CORROSION ENGINEERING

Transcription

1 CORROSION ENGINEERING Control of Lead Corrosion by Chemical Treatment* B.P. Boffardi and AM. Sherbondy* ABSTRACT Laboratory corrosion studies with lead pipe in the presence of orthophosphate or polyphosphate show that orthophosphate ions are most effective in control/ing lead leaching. XPS analysis indicates that the mechanism differs for the two phosphate species; orthophosphate interacts directly with the lead surface whereas polyphosphates require calcium ions to develop a self-limiting protective barrier. Pilot plant testing using a pipe loop fabricated from lead, soft so/der, copper, and a brass faucet simu/ating household water usage with typical on-off cycles established equilibrium corrosion rates for the above materials over live to eight months in two waters. Both high-hardness high-alkalinity and low-hardness low-alkalinity waters in the Pittsburgh, PA, area were treated with zinc-orthophosphate. Both lead and copper materials meet the U. S. Environmental Protection Agency (EPA) proposed maximum contamination levels (MCLs) of.15 mg Pb/L and 1.3 mg Cu/L. KEY WORDS: lead corrosion, orthophosphate, potable water INTRODUCTION The Safe Drinking Water Act (SDWA) of 1974 required the U.S. Environmental Protection Agency (EPA) to set drinking water standards to protect the public health. In 1986, major amendments to the law were passed, one of which banned the use of lead Submitted for publication July Presented as paper no. 445 at CORROSION/91 in Cincinnati, OH, Calgon Corp., P.O. Box 1346, Pittsburgh, PA materials in new plumbing and in plumbing repair and required water utilities to notify the public about lead in drinking water. In August 1988, the EPA proposed new regulations limiting lead concentration in drinking water to.5 mg/l at the source and.1 mg/l at the consumer's tap with 95 percent of the samples below.2 mg/l.' Much discussion directed at reducing the proposed lead levels has occurred since the issuance of the proposed regulation. Discussions have centered on where and how to collect samples, the level of the lead contaminant, treatment options or techniques, and who is responsible for controlling corrosion by-products in household plumbing systems. These very complex issues are controversial and difficult to resolve. The EPA finalized the regulations on June 7, 1991, with a defined maximum contamination level of.15 mg Pb/L for lead and 1.3 mg Cu/L for copper and corrosion-control treatments required to achieve these levels. There are several treatment strategies that can be used to control lead corrosion that encompass both nonchemical and chemical treatments. 2 The essence of nonchemical treatment options consists of: Using lead-free solders to join copper connections, (95:5 tin:antimony; 96:4 tin:silver; 97:3 tin:copper solders). Replacement of lead service lines or lead plumbing in both the distribution and home plumbing system. Flushing the lines for a period of time to eliminate "first-draw" water prior to human consumption. Chemical treatment options are under the discretion of the water utility. These options include ph/ /91 /255/$3./ , National Association of Corrosion Engineers CORROSION DECEMBER 1991

2 alkalinity adjustments or the addition of chemical corrosion inhibitors. ph/alkalinity control is effective in reducing lead solubility when the ph of the finished water is above eight. However, at this ph, the effectiveness of chiorine disinfection is reduced while the formation of trihalomethanes increases. High alkaline waters will form basic lead carbonate, Pb3(OH) 2(CO3) 2. lts solubility is relatively insensitive to ph over the range 6.5 to 8, while maintaining a lead level in excess of.2 mg/l. Chemical inhibitors provide the most effective means to reduce lead solubility and control corrosion in the overall distribution system. There are essentially three basic building blocks used to inhibit corrosion in potable water systems: orthophosphate, polyphosphate, and silicates; all with or without zinc. The use of phosphorous-based inhibitors to minimize lead solubility was investigated and found to be effective. PRELIMINARY EVALUATIONS The effects of orthophosphate and polyphosphate ions on lead solubility were determined under accelerated conditions to rapidly ascertain which of the two chemical additives was most effective in controlling lead corrosion. Experiments were performed in an experimental soft water, the composition of which is shown in Table 1. Lead pipe, obtained from Pennsylvania American Water Company, was sectioned into 8-in. lengths and cut longitudinally. These sections were placed in a 5 g/l ammonium acetate solution and boiled for five minutes. This cleaning procedure removed all deposits adhering to the lead surface. Cut edges and outside surfaces of the lead pipe were coated with a water impervious epoxy paint. Thus, only the internat surface was exposed to the experimental solution. Thirty-seven square inches of exposed lead surface area was placed in each of three separate 1 -L water baths controlled at 6 C and ph statted at 7.3. The solutions were air sparged. Two chemical inhibitor treatments were evaluated besides an untreated control: 1 mg/l zinc with 5 mg/l orthophosphate ions and 1 mg/l zinc with 5 mg/l polyphosphate (1.1:1 Na2:P25). These treatment levels are extremely high for use in potable water systems but will rapidly indicate efficacy within a short time. The temperature controller contained a built-in circulator pump that circulated the fluid contents of the bath. Every seven days during a 21-day period, test samples were collected and analyzed for soluble lead. The samples were filtered through a 2.5-g filter, acidified, and analyzed using atomic absorption graphite furnace techniques. The results of this accelerated study are shown in Figures 1-3. Lead solubility increased with time in the TABLE 1 Experimental Soft Water Species Concentration, mg/l Calcium 8 Magnesium 2.7 Total hardness as CaCO 3 31 Chloride 8 Sulfate 19 Bicarbonate 25 Alkalinity as CaCO ph 7.3 LEAD mg Pb/L FIGURE 1. Lead solubility, soft water, ph 7. 3, 6'C, control. LEAD mg Pb/L.25 r FIGURE 2. Lead solubility, soft water, ph 7.3, 6 C, 1 mg/l zinc-5 mg/l orthophosphate. LEAD mg Pb/L FIGURE 3. Lead solubility, soft water, ph 7.3, 6 C, 1 mg/l zinc-5 mg/l polyphosphate. CORROSION Vol. 47, No

3 control bath, as expected, since the solution concentration was cumulative. After three weeks, lead solubility was 3 mg/l. Lead surface deposits consisted of a mixture of lead carbonate PbCO3 and basic lead carbonate Pb3(OH) 2(CO 2 3). The effect of high concentrations of zinc-orthophosphate is shown in Figure 2. This treatment scheme rapidly controlled lead solubility. After seven days, lead solubility was.22 mg/l compared to the control of 1.25 mg/l. At the end of the third week, lead levels were at the limit of detection, less than.5 mg/l. Deposit on the lead pipe was amorphous. The zinc-polyphosphate treatment was also effective in reducing lead solubility but not as effective as orthophosphate. The first week's results were 1.65 mg/l of lead decreasing to.23 mg/l after three weeks. The lead surface deposit was amorphous. Data for the inhibitor treatments and control are shown in Table 2. SURFACE FILM A second series of experiments were performed to determine the chemical composition of the film formed on the lead. These experiments consisted of placing five 1/2-in. by 3-in. lead coupons in 42 L of experimental soft water controlled at ph 7.3 and 25 C. The large volume of water required a significantly reduced level of chemical treatment normally maintained in distribution systems at.5 mg/l zinc - 1. mg/l PO4 and.5 mg/l zinc mg/l polyphosphate. The lead coupons were removed from the two baths after 1, 4, 7, 11, and 13 days of immersion. The 1 coupons were subjected to x-ray photoelectron spectroscopy (XPS) surface analysis, providing chemical identification on the outermost layers of the surface up to 5 Á in depth. Table 3 shows the chemical composition of the lead surface as a function of time in the presence of the zinc-orthophosphate treatment. Note that the zinc constituent on the surface vanishes after four days, while the phosphorous content increases, decreases, and increases, suggesting a building and repair of a phosphorous film. The calcium content is essentially negligible indicating calcium ions do not participate in forming the protective film. Finally, the lead concentrations vary inversely to phosphorous. As the protective film develops, the concentration of lead decreases and is more toward the inner surface, reducing the concentration of available lead at the solution/metal interface. Zinc-orthophosphate exists in solution as distinct entities, and therefore, only the phosphate ion interacts with the lead surface forming the protective film. The divalent metal ions, zinc and calcium, do not participate in the protective film on lead (although these divalent metal ions do participate in forming a protective film on steel). The surface film developed on lead by zinc-polyphosphate is characterized by XPS analysis (Table 4). Note that zinc and calcium ions are now incorporated into the film with this treatment compared to their absence when zinc-orthophosphate is used. The zinc level increases during the first few days of immersion and seems to plateau after seven days. The phosphorous concentration seems to remain constant, as does the calcium concentration. The lead levels decrease as expected due to the build-up of the film away from the lead surface. These results can be explained by visualizing the polyphosphate molecule in solution as a negatively charged linear chain species (Figure 4). The zinc and calcium ions in the water interact with the polymer to form a multimetal ion complex. This species attaches to the lead surface and builds upon itself by bridging to the adjacent inhibitor metal complex such that the next inhibitor complex forms on top. Thus, a self-limiting film TABLE 2 Lead Solubility in Experimental Soft Water ph 7.3, 6 C Lead Solubility mg/l Week 1 Week 2 Week 3 Control mg/l zinc 5 mg/l orthophosphate.22.9 <.5 1 mg/l zinc 5 mg/l polyphosphate TABLE 3 XPS Surface Analysis on Lead for Zinc-Orthophosphate Treatment Atomic Concentration Days Zinc Phosphorus Calcium Lead Days TABLE 4 XPS Surface Analysis on Lead for Zinc-Polyphosphate Treatment Zinc Atomic Concentration Phosphorus Calcium Lead , CORROSION-DECEMBER 1991

4 o^!^o aro!^ FIGURE 4. Representation of a stretched-outpolyphosphate chain. develops over time, the lead surface being excluded from the aqueous phase by the inhibitor barrier. This film inhibits the migration of lead ions into water systems. LEAD TEST ASSEMBLY A field test assembly was built to evaluate the effectiveness of a 3:1 orthophosphate: zinc treatment under normal operating conditions in a distribution system. Figure 5 is a schematic of the test assembly. The assembly consisted of four flow paths. The first three paths were made of 3/4-in. PVC piping, and each contained an insert of a given pipe material. The first path held two sections of lead pipe, a 6-in, and an 18-in. length. The lead pipe was virgin material with the dimensions of 1-in. OD, 5/8-in. ID. This leg simulated a lead service connection from a water service main. The second path held two 6-in. sections of 1/2-in. copper water pipe internally coated with 5:5 tin:iead solder. The coated copper was used to simulate the behavior of solder at a jointed copper connection in the home. The third leg was identical to leg two without the tin:lead coating. Both legs two and three were made from virgin copper. In addition to the three flow paths, a fourth path consisting of a 5-ft coil of 1/2-in. copper tubing with five soldered connections was placed at 1 -ft intervals. Also included in the assembly was a brass utility faucet. The pipe inserts were held in the PVC line such that no turbulence occurred at the junctions. This connection system, shown in Figure 6, prevented the occurrence of any localized attack at the edges of the specimen inserts. Time solenoid valves were placed at the end of each flow path to assure that water flowed through only one path at a time. The timing cycle was 2-3 min in duration. No solenoid valve was connected to the fau- FIGURE 5. Lead test assembly. THREADED am PIPE SAMPLE P1PE FLUSH WRH HOLDER " RING f---^ POLWINYL CHLORIDE FIGURE 6. Connection system developed to pre vent crevice corrosion. cet. A Dole valve having 1.5 gpm flow control was connected after the solenoid valve to regulate flow rate through each leg. A Dole valve was also connected to the spout of the utility faucet. The timing cycle through the assembly simulated household water usage consisting of 16 h of flow followed by 8 h of stagnation. Total flow through the assembly was 1,44 gal per 16 h. Flow through the faucet occurred for 8 h per day. Sample taps were placed on each flow segment just prior to the solenoid valves so that water samples could be collected after the stagnation period and prior to reactivations of the timing cycle. The one-l samples were collected from each of the first three paths after an 8 h stagnation period, whereas, two one-l samples were collected from the 5-ft copper coil. A 125-mL sample was collected from the utility faucet. Both first and second draw samples were analyzed for lead and/or copper depending on the metallurgy of the specimen. A 2-3 min flush occurred between first and second draw samples. This water flow between CORROSION Vol. 47, No

5 TABLE 5 Comparison of Graphite Furnace AA and "Leadtrak" for Soluble Lead Sample Furnace Atomic Adsorption, Leadtrak Leadtrak Number µg/l µg/l 9.1 µg/l Spike, µg/l < < < < < <2 9.2 sample collection resulted in very low values for lead and/or copper in the second draw and were at the limit of detection. Thus, only first draw samples were analyzed throughout the test. Solution analysis for lead was performed using both atomic absorption, graphite furnace, and the Hach "Leadtrak" procedure. The Hach procedure was used to determine intermediate data values whereas the atomic absorption procedure was used to verify results. Comparative data developed by Hach is shown in Table 5. Copper levels were determined by the atomic absorption technique. RESULTS AND DISCUSSION Two different pumping stations, which are part of Pennsylvania American Water Co., were sites for evaluating a 3:1 orthophosphate:zinc chemical treatment formulation. As previously discussed, the orthophosphate treatment was selected as the product of choice since it was shown to be more effective than polyphosphate. The two sites were the Oneida pumping station in Butler, Pennsylvania, and the Canonsburg station in Canonsburg, Pennsylvania. Surface water is the raw water source at both sites. The pumping stations are approximately 7 miles apart. Water compositions are significantly different, as shown in Tables 6 and 7. Three test assemblies were placed at each test site. At the Butler site, two assemblies were connected to the filtered effluent and the remaining assembly to the finished effluent. Chemicai treatment of 1 mg/l PO4 and.3 mg/l zinc was added to one of the filtered effluent systems. The other filtered effluent system was a control. Note that the filtered effluent was at a ph of 7.2. The third assembly connected to the finished water was at ph 8.5. The Butler site adds lime to the filtered effluent increasing the hardness and ph (Table 6). The impact of the above treatment on lead pipe is dramatically shown in Figure 7. The lead levels decrease rapidly within the first 2 days and continue a gradual decline to a.1 mg/l range over 16 days of operation. This phenomenon also occurs with the ph 8.5 system. Thus, chemical treatment to control lead with 1 mg/l PO4 and.3 mg/l zinc at is equivalent to control at high ph. Note that at, in the absence of chemical treatment, lead remains high, in the vicinity of.4 mg/l. Although the curves in Figure 7 appear to be smooth and continuous, there are pertur- Species TABLE 6 Butler, Pennsylvania Filtered Finished Effluent mg/l Effluent mg/l Calcium 34 Magnesium 5.9 Total hardness as CaCO Manganese.6.5 Alkalinity as CaCO, Chloride 22 1 Sulfate TABLE 7 Canonsburg, Pennsylvania Species Finished Effluent mg/l Calcium 7 Magnesium 13 Total hardness as CaCO 3 23 Manganese <.1 Alkalinity as CaCO3 135 Chloride 6 Sulfate CORROSION-DECEMBER 1991

6 bations above and below the curves. These higher lead levels or spikes may be caused by minor erosion of the lead film. Only leg 1 had spiked lead levels. The lead analysis for the remaining paths had minimal variances, indicating much tighter surface film. The experimental data are shown in Table 8. The effect of chemical treatment on tin:lead solder is shown by Figure 8. Within five days, lead dissolution from the solder is controlled to.1 mg/l. Filtered effluent, with or without chemical treatment and high ph are equivalent. The influence of galvanic attack to dissolve lead from the solder joints in the 5-ft coil of copper tubing is shown by Figure 9. Both chemical treatment and high ph controlled galvanic attack, suppressing lead dissolution to.1 mg/l, whereas filtered effluent at produced a lead concentration of.2 mg/l. Galvanic corrosion is a concern because it accelerates attack on the solder by copper. Since the lead solder is anodic to copper and the surface area of the solder is extremely small compared to the larger cathodic copper surface, accelerated galvanic attack predominates. The attack is suppressed by chemical treatment or high ph. The lead concentration from the first 125 ml from the brass utility faucet is shown in Figure 1. Chemical addition reduced the lead to.1 mg/l after 16 days, while high ph achieved this value after 1 days. The untreated filtered effluent never achieved a low lead concentration. In fact, after 16 days, the lead concentration was.14 mg/l. Copper levels from the 5 ft of copper tubing are shown by Figure 11. Over the 16 days of operation, the copper concentration decreases to less than 1 mg/l. The negative slope with 1 mg/l PO mg/l zinc treatment is more gradual than the untreated filtered effluent at. The finished water remained essentially constant at.8 mg/l. The EPA has promulgated a level of 1.3 mg/l3, which was attained at the Butler test site. It FILTERED EFFLUENT 3 FILTERED EFFLUENT 35 : 2 FATERED EFFLUEKT Gi a1. M91L PO,-.3 mg/l ^nc 4 J RNTSHéD WATER ph 8.5 FILTERED EFFLUENT 1. mg/. PO. -.3 mg/l zinc 1 ph FIGURE 7. Butler, leg 1, 2 ft of lead pipe FIGURE 8. Butler, leg 2, 1 ft of 5:5 tin:lead solder. Date TABLE 8 Lead Concentration in 1 Liter First-Draw Sample From Leg 1 of Butler, Pennsylvania, Test Assemblies Elapsed Time (days) Control Filtered Effluent 1. mg/l PO 4.3 mg/l Zinc Finished Water ph 8.5 3/6/ /9/ /17/ /23/ /4/ /11/ /17/ /24/ /1/ /8/ /16/ /23/ /7/ /2/ /11/ /26/ /4/ CORROSION-Vol. 47, No

7 , L1 i 1 1 DAMS FIGURE 9. Butler, leg 4, 5 ft of copper tubing with five solder joints. J FIGURE 1. Butler faucet, 125-mL sample RLTERED EFFLUENT ph F1LTEHED EFFLUENT E 2. w 1.5 O V 1. FTIIS* D WATER ph a E 1.5 ȧa O 7, FlLTERED EFFLUENT 1. mgll P.-..3 mga. zi.5 FlLTERED EFFLUENT 1. mg/l. PO -.3 mo/l FMLSHED WATER pm FIGURE 11. Butler, leg 4,5 ft of copper tubing with five solder joints. FIGURE 12. Butler faucet, 125-mL sample. should again be stated that the copper tubing was virgin material and will have higher copper release rates compared to seasoned tubing. The copper levels from the utility faucet are shown in Figure 12. The pronounced hump in the data is the result of initial copper corrosion to form the protective copper oxide film: 2Cu* + H 2O -+ Cu 2 + 2H+. The copper release rate is higher at the lower phof 7.2 than ph 8.5. The chemical treatment reduces copper corrosion at. After 16 days, the copper concentration levels off at.5 mg/l. The Canonsburg site also had three test assemblies, all using finished effluent. The assemblies consisted of a control (no treatment), 1 mg/l PO mg/l zinc chemical treatment, and sodium hydroxide addition to raise the ph of the finished effluent to 8.5. Note that the water quality at Canonsburg has very high hardness (23 mg/l as CaCO3) and high alkalinity, whereas finished effluent from the Butler site was moderately hard (11 mg/l as CaCO3) with low alkalinity. A calcite saturation index as a function of ph and calcium hardness is shown in Figure 13. The index is defined as the ratio of the ion activity product for calcium and carbonate divided by the thermodynamic UNDERSATURATED CALCIUM HARDNESS FIGURE 13. Calcite saturation index, 25 C. equilibrium constant: CSI = ACa. ACO 3 KCaCO 3 The ionic concentrations are corrected for temperature and ionic strength. When the ratio is less than one, then the system is undersaturated. When the ratio equals one, it is saturated and at equilibrium; and when the ratio is greater than one, precipitation of CaCO3 can occur. The solid line in Figure 13 represents a ratio equal to one. Below and above the line are undersaturation and precipitation regions. 972 CORROSION-DECEMBER 1991

8 The finished effluent at Canonsburg is essentially at saturation. Thus, increasing the ph of this water to 8.5 should cause calcium carbonate to precipitate. In fact, within a few weeks, the test assembly with ph 8.5 control became inoperative because of calcium carbonate precipitation preventing water flow through the system. It should be mentioned that one of the treatment recommendations of the EPA is to adjust the ph of the finished effluent above eight as a means of controlling lead solubility. Although we know high ph is totally adequate in some waters, for example, the Butler site, it is not universally applicable. Thus, ph adjustment must be viewed in light of the hardness and alkalinity of the water. Lead concentrations from the first flow path containing two ft of lead pipe is shown in Figure 14. Lead levels continue to decrease in the presence of 1 mg/l PO4 -.3 mg/l zinc reaching.1 mg/l lead after 22 days of operation. Note that similar lead values were achieved at the Butler site after 16 days. The impact of water quality significantly influences the time to achieve a low lead concentration. The finished water without chemical treatment maintained a lead concentration averaging.1 mg/l. As with the Butler site, high and low lead concentrations were found, resulting from film disruption. Table 9 shows the progressive decrease in lead levels with time. Only this leg had spiked lead levels similar to leg 1 at the Butler site. The decrease in lead concentrations with time for the one ft of 5:5 tin:lead solder is shown in Figure 15. The orthophosphate-based treatment rapidly estab FIMSHED WATER 1.o m91l PO4 -.3 mgll zint BO FIGURE 14. Canonsburg, leg 1, 2 ft of lead pipe. lished a protective film on the solder such that the lead concentration was.1 mg/l for the 24 days duration. The finished water gradually met the.1 mg/l lead level. Within the first 4 days of the test, lead levels decreased from.1 mg/l to.2 mg/l, then slowly decreased to.1 mg/l. There was no indication of accelerated galvanic attack between lead solder and copper joints with the above treatment (Figure 16). A protective film was rapidly established within the first few days of operation; lead concentrations were.1 mg/l. This film was not a deposit of CaCO 3 since the untreated path required 4 days to achieve a.1 mg/l lead concentration. Soluble lead coming from the utility faucet is shown by Figure 17. Concentrations in the finished water with and without chemical treatment track one another. Lead concentrations decreased rapidly after Date TABLE 9 Lead Concentrations (mg/l) in 1 Liter First-Draw Sample from Leg 1 of Canonsburg, Pennsylvania, Test Assemblies Elapsed Time (days) Finished Water, Control 1. mg/l PO,.3 mg/l Zinc 2/16/ /27/ /1/ /14/ /28/ /4/ /18/ /2/ /9/ /23/ /2/ /18/ /16/ /6/ /12/ /2/ /3/ /11/ CORROSION-Vol. 47, No

9 J 8 6O. 6 4 J 2 1. mg/l PO..3 mg/l zinc 4 2 \ ph T.2 - ^1.mg-LPO.-.3mg1Lzlnc FIGURE 15. Canonsburg, leg 2, 1 ft of 5:5 tin:lead solder FIGURE 16. Canonsburg, leg 4,5 ft of coppertubing with five solder joints J 8 a 4. FOOSHED WATER 3 6 Ot 4 2 F1514 WATER 1. mga. PO..3 mg1l zloc É !.RMI.PO.-.3 mg/l zine DAMS FIGURE 17. Canonsburg faucet, 125-mL sample FIGURE 18. Canonsburg, leg 4,5 ftofcoppertubing with five solder joints. the first 2 days then gradually decreased to.18 mg/ L lead at the termination of the test. The low lead values indicate minimal lead leaching from the faucet in a very hard water. Copper loss from the 5 ft of tubing is shown in Figure 18. Very obvious from this figure is the beneficial nature of the chemical treatment in suppressing the release of copper. Significantly higher copper levels were attained from the control system,.8 mg/l with chemical treatment compared to 3.2 mg/l copper for the control. Although both copper coils were virgin material, the orthophosphate-based treatment suppressed copper corrosion by forming a protective barrier between the water and the meta) surface. The untreated or control system must first corrode before the cuprous oxide film can form. Copper concentration from the faucet is shown in Figure 19. As with the Butler site, only the first-draw 125-mL sample was analyzed. However, no hump in the data was found as in the Butler site. The higher hardness water slows the loss of copper from the brass fixture. After 22 days, both the chemically treated and control systems had essentially the same level of copper,.8 mg/l. One final point should be made regarding an orthophosphate-based treatment to control lead solubility. That is the formation of tricalcium phosphate. For waters that are high in hardness, the potential to form either a phosphate deposit or complex increases with ph. If the orthophosphate ion interacts preferentially with calcium ions to form an ion pair, the orthophosphate ion is not available to form a protective film on the lead surface. The minimum level of calcium hardness required to form tricalcium phosphate as a function of ph at 1 mg/l orthophosphate concentration is shown in Figure 2. The utility of the curve is similar to Figure 13, calcite saturation. As the calcium hardness increases, the ph of the system must decrease to prevent calcium deposits. Therefore, the use of orthophosphate ions to control lead solubility may not be applicable in high-hardness waters. Figure 2 indicates only the formation of tricalcium phosphate, but the formation of phosphate ion pairs can occur at much lower phs than the formation of tricalcium phosphate. Surface analysis of the lead pipe from leg 1 of each test assembly from the two test sites were characterized using scanning electron microscope/energy dispersive spectroscopy (SCM/EDS). The treated assemblies all showed a major concentration of phosphorous on the lead surface but no indication of any zinc. The protective phosphate film formed on the lead surface was of the order of 1 micrometer (1 g) and extremely tenacious. 974 CORROSION DECEMBER 1991

10 U 2. É U 1..5 \\\ 1. mg3 ^lr mg/1 zlnc R = 7.6 n UNDERSATURATED CALCIUM HARDNESS FIGURE 19. Canonsburg faucet, 125-mL sample. FIGURE 2. Tricalcium phosphate saturation, 25 C, 1 mg/l PO4. SUMMARY Orthophosphate-based chemical treatment will control lead concentration in potable water to levels promulgated by the EPA.15 mg/l. The rate at which protection is achieved is a function of water quality. Soft, low-alkalinity waters can be protected to.1 mg/l lead within a few months after initiation of chemical treatment. Chemical treatment protection is equivalent to high ph adjustment but without the negative aspects of poor disinfection and increased formation of trihalomethanes. In higher hardness alkaline water, lead control can be achieved with chemical treatment. Adjusting the system ph above the vicinity of neutral ph can cause calcium carbonate precipitation, which can result in deposit formation, higher pumping costs, and accelerated underdeposit corrosion. ACKNOWLEDGMENTS The authors acknowledge American Water Works Service Co., especially personnel at Butler and Canonsburg sites. REFERENCES 1. Public Law , Safe Drinking Water Act, Amendment of 1986 (June 19, 1986). 2. B.P. Boffardi, Materials Performance 29, 8 (199): p Drinking Water Regulations; Lead and Copper Proposed Rule, Fed. Reg. 53:16:31516 (August 18, 1988). CORROSION VoI.47, No