Electroslag refining flux is a fused mixture of calcium fluoride, lime, magnesia, alumina, and other oxides used in metal refining to facilitate removal of impurities and prevention of oxide formation during production of high-quality alloys and superalloys. In order to calculate both CaF 2 and CaO contents in the flux one needs to know both calcium and fluoride concentrations. A method for a quick routine determination of fluoride content using potentiometry in the presence of an ion-exchanging resin has been described. Removal of calcium and other heavy metals from the sodium carbonate/borate fusate remedies the method developed by Yeager at al., and makes it usable not only to a wide variety of freshly fused fluxes but also slags and recycled flux materials. The method has been tested on several flux formulations and the results were as accurate and precise as obtained using conventional technology.
Electroslag melting of metals places great restrictions on possible suitable flux compositions. Appropriate flux compositions for melting reactive metals are thus practically limited to alkalineearth and rare-earth plus yttrium fluoride mixtures with calcium, aluminum, magnesium,
titanium, and silicon oxides. Properly composed flux has the melting point just below that of the
metal, the density of flux is lower than that of the metal at its melting point, the electrical
conductivity low enough for economical operations, and such flux should not react with molten
metal. Fluxes that decrease in CaF content have a decreasing value in electrical conductivity. Meanwhile viscosity and density begin to increase. Eutectic melting which controls freezing of flux and ingot are affected by CaF 2 concentration, melting points decrease with higher concentrations of CaF 2 2 . As shown in the Figure 1 differ+ent compositions of the tertiary fluxes (CaF 2 -CaO-Al 2 O ) lead to several regions rich in different compositions of the flux. This is why the concentration of fluoride is critical for the properties of flux. Verification of the chemical composition of fluxes is a difficult and time consuming process. The mixture of oxides, fluorides and silicates in fluxes presents analytical challenges when quantitative analyses for major constituents and trace impurities are sought. Thus, the need for rapid, accurate and cost-effective method for fluoride determination in the presence of different calcium derivatives, metal oxides (Al 2 O 3 , MgO, TiO 3 ) and silicates is of obvious importance. Several wet methods for fluorine analysis exist, but these are generally time-consuming, expensive, and operator-dependent, and require modifications to be useful for analysis of ESR flux. Modification of the FISE method  2 developed by Yeager at al  was a major step in development of a quick and reliable method for fluorine determination. In this method mixture of ethylenediaminetetraacetic acid (EDTA), ammonium citrate and ammonium tartarate is used as chelating agent in order to complex metal cations. An excess of these complexing agents masks the effect of cations on fluoride activity and drives complex formation equilibria toward coordination molecule formation. 303 Figure 1. Phase diagram for CaF 2
-Al 2 O -CaO system. After Hoyle  (Courtesy: Metals Society). 3
The method was proved to be well-suited for analysis of fluxes when the sample CaF content is < 63% by mass. Analysis of the unknown samples by this method can be problematic because in order to use it a minimum roughly qualitative knowledge of the sample composition is necessary. Magnesium also is expected to be a significant interference due to its lower stability when complexed. In our work we used different ion exchanging resins in order to decrease concentration of calcium, magnesium and other heavy metals.
Materials and Methods
In this work in progress for testing our hypothesis that use of an ion exchanging resin for cation removal expands capability of the FISE technique, five test samples were used. Two control samples containing no fluoride and maximum concentration of fluoride one can expect in the ESR flux (lime and fluorspar, respectively), and three typical non-proprietary flux compositions, containing different concentration of fluoride and calcium. The samples were characterized by LOI (Loss On Ignition), calcium, aluminum, and fluoride measurements. Concentrations of Ca, Al, Mg, Ti, and Fe were measured using ICP. The concentration of fluoride was determined by the “Foote method” . The composition of the test samples, and actual concentrations of calcium and fluoride are shown in the Table I. 304 2 All ICP control standards contained a mixture of ACS grade reagents from Alfa Aesar. 1g of each sample was fused in a platinum crucible with 2.5 g of Na 2 B 4 O 7 x 10 H O and 3.5 g of Na 2 CO . The solid matrix was then digested in diluted nitric acid then transferred to a volumetric flask and diluted up to 200ml with 18 ΜΩ DI H 3 O.
Accumet Excel 25 potentiometer/pH bench-top dual channel meter with an Accumet BNC LaF3 glass-body fluoride ion-selective combination electrode was used for the measurements. The pH meter was calibrated via three point calibration using pH 4, 7, and 10 certified pH buffers from VWR. For fluoride analysis the instrument was calibrated with five calibration fluoride standards ranging from 5.00x10 -4 M to 4.00x10 -3 M.
Thermo Jarrell Ash IRIS ADVANTAGE ICP-OES spectrometer was used for calcium, magnesium and aluminum determination. The LECO HF-400 Carbon/Sulfur analyzer was used for carbon sulfur measurements. The Alpha Resources standard AR-961 (C= 0.0194 +/- 0.0008 and S= 0.0161 +/- 0.0010) was used for the machine calibration.
The Jernkontoret JK S 10 ESR-Slag from Swedish Institute for Metals Research ( F = 34.4, CaF
(calc) = 70.7, Ca (tot) = 50.8, CaO (calc) = 23, Al 2 O 3 = 0.54, SiO = 7.8, MgO = 0.30, FeO = 0.10) was used as the certified reference material for Fluorine measurement.
Statistical tests were carried on using Pro-Stat software from Poly Software International.
2 Results and discussion
The ion exchanging resine was packed into a 5ml column and 50ml of solution prepared for the modified FISE method (4ml of the borate/carbonate fusate solution, 10ml of 2M citric acid, 20 ml of 2M ammonium citrate, 10ml of 2M ammonium tartarate, 3 ml of ethylenediamine, 25 ml of 0.1M EDTA volumized up to 100ml with 18 ΜΩ DI H2O) was quickly filtered through. The concentration of Ca2+, Al3+, Mg2+ was measured using ICP-OES. The concentration of F- was determined using modified FISE method . Two commercially available strong acid cation exchanging resins: DOWEX HCR-S and metal chelating sepharose, were tested. The better results for quick removal of unwanted cations (the concentration of calcium dropped at least 30%) without affecting the measured concentration of fluoride, and better cost/performance ratio were obtained using DOWEX HCR-S resine. The results (average values of three measurements) are shown in the Table II. 305 2
TABLE II. Removal of calcium and fluorine values no IE Dowex Sepharose Flux Ca F Ca F Ca F fluorspar 51.21 46.73 22.02 46.91 20.12 45.12 lime 69.00 -0.08 20.69 0.03 20.88 0.12 C-03 46.01 34.24 18.08 34.53 16.33 33.24 C-04 45.12 28.81 23.11 28.73 19.19 28.06 C-89 40.00 19.76 24.91 20.01 20.00 18.18
The precision of the method was determined using 32 different measurements of the fluoride concentration in the Jerncontoret S-10 with nominal concentration of fluoride 34.4%. As shown in the Figure 2 the distribution of measured values is Gaussian. The mean value obtained is 34.86% and the variance is 1.76.
Figure 2. Normal distribution of data points obtained for S-10 Jernkontoret fluoride standard with nominal concentration 34.4% of fluoride. Number of points n=32, the mean value is 34.86% and the variance is 1.76
As seen by the results, quick, cheap, and reliable analysis of the fluoride content ion ESR flux is possible using modified FISE method  for the ion exchanger filtered samples. Many samples can be run daily. Calibration of the meter takes roughly 20 min, and each sample thereafter takes only ca. 10 min to filter and analyze. Comparatively speaking the traditional method (developed by now defunct Foote Mineral Co.)  takes hours of preparation and ca. 40 min for each sample. 306 Conclusions
As seen by the results, quick, cheap, and reliable analysis of the fluoride content ion ESR flux is possible using modified FISE  method on the ion exchanger filtered samples. Many samples can be run daily. Calibration of the meter takes roughly 20 min, and each sample thereafter takes only ca. 10 min to filter and analyze. Comparatively speaking the traditional method (developed by now defunct Foote Mineral Co.)  takes hours of preparation and ca. 40 min for measurement of each sample. The new method was tested, and proved useful so far on the most popular group of fluxes containing only: CaF 2 , CaO and Al 2 O . The further tests will be done on more complicated compositions containing also Mg, Mn, Y, Zr, and Si.
Authors want to thank Joachim Rudoler for his financial support and Cathy Sullivan, Bob Martin, and Mark Scherrard for technical help.
References 1. Yeager J.L., Miller M.D., Ramanujachary K.V., “Determination of Total Fluoride Content in Electroslag Refining Fluxes Using a Fluoride Ion-selective Electrode” Industrial and Engineering Chemistry Research, 45 (2006), 4525-4529. 2. McQuaker, N.R., Gurney M. “Determination of Total Fluoride in Soil and Vegetation Using an Alkali Fusion-selective Ion Electrode Technique” Analytical Chemistry, 49 (1997), 53-56. 3. G. Hoyle , Electroslag Processes (London and New York : Applied Science Publishers, 1981), 12. 4. J. L. Krevel “Deteremination of Fluorine in Prefused Flux by Pyrolysis” Foote Mineral Company Method No. 32-10-0306, 1987). 307