Graduation Year


Document Type




Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department


Major Professor

Mark C. Rains, Ph.D.

Committee Member

Len Vacher, Ph.D.

Committee Member

Mark T. Stewart, Ph.D.

Committee Member

Jeffery Ryan, Ph.D.


ASR, Arsenic, Floridan Aquifer, Pyrite, Geochemical Modeling, Injection/Recovery


Aquifer storage and recovery (ASR) is a strategy in which water is injected into an aquifer when it is plentiful and pumped from the aquifer when water is scarce. An impediment to ASR in Florida is leaching of naturally-occurring arsenic from limestone of the Upper Floridan Aquifer System (UFAS) into stored water. The concentration of arsenic in surface water, which serves as the recharge water for many ASR systems, and native groundwater is usually much less than 3.0 µ/L. However, data from ASR wells in Florida show that arsenic in recovered water frequently exceeded the 10 µg/L maximum contaminant level (MCL) established by the Environmental Protection Agency and were as high as 130.0 µg/L. The cause of elevated arsenic concentrations is displacement of reduced native groundwater with oxygenated surface water that dissolves arsenic-bearing pyrite in limestone. Although arsenic can be removed from recovered water during final treatment, mobilization of arsenic in the aquifer at levels that exceed the MCL is problematic under federal regulations.

This dissertation investigated a number of aspects of the ASR/arsenic problem to provide additional insights into the mechanisms of arsenic mobilization and measures that could be taken to avoid or reduce the release of arsenic during ASR operations.

Chapter 2, involved development of a geochemical model to simulate an ASR system’s injection of oxygenated surface water into reduced groundwater to determine whether aquifer redox conditions could be altered to the degree of pyrite instability. Increasing amounts of injection water were added to the storage-zone in a series of steps and resulting reaction paths were plotted on pyrite stability diagrams. Unmixed storage-zone water in wells plotted within the pyrite stability field indicating that redox conditions were sufficiently reducing to allow for pyrite stability. Thus arsenic is immobilized in pyrite and its concentration in groundwater should be low. During simulation, as the injection/storage-zone water ratio increased, redox conditions became less reducing and pyrite became unstable. The result would be release of arsenic from limestone into storage-zone water.

Chapter 3 examined the importance of maintaining a substantial volume of stored water around an ASR well to prevent recovery of reduced native groundwater to the vicinity of the well. Depleting the stored water and recovering reduced native groundwater would result in dissolution of arsenic-bearing hydrous ferric oxide (HFO) and release of arsenic into water recovered from the ASR well. Injection/recovery volumes for each cycle for each well were tracked to determine if a substantial volume of stored water was maintained for each cycle or if it was depleted so that reduced native groundwater was brought back to the well. Each well was assigned to either the “storage zone maintained group” where a zone of stored water was established in early cycles and largely maintained through the period of investigation, or the “storage-zone depleted group” where a zone of stored water was either established in later cycles and/or was depleted during the period of investigation. Graphical and statistical analyses verified that maximum arsenic concentrations for storage-zone maintained wells were nearly always lower in each cycle and declined below the MCL after fewer cycles than those of storage-zone depleted wells.

Chapter 4 was a mineralogical investigation of cores located at 20 m (ASR core 1), 152 m (ASR core 2), and 452 m (ASR core 3) from operating ASR wells to determine where mobilized arsenic in limestone is precipitated during ASR. If arsenic is precipitated distally, reduced concentrations of elements in pyrite, (iron, sulfur, arsenic, etc.) would be expected in ASR core 1 relative to more distant cores and there would be noticeable changes in appearance of pyrite crystals due to enhanced oxidation. The results showed that mean concentrations of the elements were lowest in ASR core 2, which did not support distal precipitation. However, scanning electron microscopy identified well-defined pyrite framboids only in core 3 while framboids in ASR cores 1 and 2 were less clear and distinct, indicating pyrite oxidation in cores closest to ASR wells.

Statistical comparison of concentrations of iron, sulfur, and arsenic between the three ASR cores and 19 control cores not subject to ASR, showed that mean concentrations in ASR cores 1 and 2 were statistically similar to concentrations in control cores. This indicated that concentrations in ASR cores 1 and 2 had not been significantly reduced by ASR. The concentrations of elements were higher in ASR core 3 than in ASR cores 1 and 2 and control cores and statistically dissimilar to all but one control core. This indicated natural heterogeneity in core 3 rather than diminution of elements in ASR cores 1 and 2 due to ASR. The statistical analysis supported local precipitation. Once arsenic is mobilized from dissolved pyrite, it is rapidly complexed with precipitated HFO near the well. As long as all of the stored water is not removed during recovery so that reduced native groundwater is brought back to the well, HFO remains stable and complexed with arsenic. The concentration of elements would not have been lowest in ASR core 1 for this reason and because calculations showed that the mass of arsenic removed during recovery events prior to coring was minor compared to the total in limestone surrounding the well. The implications of this are that while large quantities of arsenic are present near the ASR well, only a small percentage may be available for dissolution. Most arsenic occurs with pyrite in limestone, which may insulate it from exposure to oxidized injection water. Water recovered from ASR wells may continue to have low concentrations of arsenic indefinitely because as limestone is dissolved, more pyrite becomes exposed and available for dissolution.

The primary contribution of this dissertation to understanding and overcoming the arsenic problem in ASR systems is the empirical data developed to support or challenge important ASR/arsenic hypotheses. These data were used to 1) establish that background concentrations of arsenic in groundwater of the Suwannee Limestone were less than 1µg/L, 2) demonstrate that redox conditions necessary for pyrite in limestone to become unstable and dissolve occur when oxygenated surface water is injected into the aquifer, 3) demonstrate that the concentration of pyrite in the Suwannee Limestone is spatially variable to a high degree, 4) support the hypothesis that following injection of oxygenated surface water, pyrite in limestone dissolves and releases arsenic into solution and HFO forms and complexes with the arsenic near the ASR well, 5) propose that only a small percentage of pyrite near an ASR well may be available for dissolution during each cycle because most occurs in the limestone matrix and is isolated from injection water, 6) propose that as a result of the previous conclusion, water recovered from ASR systems may continue to have low concentrations of arsenic indefinitely because as limestone that contains pyrite is dissolved with each cycle, additional pyrite is exposed and is available for dissolution, and 7) support the effectiveness of maintaining a zone of stored water in an ASR well as an effective means of minimizing arsenic in recovered water during ASR.