Lecture No. 5. Measuring Metals in Sediments

Review of Stream Sedimentology

See handouts.

A. Fundamentals

Sediment in streams in carried as

• Dissolved load
• Suspended load
• Bedload

Dissolved load moves in solution. Suspended load moves in suspension, of course. Bedload moves by saltation and traction.

For now we will consider suspended load and bedload.

B. Controls on Settling Velocity

There is a strong relationship in some systems between stream discharge and sediment transport. The correlation is stronger for bedload than for suspended load.

In Figure 1, we see the correlations between discharge, stream power, and bedload transport for a river in Wyoming in May and June.

In Figures 4 and 5 (Powder River), the relationships for suspended load vs. discharge are very noisy.

Figure 2 (Niobrara River) shows bedload discharge per unit width in a direct relationship with velocity. (The y-axis in upside down.)

In Figure 3, for two rivers, the suspended load leads the discharge.

Figure 6 shows depth profiles for discharge-weighted concentrations of suspended sediments. The bottom of the figure is bedload. Not surprisingly, sands tend to stay near the bottom. Clays and silts show much more uniform depth profiles.

Figure 7 shows correlations across a stream cross-section between discharge and sediment load.

Figure 8 is a familiar one. It shows much better correlation between velocity and load for bedload, mostly because the entrainment and settling curves are close for sand and gravel but very far apart for clay and silt.

C. Solute-Particle Relationships

1. The relative transport rates of different phases in rivers are very difficult to measure because of variability in space and time.

2. Bedload is rarely considered in papers because it is so hard to measure.

3. There are some reasonable relationships between solute and particulate phases if we consider suspended load. (But they aren’t all that good.)

See the second series of figures.

Figure 1 shows metals transport in two "clean" rivers, the Amazon and Yukon. Note that most of the metal (Cr, Mn, Fe, Co, Ni, and Cu) moves in suspension.

Figure 2 shows some dirtier rivers. Again, most metal moves in suspension.

In order to look at metal transport quantitatively,

1. We have to separate the sediment from the water.

2. We need to know the discharge of the stream (fairly straightforward) and the discharge of sediment (very hard to do because of variability in space and time). Most information on sediment transport is calculated from sediment-rating curves. (See Figures 4 and 5 in the first series.) One river, one rating curve — or maybe one river, multiple rating curves for different times of year.

Sediment Separation from Dissolved Load

Whenever possible, this is done in the field. A water sample is collected in a clean container and filtered (if desired), then treated with a preservative if appropriate. Sample treatment varies depending on what is being determined. It usually breaks down like this:

a. Samples to be analyzed for dissolved metals are filtered, then adjusted to pH 2 with either nitric or hydrochloric acid.

b. Samples to be analyzed for total metals (dissolved and in sediment) are acidified without filtration.

c. Samples to be analyzed for anions are filtered and not acidified.

d. Samples to be analyzed for dissolved organic carbon are normally filtered, then acidified with hydrochloric acid and usually transported on ice.

The containers used for most samples are polyethylene. Cleaning is a very thorough process. The less stringent cleaning process for containers used for samples analyzed for metals involves

• washing with Liquinox ¨,

• rinsing thoroughly, first with tap water, then with deionized water,

• soaking for > 1 hour with 5 N HCl (reagent grade),

• rinsing 3-4 times with deionized water and 1-2 times with double-deionized water, and

• drying in a dust-free environment.

The containers used for samples to be analyzed for organic carbon are glass. These containers go through the same acid-washing steps described above and then are baked out for > 1 hour in a muffle furnace at 400¡ C.

Containers used for samples analyzed for anions are thoroughly washed with Liquinox ¨ and rinsed with double-deionized water but are not usually acid-washed because of the chance for chloride contamination.

Next: More on sample treatment and liquid-solid separation.

Here are a few more optional exercises in balancing redox equations for those who would like to get more practice.

1. Methane reacting with hematite in anoxic sediments (neutral):

CH₄ + Fe₂O₃ ¨ CO₂ + Fe²⁺

2. Fermentation of carbohydrates by anaerobic bacteria (neutral):

CH₂O ¨ CO₂ + CH₄

3. Nitrification by autotrophic bacteria under neutral conditions:

NH₄⁺ + O₂ ¨ NO₃^–

4. Denitrification in acidic soil:

NO₃^– + CH₂O ¨ N₂ + CO₂

5. Dissimilatory nitrate reduction in acidic soil:

NO₃^– + CH₂O ¨ NH₄⁺ + CO₂

The filtration (or sediment separation) methods include the following:

1. Membrane filters, with or without coarse prefilters or centrifuging, are used. These filters are well suited to field sampling. Unless the samples are going to be very large, all that is needed is a clean 60 mL syringe and a disposable filter for each sample.

As I mentioned earlier, the definition of "dissolved" vs. "suspended" load is an operational one, and USGS has customarily used 0.45 m m as the cutoff. (Why? Because 0.45 m m filters are widely available, and because filtration through finer filters gets increasingly hard.)

Look at Figures IX-1 and IX-2 in the second set. For iron in these two well-aerated streams (forks of the Madison River in Yellowstone Park), the finer the filter you use the less dissolved iron there appears to be. On the other hand, arsenic in the Madison River system is mostly in solution irrespective of filter size (Figure IX-3).

One problem is that filtration doesn’t get you a decent amount of sediment unless you filter a lot of water. Moreover, the filters most commonly used are designed for the exclusion of sediment, not for its recovery.

2. Tangential-flow filters are somewhat better at recovery of sediment-free water, since recovery is traded off to get purity. They also yield a sediment-enriched slurry that can be filtered or centrifuged to recover sediment. (See overheads.)

The trouble with tangential-flow filters is that they are very expensive (upwards of $1200 each), take lots of water, and are a bear to clean out between samples. They are also slow and subject to clogging. And they are not well suited for field use. The large volumes of water used with them require something like a peristaltic pump and a fresh length of acid-washed tubing for each sample.

3. Hollow-fiber filters are another approach to getting sediment-free water by trading off recovery for purity. They are even more trouble to clean than tangential-flow filters, require pumps and tubing changes, and are generally a pain if you are hiking out to a stream carrying everything on your back. (See the last figure.)

4. High-Speed Centrifugation

It can involve batch operations (bottles or tubes in a centrifuge) or a continuous operation akin to a cream separator. Size range of particles separated can be from 0.5 m m to 0.01 m m. This approach works pretty well for a lot of solids. Density is crucial, since centrifugation relies on density differences. Materials too close to water in density or those lighter than water (like most organics) can’t be separated this way.

The main drawback is that you can’t do this in the field, because you can’t lug the centrifuge around, and streams don’t usually come equipped with 115 VAC outlets.

5. Dialysis

This is also slow, but it really works. Water is forced through a membrane that only allows molecules below a certain size through. The cutoff can be almost any molecular weight. For our purposes (i.e., for purposes of geochemistry), a fairly large cutoff size can be used — say 10,000 to 20,000 daltons. (1 Dl = 1 amu) For other purposes, such as protein purification, lower cutoff sizes are used. One drawback: soluble fulvic acids may be held up with the solids.

Pierce chemical makes a really nice device consisting of a small funnel-like container with dialysis membrane at the bottom. This fits into a slightly larger receiving cup. The apparatus is assembled, a water sample is poured into the upper member, and the whole thing is centrifuged at a few thousand gravities for a few minutes. When it is taken out of the centrifuge, most or all of the water is in the bottom, and all the particles and large molecules (fulvic acids, for instance) are in the top.

Again, this generally cannot be done in the field.

Assumptions

We frequently make the following assumptions:

1. We can measure solute and particle concentrations, and we use the procedural definition with its 0.45m cutoff.
2. It is hard to measure river loads because of variability.
3. But we need to know relationships between particles and solutes.

Distribution Coefficient, D

This is a way to make comparisons between solute and sediment loads. It is an equilibrium constant. It assumes that a solute has a certain affinity for being dissolved in water, and another affinity for being adsorbed to a given surface (e.g., FeOOH). Thus the ratio of adsorbed concentration (mols per square meter of surface) to dissolved concentration (mols per liter) is a constant at a given temperature. In environmental work, grams are used instead of mols, and adsorbed concentration is expressed in m g metal per g sediment. (If the adsorbed metal coprecipitated with the sediment, this may actually be a more accurate measurement.)

For those who have worked with liquid or gas chromatography, this distribution coefficient is similar to a partition coefficient.

Here are some examples from the Madison River:

1. Upper Madison

2. Lower Madison

3. Missouri River

Here are some examples from another study that show how the distribution coefficient depends on how the sample is filtered (for some elements more than others):

What does this tell us? The finer particles of sediment seem to be enriched in iron, and the value of D depends strongly on what size we use for our cutoff; but As concentrations are pretty evenly distributed across the different particle sizes.

The important point here is that your filter size is part of your definition of D.

Total Recoverable Metals

This is a common USGS method in which an unfiltered sample is heated with dilute hydrochloric acid. This treatment tends to dissolve amorphous and poorly crystallized iron and aluminum oxyhydroxides and free up the metals adsorbed to or coprecipitated with them. It also tends to desorb metals that are adsorbed reversibly to sand, clay, and other silicate minerals, and it dissolves certain acid-soluble sulfides (but not pyrite). Thus it measures the sediment-bound metals that tend to exchange easily with the solute phase. It does not dissolve silicates, quartz, or refractory oxides (corundum, hematite, magnetite, etc.)

This method measures dissolved metals, plus metals that may conceivably become dissolved as a result of minor changes in water chemistry, ingestion by aquatic life, etc. It separates that fraction of metals from the fraction that will not readily come into equilibrium with the dissolved fraction.

A sample is split according to the following scheme:

Sample S is split into S1, S2, and S3.

Stream discharge is also measured.

Total suspended sediment is measured by filtering the water, drying the filter, and weighing the residue. The metals in sediment and metals in solution can then be converted to suspended and dissolved metal load per day.

Example of the use of total recoverable metals: Clark Fork River at Deer Lodge, 11 April 1985.

Total suspended solids = 63 mg/L

Total recoverable Cu = 130 m g/L

Dissolved Cu = 12 m g/L

Q = 358 ft³/s = 10142 L/s

Sediment-bound Cu in water = 130g/L – 12 g/L = 118 g/L.

Cu concentration in sediment = (118 g Cu/L H₂O)/(0.063 g sediment/L H₂O)

= 118 m g Cu/0.063 g sediment = 1873 m g Cu/g sediment = 1873 ppm Cu in sediment

Dissolved Cu load per day = 12 m g Cu/L « 10142 L/s « 3600 s/h « 24 h/d = 10512 g Cu/d = 10.5 kg Cu/d

Suspended Cu load per day = 118 m g Cu/L « 10142 L/s « 3600 s/h « 24 h/d = 103368 g Cu/d = 103.4 kg Cu/d

Bedload transport is neglected because of the difficulty in measuring it. (Consider that each size and density fraction of the bedload moves at a different rate and contains a different concentration of each metal.)

In general, total recoverable metals tracks sediment concentration (naturally, because there tends to be much more metal in sediments than in solution). (See handouts.)

The second handout shows some interesting data. Note in Figure 18 that arsenic in suspended sediment decreases steadily downstream of the Butte area. What is happening is that cleaner sediment is being added to the river. The river’s total arsenic load is not dropping — it is probably still increasing — but the average spoonful of sediment contains less arsenic. The tributaries tend to be low in sediment arsenic, except for Flint Creek, which drains the Phillipsburg mining district.

In Figure 19, copper decreases downstream; at Missoula it probably is close to natural background levels. The tributaries are all pretty low in copper.

The next page of the handout shows a formula for converting stream discharge and metal concentration to metal discharge in tons per day. It is nothing complicated. K is just the product of 5 constants:

See Figure 25; total recoverable Cu discharge at Deer Lodge correlates well with suspended sediment discharge.

Next: More on sediments.

To previous lecture: Lecture No. 4. Chemical Reaction Types

To next lecture: Lecture No 6. Grain Size Controls

Back to list of lecture notes