Old dog, new tricks?

It's the end of another perfectly good summer, and what do you have to show for it? Hopefully some good research data and a pounding hangover. For me, the summer research has been quite a learning experience.

You might recall that my original objective in the lab was to extract all of the phytoliths (more generically called biogenic silica or BSi) from my Nebraska soil samples. This hasn't changed, but the methods have. The old BSi extraction method follows that used by Piperno (2006). It is called the density extraction method, and it goes something like this:

1. Dry and weigh your sample.


2. Add hydrogen peroxide (H202) and hydrochloric acid (HCl) to remove organics and carbonates, respectively.


3. Add sodium metaphosphate to deflocculate the sample.


4. Wet seive the sample through a 53 micron seive to remove larger particles. [There are phytoliths which are larger than 53 microns, but most researchers tend to ignore these, as they are rare.]


5. Perform gravity sedimentation to remove particles smaller than 4 microns. [Again, there are phytoliths and phytolith fragments smaller than 4 microns, but researchers ignore these because it is very difficult to extract them in the density extraction method. Only later did I discover that a large percentage of BSi is actually smaller than 4 microns. This is one of the reasons I ultimately decided to abandon this technique.]


6. Float BSi in heavy liquid, such as sodium polytungstate, set at 2.3 g per cubic cm. Extract with pipette or peristaltic pump.


7. Dry BSi extract and weigh.

There are many reasons why this method should not be used for quantifying total BSi. First, seiving and gravity sedimentation exclude BSi which is >53 or <4>

Based on these drawbacks, I decided to turn elsewhere in my BSi quest. There is another method out there, termed alkali dissolution, which is promising (see my reviews of Jones 1969, Herbauts et al 1994, Saccone 2005). The basic method is as follows:

1. Add H2O2 and HCl to remove organics and carbonates. [This step is recommended by Saccone et al 2005, since it allows for easier dissolution of the silica.]


2. Add sodium carbonate or sodium hydroxide (both alkali solutions) to the sample, and heat for about 5 hours. [Heating and digestion times vary depending upon the researcher.]


3. Once an hour, remove a subsample of the supernatant and run it through the spectrophotometer to determine the dissolved silica concentration.

As time progresses in the digestion process, the concentration of dissolved silica will increase. If this were a perfect world, all of the dissolved silica would be from BSi. But unfortunately silica is a very common mineral, and can be found in many different inorganic forms, such as quartz, allophane, feldspars, aluminosilicates, and many others. These inorganic silicates usually have a lower dissolution rate than BSi. Moreover, the inorganics tend to dissolve at a linear rate through time. BSi, on the other hand, tends to dissolve quickly within the first two hours. This is shown graphically above (this graph is taken from DeMaster 1981). The x-axis is time, while the y-axis is the concentration of dissolved silica. Thus, the graph shows the increase of dissolved silica through time. As you can see, there is a large increase in silica in the first two hours, meaning that something is dissolving quickly (duh, right?). After two hours, the dissolution slows down and progresses at a linear rate. With this graph, DeMaster was able to estimate accurately the amount of BSi. Up to that point, nobody had been able to do this, since dissolved silica is all the same, regardless of the source. But DeMaster worked off of a few assumptions. First, that the BSi would dissolve quickly. Second, the inorganic silicates are going to be dissolving throughout the whole process, even in the beginning. But since the silicates dissolve at a linear and predictable rate, it was easy to determine how much dissoved silica was added from them. All one needs to do is follow the flat part of the curve back to time zero. All of the silica added below this extrapolation would presumably be from the BSi, while all that above the line would be from silicates.

This is a novel approach to determining total BSi, and quite a few researchers have used it, in both soil and aquatic sciences. However, it is not without its drawbacks. The DeMaster graph works best when there are very few inorganic silicate mineral types present. The graph above illustrates the dissolution of only one silicate. But what happens when there are multiple silicates, as is the case in soils?

This is where accuracy can take a back seat to averaging. In the case of soils, one must look at the dissolution curve and try to find the average silicate dissolution curve (hopefully there will be one dominant silicate which controls the curve). There are other options worth looking into as well. First, I wonder if it would be possible to document all of the silicates in a sample, and their relative proportions. If the dissolution rates are known for each silicate, then maybe it would be possible to get more accurate results. I guess all you would really need to know is the most reactive silicate, right after the BSi.

Second, aquatic researchers have made use of silicon-aluminum ratios (Si/Al) to estimate the amount of silica added from silicates. Koning et al (2002) dissolved various aquatic samples in sodium carbonate, and simultaneously measured dissolved silica and aluminum concentrations through time. The underlying idea is that BSi has a very high Si/Al (meaning it has very little Al), while silicates tend to have much lower Si/Al (alot more Al). As with the DeMaster graph above, Koning et al. displayed the change in dissolved silica through time. But they also added the change in dissolved Al as well. Armed with this type of graph and some really scary looking equations, they were able to differentiate BSi from up to three different types of silicates in a sample. The researchers were trying to show that they could identify individual silicates in their samples; they really didn't care about the BSi. But Si/Al ratios can still be used to quantify the total BSi: one simply needs to find the shift from high to low Si/Al ratios in the dissolution curve. It doesn't matter how many silicates are in the sample, so long as they all have lower Si/Al ratios than the BSi.

A third possible method is the use of stable isotopes. Derry et al. (2005) and Kurtz et al. (2002) looked at the ratio of germanium to silica in soils (see my earlier reviews of these articles). In a nutshell, higher Ge/Si indicates a more weathered soil. While I'm not concerned about soil weathering, I am interested in the Ge/Si of soils, because it may be vastly different than that found in BSi. There is no study that I am aware of which has documented the Ge/Si in BSi. Combined with Si/Al, Ge/Si could be a valuable tool. It's only drawback would be the expense.

So all in all, that is what I've been working on this summer. Joe and I have ordered the supplies to begin the dissolution extraction method. A few of the supplies are on backorder, so here I sit and wait. And blog.