Lecture No. 6. Grain Size Controls
Total-in-Sediment Metals
This is a USGS method. A dried and weighed sediment sample is
treated with several very strong chemical reagents. Nearly all
the sediment, including quartz and silicates, dissolves. This
method is entitled "Metals, minor, total in
sediment."
_________________________________________________________________________________________
Assignment (due Thursday, February 25): Find out
what the "strong chemical" digestion method is that the
USGS prescribes for the total-in-sediment trace metals method.
Turn in a brief (1 paragraph) account of the digestion.
_________________________________________________________________________________________
Grain-Size Control of Elemental Concentration
There commonly is a strong correlation between sediment grain size and metal concentration. More specifically, the % fines in the sediment and the concentration of metal in the sediment [Me]sed are strongly correlated. [Me] tends to be higher in the finer fractions. This is especially true for Cu, Cd, Fe, and Zn.
Why is this?
As grain size decreases, the fraction of silica decreases, surface area increases, ion exchange capacity increases, and clay mineral content increases.
Now here is a kicker of sorts: Sampling only the small grain sizes greatly reduces variation. I call it a kicker, because it could be viewed as cutting out statistical variation by throwing out some valid data, which is also called fudging. However, since we know that for most purposes almost all of the adsorption, ion exchanging, and chelating take place on the smaller particles, it actually is justifiable.
In the next page of the handout, there is a figure (A) that shows how surface area per gram increases as particle size decreases. Table B lists specific surface areas for various grain sizes. Note that spherical particles are assumed. That assumption is generally not good for clay sizes, since clay crystals tend to be flat and have larger specific surface areas; surface area actually increases faster at small grain sizes. Table C lists specific surface areas for various minerals in the clay size range, confirming what I just said. The next page is a plot of surface per unit volume against grain size classes. Note that the x-axis is categories, not linear numbers. Particles get smaller to the right.
On the next page, Table A shows cation exchange capacities of various minerals, all of which are enriched in the smallest size range. Table B shows the increase in cation exchange capacity with decrease in grain size for kaolinite, and Table C does the same for illite.
The next page shows the dependence of metal concentrations in Clark Fork sediments on grain size. The horizontal scale starts at 0 m m on the left. After 162 m m, the next number is 1 mm. Interestingly, the different metals behave differently, depending on their adsorption and ion-exchange properties. Cadmium is richest in the clay-sized range. Copper, iron, manganese, nickel, and zinc seem to peak in the silt size range. Nickel and zinc have bimodal curves, with second maxima in the clay range. But note that all of these metals are richest in the mud fractions.
The best approach to correcting for grain size effects is still to limit the particle size. Different researchers use different maximum grain sizes.
a. The 4f fraction and smaller is always carried as suspended load. (And recall that we can easily measure suspended solids, but bedload is very hard to measure meaningfully.)
b. Mud accumulates and has high biological import. Bugs, fish, and other critters ingest it. It gets onto gills.
The next page shows trace element concentrations in the River Ems (Germany) vs. the percentage of fine (<16 m m) grains. Manganese, iron, cobalt, and mercury are all strongly enriched in the fine fraction. (Note the scales are different for the different elements.)
On the same page is a table from a paper by Horowitz and Elrick comparing predicted concentrations of various metals with measured concentrations in the mud fraction (< 63 m m) with measured ones. This requires a little explanation. The assumption here is that all, or essentially all, of the recoverable sediment-bound metal is in the mud fraction. If that is true, then the concentrations of those elements in the mud fraction would have to be larger by a factor of 100% divided by the percentage of the sediment smaller than 63 m m. For example, if the sediment is 40% mud, then the concentrations in the mud would be higher than those in the bulk sediment by a factor of (100%/40%) = 2.5 in order to account for the bulk concentrations. (This is obfuscated a little in the key to the table referring to this ratio as 100%/(100%-% sand).) Metal concentrations were measured in the bulk sediment and in the mud fraction. Measured concentrations in the mud were then compared with the concentrations calculated based on the assumption.
Some metals gave better fits than others to the prediction, and some sampling locations fit better than others did. The Columbia Slough samples generally did not fit. Everything fit pretty well at Lake Bruin. Copper gave a good fit for most sediments, while it was only fair for the Nemadji River. Zinc, cobalt, and titanium were mixed bags. The other elements fit fairly well. It looks as though we are onto something, but we don’t have the entire picture.
The Horowitz-Elrick calculation looks interesting, but it is not really reliable, so it is not that useful.
To find the variability due to grain size, you have to separate out fractions and analyze them separately. Horowitz-style normalization only works under limited circumstances.
What is really happening? Metal concentrations on/in sediments are controlled by the physical environment and the geochemical environment.
The geochemical environment controls just what sticks to the coarse and fine grains:
The next page shows copper concentrations in mud vs. bulk sediment at various distances from a contaminant source on the Clark Fork. Concentrations in mud are invariably higher than the bulk concentration, and they stay high longer. Why? Mud (suspended sediment) travels farther and faster than bedload. In a normal year, the suspended sediment travels as fast as the water — miles per hour. Sand travels at most a few hundred feet per year. Gravel travels maybe 10 ft/y. So, at greater distances from the contaminant source, the effect of contaminated mud is still felt. Other phenomena may also be at work. For example, as mud clasts break up, more of the material becomes suspended as fine sediment. Coatings get worn away from sand and gravel with travel, too, adding more to the fine fraction downriver. In any case, the Horowitz calculation obviously doesn’t work for the Clark Fork, since the ratio of mud concentration to bulk concentration keeps changing.
The next three pages are figures from a report Johnnie Moore and Edward Brook did in 1987 on the controls on metal transport in the Clark Fork River. On the first page, total metal (Cu, Mn, Fe, or Zn) in sediment is plotted against the percentage of the sediment smaller than 63m m in size. The correlation is poor. On the next page, normalized sediment metal calculated from the Horowitz formula is plotted against metal concentration in fine sediment. Correlation is even worse. On the third page, metal concentrations in fine sediment and the normalized metal concentrations are plotted vs. river kilometer. They do not track each other very well.
Something to keep in mind about the Clark Fork River is that its sediment is unnatural. Much of the transported sediment, both fine and coarse, did not weather from igneous or sedimentary rocks. It was pounded in ore mills, and the part that did not contain enough metal to be smelted was dumped on the flood plain. (That’s where a major portion of the Clark Fork tailings deposits came from.) We might therefore expect the tailings’ size distribution to be a little odd. Some of the tailings are sand-sized, and others are finer. Small wonder that the metal analyses from the river sediment do not follow any theoretical distribution.
What we have are two situations.
In fact, if the samples are >40-50% mud and there is not much variation among samples, Horowitz’s normalization works pretty well. Sea-bottom muds are a good example.
Geochemical Classification
The geochemical environment is based on pH and Eh (or pe). It is characterized by the Eh-pH diagram, which is based on activities of H+ and e-. Thus it is constructed from thermodynamic data. It predicts partitioning of phases and mineralogical separation.
See the example diagram below. Note that the transition from species 1 to species 2 is strictly pH-controlled, while the transition from species 3 to species 4 is strictly Eh-controlled. Both pH and Eh influence the transition between species 2 and 4.
The O2 and H2 lines define the redox region in which water is stable.
The boundaries are equilibrium conditions. A given diagram is valid
Next: We will construct an Eh-pH diagram for aluminum and study more Eh-pH diagrams.
To previous lecture: Lecture No. 5. Measuring Metals in Sediments
To next lecture: Lecture No7. Eh-pH Diagrams
Back to list of lecture notes