Phosphate plume persistence at two decommissioned septic system sites

Phosphate plume persistence at two decommissioned septic system sites

Robertson, W D

Introduction

Laboratory studies of phosphate transport through sediments invariably note a high degree of P043- retardation. This is generally attributed to PO^sub 4^^sup 3-^ adsorption onto the sediment solids, particularly the metal oxyhydroxide minerals that possess positive surface changes at normal pH ranges (e.g., Parfitt et al.1975; Rajan 1975; Goldberg and Sposito 1984). However, in order to successfully model PO 3- behavior in some column studies, it has been necessary to invoke a two-stage process in which linear equilibrium sorption, which occurs rapidly and is reversible, is accompanied by a second “sorption” process that occurs more slowly and is less readily reversible. The latter process is referred to as the kinetic or ratelimited sorption step and is often manifested as a consumption process that prevents column effluents from achieving C/Co values of 1.0 or leads to pronounced asymmetry of breakthrough curves (e.g., Grove and Stollenwerk 1985; Nagpal 1986; van der Zee et al. 1989). These slower reactions are poorly understood and have been attributed to a variety of processes such as intra-particle diffusion, variability in the energetics of the sorption sites, slow recrystalization of adsorbed P043- into new mineral phases, or the direct precipitation of P minerals from solution (e.g., Chen et al. 1973; Martin et al. 1988; Davis and Kent 1990; Isenbeck-Schroter et al. 1993; Saiers and Hornberger 1996). The importance of these slower secondary processes is that they can be less readily reversible and in some cases may constitute consumption mechanisms that can permanently immobilize P.

The degree of reversibility of PO^sub 4^^sup 3-^ attenuation reactions can have important implications when P-rich contaminant sources such as septic systems are located close to sensitive surface water bodies that may be at risk from excessive phosphorus loading. Nutrient loading from all sources currently is receiving increased scrutiny in many fresh water and marine environments as a consequence of deteriorating water quality conditions. Phosphorus levels in septic tank effluent (3 to 20 mg/L P; Whelan 1988; Cogger et al. 1988; Beek et al. 1977; Robertson et al. 1998) are several orders of magnitude higher than the minimum levels necessary to stimulate algae growth in aquatic ecosystems (Dillon and Rigler 1974). While the initial fast sorption step is capable of substantially retarding PO^sub 4^^sup 3-^ migration velocity, because it is reversible, it does not necessarily prevent PO^sub 4^^sup 3-^ from migrating downgradient and eventually impacting nearby water bodies. As a consequence, some jurisdictions, such as the Province of Ontario, have adopted a conservative approach when calculating lakeshore P loadings and presently assume that all of the P mass loaded to septic systems is ultimately capable of migrating downgradient to adjacent surface water bodies (Dillon et al. 1986). However, if the secondary slow step is indeed irreversible, it may represent a consumption mechanism that is capable of permanently immobilizing PO^sub 4^^sup 3-^, hence allowing estimates of P loading from septic systems to be revised downward.

In this study, PO^sub 4^^sup 3-^ behavior was monitored in two septic system plumes for extended periods, including two to four years after decommissioning of the tile beds. Our objective was to observe PO^sub 4^^sup 3-^ behavior after the sewage source was removed, particularly to gain insight into the importance of any secondary consumption mechanisms that might be active. If such processes are important in the ground water zone, then an observation of declining PO^sub 4^^sup 3-^ concentrations should be apparent after the sewage source is removed. On the other hand, if reversible sorption is the dominant mechanism, then pre-existing P043- values may persist for a considerable period after decommissioning, owing to the substantial P mass that is likely to be adsorbed on the sediment solids.

Field Sites

Both of the field sites are situated on calcareous sand aquifers at locations where conventional septic systems are generating distinct ground water plumes. Multiple piezometer bundles of the type described by Cherry et al. (1983) and consisting each of six to 21 polyethylene tubes attached at varying depth intervals to a PVC center pipe, were installed using a portable percussion hammer and removable 5 cm diameter casing. These devices allowed economical installation of detailed monitoring networks so that flowpaths emanating from the tile beds could be traced downgradient with confidence.

At the Langton site, sewage from a public school with about 200 students was discharged to a conventional gravity fed septic system for a 47-year period from 1947 to 1994. A 325 m2 tile bed is located on an unconfined aquifer of medium sand at a location where the water table is at 3 m depth. In 1991, a monitoring network consisting of 45 piezometer bundles with a total of 400 monitoring points was installed in the area below the tile bed and for a distance of up to 110 m downgradient. Extensive sampling in 1991 enabled investigators to clearly define the sewage plume based on elevated levels of electrical conductance (EC), Na+, Ca2+, HCO^sub 3-^, C1-, P043-, and NO3- (Harman et al. 1996). A Br tracer test and calculations using the Darcy equation established that the sewage effluent resides for about one to two weeks in the vadose zone and then forms a horizontally migrating plume in the ground water zone traveling at a velocity of about 100 m/yr.

The plume contains a zone of enriched P043- which extends about 70 m downgradient from the tile bed and has P043- concentrations in the range of 0.3 to 2 mg/L P. Beyond this distance PO43levels decline abruptly to background values (

Although a distinct phosphate zone was present in the Langton plume, it was noted that ground water PO43- concentrations, even in the monitoring points located closest to the infiltration bed, were substantially lower than effluent values. Concentration differences between the sewage effluent and the ground water suggested that about 85% of the P043- mass was being retained in the 3 m thick vadose zone even after more than four decades of sewage loading at this site (Harman et al. 1996).

In May 1994, a replacement tile bed was constructed at the Langton site and sewage flow to the old bed was terminated. The monitoring network at the old tile bed was left intact, and ground water monitoring continued after decommissioning.

At the Long Point site, sewage effluent from a seasonal-use campground that has about 200 overnight camping sites is discharged to a conventional septic system located on an unconfined aquifer of medium sand. Effluent has been pumped at various times to two tile beds that are present at the site, positioned about 300 m apart. From 1971 to 1990 effluent was pumped exclusively to tile bed 1; from 1990 to 1995 it was pumped to tile bed 2; then in 1996 it was diverted back to tile bed 1. In 1990, a monitoring network consisting of 135 monitoring points in eight multiple piezometer bundles was installed underneath tile bed 2 at the same time that sewage flow was diverted to this tile bed. This study focuses on PO43- behavior below tile bed 2. Tile bed 2 lies close to a water table divide and occasionally lies directly over the divide particularly during periods of heavy sewage loading (Cherry et al. 1990). Because of this, a substantial downward component of flow was imparted to the septic system plume such that it invaded the entire thickness of the aquifer to 7 m depth after six months of effluent loading. A horizon of N03- depletion occurred at mid-depth in the plume and Aravena and Robertson (1998) used isotopic evidence to deduce that attenuation was the result of denitrification using trace quantities of pyrite, in addition to carbonaceous material present in the aquifer, as the electron donors. N03- behavior in the older tile bed plume (tile bed 1) was also the focus of a previous study (Robertson and Cherry 1992).

In the P043- zones at both Langton and Long Point, nitrogen occurred primarily as N03- rather than as NH4+ throughout the period of active sewage loading and Fe concentrations remained low (

Both the Langton and the Long Point 2 plumes are included in a recent review of phosphate behavior at 10 septic system sites in Ontario (Robertson et al. 1998). The study deals only with the inuse period, however, and does not contain data from the decommissioned period.

Methods

At the Langton site, sampling for major plume constituents (Na+, Ca2+, Cl-, PO43-, N03-) was generally continued at one- to fourmonth intervals, for a four-year period after decommissioning. Sampling was focused on three primary monitoring points: piezometer B31-2 located in the shallow ground water zone beneath the center of the tile bed (proximal monitoring point); piezometer B 1-8 located in the plume core 16 m downgradient from the center of the tile bed (medial point); and piezometer B7-8 located in the plume core 65 m downgradient, near the frontal part of the P plume (distal point). In addition, a more detailed sampling episode was undertaken about one year after decommissioning (September 1995) that included sampling 61 monitoring points from along the plume centerline for Na+, Ca2+, Cl-, PO43- and NO3-,

At the Long Point 2 site, detailed P043- distribution was established after six seasons of effluent loading by sampling undertaken during October 1995 to March 1996. Although some of these samples were retrieved after the campground was closed in November 1995 for the winter season, for the purposes of this discussion these samples are considered representative of the in-use period. No further effluent was then discharged to tile bed 2 until after a second detailed sampling for PO43- was completed in August 1997. Field measurements of electrical conductivity were completed at that time to check for the presence of the plume and six samples were analyzed for C1- content to provide a comparison with the earlier data. Otherwise, no additional ion analyses were undertaken during this sampling episode.

Samples were retrieved using a peristaltic pump and were field filtered (0.45 pm) prior to atmospheric exposure. Samples for P043- and major ions were acidified (pH

During earlier investigations at Langton (Harman et al. 1996), the content of adsorbed P on the sediment solids was determined by NaHCO3 leaching using the method of McBride (1994).

Results

PO^sub 4^^sup 3-^ Response to Decommissioning

Figure 1 shows distribution of the major plume solutes (Na+, Ca2+, PO^sub 4^^sup 3-^, NO3-) along the plume centerline at Langton before and after decommissioning. Although the plume also contained high C1concentrations (-80 mg/L, Harman et al. 1996), this constituent was also elevated in some of the background ground water and thus was not used as one of the primary plume tracers. Also note that as a result of modifications to the monitoring network that occurred during 1991 to 1995, the sampling points shown on Figure 1 may not be the same for the in-use and decommissioned periods. However, the sampling detail was sufficient during both sampling episodes to accurately represent the distribution of the plume. One year after decommissioning, the mobile solutes (Na+, Ca2+, N03-) are shown to have migrated a distance of 60 to 100 m downgradient from the tile bed, consistent with the estimated ground water velocity of about 100 m/yr. The P043- plume, in contrast, showed virtually no change in distribution or concentration levels. The remarkably invariant nature of the PO^sub 4^^sup 3-^ concentrations at Langton are more readily observed in Figure 2, which shows the breakthrough of the major solutes (Na+, K+, Ca2+, Cl-, PO^sub 4^^sup 3-^, N03 ) that are substantially elevated in the plume compared to the background ground water. At each of the three primary monitoring locations decommissioning caused little change in PO^sub 4^^sup 3-^ concentrations, whereas all other plume solutes returned to near background values within one year of decommissioning. Table 1 compares the detailed chemical composition at each of the primary monitoring locations at Langton before and after decommissioning. Similar detailed geochemical data is not available at the Long Point 2 site for the decommissioned period.

At the Long Point 2 site, P043- behaves in a similar manner, exhibiting only a minor degree of change after decommissioning (Figure 3). Comparison of P043- profiles below the center of the tile bed before and after decommissioning (Figure 4) further confirms the consistency of P043- values in the shallow water table zone and also provides strong evidence that the frontal (deep) part of the phosphate plume has continued to migrate downward, in an unencumbered manner, after sewage loading has ceased. Because the Long Point tile bed is located close to a water table divide, the horizontal component of ground water flow is less vigorous than at Langton. Nonetheless, lower EC values (~700 vs. ~1200 uS; Figure 4) indicated that by 1997 much of the sewage plume had migrated away from the tile bed area. Lower Cl- concentrations were also noted in this zone in 1997 (7 to 36 mg/L) compared to 1995/1996 (40 to 53 mg/L). The Long Point plume contrasts with the Langton plume in that P043- levels are higher (3 to 5 mg/L P) and also a larger number of monitoring points (20) are positioned in the high PO^sub 4^^sup 3-^ zone located immediately underneath the infiltration bed. P043- levels in these piezometers remained similar before and after decommissioning, averaging 4.4 mg/L P in 1995/1996 compared to 3.8 mg/L P in 1997.

Discussion

The authors are not aware of other studies that have documented P043- behavior in decommissioned sewage plumes. However, laboratory column studies were carried by Walter et al. (1996) in order to predict PO^sub 4^^sup 3-^ behavior in a sewage plume in a sand and gravel aquifer on Cape Cod, Massachusetts, after eventual decommissioning of the sewage source at that site. Tests using uncontaminated sediments from the site generated asymmetric PO^sub 4^^sup 3-^ breakthrough curves that could not be accurately simulated using a diffuse layer surface – complexation model (HYTEQ). In other tests, when contaminated sediments were eluted with uncontaminated, oxygenated water, it was observed for sediments from the anoxic zone that P043- declined to near background levels (

The persistence of high P043- concentrations in the ground water zones at Langton and Long Point, long after the sewage source has been removed, requires that PO^sub 4^^sup 3-^ is being liberated from the sediment solids. A variety of potential release mechanisms are possible, including desorption, intra-particle diffusion, and mineral dissolution. To gain insight into the relative importance of these processes, it is of interest to inventory the subsurface phosphorus mass at these sites and define the modified flow system that is now in place after sewage loading has ceased. Figure 5 presents a conceptual view of the Langton flow system, before and after decommissioning, and identifies the major zones of P mass accumulation. Long-term monitoring at Langton had previously shown that even after many decades of sewage loading, PO^sub 4^^sup 3-^ concentrations arriving at the water table below the infiltration bed remained about 85% lower than the effluent value (~ 9 vs. 1.5 mg/L P; Harman et al. 1996; Robertson et al. 1998). This required that at least 85% of the total sewage P mass was retained in the vadose zone sediments. Sediment analyses revealed that the sub-tile sediments were indeed enriched in P, but that the zone of enrichment was confined to a 30 cm depth interval occurring immediately below the infiltration pipes (Harman et al. 1996; Zanini et al. 1998). Sediment acid extractable P content in the enriched zone (-1000 to 1800 lag/g) was about two to four times higher than the background value (- 450 pg/g), whereas vadose zone sediments lying at greater depth below the infiltration bed had P contents that were similar to background values. Zanini et al. (1998) noted that, at each of four septic system sites investigated – including the Langton site – P solids accumulated in the vadose zone in close proximity (

A second major component of the P inventory at Langton is that which is adsorbed onto the aquifer solids in the ground water zone. Desorption tests carried out on core samples from the shallow water table zone below the tile bed measured adsorbed PO^sub 4^^sup 3-^ -P of 11 mg/kg in this zone (Harman et al. 1996). Considering the average pore water P043- concentration in this zone (- 1.5 mg/L P), a distribution coefficient (Kd) of 7.3 is indicated from the desorption test, where Kd is defined as the mass of P043- adsorbed on the sediment solids / the concentration in solution. Although the previous measurements of adsorbed P043- reported by Harman et al. represented a limited data set (one analysis from the plume core zone below the tile bed), if the resulting Kd value is considered representative of the aquifer, then the approximate P mass distribution at the site is as follows: 85% retained in the vadose zone, the bulk of which appears to be associated with the rapid transformation zone; 13% adsorbed on the aquifer solids; and 2% present in solution. Substantially higher Kd values (25 to 400) were indicated for the Cape Cod aquifer based on the P043- extracted from contaminated zone cores at that site (Walter et al. 1996).

Figure 5 also depicts the pre- and post-decommissioned flow system at Langton and shows the distribution of sewage-derived pore water after one year with no sewage loading. As a result of the approximate 15 times lower loading rate through the tile bed after decommissioning (rainfall recharge of – 0.2 m/yr vs. sewage recharge of ~ 3m/yr), the tile bed “plume” should now be only onefifteenth the thickness it was during active sewage loading, which was about 1.5 m. Therefore, flowpaths that originate from the tile bed area, which are those that can interact with the P solids in the rapid transformation zone, will occupy only a thin zone on the upper fringe of the pre-existing PO^sub 4^^sup 3-^ plume. Flowpaths occupying the remainder of the old plume zone will originate from some upgradient location and will be influenced only by the P mass present in the ground water zone (primarily P043- adsorbed on the aquifer solids; i.e., about 13% of the total P mass). This change in flowpath origin for the proximal monitoring point (B31-2) is depicted in Figure 5. After decommissioning, flow at this point originates from the upgradient flow system; thus, P043- concentrations will be affected only by the P solids present in the aquifer between the monitoring point and the upgradient plume boundary.

It is of interest to compare PO^sub 4^^sup 3-^ behavior at B31-2 to that which would be expected based on liberation of the amount of adsorbed P043- indicated by the laboratory desorption test. Using the retardation equation (Freeze and Cherry 1979):

and assuming sediment bulk density (pb) of 1.7 and porosity (0) of 0.35, the laboratory Kd value leads to a P043- retardation factor (R) of about 40. This is close to the field apparent retardation factor (60) indicated from the P043- plume length in 1991 (70 m), the history of usage (44 years), and the ground water velocity (100 m/yr) (Harman et al. 1996). Considering a retardation factor of 40 and assuming that sorption is rapid and reversible, the PO^sub 4^^sup 3-^ plume should now be migrating downgradient at a rate of about 2 m/yr and declining P043- concentrations should be evident near the upgradient edge of the plume. Although the exact position of the upgradient P043- plume boundary was not established in the earlier 1991 study, Na+ and N03- analyses (Figure 1) indicated that the plume boundary was positioned about 5 to 10 m upgradient of B31 at that time. Considering a P043- migration rate of 2 m/yr, declining P043concentrations should be apparent at B31 about three to five years after decommissioning. Thus, declining P043′ levels should occur within the next few years at this location if the velocity estimate is correct.

The possibility also exists that P043- is being liberated in the ground water zone by the dissolution of other secondary phosphate minerals that may have accumulated downgradient of the rapid transformation zone as a result of their slower precipitation rates or as a result of the geochemical evolution of the plume in the ground water zone. Two such potential minerals are the calcium-phosphate minerals, hydroxyapatite (Ca5 (PO4)3 OH), which consistently exhibits supersaturation in the plume (S.I. = 0.29 to 2.1; Table 1), and 0 tricalcium phosphate (3 Ca3(PO4)2), which approaches or exceeds saturation (S.I. = -0.96 to 0.12; Table 1). However, it seems unlikely that a condition of equilibrium with these minerals would allow PO43′ concentrations to remain so invariant considering that large changes in cation concentrations (i.e., Ca2+) and in pH have occurred upon decommissioning (Table 1 ). It is interesting to note that in the previous study of P solids in the vadose zone (Zanini et al. 1998), Ca-P solids were absent or were a minor constituent even though a condition of hydroxyapatite supersaturation was also observed in that study. For a more thorough assessment of P043geochemistry at these sites, the reader is referred to the previous studies (Zanini et al. 1998; Robertson et al. 1998).

Overall, the field evidence at Langton suggests that Po43behavior in the subsurface is influenced by mineral precipitation reactions in the near-tile rapid transformation zone, but then these reactions diminish in importance farther downgradient. In the ground water zone, PO^sub 4^^sup 3-^ behavior appears to be dominated by sorption reactions that are both readily reversible and rapid with respect to the time frames (years) normally associated with ground water flow systems. The ultimate geochemical stability of the P solids retained in the rapid transformation zone cannot be ascertained from the present data, but will be the focus of future work at the site.

At Long Point, the field evidence suggests that similar processes are controlling PO^sub 4^^sup 3-^ in the ground water zone; however, concentrations are substantially higher (3 to 5 mg/L P), indicating that PO^sub 4^^sup 3-^ immobilization in the rapid transformation zone has been less effective at this site. Only about 30% of the P043-is attenuated in the vadose zone at Long Point (Robertson et al. 1998), which may be a consequence of the slightly higher pH regime at that site (6.9 vs. 6.6) or may be the result of some other factor, such as the seasonal nature of the loading. This suggests that a much larger proportion of the total sewage P043- mass (- 60 to 70%) likely exists as adsorbed material in the ground water zone.

Implications

This evidence indicates that after PO 3- enters the ground water zone at these sites, it is little affected by secondary irreversible adsorption processes or other slow consumption reactions. As a result, most of the PO^sub 4^^sup 3-^ mass present in the ground water zone is expected to be ultimately mobile in the flow system, albeit at a highly retarded migration rate, such that it will be potentially capable of contributing to downgradient P loading. The PO^sub 4^^sup 3-^ retained in the rapid transformation zone, although it can represent the bulk of the P mass present, will likely contribute only a minor amount to the downgradient P flux for many years after decommissioning, because it is confined to a more localized source area. It could be the most longlived P component, however. It is also the component that is most easily remediated since this material could be readily excavated and removed from the site. The adsorbed material in the ground water zone is much more dispersed and consequently would be more difficult to remediate. Thus, when assessing PO^sub 4^^sup 3-^ mobility at septic system sites it is important to establish the amount that is likely to be retained in the rapid transformation zone. An assessment of this zone in a recent review paper (Robertson et al. 1998) showed that its PO^sub 4^^sup 3-^ removal effectiveness varied widely from 23 to 99% depending on site conditions. If site investigations determine that the native sediments present do not have a substantial natural ability to retain P in the rapid transformation zone, then alternative disposal methods may be warranted in some sensitive areas in order to protect downgradient surface water quality.

Acknowledgments

Funding for this study was provided by the Waterloo Centre for Groundwater Research, the Ontario Ministry of Environment and Energy, and the Procter and Gamble Co. Access to the study sites was kindly provided by the Haldimand Norfolk Board of Education and the Ontario Ministry of Natural Resources. Chemical analyses were completed by the Water Quality Laboratory, University of Waterloo, and Philip Analytical Services, Halifax, Nova Scotia.

Copyright Ground Water Publishing Company Mar/Apr 1999

Provided by ProQuest Information and Learning Company. All rights Reserved.