flux of zinc to
Table 5 Average seepage rates used in the calculation of solute fluxes Period
Seepage rate, S Žm3 dayy1 .
July 25᎐August 1, 1993 August 1᎐September 13, 1993 September 14᎐October 12, 1993 October 13᎐November 3, 1993
46.1 63.5 29.7 7.7
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
303
Fig. 4. Weekly average seepage rates vs. time.
the water cover during the 28 days between 13 Sept and 12 Oct, 1993, is calculated to be y0.0031 g my2 dayy1 as shown in Table 6. The negative sign means that zinc was actually removed from the water cover. Table 7 presents net fluxes for the various solutes between sampling events in 1993, calculated using the above procedure. Net solute fluxes were not calculated for 1992 due to lack of sufficient concentration data. A sensitivity analysis on net solute fluxes with respect to the seepage rates was performed, since seepage rates carry the most uncertainty among the parameters required to calculate net solute fluxes. Table 7 also shows net solute fluxes obtained assuming both an underestimation and an overestimation of the seepage rates by 10%. The results indicate that the order of magnitude and the sign of the net solute fluxes would generally not be affected by such an error in the seepage rate. The only exception is for Mg between 12 October and 3 November, 1993. The flooding of 1993 resulted in a net release Žby dissolution or desorption. of Al, Ca, Mg, Mn, Na and S Žas sulfate . to the water cover, as shown by the positive Table 6 Sample calculation of net solute fluxes a Date
Test cell water volume Žm3 .
Zn Žmg l -1 .
Zn Žg.
September 13, 1993 October 12, 1993
2994 2095
0.74 0.66
2215.6 1382.7
a Average seepage rate s 29.7 m3 dayy1 ŽTable 1.; Area of tailings surface s 54 m= 54 m. Net flux of zinc to water cover during 13 Sep᎐12 Oct, 1993 Ž28 days. using Eq. 5: wŽ138.27y 2215.6.grŽ28 days. q Ž29.7 m3 dayy1 . Ž0.74 q 0.66.r2 g my3 4rŽ54 m.rŽ54 m.x, or y0.0031 g my2 dayy1.
value of the corresponding net fluxes during the period 25 July to 1 August, 1993. In contrast, Fe and Zn displayed negative fluxes of y8.8 g my2 dayy1 and y0.007 g my2 dayy1, respectively, and were likely removed from the water cover by precipitation or sorption. As time progressed, however, the majority of fluxes decreased steadily. In the 2᎐3 months that followed, precipitation andror sorption reactions began removing the solutes that had previously been released to the cover during flooding. This is demonstrated by the fact that all the net fluxes were negative between 13 September and 12 October. Iron continued to be precipitated, but at a smaller rate than during flooding. Zinc appears to have been dissolved or desorbed at a small rate of 0.006 g my2 dayy1 just after flooding Ž1 August᎐13 September period., but the Zn net flux became negative again during the next period. During the last period Ž12 October᎐3 November., all solutes except aluminum displayed a positive, albeit small, net flux. The reason for this is not obvious. Table 8 compares the net solute fluxes to the water cover during the 1993 flooding Ž25 July to 1 August. with the solute concentrations both in the water cover and in the shallow pore water on the eve of flooding Ž25 July.. The data suggest that the net fluxes of some solutes to the water cover were not primarily controlled by diffusion to or from the pore water. For example, although the concentrations of Al, Mg, Mn and S were higher in the water cover than in the shallow pore water, the net flux of these solutes was towards the water cover Ži.e. positive.. In the case of Zn, the net flux was negative, although Zn concentrations were higher in the shallow pore water than in the water cover. These observations are consistent with solute
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
304
Table 7 Calculated net solute fluxes Žin g my2 dayy1 . for 1993 Solute
July 25᎐August 1, 1993
August 1᎐Sept 13, 1993
Sept 13᎐October 12, 1993
Oct 12᎐Nov 3, 1993
Al
0.18 a Ž0.16 b , 0.20 c . 118.8 Ž118.1, 119.4. y8.8 Žy9.2, y8.3. 39.9 Ž39.6, 40.2. 0.40 Ž0.38, 0.42. 5.40 Ž5.36, 5.43. 157 Ž155, 158. y0.007 Žy0.010, y0.003.
y0.033 Žy0.042, y0.024. 6.6 Ž5.3, 7.9. y1.51 Žy1.57, y1.45. 1.3 Ž0.9, 1.7. y0.063 Žy0.073, y0.052. 0.13 Ž0.08, 0.19. 4.3 Ž2.7, 6.3. 0.0060 Ž0.0045, 0.0076.
y0.078 Žy0.080, y0.077. y1.5 Žy2.1, y0.8. y0.015 Žy0.015, y0.014. y0.35 Žy0.55, y0.15. y0.009 Žy0.012, y0.006. y0.09 Žy0.11, y0.06. y2.0 Žy2.8, y1.2. y0.0031 Žy0.0038, y0.0023.
y0.0039 Žy0.040, y0.037. 1.9 Ž1.8, 2.1. 0.0065 Ž0.0065, 0.0066. 0.03 Žy0.02, 0.08. 0.0046 Ž0.0038, 0.0054. 0.029 Ž0.022, 0.034. 1.7 Ž1.5, 1.9. 0.0011 Ž0.0009, 0.0012.
Ca Fe Mg Mn Na S Zn a
Flux corresponding to the estimated seepage rate. Flux corresponding to 90% of the estimated seepage rate. c Flux corresponding to 110% of the estimated seepage rate. b
concentrations in the water cover being strongly affected by precipitation᎐adsorption᎐dissolution reactions within the hydrous ferric oxide precipitate covering the tailings surface. The pH and solute concentration data measured in the water cover on 12 October 1993 were used in geochemical equilibrium modeling of the water cover to calculate the saturation indices ŽSI. of various minerals. Minerals with saturation indices close to zero are possible candidates for controlling the concentration of the associated ions in solution. A large negative value indicates that the water is undersaturated with respect to the mineral and therefore the mineral will likely not precipitate. A high positive value is usually an indication that the mineral could theoretically form, but the
kinetics of the reactions controlling its formation are not favorable. Saturation indices for key minerals computed using the computer program MINTEQA2 ŽAllison et al., 1990. are presented in Table 9. SI values for zinc minerals were not tabulated, as they all indicated undersaturation ŽSI values of y6 or more negative.. The modeling results suggest that aluminum concentrations in the water cover were possibly controlled by AlŽOH.3Ža. or jurbanite ŽAlOHSO4 .. Saturation indices for gypsum ŽCaSO4 ⭈ 2H2 O. and ferrihydrite ŽFeŽOH.3 ⭈ 3H2 O. are positive and low, indicating that these minerals are likely present and could be controlling the concentrations of Ca, SO42y and Fe3q in the water cover. Zinc did not form any secondary minerals, and negative net
Table 8 Comparison of 1993 net solute fluxes to the water cover with solute concentrations in the shallow pore water and in the water cover Ž25 July. just before flooding
Al Ca Fe Mg Mn Na S Zn a
Pore water in top 20 cm of tailings on July 25, 1993 Žmg ly1 . a
Water cover on July 25, 1993 Žmg ly1 .
Average flux for July 25᎐Aug 1, 1993 Žg my2 dayy1 .
0.27, 0.30 543, 595 159, 379 85, 147 11, 21 26, 28 677, 879 11.7, 4.6
21.8 302 466 182 24 17 951 3.8
0.18 118.8 y8.8 39.9 0.40 5.40 157 y0.007
Values represent concentrations measured in pore water extracted from two core samples
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306 Table 9 Calculated mineral saturation indices Mineral
Saturation index ŽSI.
AlŽOH. 3 Ža. Jurbanite ŽAlOHSO4 . AlŽOH.10 SO4 Boehmite ŽAlŽOH. 3 . Ferrihydrite Gypsum Gibbsite Goethite Na-jarosite Diaspore
y0.783 0.309 6.32 0.978 0.591 0.313 0.992 4.420 1.679 2.818
fluxes of Zn are probably explained by adsorption onto hydrous ferric oxide precipitate.
4. Discussion and conclusions The field test cell results suggest that directly flooding previously oxidized sulfide mine tailings can lead to the release of metals and sulfate to the water cover. Iron and zinc concentrations in the water cover reached 466 and 3.8 mg ly1 , respectively, one year after the flooding of July 1992. After the second flooding, however, dilution of the water cover by rain and snowmelt, flushing of solutes by water infiltration from the cover to the tailings, and removal of some metals by precipitation and sorption progressively reduced the concentrations in the water cover. In July 1994, i.e. 2 years after the first flooding, metal concentrations in the water cover met regulatory discharge limits: As - 0.5 mg ly1 ; Cu - 0.3 mg ly1 ; Fe - 3 mg ly1 ; Pb - 0.2 mg ly1 ; and Zn - 0.5 g ly1. However, the water cover pH was lower than the minimum of 6.0. Because the pore water of the Mattabi tailings was rich in ferrous iron, the establishment of a water cover was accompanied by the precipitation of a thin layer of hydrous ferric oxide precipitate on the surface of the tailings. This was due to the oxidation of ferrous iron by dissolved oxygen, and subsequent hydrolysis of ferric iron. The precipitate contributed to the removal of zinc and other metals from the cover by sorption reactions. This reasoning is supported by the steady decline in the calculated net solute fluxes after the flooding of 1993. In the long term, as organic-rich sediment accumulates at the bottom of the water cover and uses up oxygen, conditions in the interface might turn from oxic to anoxic. Pedersen et al. Ž1997. describe a situation occurring in Buttle Lake, British Columbia, where a layer of organic-rich natural sediment covers pyrite-
305
rich tailings. The first 1᎐2 cm of the natural sediments remain aerobic, but anoxic conditions immediately below cause the dissolution of Fe and Mn oxides and the release of adsorbed or co-precipitated metals such as Zn to the pore solution. Below a 5-cm depth, Zn is consumed, presumably by sulfide precipitation. Upward diffusion of metals in the pore water leads to reprecipitation of oxyhydroxides near the surface water᎐ sediment interface, and possibly to release of Zn in the lake water ŽPedersen et al., 1997.. Hence at sites with little or no seepage, organic sediment deposition could affect the water cover quality in the long term. However, at sites with significant seepage as in the Mattabi test cell, upward diffusion of metals from the pore water to the water cover is unlikely. The results of the study suggest that some amount of water treatment will be required following flooding of pre-oxidized tailings. The extent of the treatment will depend on the accumulation of soluble oxidation products near the surface of the tailings prior to flooding and the seepage rate. Once oxidation products are flushed out of the water cover, surface water treatment costs will quickly decrease. If the seepage rate is high, as was the case in the present study, large volumes of tailings pore water will need to be collected for treatment. Treatment costs for the seepage will initially increase after flooding, but can be expected to eventually decrease as the contaminated pore water is replaced by cleaner water infiltrating from the water cover. Economic considerations suggest that at sites with high seepage rates, a water cover may be suitable only if it can be supplemented with gravity-fed fresh water from a nearby source to the cover, and thus compensate for seepage losses.
5. Nomenclature Cw : E: FA: FN : FT : H: Mw (t): P: S: SL: ⍀:
Solute concentration in the water cover Žg my3 . Factored evaporation Žmm. Advective flux to the water cover Žg my2 dayy1 . Net flux to the water cover Žg my2 dayy1 . Total flux to the water cover Žg my2 dayy1 . Change in storage or cell water level Žmm. Mass of a given solute Žg. in the water cover determined at time t Ždays. Factored precipitation Žmm. Seepage rate Žm3 dayy1 . Seepage loss Žmm. Area of flooded tailings in the test cell Žm2 .
306
L.J. Catalan et al. r Ad¨ ances in En¨ ironmental Research 4 (2000) 295᎐306
Acknowledgements The test cell was constructed by Noranda Inc and monitored by personnel from Mattabi Mines and Noranda Technology Center. Roger Freymond, Phil Tibble, and Mark Woyshner were also involved in data collection and processing. The authors wish to thank Noranda Inc. for permission to publish this paper. References Allison, J.D., Brown, D.S., Novo-Gradac, K.J., 1990. MINTEQA2_ PRODEFA2. A geochemical model for environmental systems: Version 3.0 user manual. US EPA, Environmental Research Laboratory, Athens, GA. Amyot, G., Vezina, S., 1997. Flooding as a reclamation solu´ tion to an acidic tailings pond ᎏ the Solbec case. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, II, 681᎐696. Blowes, D.W., Jambor, J.L., 1990. The pore water geochemistry and the mineralogy of the vadose zone of sulfide tailings, Waite Amulet, Quebec. Appl. Geochem. 5, 327᎐346. Dave, ´ N.K., Lim, T.P., Horne, D., Boucher, Y., Stuparyk, R., 1997. Water cover on reactive tailings and waste rock: laboratory studies of oxidation and metal release characteristics. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, II, 779᎐794. Fisheries and Environment Canada, 1978. Hydrological Atlas of Canada. Surveys and Mapping Branch, Department of Energy, Mines and Resources, Ottawa, Ontario. Freymond, R., 1994. The effect of an overlying water body on sulfidic tailings pore-water ŽMattabi mine case study., B.A.Sc. Thesis, Department of Geological Engineering, University of Waterloo, Waterloo, Ontario. Geocon, 1993. Construction, evaluation and seepage test analysis of test cell on tailings beach Mattabi Mine Site, Ontario, Geocon Report T11498r15821, Montreal, Quebec, Canada. Kam, S.N., Knapp, R.A., Balins, J.K., Payne, R.A., 1997. Interim assessment of flooded tailings performance ᎏ
Quirke mine waste management area. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, II, 853᎐870. Karam, A., Guay, R., 1994. Inondation artificielle du parc ` a ´ residus miniers Solbec Cupra: Etudes microbiologiques et ´ chimiques, MEND Report 2.13.2c, CANMET, Natural Resources Canada, Ottawa, Ontario. ¨ Ljungberg, J., Lindvall, M., Holmstrom, B., ¨ H., Ohlander, 1997. Geochemical field study of flooded mine tailings at Stekenjokk, northern Sweden. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, III, 1401᎐1417. MEND, 1994. Flooding of a mine tailings site ᎏ Solbec Cupra. Suspension of Solids: Impact and Prevention, Report 2.13.2a, CANMET, Natural Resources Canada, Ottawa, Ontario. Pedersen, T.F., McNee, J.J., Flather, D., Mueller, B., Sahami, A., Pelletier, C.A., 1997. Geochemistry of submerged tailings in Buttle lake and the Equity Silver tailing pond, British Columbia, and Anderson Lake, Manitoba: what have we learned? Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, III, 989᎐1005. Rescan Environmental Services Ltd., 1989. Subaqueous disposal of waste rock and tailings, MEND Report 2.11.1a, CANMET, Natural Resources Canada, Ottawa, Ontario. Robertson, J.D., Tremblay, G.A., Fraser, W.W., 1997. Subaqueous tailings disposal: a sound solution for reactive tailing. Proceedings of the 4th International Conference on Acid Rock Drainage, Vancouver, BC, Canada, May 31᎐June 6, III, 1029᎐1044. Sobek, A.A., Schuller, W.A., Freeman, J.R., Smith, R.M., 1978. Field and Laboratory Methods Applicable to Overburdens and Minesoils, EPA Report EPA-600r2-78-054, US EPA, Cincinati. US EPA, 1986. Hazardous Waste Management System; Identification and Listing of Hazardous Waste; Notification Requirements; Reportable Quantity Adjustments; Proposed Rule. Federal Register Volume 51 Ž 114 . , 21648᎐21693. Yanful, E.K., Simms, P., 1997. Review of water cover sites and research projects, MEND Report 2.18.1, CANMET, Natural Resources Canada, Ottawa, Ontario.