UO2 Dissolution Modeling


Scott Dixon

Aquatic Chemistry Lab

Washington University in St. Louis

May 1st, 2007


Advisor: Daniel Giammar

Associate: Kai-Uwe Ulrich



 Knowledge of the dissolution kinetics of uranium dioxide, UO2, is vital in the remediation of DOE radioactive waste sites.  A flow-through reactor is used to test UO2 dissolution rates for varying levels of oxygen, pH and carbonates.  To properly understand data from these experiments, a model was created to fit the expected dissolution curve to the experimental curve by adjusting a dissolution rate constant.  By adjusting the oxygen, pH, and carbonate levels in different experiments and matching the data to the model, a dissolution rate constant can be found for each condition.  After enough experiments, a relationship will develop between experimental conditions and UO2 dissolution kinetics.  These relationships will increase the scientific understanding of uranium mobilization, and hopefully lead to a scientifically sound bioremediation process when the time comes.



The Department of Energy (DOE) estimates that 475 billion gallons of ground water has been contaminated with uranium from past waste disposal practices.  There are another 75 million cubic meters of contaminated sediment, and 3 million cubic meters of leaking sites such as landfills, trenches and spill-tainted soils.


Research teams here at Washington University in St. Louis as well as at the Stanford Synchrotron Radiation Laboratory, the Ecole Polytechnique Fédérale and the Oregon Health and Sciences University, are working to better understand the dynamics of uranium immobilized through in situ bioremediation.  The name of the project is “Coupled Biogeochemical Processes Governing the Stability of Bacteriogenic Uraninite and the Release of U(VI) in Heterogeneous Media: Molecular to Meter Scales.”


The focus of the project is to analyze how biological and chemical interactions change the speciation and mobility of uranium dioxide (UO2).  The hope is that if the uranium dioxide chemistry could be manipulated to a more stable form, this would minimize the risk of transport off site and subsequent human exposure.  For instance, if the particles were less water soluble, this could lower or stop the spreading of contamination into the soil and ground water.


Bioremediation has the potential to transform uranium from mobile species into less mobile ones, decreasing the transport rate. The project goal is to fully understand the behavior of uranium left at a waste site and, more specifically, to understand how biological and chemical interactions change the dissolution kinetics of uranium.  Here at Washington University, Dr. Kai-Uwe Ulrich and I worked on how fast uranium makes the transition from U(IV), uranium in the +IV oxidation state, to U(VI), uranium in the +VI oxidation state.  We were concerned with this relationship because these two oxidation states show significantly different equilibrium solubility.  U(VI) is highly water soluble, so it travels quicker underground and can reach the water table.  U(IV) is much less water soluble, and is therefore a more ideal state for uranium burial. 


We wanted to find out how stable U(IV) was under different environmental conditions.  The conditions we were tested for were oxygen, pH, and carbonate levels. In the presence of oxygen, U(IV) oxidizes to U(VI) at a rapid rate.  Oxidation is one type of reaction that leads to U(IV) dissolution, but not the only one.  Dissolution can also occur in an oxygen free environment.  By altering the pH or carbonate levels, the equilibrium solubility will change.  Carbonate forms complexes with both oxidation states.  It increases the water solubility of U(VI) by orders of magnitude, but it increases the solubility of U(IV) only slightly.  This comes into play because the uranium form we used in our experiments was UO2, which when in the +IV oxidation state still has low levels of U(VI).  So by adding carbonate to UO2 in the +IV oxidation state, there is a dramatic increase in the water solubility for the portion of U(VI).  This creates dissolution even in an oxygen free environment.



In order to measure the dissolution rate of U(IV) to U(VI), we used a flow-through reactor.  A flow-through reactor consists of an airtight container with influent and effluent tubes.  In the reactor is a mixed aqueous suspension of uranium solid.  The reactor worked by pumping a solvent into the container, which then forces a new solution out of the reactor.  A filter is placed in front of the effluent tube, thus preventing unwanted solids from exiting the container.  In our case, we wanted to stop UO2 particles from flowing out. 


To run an experiment, we filled our reactor with an aqueous suspension of U(IV).  U(IV) remains in particle form because it is not very water soluble.  Then, we would pump a solvent with known pH, oxygen content and chemical makeup into the reactor.  The U(IV) particles would remain in the reactor because of the filter on the effluent line.  However, if any of the U(IV) was oxidized to U(VI), a portion of the particles would dissolve and that uranium could exit the reactor.  However, some U(VI) could be retained within the solids and may not be released to the dissolved phase.  From that point, we measured the uranium concentrations in the effluent solution by inductively coupled plasma mass spectroscopy (ICP-MS).  Knowing how much U(IV) was initially in the reactor and how much dissolved uranium flowed out of the reactor we could calculate a dissolution rate for the reaction.



In each one of our experiments we used a model to estimate the dissolution rate k.  For one of our first experiments, we filled the 13.87 ml reactor with 35.2 mg of uraninite, a uranium rich material.  We periodically collected samples of the solution flowing from the reactor to test the uranium levels.  A plot of the results is shown on the figure below.


            To find the dissolution rate k for this reaction we needed to fit the concentration curve we got from the model to the data.  Figure 4.2 shows the model’s approximation for the uranium concentration for three different values of k.  The blue curve corresponds to a dissolution rate of , which although it matches the first four data points very well, the steady-state concentration is .0017 mM off.  The orange line represents the concentration when  and we see the dissolution rate is too low.  Finally, the green curve corresponds to .  Here the initial estimate is low, but the steady-state concentration is accurate.  We put more weight on the accuracy of the steady-state concentration than on fitting the initial curve because the initial reactions may not occur as methodically as predicted, but the concentration should always level off at the true steady-state value.  Thus, the model approximates the dissolution rate for these conditions to be .


We learned early that no two experiments are alike.  The model had to be adjustable to new circumstances that emerged and fortunately, it was.  We varied the model to allow for multiple dissolution rates at one time.  This meant we could have 98% of the particles have a dissolution rate of  and 2% with a dissolution rate of .  In one experiment we had uranium contamination in the feeding solution, thus we had to adjust the model to allow for a non-zero inflow of dissolved uranium.  Often, we wanted to switch feeding solutions while the reactor was running, causing a change in dissolution rate midway through the experiment.  The figure below is an example of this.  For this particular experiment, we put an alternative form of uranium called schoepite in the reactor.  We ran the reactor with a feeding solution of .1 M potassium nitrate buffered to pH 6.  Then after fifteen hours, we switched the feeding solution to a .005 M sodium bicarbonate, .095 M sodium nitrate solution at pH 7.5.  The feeding solution switch caused the dissolution rate k to increase by one order of magnitude.


As is evident from the last figure, the flow-through reactor model can be adjusted to fit different types of experiments.  Another option is available by changing the flow rate to .  This will change the flow-through reactor model to a batch experiment model.  Also, the model is not only restricted to uranium.  If the molecular weight and particle density is available for the compound of interest, the model should accurately capture the dissolution dynamics of that compound in a flow-through reactor.



As demonstrated in the last section, the model can be adjusted to fit many experiments.  It became clear early on that this tool would be useful for all flow-through reactor testing, independent of the material in question.  We created a generic model for this purpose.  With this version of the model, the inputs can be adjusted for two conditions.  This is good for two reasons.  One, this makes comparing inputs easier.  For example, in the figure below, the inputs for these plots are identical except condition two does not include the Gibbs term.  The plot clearly shows that the Gibbs term will limit the dissolution considerably.


The second reason is in case there is a change in the reactor midway through the experiment.  For instance, if the feeding solution is changed, there could be a change in dissolution rate.  By putting a “Y” next to the condition change option and inserting at what time this takes place, the model will adjust from a comparison plot to a single two-condition plot.  The final values for solids concentration, particle diameter, specific surface area and dissolved concentration for condition one are then transferred to be the initial values for condition two at the time of the switch.


Now, after inserting the experimental data into the specified area, it is possible to fit the model to the data by adjusting the dissolution rates.  The figure below is an example of a two-condition reaction depicted properly with the generic model.


Another generic model was created using Matlab.  The file size is much smaller and more efficient, but many issues that make the Excel model good are not available.  For instance, the ability to compare experimental values to the modeled values becomes more complicated and with it loss, the ability to determine dissolution rates.  Also, the ease of adjusting the model is then convoluted with coding, whereas the Excel model is more user-friendly.  The Matlab script is available in the appendix.  Figure 5.5 is an example of the output of the file.  Note that experimental data is not included. 



The contamination of our water and soil from radioactive waste sites is a considerable problem that will not diminish without intervention.  We need to develop processes that can eliminate this problem, or at least confine it.  Research being conducted today at universities and national labs is the first step in solving this issue.  When we begin to treat this problem in the field, we need to be confident of the proper tactics to use.  Therefore, we need an abundance of experiments to increase our knowledge of when and how uranium becomes immobilized.  


The model I created is a step in the right direction.  It allows us to experiment more often while providing us with a greater wealth of information per trial.  It is also flexible enough to capture the different types of experiments we have been running relating to uranium dissolution.  It does not provide all the answers, but it has been a valuable tool.  It has aided us on our path for a greater understanding of how oxygen, pH, and carbonates affect UO2 dissolution kinetics.