Evaluating Orthophosphate-Silicate Blend as an Alternative to Blended Phosphates for Corrosion Control and Sequestration

Abstract

The presence of iron and manganese in drinking water distribution systems can contribute to discoloration, taste and odor issues, scale buildup, deposition corrosion, and the adsorption and transport of lead. Sequestrants can minimize aesthetic concerns and scale buildup, but they are risky due to increased lead/copper solubility. Here, we used a bench-scale batch reactor to evaluate orthophosphate with sodium silicate or polyphosphate for simultaneous lead corrosion control in waters with and without iron/manganese. Consistent with previous work, ortho-trimetaphosphate increased the total lead (23%). Increased dissolved lead was also observed for both ortho-trimetaphosphate (50%) and ortho-silicate (30%) treatments. When iron/manganese was present, orthophosphate-silicate was associated with 5–12% less total lead relative to orthophosphate under the same conditions, and the effect of ortho-trimetaphosphate was pH dependent. The addition of silicate and trimetaphosphate also reduced water discoloration compared to orthophosphate, as measured by apparent color. Here, the orthophosphate-silicate blend did not significantly increase total lead in the absence of high iron and manganese but increases to the highly mobile, dissolved fraction should be noted. Utilities seeking to control lead for compliance purposes, while simultaneously managing iron and manganese for consumer confidence, should explore orthophosphate-silicate as a possible solution in their corrosion control assessments.

Short abstract

Silicate blended with orthophosphate can mitigate the lead increase and aesthetic concerns caused by iron and manganese.

1. Introduction

The presence of iron and manganese in drinking water distribution systems can contribute to aesthetic, infrastructure, and, in some instances, health concerns. Common aesthetic concerns are discoloration, taste, and odor issues, but scale buildup, deposition corrosion, and heavy metal transport can also be challenges for utilities. Aesthetic or secondary regulatory limits in drinking water range from 0.02–0.05 mg Mn/L and 0.3 mg Fe/L.¹⁻³ More recently, Health Canada has adopted a maximum acceptable concentration of 0.1 mg Mn L^–1 as a health objective.²

Iron and manganese in drinking water distribution systems may come from source water or distribution materials. A recent study of over 5000 water systems in the United States found that over 20% of systems (15% of surface water systems) had distribution system entry point manganese exceeding regulatory limits. Manganese was widespread and naturally occurring in both ground and surface waters, but the highest levels were found in groundwater systems and in the Northeast.⁴ While effective iron corrosion control can minimize the presence of iron in distribution systems,^5,6 manganese removal during treatment is challenging. Manganese removal can be accomplished through physical, chemical, and biological processes, but selecting appropriate treatment processes is dependent on the specific water chemistry and the form of manganese present. This is particularly challenging for utilities drawing from lakes, where manganese speciation can change throughout the year due to chemical and biological shifts in the lake. Similarly, intake depths influence the form and concentration of manganese, which can vary in a stratified lake.⁷

When removal cannot be achieved or is too costly, utilities can control precipitation and deposition of manganese or iron using common sequestrants like polyphosphates to minimize the aesthetic issues or scale buildup. Sequestration can mitigate the formation of particulates that contribute to discolouration by complexing or dispersing the metals. Lytle and Edwards⁸ describe three primary mechanisms of effect using polyphosphate: a polyphosphate–metal complex may inhibit the (1) oxidation of soluble iron/manganese, (2) precipitation of iron/manganese solids, and (3) aggregation of particulate iron/manganese. Further, complexation of these metals can prevent deposition and scale buildup.

Recently, growing attention has been given to the risk of increased lead release with polyphosphate.⁹⁻¹³ Polyphosphates, which form strong complexes with metals, can increase lead solubility, but complexation capacity depends on water chemistry.¹⁴ Corrosion control via polyphosphate treatment is achieved through the product’s reversion to orthophosphate in the distribution system.¹³ Orthophosphate is widely used for controlling lead corrosion by forming a relatively insoluble lead-phosphate scale,^12,15 however, orthophosphate is relatively ineffective at managing aesthetic concerns related to iron and manganese.

Considering the revised Lead and Copper Rule and the risks associated with polyphosphate, here, we explore sodium silicate blended with orthophosphate as an alternative to blended phosphates for simultaneous lead corrosion control and sequestration of iron and manganese. Sodium silicates are infrequently used in North America,¹⁵ but they can effectively sequester iron and manganese.¹⁶ While limited, research suggests that the effect of silicates ranges from negligible to increased lead release and that reductions in lead are primarily a result of the increased pH associated with sodium silicate addition.¹⁰ While proprietary silicate/polyphosphate blends exist for corrosion control and scaling,¹⁷ sodium silicate has not been blended with orthophosphate for the purposes of lead corrosion control and simultaneous sequestration to the best of our knowledge. Here we compared orthophosphate alone, and orthophosphate blended with either sodium silicate or polyphosphate for

• 1)lead corrosion control in waters with and without iron and manganese, and
• 2)sequestration using treated water.

2. Methods

2.1. Test Water

Water for this experiment was sourced from Halifax Water’s J.D. Kline Water Supply Plant (JDKWSP)—a direct filtration drinking water treatment plant and the largest in Atlantic Canada. Manganese removal is seasonally challenging for some of their utilities, resulting in consumer and deposition concerns. Furthermore, cast iron mains and lead service lines are present within the distribution system.¹⁸ The source water has been characterized previously.¹⁹ Clearwell water from the JDKWSP was collected weekly and stored at 4 °C.

2.2. Corrosion Cells

New lead coupons (Canada Metals) (n = 52) were cut (2.7 cm × 1.5 cm x 0.2 cm) and washed in a covered beaker, filled with a 10% nitric acid bath, and placed in a sonicator for 30 min. Coupons were moved in the bath every 5–10 min. Once clear, coupons were rinsed three times with Milli-Q water, dried, and polished with paper towel. The coupons were then suspended in 125 mL glass jars through holes in a screw-top lid and secured on the outside with a silicone sealant. Silicone sealant was not a source of Si contamination in the present study: median concentrations in the no-silicate conditions (median Si = 0.7 mg/L, IQR: 0.55–0.8) were comparable to that of JDKWSP’s unfinished clearwell water (median Si = 0.61 mg/L, IQR: 0.46–0.71).

Test water was prepared weekly, with initial measurements of phosphate, chlorine, and pH taken from the clearwell water. Each condition was then sequentially dosed with orthophosphate, trimetaphosphate, sodium silicate, iron/manganese, and chlorine and pH adjusted, as specified. Prior to each subsequent water change, free chlorine and pH levels were measured and adjusted as needed to maintain the target conditions. Cells were kept in the dark and at room temperature. Each cell was filled to minimize headspace, and water was changed daily, Monday–Friday, providing 24- and 72-h stagnation periods. Samples were drawn weekly with 24-h stagnation periods to monitor lead release over time.

2.3. Study Design

2.3.1. Phase One: Preconditioning

In phase one (Figure 1), all coupons (n = 52) were preconditioned in 0.66 mg/L (as P) orthophosphate, a free chlorine residual of 1 mg/L, pH 7.5, and dissolved inorganic carbon concentration of 5.2 mg/L. The conditioning phase spanned 15 weeks. One pair of coupons was extracted from the study at this time to be used as a baseline assessment of scale prior to the treatment phase. The distribution of total and dissolved lead at the end of Phase 1 for each cell is provided in Figure S3 (Supporting Information). The median dissolved and total Pb concentration were 23.5 ppb (IQR: 10.1–86.5 ppb) and 98.8 ppb (49.1–205.2 ppb), respectively.

Figure 1.

Experimental Conditions and apparatus for (A) phase one: conditioning and (b) phase two: treatment. (C) Photo of a lead coupon after the conditioning phase.

2.3.2. Phase Two: Treatment

Corrosion cells (n = 50) were randomly paired and assigned treatment conditions. The treatment phase evaluated orthophosphate blended with either trimetaphosphate or sodium silicate against orthophosphate using only filtered water from the JDKWSP at high and low pH and iron and manganese concentrations (Table 1). To achieve this, all cells were dosed with zinc-orthophosphate at 0.33 mg/L as P. The sequestrant doses are listed in Table 1.

Table 1. Experimental Conditions for 2^4 Factorial Design^a.

factor	level
	high	center	low
Mn & Fe concentrations	1 mg/L (each)	0.5 mg/L	0 mg/L
pH	9.5	8.5	7.5
sodium silicate	20 mg/L	10 mg/L	0 mg/L
trimetaphosphate	3 mg/L as P	1.5 mg/L	0 mg/L

^aEach test condition had an orthophosphate dose of 0.33 mg/L (as P) and a free chlorine residual of 1.0 mg/L

This experiment employed a 2⁴ factorial design with center points (Table S1).²⁰ pH, manganese and iron concentrations, silicate dose, and polyphosphate dose were the factors, and each condition was duplicated. The design generated 52 corrosion cells: 2⁴ runs +2 × 4 axial runs +1 center point +1 point for analysis following phase 1, duplicated [(16 + 8 + 1 + 1) × 2]. Only 50 corrosion cells advanced to the treatment phase; one pair was removed so that scale analysis could be conducted on coupons as a baseline assessment prior to treatment.

Test water was changed 5 times per week, resulting in approximately 200 total water changes (Figure 2). Total metals were monitored weekly and dissolved biweekly. The color (true and apparent) and size data were collected near the end of the experiment to best capture the effect of each condition.

Figure 2.

Time series of total lead release by cell (n = 50) as a function of water changes, inhibitor:sequestrant treatment, pH, and iron/manganese dose. Points represent measured total lead; bold lines represent the smooth term.

2.4. Reagents

Solutions were prepared with ultrapure water. The test water pH was adjusted using NaOH. Stock solutions of manganese and iron were prepared by using MnSO₄ and FeSO₄, respectively. Chlorine was dosed using 5–6% NaOCl. Trimetaphosphate was prepared using Na(PO₃)₃ (Alfa Aesar). Zinc orthophosphate and sodium silicate (SiO₂:Na₂O = 3.22) were sourced commercially. Trace metal grade nitric acid (HNO₃) was used for the acid digestion.

2.5. Element Detection

Samples were acidified and inductively coupled to plasma-mass spectrometry (ICP-MS, Thermo Scientific) was used to quantify metal concentrations according to Standard Methods 3125 and 3030.^21,22 Following a 24-h stagnation period, corrosion cells were inverted three times to resuspend settled colloidal/particulate metals, and 10 mL aliquots were drawn. The first 10 mL were decanted directly into a polypropylene tube for total metals quantification. Another 10 mL aliquot was passed through a 0.2-μm cellulose nitrate membrane filter with a syringe filter cartridge to waste to minimize absorptive losses before passing another 10 mL through to a polypropylene tube for a filtered sample. Both aliquots were acidified with HNO₃ to achieve a pH less than 2.0 and stored for a minimum of 24-h before being analyzed with ICP-MS.

2.6. X-ray Diffraction

After the experiment was complete, coupons were left to dry before being mounted on a piece of clay in a hollow XRD mount. A Proto AXRD Benchtop with a copper Kα radiation source was used to identify the crystalline phases.

2.7. Statistical Analysis

Experimental data were analyzed with R (version 2023.09.1) and several contributed packages (mgcv²³). Lead concentrations (total and dissolved) were modeled using a generalized additive mixed model (GAMM), which is suited for analyzing serially correlated data. Lead concentrations were log-transformed and modeled as a function of silicate and trimetaphosphate dose, pH level, and iron and manganese concentration. The temporal effect was accounted for with a thin plate regression spline, and the correlation between sample measurements was controlled using a continuous time first-order autoregressive process.²³ Models and figures for the total and dissolved lead data sets were constructed from 1501 and 688 observations, respectively. Full details of the factorial design and the model are provided in the Supporting Information. Total and dissolved lead concentrations over time for each condition are provided in the Supporting Information (Figures S3 and S4).

The log-transformation of the dependent variable makes the effect estimates multiplicative. That is, the effect of adding trimetaphosphate and iron/manganese is the product of the effects of the trimetaphosphate, iron/manganese, and trimetaphosphate × iron/manganese terms. Products greater than or less than one represent effects that increase or decrease lead release, respectively. We compared lead release for the inhibitor/sequestrant blend systems at high and low iron/manganese systems via a ratio. The effects of pH, silicate, and trimetaphosphate on color and turbidity were evaluated using a linear model with two-way interactions. Spearman’s rank correlation was used to assess relationships among turbidity, color, particle size, and particulate lead.

Data and R code necessary to reproduce the analysis are available on GitHub at https://github.com/kalli-hood/ortho-polyphosphate-silcates-sequestration-CCT/tree/main/blendedsequestrants-main.R and a set of model diagnostics are included in the Supporting Information.

3. Results and Discussion

3.1. Polyphosphates Increase Total Lead

Adding trimetaphosphate increased total lead by 22% (95%CI: 11–35%). The sodium silicate-orthophosphate condition was linked to a nonsignificant increase (8%, 95%CI: −2–19%) in total lead (Figure 3). The effect on the dissolved fraction was more pronounced: dissolved lead increased by 54% (95%CI: 25–91%), and 30% (95%CI: 5–61%) for trimetaphosphate and sodium silicate, respectively, likely due to dispersion of small colloids through charge repulsion and, in the case of trimetaphosphate, possibly through complexation.^24,25 Previous work²⁶ showed that polyphosphates can inhibit the formation of hydroxypyromorphite, particularly at high ratios of poly- to ortho-phosphate like that used in the present study. Silicates are not thought to control lead release through the formation of insoluble minerals. One study²⁷ has suggested that silicate can mitigate lead release through the uptake of silicate in an amorphous layer rich in aluminum and lead.

Figure 3.

(A) Percent change in geometric mean lead release relative to the control condition based on the GAMM coefficients. Total and dissolved lead fractions were modeled separately; total lead is depicted with triangles, and dissolved with circles. Negative values indicate that condition was associated with a reduction in lead, whereas positive values indicate an associated increase. (B) Modeled total lead release and (C) Modeled dissolved lead release across a continuous range of inhibitor/blend dose.

Increasing pH to 9.5 decreased total lead by 24%, and dissolved lead by 34% (95%CI: −15 – −48%). This suggests that lead–carbonates were at least partially responsible for the lead solubility control in the system. Based on the DIC and pH targets of the present water conditions, the decrease in total lead associated with increasing pH is consistent with lead carbonate solubility models.¹² However, XRD spectra show that cerrussite (PbCO₃), hydrocerussite (PbCO₃(OH)₂), and hydroxypyromorphite (Pb₅(PO₄)₃OH) were present (Figure S6). If hydroxypyromorphite were the solubility-controlling phase, one would expect dissolved lead to be lower in pH 7.5 and 8.5 conditions relative to pH 9.5.¹²

At pH 9.5, trimetaphosphate was associated with a reduction in total lead release, but this effect appeared to be dominated by the effect of pH: the 21% reduction is comparable to that of higher pH alone (24%). Previously, Edwards and MacNeill¹⁴ observed that the effect of polyphosphate on lead release was less pronounced at pH 9.5. Another study found less adsorption of polyphosphates to lead carbonate surface at pH 9 versus 7, perhaps due to electrostatic repulsion as lead carbonate becomes more negative at higher pH.²⁶ It may be that trimetaphosphate is less effective at dispersing colloids at high pH owing to their more negative charge.

3.2. Iron and Manganese Increase Particulate/Colloidal Lead

Adding iron and manganese to the system appeared to influence the dissolved and total lead concentrations differently. Cells with high iron and manganese had 41% more total lead than those without, but their presence reduced the dissolved fraction of lead by 44% (95%CI: −31 – −55%). The observed increase in total lead is expected and consistent with previous work.²⁸⁻³⁰ With respect to the dissolved fraction, the associated decrease in lead when iron and manganese were present may be explained by the adsorption of dissolved lead species to iron- and manganese-rich colloids that were too large to pass through the filter.

3.3. Interactions between Orthophosphate-Silicate for Lead Control and Iron/Manganese Mitigation

3.3.1. Total Lead

The orthophosphate-silicate treatment resulted in less total lead compared to the orthophosphate-only treatment when iron and manganese were present (Figure 4, Tables S2 and S4). That is, the addition of silicate at both the high/low pH levels was associated with lower lead release than in the matched conditions with orthophosphate only (Table S2).

Figure 4.

Estimated concentration of lead as predicted by models of total (n = 1501) and dissolved (n = 688) lead. Estimates were calculated by taking the product of the control estimate and the effect estimate for each term. The ratio for interaction terms were calculated as the product of all the relevant factors (more details in the SI).

The effect of ortho-trimetaphosphate treatment on lead release in high iron/manganese water differed by pH: at low pH, it was associated with higher lead release than the matched control, whereas at pH 9.5, it was associated with lower lead release. The trimetaphosphate × pH interaction has been discussed previously. In summary, both sodium silicate and trimetaphosphate had negative interactions with iron/manganese: individually, each of these factors increases lead release, but when silicate or trimetaphosphate co-occurred with iron/manganese, less lead was released than would be expected (Figure 5).

Figure 5.

Visual depiction of the interaction effects between the sequestrant/inhibitor blend and iron and manganese for high/low pH on total lead release. Example description for silicate × Mn/Fe at pH 7.5: in the reference condition (orthophosphate), the mean lead was 47 ppb (blue point). The main effect of adding Mn/Fe to this condition increased, leading to 67 ppb (blue point). The main effect of silicate to the reference condition increased to 51 ppb (red point). If there were no interaction effect, we would expect the effect of adding silicate and Mn/Fe to increase lead along the dashed line (to ∼71 ppb); however, we see that when both are added, the concentration of lead increased only to 63, indicating a negative interaction.

3.3.2. Dissolved Lead

Dissolved lead was generally lower in the spiked iron/manganese cells relative to the unspiked cells, as previously discussed. However, the presence of sodium silicate or trimetaphosphate increased the dissolved fraction (Figure S2, Table S5). Specifically, the orthophosphate-silicate treatment was associated with higher dissolved lead release than matched controls at both pH 7.5 and 9.5. The effect of ortho-trimetaphosphate was split: at pH 7.5, dissolved lead was higher than the matched control, whereas at pH 9.5 it was lower.

The presence of silicate or trimetaphosphate in iron-/manganese-rich matrices may reduce the concentration of total lead by reducing the number of potential adsorption sites for lead in two ways. First, the presence of these sequestrants may inhibit the precipitation of these metals in solution, thereby maintaining their soluble form to which lead cannot adsorb.³¹ A second explanation is that adsorption sites on particulate iron and manganese are complexed with the sequestrant and can less readily adsorb lead. Silicate treatment has been shown to inhibit DOM-iron interaction.³² When high levels of iron and manganese are present, the sequestrants may be interacting with the suspended iron and manganese rather than dispersing lead, but further research is warranted. Previously, polyphosphates and sodium silicate have been observed to interact with iron/manganese, promoting a highly mobile transport vector for lead.^11,25

3.4. Sequestrants Improve Discoloration

As expected, conditions spiked with iron and manganese had higher apparent and true color (Figure 6). In the linear model for apparent color, the effect estimate of spiking iron/manganese was an order of magnitude greater than all other factors and associated with an increase of 14 Pt–Co units. Increasing pH from 7.5 to 9.5 was also associated with an increase in color of approximately 5 Pt–Co units. Higher pH supports the precipitation of larger iron/manganese particulates that would have been retained on the filter prior to true color analysis. In the true color model, spiked iron/manganese was linked with an increase of 1.3 Pt–Co.

Figure 6.

(A) True and apparent color and (B) turbidity by inhibitor/sequestrant treatment and pH level. Color, which was strongly influenced by iron/manganese, is depicted at the high levels only. The dashed red line indicates the Health Canada aesthetic objective for color in drinking water, 15 Pt–Co units. Turbidity, true and apparent color were each measured twice in all cells (n = 50) at the end of the study.

Both trimetaphosphate and sodium silicate addition statistically significantly improved apparent color. When iron and manganese are present, the orthophosphate-silicate treatment was associated with a 3.6 Pt–Co reduction in color, and the ortho-trimetaphosphate treatment was linked with a 1.7 Pt–Co reduction compared to orthophosphate treatment alone.

3.4.1. Turbidity

The effects of silicate, trimetaphosphate, pH, and iron/manganese, and their two-way interactions, on turbidity were evaluated with a linear model. Like color, iron/manganese and pH levels were the strongest predictors of turbidity, with higher levels associated with statistically significant increases in turbidity (Table S8, SI). Across all conditions, silicate addition was associated with a small but statistically significant decrease (−0.1 NTU, p = 0.015) in turbidity. The two-way interaction between silicate x iron/manganese (−0.1, p = 0.003) provides better insight: silicate reduces the turbidity caused by precipitated metals (Figure 6b). No significant change was detected with trimetaphosphate addition.

3.4.2. Particle Size, Color, Turbidity, and Lead Release

Turbidity and apparent color were strongly correlated (ρ = 0.86). The linear models suggested that high levels of iron/manganese were the main contributors of color and turbidity and that this effect was stronger at higher pH, due to increased metal precipitation. Particle size had a moderate correlation with turbidity (ρ = −0.42) and color (ρ = −0.47), indicating that smaller particles contribute more to these parameters. However, particulate lead showed little correlation with turbidity (ρ = −0.01) and color (ρ = −0.004), indicating these relationships are not monotonic and that factors beyond the presence of particles are influencing particulate lead release.

4. Conclusions

Both polyphosphate and sodium silicate were inferior to orthophosphate for total and dissolved lead control as individual chemical agents. However, when high levels of iron/manganese are present, these sequestrants can be used alongside orthophosphate to address discoloration concerns. Our findings suggest that for systems with elevated iron/manganese, silicate blended with orthophosphate can mitigate the lead increase caused by iron and manganese. Dissolved lead increased under treatment with silicates blended with orthophosphate, which would necessitate further exploration for specific circumstances. This work suggests that sodium silicates are superior to trimetaphosphate for addressing trade-offs between sequestration and lead release control in comparable water chemistries. While the present study evaluated these formulations in relatively soft waters (typical hardness <20 mg/L as CaCO₃), other recent work has shown that hard water can hinder the sequestration capacity of polyphosphate.⁸ Further research is warranted to optimize dosing for both formulations, particularly with trimetaphosphate dosing, which was high in this work. Utilities seeking to control lead for compliance purposes, while simultaneously managing iron and manganese for consumer confidence, should explore ortho-silicate blends in their corrosion control assessments to understand how these formulations interact in waters with different pHs, alkalinities, and hardness. Nevertheless, there is potential for sequestrant/orthophosphate blends to comanage scaling and corrosion issues related to iron, manganese, and lead in distribution systems.

Data Availability Statement

The data and code needed to reproduce this analysis available on GitHub, https://github.com/kalli-hood/ortho-polyphosphate-silcates-sequestration-CCT.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsestwater.5c00159.

• Additional details on the experimental design, statistical modeling, and supporting data for the study; description of the factorial design, coding of variables, and the log-transformation approach used to model lead release; generalized additive mixed model (GAMM) results present coefficients and estimates for total and dissolved lead release, highlighting statistical significance and interaction effects; lead release patterns over time are visualized, including the effects of inhibitors and environmental conditions; additional supporting data include orthophosphate stability, XRD spectra, and zeta potential measurements (PDF)

Author Contributions

K.M.H.: conceptualization, data curation, formal analysis, investigation, visualization, writing—original draft. B.F.T.: validation, supervision, writing—review and editing. G.A.G.: supervision, funding acquisition, writing—review and editing. CRediT: Kalli M Hood conceptualization, data curation, formal analysis, investigation, visualization, writing - original draft; Benjamin F Trueman supervision, validation, writing - review & editing; Graham A. Gagnon funding acquisition, supervision, writing - review & editing.

The authors declare no competing financial interest.