Alternative Aircraft Anti-Icing Formulations with Reduced Aquatic Toxicity and Biochemical Oxygen Demand (2010)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5-1 SECTION 5 TIER 2 RESULTS Aircraft Type IV Anti-Icing Fluids Table 5-1 lists the FPDs, surfactants, thickeners, and corrosion inhibitors that were selected for evaluation during Tier 2 testing of Type IV aircraft anti-icing formulations. Based on the results of Tier 1 testing, the FPDs that were selected showed improvements in COD, BOD and/or aquatic toxicity over propylene glycol. Many surfactants tested in Tier 1 had improvements in toxicity over current-use surfactants. Surfactants were selected to take advantage of these toxicity improvements as much as possible while reducing the contact angle and surface tension to ensure that the formulations completely coat the aircraft surfaces. Thickeners were selected based on their aquatic toxicity and their ability to shear in a manner similar to commercial Type IV anti-icing agents. Corrosion inhibitors were down- selected based on aquatic toxicity. TABLE 5-1. Candidate components of Type IV aircraft anti-icing fluids evaluated in Tier 2. FPDs Surfactants Thickeners Corrosion Inhibitors Glycerol DEG Tergitol L-64 Tergitol TMN-10 Triton CG-110 with 10% Ridafoam NS 221 Kelzan HP K1A96 Carbopol EZ-4 with TEA TEA Mazon RI 325 Tier 2 experiments involved the testing of more complex mixtures as compared to Tier 1, which focused on mixtures of the components and water: • FPDs + water + thickeners; FPD:water = 1:1 by weight • FPDs + water + surfactants; FPD:water = 1:1 by weight • FPDs + water + thickeners + surfactants; FPD:water = 1:1 by weight • FPDs + water + thickeners + surfactants + corrosion inhibitors; FPD:water = 1:1 by weight Tables 5-2 through 5-5 summarize the methodology used to identify Type IV anti-icing formulations with very low aquatic toxicity. The tests shown in Tables 5-2 and 5-3 were done in parallel. For example, in Table 5-2, concentrations were experimentally determined to match the viscosity/shear rate curve of a commercial Type IV anti-icing formulation. The thickener was down-selected for each FPD based on the aquatic toxicity of the formulation and the cost of the thickener. The aquatic toxicity of these mixtures was calculated based on the toxicity of the individual components previously measured in Tier 1.

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-2 TABLE 5-2. Method of down-selecting thickeners for selection of FPD and thickener with minimum aquatic toxicity and cost. FPD Thickener Concentrationa (%) Aquatic Toxicity Down Select DEG Kelzan HP A1 Estimate screening test toxicity based on individual components A K1A96 A2 Carbopol EZ-4 with TEA A3 Glycerol Kelzan HP B1 Estimate screening test toxicity based on individual components B K1A96 B2 Carbopol EZ-4 with TEA B3 aConcentrations selected to match commercial Type IV viscosity/shear rate curve. TABLE 5-3. Method of down-selecting surfactants for selection of FPD and surfactants with minimum aquatic toxicity and cost. FPD Surfactant Concentrationa (%) Aquatic Toxicity Down Select DEG Tergitol L-64 C1 Estimate screening test toxicity based on individual components C Triton CG-110 with 10% Ridafoam C2 Tergitol TMN-10 C3 Glycerol Tergitol L-64 D1 Estimate screening test toxicity based on individual components D Triton CG-110 with 10% Ridafoam D2 Tergitol TMN-10 D3 aConcentrations were selected to match commercial Type IV contact angle. TABLE 5-4. Method of verifying selection of thickeners and surfactants. FPD Thickener + Surfactant Testa Aquatic Toxicity DEG A + C Viscosity/Shear Rate Contact Angle Measure screening test aquatic toxicity and compare to estimated screening test toxicity based on individual components as measured in Tier 1. Glycerol B + D Viscosity/Shear Rate Contact Angle Measure screening test aquatic toxicity and compare to estimated screening test toxicity based on individual components as measured in Tier 1. aCompare test results to commercial Type IV formulation; determine if thickening agents absorb surfactants and if surfactants affect viscosity/shear rate behavior.

SECTION 5—4BTIER 2 RESULTS 5-3 TABLE 5-5 Method of down-selecting corrosion inhibitors. FPD Thickener + Surfactant Corrosion Inhibitora Down Selectb Aquatic Toxicity DEG A + C TEA Mazon RI 325 E Perform complete aquatic toxicity tests on down-selected mixture (FPD+A+C+E) Glycerol B + D TEA Mazon RI 325 F Perform complete aquatic toxicity tests on down-selected mixture (FPD+B+D+F) aSandwich and immersion corrosion testing with 0.2% by weight of corrosion inhibitors. bIf both pass corrosion testing, select corrosion inhibitor with lowest screening test aquatic toxicity. If only one passes corrosion testing, use that corrosion inhibitor. Similarly, in Table 5-3, the contact angles for the various surfactant concentrations in FPD/water mixtures were measured. For each surfactant, the concentration that matches the contact angle for the commercially available Type IV formulation was selected. The contact angle is related to the surface tension, which should be less than 35 dynes/cm for a Type IV anti-icing formulation (33). The surfactant was down-selected for each FPD based on the aquatic toxicity of the formulation and the cost of the surfactant. The aquatic toxicity of these mixtures was estimated based on the toxicity of the individual components previously measured in Tier 1. Anti-icing formulations with the thickener and surfactant concentrations found in the tests shown in Tables 5-2 and 5-3 were tested for contact angle and viscosity, as shown in Table 5- 4. The primary aim of these tests was to determine if there is any interaction between the thickeners and surfactants that will substantially affect the viscosity and contact angle. Finally, corrosion inhibitors were added to the formulations, and corrosion and full aquatic toxicity testing was carried out to develop a final formulation with minimum aquatic toxicity. Testing FPDs/Water and Thickeners The three candidate thickeners were mixed at different concentrations with 1:1 mixtures by weight of the two candidate FPDs and water. A thickener concentration was found for each of the two FPD formulations to match the viscosity/shear rate curve of a commercial Type IV aircraft anti-icing formulation. The viscosities of the candidate FPD/water and thickener solutions were measured at a temperature of 20°C (AMS 1428 Type IV Fluid ) and multiple shear rates (0.3 revolution per minute [rpm], 6 rpm and 30 rpm per AMS 1428) on a Brookfield Programmable DV-II+ Viscometer. The measurements were compared to the commercial Type IV AAF. Procedures for mixing the thickeners (in particulate form) with liquid solutions were followed from the manufacturer’s literature. The following procedures were used:

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-4 Kelzan HP and K1A96 • Formulation was done with deionized water at room temperature. • A mixer or blender was used to ensure adequate dispersion of the thickener particles. • The FPD/water solution was placed in the mixer and the stirring rate was set to a high value. • The thickener was slowly added directly into the vortex. Enough thickener was added to produce the desired thickener concentration. • After all the thickener was added, the mixer was operated for 10 minutes. • The solution was stored at room temperature. • The solution was tested after a 24-hour period. • The appearance of any particles or phases in the solution before use was noted. Carbopol EZ-4 with TEA This formulation consists of a mixture of Carbopol EZ-4 in FPD/water neutralized with TEA. • Formulation was done with warm temperature water. The optimal temperature was between 40 and 50°C. • A mixer or blender was used to ensure adequate dispersion of the thickener particles. • The FPD/water solution was placed in the mixer and the stirring rate was set to a medium speed. Setting the mixer to a high speed (greater than 5,000 rpm) could degrade the polymer matrix. • The thickener was slowly added directly into the vortex. Enough thickener was added to produce the desired thickener concentration. • After all the thickener was added, the mixer was operated for 10 minutes. • The initial pH of the solution was measured and recorded. • Sufficient TEA was added to raise the pH of the solution to 8.0. • Upon neutralization, the solution had a grainy “applesauce” appearance. This was normal and disappeared within 1 hour. • The solution was stored at room temperature. • The solution was tested after a 24-hour period. • The appearance of any particles or phases was noted before use. The first set of experiments evaluated the effect of thickener concentration on the viscosity of mixtures of Type IV FPDs and water. Table 5-6 lists the FPDs, thickeners and thickener concentrations that were initially evaluated. The range of thickener concentrations was selected based on the Tier 1 results, in which selected concentrations of the thickeners were

SECTION 5—4BTIER 2 RESULTS 5-5 mixed with water and the viscosities compared to a commercially available Type IV formulation. The change in concentration of the thickeners to match the viscosity of the commercially available Type IV formulation was evaluated for each thickener. TABLE 5-6. Test matrix to evaluate the effect of thickener concentration on viscosity. FPD/Water Thickener Thickener Concentration, wt % Glycerol/watera DEG/watera Kelzan HP 0.5 0.375 0.25 K1A96 0.75 0.563 0.375 Carbopol EZ-4 with TEA 0.1 0.074 0.05 Commercial Type IV formulation a50:50 wt % of FPD/water. Figures 5-1 and 5-2 show the average viscosities of the three thickeners at different shear rates for each of the FPD/water mixtures. For each thickener concentration, four viscosity measurements were made. For some cases, on the semi-log scale, the data points were not differentiable. In addition, data for water and the commercial Type IV formulation are also shown. A shear rate of 0.084/sec corresponds to 0.3 rpm for a No. 34 Brookfield viscometer spindle; similarly, 1.68/sec and 8.4/sec for 6 and 30 rpm, respectively. Table 5-7 summarizes the concentrations of the thickeners at different shear rates that give the same viscosity as the commercial Type IV formulation.

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-6 Figure 5-1. Viscosity of Kelzan HP and K1A96 at different concentrations and shear rates for glycerol/water and diethylene glycol/water mixtures. Kelzan HP 100 1,000 10,000 100,000 1,000,000 0.20 0.30 0.40 0.50 Thickener concentration, % V is co si ty , c P Shear Rate, 0.084/sec K1A96 0.2 0.4 0.6 0.8 Thickener concentration, % 100 1,000 10,000 100,000 1,000,000 0.20 0.30 0.40 0.50 Thickener concentration, % Vi sc os ity , c P Shear Rate, 1.68/sec 0.2 0.4 0.6 0.8 Thickener concentration, % 100 1,000 10,000 0.20 0.30 0.40 0.50 Thickener concentration, % V is co si ty , c P Shear Rate, 8.4/sec 0.2 0.4 0.6 0.8 Thickener concentration, % Glycerol Diethylene Glycol Water Comm Type IV

SECTION 5—4BTIER 2 RESULTS 5-7 Figure 5-2. Viscosity of Carbopol EZ-4 with TEA at different concentrations and shear rates for glycerol/water and diethylene glycol/water mixtures. Shear Rate, 0.084/sec 100 1,000 10,000 100,000 0.025 0.05 0.075 0.1 Thickener concentration, % Vi sc oi st y, c P Shear Rate, 1.68/sec 100 1,000 10,000 0.025 0.05 0.075 0.1 Thickener concentration, % Vi sc os ity , c P Shear Rate, 8.4/sec 100 1,000 10,000 0.025 0.05 0.075 0.1 Thickener concentration, % Vi sc oi st y, c P Glycerol Diethylene Glycol Water Comm Type IV

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-8 TABLE 5-7. Concentration of thickeners (wt %) for different FPDs to match viscosity of commercial Type IV anti- icing fluid at different shear rates. FPD/water Thickener Concentration (%) for Three Different Shear Rates (1/sec) 0.084 1.68 8.4 Diethylene glycola Kelzan HP 0.30 0.30 0.50 K1A96 0.40 0.40 —b Carbopol EZ-4/TEA 0.080 0.075 0.075 Glycerola Kelzan HP 0.25 0.25 0.32 K1A96 0.30 0.25 — Carbopol EZ-4/TEA 0.055 0.055 — Water Kelzan HP 0.38 0.40 0.50 K1A96 0.20 0.20 0.25 Carbopol EZ-4/TEA 0.025 — — a1:1 by weight. b— not determined Thickener concentrations were selected to match the viscosity/shear rate curve for the commercial Type IV formulation over most of the shear rate range (Table 5-8). Figure 5-3 graphically presents the test results over the entire shear rate range. TABLE 5-8. Thickener concentrations selected to match viscosity/shear rate curve of commercial Type IV formulation. FPD Thickener Concentration (wt %) DEG Kelzan HP 0.25 K1A96 0.38 Carbopol EZ-4 with TEA 0.076 Glycerol Kelzan HP 0.25 K1A96 0.28 Carbopol EZ-4 with TEA 0.055

SECTION 5—4BTIER 2 RESULTS 5-9 Figure 5-3. Comparison of viscosity/shear rate data for selected thickener concentrations to commercial Type IV formulation. Diethylene Glycol 100 1000 10000 100000 0.01 0.1 1 10 100 Shear Rate, 1/sec Vi sc os ity , c P 0.251% 12_18 0.251% 11_10 Octagon Kelzan HP Glycerol 0.01 1 100 Shear Rate, 1/sec 0.250% 0.250% Octagon K1A96 100 1000 10000 100000 0.01 0.1 1 10 100 Shear Rate, 1/sec Vi sc os ity , c P 0.377% 12_18 0.377% 11_08 Octagon 0.01 0.1 1 10 100 Shear Rate, 1/sec 0.276% Octagon Carbopol EZ-4/TEA 100 1000 10000 100000 0.01 0.1 1 10 100 Shear Rate, 1/sec Vi sc os ity , c P 0.0763 12_18 0.0763% 11_10 Octagon 0.01 0.1 1 10 100 Shear Rate, 1/sec 0.0552% 0.0479% Octagon

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-10 Table 5-9 summarizes the aquatic toxicity and cost of the thickeners at the concentrations shown in Table 5-4 (to match the commercial Type IV formulation). The aquatic toxicity is based on the measured values of the neat components determined in Tier 1. Except for one case (DEG/water/Carbolpol EZ-4), the aquatic toxicity of the FPD/water/thickener formulation was determined primarily by the FPD and not by the thickener itself. An additional consideration used in down-selecting the appropriate thickener was the cost of the thickener in the formulation. The unit costs of the thickeners are given in the sixth column of Table 5-9 for large (bulk) quantities of the thickeners. Although the toxicity of Carbopol EZ-4 (with TEA) is higher than the other two thickeners, the required concentration of Carbopol EZ-4 is much lower. As a result, the cost of the Carbopol EZ-4 thickener in the anti-icing formulation is a factor of two to three lower than Kelzan HP and K1A96 thickeners. Carbopol EZ-4 with TEA was selected as the preferred thickener for the remainder of the Tier 2 test series. TABLE 5-9. Toxicity and costs of thickeners in FPD/water formulations. FPDa Thickener Concentrationb (wt %) P. promelas LC50 for pure product (mg/L) Predicted P. promelas LC50 in Formulation (mg/L) Unit Cost of Thickener ($/lb) Cost of Thickener in Formulation ($/lb) DEG — — 56,900 113,800 — — Kelzan HP 0.25 2,400 949,800 6.52 0.0163 K1A96 0.38 3,200 843,500 10.09 0.0380 Carbopol EZ- 4 with TEA 0.076 270 347,300 10.68 0.0081 Glycerol — — 46,000 92,000 — — Kelzan HP 0.25 2,400 950,300 6.52 0.0163 K1A96 0.28 3,200 1,153,100 10.09 0.0279 Carbopol EZ- 4 with TEA 0.055 270 479,900 10.68 0.0059 aMixture of FPD and water (1:1 by weight). bConcentration selected to match commercial Type IV viscosity vs. shear rate curve. FPDs/Water and Surfactants Surfactants evaluated in Tier 1 were down-selected based on pure product toxicity and contact angle. The surfactants selected for testing in Tier 2 were Tergitol L-64, Tergitol TMN- 10 and Triton CG-110 mixed with 10 percent (by weight) of Ridafoam NS 22, an anti- foaming agent (Table 5-1). For a given surface, the contact angle is directly related to the liquid surface tension. Surfactants reduce the surface tension of the applied fluids to ensure they completely coat the aircraft and the runway surfaces. In Tier 2, the down-selected surfactants were added to FPD/water mixtures and their concentrations selected such that their contact angle matched the contact angle of the commercial Type IV deicing formulation. As in Tier 1, the contact angle measurements were carried out on a Drop Shape Analysis System DSA 100 (Krüss). Initial experiments were done on a silicon/gold (Si/Au)

SECTION 5—4BTIER 2 RESULTS 5-11 surface, but the measurements were inconsistent. The experimental procedure was revised for determining an advancing contact angle, in which the contact angle is measured while the drop on a surface is increasing in volume. This volume increase is accomplished by adding liquid to the drop by use of a syringe. This procedure provided more consistent data, especially at low contact angles, enabling the identification of the minimum concentration of candidate surfactant needed to achieve complete wetting of aircraft surfaces. The liquid surface tension is a unique value for a given liquid, independent of the surface and the contact angle. Table 5-10 lists the liquid surface tension for different liquids (34). For a given surface, the liquid surface tension is directly related to the contact angle. The contact angle for the liquids on a Si/Au surface is also given in Table 5-6, together with the measured value of the commercial Type IV formulation. The cosine of the contact angle is shown in Figure 5-4 as a function of the liquid surface tension. For the Si/Au surface, the surface tension for the commercial Type IV formulation is approximately 44 dyne/cm, higher than the reported value of approximately 35 dyne/cm for a Type IV AAF (33). TABLE 5-10. Liquid surface tension and surface contact angle for neat liquids. Liquid Surface Tension, mJ/m2 (dyne/cm) Contact Anglea (Deg.) Decane 23.43 — Hexadecane 27.76 — Dimethylsulfoxide 43.58 11.83 DEG 45.04 27.53 Ethylene glycol 47.99 36.33 Formamide 57.49 36.43 Glycerol 63.11 47.97 Water 72.75 73.70 Comm Type IV 24.9 aFor Si/Au surface.

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-12 Figure 5-4. Surface tension for Si/Au surface using data from Table 5-10. 0.0 0.2 0.4 0.6 0.8 1.0 20 40 60 80 Surface Tension, mJ/m2 C os (c on ta ct a ng le ) Si/Gold Comm Type IV The contact angles were measured as a function of surfactant concentration for each FPD/water mixture and matched to the contact angle of a commercial Type IV anti-icing formulation. Each contact angle corresponds to a specific surface tension. Figures 5-5 and 5- 6 show the test results for DEG/water (1:1 by weight) glycerol/water (1:1 by weight), respectively. The contact angle for a commercial Type I formulation was also measured and was found to be within 3 degrees of the commercial Type IV formulation. Figure 5-5. Contact angle measurement results for DEG/water and surfactants. Diethylene Glycol 0 10 20 30 40 50 0.00 0.25 0.50 0.75 1.00 Surfactant concentration, wt % Co nt ac t a ng le , d eg re es Commercial Type IV Tergitol L-64 Commerical Type I Tergitol TMN-10 Triton CG- 110/Ridafoam

SECTION 5—4BTIER 2 RESULTS 5-13 Figure 5-6. Contact angle measurement results for glycerol/water and surfactants. Glycerol 0 20 40 60 80 0.00 0.25 0.50 0.75 1.00 Surfactant concentration, wt % Co nt ac t a ng le , d eg re es Commerical Type IV Tergitol L-64 Commercial Type I Tergitol TMN-10 Triton CG- 110/Ridafoam Table 5-11 summarizes the surfactant concentrations for each FPD/water and surfactant combination that results in a contact angle that matches the contact angle for the commercial Type IV anti-icing formulation. The contact angle for the DEG/water/Triton CG-110 formulation was always greater than the commercial Type IV formulation; this formulation was not considered further. Surfactant concentrations shown to have a higher contact angle than the commercially available formulation would need to be increased to reduce the surface tension, thereby increasing aquatic toxicity. Similarly, surfactant concentrations shown to have a lower contact angle than the commercially available formulation would need to be decreased to increase the surface tension, thereby decreasing aquatic toxicity. TABLE 5-11. Properties of FPD/water/surfactant formulations having the same surface tension as a commercial Type IV formulation. Formulation Surfactant Concentration, wt % P. promelas LC50 for Pure Product (mg/L) Predicted P. promelas LC50 in Formulation (mg/L) Price ($/lb) Relative Price DEG 50 56,900 113,800 — — Tergitol L-64 0.250 14,900 5,960,000 1.5 1.00 Tergitol TMN-10 0.150 90 60,000 2.3 0.92 Triton CG-110 + 10% Ridafoam — 740 — 2.1 — Glycerol 50 46,000 92,000 — — Tergitol L-64 0.100 14,900 14,900,000 1.5 1.00 Tergitol TMN-10 0.150 90 60,000 2.3 2.30 Triton CG-110 + 10% Ridafoam 0.200 740 370,000 2.1 2.79 Table 5-11 also shows the Pimephales promelas LC50 for the pure solutions as obtained from the Tier 1 results and the expected LC50 for the concentrations in the formulation. Tergitol L-

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-14 64 has a significantly lower toxicity than the two other surfactants. The LC50 for the rest of the surfactants was close to those of the FPD. Based on these data, Tergitol L-64 was tentatively down-selected for further testing. Another criterion for down-selection was the relative cost of the surfactants. Table 5-11 shows the market price of the surfactants and the price of the surfactants in the formulation (concentration x price) relative to Tergitol L-64. For the DEG formulation, the price of Tergitol TMN-10 is slightly lower than that of Tergitol L-64, but Tergitol L-64 is the final choice because of the difference in aquatic toxicity. For glycerol, the price of Tergitol L-64 in the formulation is lower than the other two surfactants, so Tergitol L-64 is still the final choice. FPDs/Water, Thickeners and Surfactants Surfactants were added to the FPD, water, and thickener mixtures at the concentrations found in the previous section, and the resultant mixtures were subjected to viscosity/shear rate and contact angle tests. The primary aim of these tests was to determine if there was any interaction between the thickeners and surfactants that will substantially affect the viscosity and surface tension of the formulations. Foaming tests were also carried out for the formulations. One concern was the effect of the surfactant on the viscosity/shear rate characteristics. DEG/ and glycerol/water formulations were prepared with the Carbopol EZ-4/TEA thickener at the concentrations shown in Table 5-12. The samples were split into two portions and Tergitol L-64 was added to the samples at the concentration shown in Table 5- 12. Viscosity was measured for both samples at the same shear rates. The viscosity of the DEG/water formulation without the Tergitol L-64 surfactant was 10 to 15 percent higher than with the surfactant (Figure 5-7), but for glycerol, the viscosity with the surfactant was 60 to 72 percent higher than without the surfactant. There is an uncertainty in the measurements of about 20 percent, so that there is very little effect of the surfactant on the DEG formulation, but there is a measured increase in the viscosity of the glycerol formulation with the addition of the surfactant. TABLE 5-12. Type IV anti-icing formulations containing a thickener and a surfactant. Formulation Components Concentration (wt %) DEG 50.0 Thickener: Carbopol EZ-4 with TEA 0.076 Surfactant: Tergitol L-64 DEG Thickener: Carbopol EZ-4 with TEA 0.25 50.0 0.076 Surfactant: Tergitol TMN-10 0.15 Glycerol 50.0 Thickener: Carbopol EZ-4 with TEA 0.055 Surfactant: Tergitol L-64 Glycerol Thickener: Carbopol EZ-4 with TEA 0.10 50.0 0.055 Surfactant: Tergitol TMN-10 0.15

SECTION 5—4BTIER 2 RESULTS 5-15 Figure 5-7. Comparison of Viscosity of DEG/ and glycerol/water/thickener (Carbopol EZ-4/TEA formulations with and without the addition of surfactant. Diethylene Glycol - Carbopol EZ-4/TEA 100 1000 10000 100000 0.01 0.1 1 10 100 Shear Rate, 1/sec Vi sc os ity , c P No surfactant with Tergitol L-64 Glycerol - Carbopol EZ-4/TEA 100 1000 10000 100000 0.01 0.1 1 10 100 Shear Rate, 1/sec Vi sc os ity , c P No surfactant with Tergitol L-64 There is also concern that the thickener in the formulation may absorb some of the surfactant, essentially decreasing the effective surfactant concentration, increasing the surface tension, and inhibiting the spreading of the fluid over the aircraft. Surface angle measurements were made on the DEG/water and glycerol/water formulations containing the thickener with and without the surfactants at the concentrations shown in Table 5-12 and compared to the FPD/water formulations containing just the surfactant (see Figure 5-8). The thickener concentration for both formulations matched the viscosity/shear rate curve for the commercial Type IV anti-icing formulation. Figure 5-8 graphically shows the results. For DEG/water/thickener mixtures containing either Tergitol L-64 or Tergitol TMN-10, the thickener tends to increase the contact angle (and surface tension) over mixtures containing the surfactant, while there seems to be very little effect of the thickener for the glycerol

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-16 mixtures. This result suggests that for the DEG/water formulations containing both a thickener and surfactant, the surfactant concentration should be increased to match the contact angle for the commercial Type IV formulation. The contact angle for the DEG/water mixture with only the thickener as an additive is 5 degrees higher than without the thickener, whereas for glycerol/water mixtures the contact angle is about 10 degrees less. There is an uncertainty in the contact angle measurements of about 3 degrees. The results indicate for the DEG formulations, the surfactant appears to be absorbed by the thickener, essentially increasing the contact angle and surface tension. There does not appear to be an effect of the thickener on the contact angle for the glycerol formulations. Figure 5-8. Effect of thickener on contact angle for FPD/water/surfactant mixtures. Diethylene Glycol 0 20 40 60 80 0.00 0.10 0.20 0.30 0.40 0.50 Surfactant concentration, wt % C on ta ct a ng le , d eg re es Tergitol L-64 Glycerol 0.00 0.10 0.20 0.30 0.40 0.50 Surfactant concentration, wt % Tergitol TMN-10 0 20 40 60 80 0.00 0.10 0.20 0.30 0.40 0.50 Surfactant concentration, wt % C on ta ct a ng le , d eg re es 0.00 0.10 0.20 0.30 0.40 0.50 Surfactant concentration, wt % Comm Type IV" Surfactant only Thickener only Thickener + surfactant

SECTION 5—4BTIER 2 RESULTS 5-17 Foaming is a potential problem for deicing aircraft because the deicing formulation may cover the windows and impair visibility for the pilot. Foaming tests have been included in AMS 1424H for SAE Type I deicing/anti-icing fluids but have not been included in the requirements for Type IV anti-icing fluids. However, because of the safety implications, foam testing with the down-selected DEG and glycerol formulations with Tergitol L-64 (see Table 5-12) was conducted. The test procedures for quantifying foaming characteristics were modified from those of AMS 1424H. An aluminum plate was placed in a dry ice chamber and reached a temperature of -10°C. The anti-icing formulation was heated to 60°C and placed in a blender for 15 seconds. The plate was then taken out of the dry ice and placed in a plastic container at an angle of 10 degrees with the horizontal. The heated fluid was then poured on the plate and observed for 15 minutes. Photographs of the film were taken throughout the test period. The photographs shown in Figure 5-9 (top row) were taken of the middle of the film. A control test was done with a commercially available Type IV fluid. For all three formulations, less than 5 percent of the surface layer had foam initially; the foam then dissipated within 15 minutes. Foam persisted at the edges of the film. Figure 5-9. Photographs of foaming tests. DEG without Ridafoam Glycerol formulation without Ridafoam DEG with Ridafoam Glycerol formulation with Ridafoam

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-18 Ridafoam, an anti-foaming agent, was placed in the two formulations containing Carbopol EZ-4/TEA and Tergitol L-64 at 10 percent of the surfactant concentration (Table 5-12). Foaming tests were performed on the two formulations. No foam formed initially and throughout the test run. Photographs taken at the edge of the film containing Ridafoam are shown in Figure 5-9 and indicate the absence of foam at the edge of the film, unlike the films not containing Ridafoam. The anti-icing formulations with Tergitol L-64 performed as well as the commercially available anti-icer with respect to foaming; less foam was present with the addition of Ridafoam. At this point in experiments, diethylene glycol (DEG) and glycerol were considered to be equally promising freezing point depressant candidates. The key screening data used to differentiate these two candidates shows that the properties of the two freezing point depressants are very similar (Table 5-13). The melting point, normal boiling point, and flash point for both FPDs are all acceptable. Glycerol’s theoretical oxygen demand and aquatic toxicity, as measured by microtox testing, are lower than DEG’s. DEG’s aquatic toxicity toward ceriodaphnia dubia and pimephales promelas is lower than glycerol’s. The mammalian toxicity, as measured by rat oral LD50, is nearly identical for both candidates. TABLE 5-13. Key freezing point depressant properties. Property DEG Glycerol 1 CAS Number 111-46-6 56-81-5 2 Melting Point, [C] -10a 170a 3 Normal Boiling Point [C] 246a 287a 3 Flash Point [F] 290b 320b 4 Theoretical Oxygen Demand [g O2/g FPD] 1.508 1.216 5 Aquatic Toxicity, Microtox [mg/l] 66,896c 129,617c 6 Aquatic Toxicity, C. Dubia [mg/l] 53,044c 34,865c 7 Aquatic Toxicity, P. Promelas [mg/l] 56,851c 46,004c 8 Rat Oral LD50 [mg/kg] 12,565d 12,600d a) NIST Chemistry WebBook. b) Aldrich Chemical Catalog. c) ACRP, 2008. d) Lewis, 2000. Figure 5-10 shows the freezing point curves for mixtures of each candidate with water. The black curve represents values for DEG (8). The gray curve represents values for glycerol (35). To the left of the eutectic point, the curves are nearly identical, indicating very little difference in each chemical’s ability to depress the freezing point of water. However, to the right of the eutectic point, glycerol’s curve rises much faster than DEG’s. This rapid increase is attributable to glycerol’s higher melting point. This rapid increase in the melting point of glycerol solutions could cause operational problems. For example, Figure 5-10 shows that at -20°C a 90 wt% glycerol solution would partially freeze into a slurry, whereas a 90 wt% DEG solution would still be completely liquid. Because glycerol solutions are known to supercool (remain a liquid below its

SECTION 5—4BTIER 2 RESULTS 5-19 freezing point), it is likely that the solution will not form an actual slurry but just a highly viscous solution. Figure 5-10. Freezing point curves for final candidates. -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 0 10 20 30 40 50 60 70 80 90 100 Composition [wt% FPD] Te m pe ra tu re [C ] Glycerol DEG Even though deicing fluids are not applied with more than 60 wt% freezing point depressant, asymmetric evaporation would cause the fluids to concentrate freezing point depressants. Figure 5-11 shows the vapor-liquid equilibrium curves for glycerol-water mixtures at atmospheric pressure (36). The experimental temperatures are significantly higher than those encountered during deicing operations, but the low temperature curves will have a form similar to the one shown. Figure 5-11. Glycerol-water vapor liquid equilibrium. 0 50 100 150 200 250 300 350 0 10 20 30 40 50 60 70 80 90 100 Weight % Glycerol Te m pe ra tu re [C ] For example, Figure 5-11 shows that a 50 wt% mixture at 200°C would separate into a vapor fraction containing 8 wt% glycerol and a liquid fraction containing 98 wt% glycerol. While data could not be located, it is likely that the curves would behave in a similar fashion at lower temperatures. As a result, DEG was considered a more promising FPD for the final formulation than glycerol.

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-20 FPDs/Water, Thickeners, Surfactants, and Corrosion Inhibitors Corrosion tests using sandwich and total immersion were conducted with the two most promising DEG formulations shown in Table 5-14 according to the specifications in SAE AMS 1428 Type IV Fluid. Sandwich corrosion testing was conducted in accordance with ASTM F 1110, Standard Test Method for Sandwich Corrosion Test. Filter papers saturated with concentrated and diluted anti-icing fluids were placed between a sandwich of aluminum alloy panels. Corrosion was evaluated by comparison with control panels using ASTM D 1193, Type IV water. The sandwich panels were placed for 8 hours in an oven with forced air circulation capable of maintaining 37.7°C and for 16 hours in a humidity test chamber capable of maintaining 95 to 100 percent relative humidity at 37.7°C. The pattern was repeated for 5 days. On the sixth and seventh days, the panels were kept in the humidity test chamber for 48 hours. The total test time was 168 hours. Table 5-15 shows the sandwich corrosion test results for 2024-T3 bare anodized (with sulfuric acid instead of chromic acid for environmental reasons) and 2024-T3 Alclad panels. After testing, the panels did not show corrosion worse than the control panels. TABLE 5-14. Formulations used in corrosion testing. Formulation component Weight percent FPD: DEG/water 50:50 Thickener: Carbopol EZ-4 with TEA 0.076 Surfactant: Tergitol L-64 0.250 Corrosion inhibitors: TEA or Mazon RI-325 0.2 0.2 TABLE 5-15. Sandwich corrosion test results. Test Solution 2024-T3 Bare Aluminum 2024-T3 Alclad Corrosion inhibitor - TEA DEG—concentrated 1* 1 DEG—diluted with water (1:1 wt) 1 1 Corrosion inhibitor—Mazon RI 325 DEG—concentrated 1 1 DEG—diluted with water (1:1 wt) 1 1 Control (ASTM D 1193, Type IV water) 1 1 * 1 denotes very slight corrosion or very slight discoloration, and/or up to 5% of area under filter paper corroded

SECTION 5—4BTIER 2 RESULTS 5-21 Total immersion testing was carried out in accordance with ASTM F 483, Standard Test Method for Total Immersion Corrosion Test for Aircraft Maintenance Chemicals. The method involves immersing three test coupons of the same alloy in concentrated and diluted anti-icing solutions for 168 hours at an oven test temperature of 38°C. The coupons were weighed before, during (end of 24 hours) and after the tests and observed for visible changes in comparison to a control (virgin) coupon. Table 5-16 shows the test results and indicates that the coupons do not have a weight change greater than the allowed weight change. TABLE 5-16. Total immersion corrosion test results. Test Panel Weight Change Allowed, mg/cm2 per 24 hrs Weight Change mg/cm2 per 24 hrs TEA Mazon RI 325 Concen Dilute (1:1) Concen Dilute (1:1) AMS 4037 Aluminum alloy, anodized 0.3 0.02 0.01 0.02 0.03 AMS 4041 Aluminum alloy 0.3 <0.01 0.01 0.01 0.01 Environmental Characteristics Aquatic toxicity was tested through the entire formulation process, and the final formulations were tested for COD and BOD. A synthesis of these results and changes as additional components were added are discussed in this section. Aquatic Toxicity Methods Alternative deicer formulations were evaluated for toxicity in a step-wise fashion after each additional component was added (Table 5-17). Screening toxicity tests were followed by definitive toxicity tests on the final formulations. Definitive toxicity tests were conducted following EPA guidelines. Briefly, each definitive acute test consisted of a 50 percent dilution series with five treatments plus a laboratory control. Moderately hard reconstituted water was used as the laboratory control and product dilution water. Each treatment consisted of four replicates with 5 C. dubia or 10 Pimephales promelas per replicate. The acute assays were conducted in an environmental chamber at 20°C with a 16-/8-hour light/dark cycle. Test treatments were renewed daily with freshly prepared solutions, and water quality parameters including dissolved oxygen, pH and conductivity were monitored daily. At 48 hours for C. dubia and 96 hours for Pimephales promelas, the survival per replicate was recorded for calculation of an LC50 using Probit or Spearman-Karber analysis, as appropriate (37). Definitive chronic tests were used to evaluate sublethal endpoints with Pimephales promelas (weight) and C. dubia (young produced). Like the definitive acute tests, chronic assays consisted of a 50 percent dilution series of five treatments and a laboratory control. Laboratory control and dilution water was moderately hard water. Each treatment was

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-22 replicated 10 times with one C. dubia or two fish per replicate. Organisms were <24 hours old at the start of the tests. Tests were maintained in an environmental chamber at 25°C with a 16-/8-hour light/dark cycle. Treatments were renewed with freshly prepared solutions daily. Water quality parameters were monitored daily. The number of young in the C. dubia test replicates were recorded and removed daily. The C. dubia chronic test was terminated when at least 80 percent of lab control individuals released their third brood (day 6 or 7). The Pimephales promelas chronic tests were terminated on day 7 when the fish from each replicate were removed for weighing. Three-brood totals and fish dry weight were used to calculate inhibition concentrations (IC25) using the EPA ICp program (38). In addition to the species used in regulatory requirements, the luminescent marine bacterium, Vibrio fischeri , also known as Microtox® (Azure Environmental, Carlsbad, CA, USA), were included in the battery of toxicity test species. While not used as a regulatory tool, the rapid results can be very useful, especially when evaluating many treatments during toxicity characterization. In this assay, freeze-dried bacteria were re-hydrated in a saline solution. A measured aliquot of re-hydrated bacteria were added to the test treatments and controls. The luminescence endpoint was determined spectrophotometrically after 15 minutes at a test temperature of 15°C. EC50s were calculated using the Microtox® software. Results Results from the stepwise toxicity testing showed how toxicity changed as each additional component was included in the formulation (Table 5-17). These results were compared to theoretical results based on toxicity of the individual components. The theoretical values were determined under the assumption that there were no synergistic or antagonistic toxicity interactions among chemicals when they were included in the mixture. Using this assumption, the theoretical toxicity endpoint was the most toxic endpoint (lowest value) of the components, given that component’s concentration in the formulation. Addition of most components resulted in toxicity endpoints similar to theoretical values. However, addition of the anti-foaming agent to the first formulation indicated a synergistic interaction for all three organisms (the formulation was more toxic than the theoretical value). ). Similarly, synergistic interactions in the Microtox® and P. promelas tests were observed from addition of TEA in the final formulation. For the second formulation, addition of the thickener resulted in toxicity endpoints similar to theoretical values for C. dubia and Pimephales promelas, but results from the Microtox® test indicate synergistic interactions with addition of the thickener and the surfactant. In most cases, it was valid to assume that individual component toxicity could be used to determine formulation toxicity. The instances where this was not true toxicity evaluation was more complicated, requiring empirical observations to understand which components were responsible for final formulation toxicity. In addition, different interactions were observed depending on the FPD. Of particular interest was the difference between addition of the thickener to DEG as opposed to that for glycerol. In the DEG formulation, results were similar to theoretical values, but a synergistic interaction for toxicity was present in this step for the glycerol formulation. This indicated that synergistic interactions for the same component were different depending on the composition of the rest of the formulation. In this case, it was only a difference in FPD that caused a difference in the synergistic interaction.

SECTION 5—4BTIER 2 RESULTS 5-23 TABLE 5-17. Comparison of theoretical values with measured test results for step-wise Type IV anti-icer formulation construction Added Component Percent of Mixture Theoretical Values Test Results Microtox® EC50a (mg/L) C. dubia LC50b (mg/L) P. promelas LC50 (mg/L) Microtox ® EC50** (mg/L) C. dubia LC50*** (mg/L) P. promelas LC50 (mg/L) DEG formulation Water 50 — — — — — — DEG 50 130,000 110,000 110,000 130,000 110,000 110,000 Carbopol EZ4 with TEAc (thickener) 0.0763 130,000 66,000 110,000 140,000 71,000 120,000 Tergitol L-64 (surfactant) 0.25 130,000 66,000 110,000 110,000 57,000 140,000 Ridafoam (anti- foaming agent) 0.025 12,000 66,000 110,000 25,000 25,000 59,000 TEA (corrosion inhibitor, does not include Ridafoam) -final formulation 0.2 110,000 66,000 110,000 43,000 53,000 89,000 Glycerol formulation Water 50 — — — — — — Glycerol 50 260,000 70,000 92,000 260,000 70,000 92,000 Carbopol EZ4 with TEA (thickener) 0.0552 260,000 70,000 92,000 43,000 66,000 71,000 Tergitol L-64 (surfactant) 0.1 260,000 70,000 92,000 18,000 75,000 63,000 Ridafoam (anti- foaming agent) 0.01 12,000 70,000 92,000 18,000 53,000 63,000 Screening toxicity results are only approximations. Screening toxicity procedures include fewer replicates, non- renewal of test solutions, shorter exposure duration for Pimephales promelas, and other procedural variances from definitive toxicity tests. Compounds are organized by least toxic to most toxic endpoint, determined by the most sensitive species. aThe Microtox® EC50 is the statistically determined concentration that would result in a 50% reduction in light emission compared to a laboratory control. bThe LC50 is the statistically determined concentration that would cause death in 50% of the population exposed. cTEA. Results of definitive aquatic toxicity tests on the final formulation indicate that acute and chronic toxicity endpoints were substantially greater (lower toxicity) than results previously published for current-use formulations (Table 5-18, (1)). Acute toxicity endpoints ranged from 219 to 13,800 mg/L in tests with current-use formulations with each of the four tested products having one or more of the three endpoints at least as low as 528 mg/L. The lowest

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-24 of the acute toxicity endpoints in the final DEG formulation in this research was 32,700 mg/L in the final DEG formulation. Chronic toxicity endpoints ranged from 79.4 to 1,350 mg/L in tests with current-use formulations with each of the four tested products having one or more of the three endpoints at least as low as 130 mg/L (1). The lowest of the chronic toxicity endpoints in the final DEG formulation from this research was 8,970 mg/L in the final DEG formulation (Table 5-18). In current-use formulations, surfactants were identified as the component with the greatest influence on toxicity (1). In the final formulation developed from this research, toxicity results indicate that the chosen surfactant had little or no influence on toxicity. The primary components influencing toxicity in the final formulation were the FPD, the thickener, and the corrosion inhibitor. Even with these influences, the test results indicate that toxicity endpoints were one to three orders of magnitude greater (less toxic) than those from current-use formulations. Table 5-18. Results from definitive aquatic toxicity testing of final Type IV DEG formulation (95% confidence interval). Microtox® EC50a (mg/L) Acute toxicity Chronic toxicity C. dubia LC50b (mg/L) P. promelas LC50 (mg/L) C. dubia IC25c (mg/L) P. promelas IC25 (mg/L) S. Capricorutum IC25 (mg/L) 54,900 32,700 126,000 8,970 60,200 42,100 (53,700–56,100) (28,600–37,400) (116,000–136,000) (4,730–13,800) (56,600–62,500) (40,100– 44,000) aThe Microtox® EC50 is the statistically determined concentration that would result in a 50% reduction in light emission compared to a laboratory control. bThe LC50 is the statistically determined concentration that would cause death in 50% of the population exposed. cThe C25 is the statistically determined concentration that would cause a 25% inhibition in growth (P. promelas) or reproduction (C. dubia). Biochemical Oxygen Demand Methods BOD5 and COD were tested on the final formulation. Tests were run in triplicate and results were reported as the average and percent relative standard deviation. Analytical methods are identical to Tier 1 methods as reported from Tier 1 testing in section 4. Results The concentration of COD in the final formulation was 752 g/Kg with a relative standard deviation of 0.96%. These results are consistent with Tier 1 testing results on neat DEG (COD = 1,500 g/Kg) considering that the final formulation contains 50% DEG. The concentration of BOD5 could not be determined. Difficulties with seed acclimation to DEG as described in Tier 1 testing remained in tests with the final formulation. The original Tier 1 testing on neat DEG was characterized by inadequate DO depletion at low DEG concentrations (up to 333 mg/L) and too much DO depletion at higher DEG concentrations (above 581 mg/L). DEG dilutions between these concentrations were attempted for neat DEG and for formulations with various components added. Depletion of DO was always

SECTION 5—4BTIER 2 RESULTS 5-25 too little or too much to determine reliable BOD5 concentrations. Results from the BOD28 tests indicated that the microorganism population eventually acclimated to DEG with accelerated biodegradation occurring approximately halfway through the 28-day test period. These results suggest that microorganism populations in receiving streams would likely acclimate to this FPD if used regularly for anti-icing purposes. Runway Deicers The effectiveness of potassium carbonate and tripotassium citrate to prevent the caking of sodium formate granules was experimentally tested. Tripotassium citrate was identified in Tier 1 as a candidate FPD having a lower oxygen demand and a lower toxicity potential than current products. Potassium carbonate was identified in Tier 1 as a potential corrosion inhibitor having a greater toxicity potential, but if used in low enough concentrations with a low toxicity FPD, the blend would likely still meet toxicity goals. It also serves as an FPD. The candidate components of the runway deicers evaluated in Tier 2 are given in Table 5-19, together with their properties. Anti-caking data on these additives for sodium formate were not found. TABLE 5-19. Candidate components and properties of runway deicers to be evaluated. Property FPD: Sodium Formate Anti-Caking Additive Tripotassium Citrate Potassium Carbonate CAS Number 141-53-7 6100-05-6 584-08-7 Formula CHO2Na C6H5O7K3·H2O CO3K2 ThOD 0.118 g O2/g 0.392 g O2/g -0.116 g O2/ga Solubility Limit in Water 46.8 wt% [1] 60.6 wt% 52.8 wt% Eutectic Temperature -23°C [1] — — aPotassium carbonate is an inorganic compound and may not exert any BOD. A testing procedure was developed to evaluate the use of additives to prevent caking for sodium formate runway deicers. The chemicals were obtained from Sigma-Aldrich. The procedure started with drying the components and mixtures of the components in a desiccator), sieving the sodium formate in an automatic sieve shaker, and weighing the individual components/mixtures. Approximately 7.5 percent by weight of the anti-caking powders were added to the sodium formate. The powders were placed in two sets of sample dishes and kept at constant temperature (30°C) and humidity (50 percent) in an environmental chamber for a period of time. After one day, one complete set of the powders were removed, weighed, desiccated, sieved and then weighed again. The second set of powders remained in the environmental chamber for another day (a total of 2 days), and the procedure was repeated as for the first day. The percentage of powder passing through the sieve was the metric used to evaluate additive effectiveness. In the first set of experiments, aluminum boats of sodium formate, sodium formate and potassium carbonate, and sodium formate and tripotassium citrate were prepared and passed though sieves to obtain their size distribution. Figure 5-12 shows the sieved weight analysis of the sodium formate (taken from the storage jar and desiccated). The anti-caking

ALTERNATIVE AIRCRAFT ANTI-ICING FORMULATIONS 5-26 additives were fine powder and passed through the smallest sieve (2.38 millimeters). Approximately 7.3 to 7.5 percent of the anti-caking agents were added to sodium formate. The boats were then placed in an environmental chamber at a temperature of 30°C and 50 percent relative humidity for 2 days. One set of boats was removed after 1 day and the second set after 2 days. The particles were agglomerated after 1 day and 2 days. After 1 day, the boat containing just the sodium formate gained 1.3 percent moisture, while the boats containing sodium formate + potassium carbonate and sodium formate + tripotassium citrate gained 8.9 percent and 3.9 percent (relative to the weight of sodium formate), moisture, respectively. After two days, the boats gained 1.4 percent, 11.7 percent and 3.0 percent, respectively; all of the materials were dissolved in water. These results were not consistent because the mixtures with the anti-caking agents gained more moisture than the sodium formate alone. After removing the boats from the humidity chamber, the boats were placed in a dessicator for 4 days until all of the moisture was removed from each boat; all of the materials were agglomerated. The sieve analysis of the sodium formate taken from the humidity chamber after 1 day, desiccated for 4 days, and then gently separated, is shown in Figure 5-12. Figure 5-12. Sieve analysis of sodium formate. 0% 10% 20% 30% 40% 50% 6.35 3.18 2.38 1.68 Bottom Sieve size, mm W ei gh t, % From jar From humidity chamber Next, the moisture pickup of individual components was determined by placing the components in the environmental chamber (30°C, 50 percent relative humidity). Table 5-20 shows the moisture pickup for each component after 1 and 2 days. After the first day, the aluminum boats containing the potassium carbonate were corroded and were partially dissolved after the second day. Further anti-caking experiments with potassium carbonate were discontinued. Tripotassium citrate was not desiccated prior to testing and lost moisture during the tests. TABLE 5-20. Moisture pickup of individual components of sodium formate and anti-caking agents. Component Moisture Pickupa (wt %) After 1 Day After 2 Days Sodium formate 1.47 3.03 Potassium carbonate 40.24 — Tripotassium citrate -1.52 -2.59

SECTION 5—4BTIER 2 RESULTS 5-27 aRelative to individual component. Further experiments with sodium formate and mixtures of sodium formate and tripotassium citrate were continued. Table 5-21 shows the composition of the mixtures that were tested. Boats 1 and 2 contained only sodium formate and tripotassium citrate, respectively. Boats 3 and 4 contained 12.5 percent and 11.2 percent of tripotassium citrate relative to sodium formate. After one day in the humidity chamber (relative humidity 50 percent; temperature, 30°C), sodium formate picked up 1.3 percent water and tripotassium citrate picked up only 0.023 percent. However, the mixtures of tripotassium citrate and sodium formate picked up from 4.5 to 7,1 percent water. Although tripotassium citrate picked up a negligible amount of water, the mixtures of tripotassium citrate and sodium formate picked up more water than the sodium formate by itself, indicating that the tripotassium citrate produces a synergistic water absorption effect. After two days, sodium formate picked up 2.5 percent of water and tripotassium citrate lost some mass, the mixtures of sodium formate and tripotassium citrate picked up about 13-14 percent water. It was hypothesized that the aluminum in the boats was contributing some exchange of the sodium or potassium ions in the FPD and anti-caking agents, enhancing the absorption of moisture. However, the experiments were replicated with plastic boats, and the moisture absorption results were similar to those found with the aluminum boats. There is no ready explanation for these observations. Further testing was discontinued. No satisfactory anti- caking solution to sodium formate was found. TABLE 5-21. Anti-caking experiments with sodium formate and potassium citrate. Boat 1, Sodium Formate 2, Tripotassium Citrate 3, KCitrate/ NaFormate 4, KCitrate/ NaFormate Potassium citrate/sodium formate, wt % 0 — 12.5 11.2 Water pickup//NaFormate + KCitrate) after one day, wt % 1.3 0.023 7.1 4.5 Water pickup/(NaFormate + KCitrate) after two days, wt % 2.5 -2.2 14.4 13.2 Incremental water pickup between first and second days, H20/NaFormate, wt % 1.1 — 8.1 9.7