White Paper ELECTRICAL CONDUCTIVITY, TEMPERATURE DEPENDENCE, & GE M9 ANALYZER Though the Conductivity is a relatively fundamental measurement both online and in the laboratory, the theory and practice of its measurement can be quite technical. This article serves to describe basic theory of Conductivity and the variables that can affect its measurement. Particular attention is given to Temperature, which is often modeled to yield a Temperature-Compensated Conductivity reading. Finally, a discussion of the specific methodology found in the GE Sievers* M9 Total Organic Carbon (TOC) Analyzer is presented. Introduction The following white paper discusses the concept of electrical conductivity in aqueous solutions, the effects of temperature on conductivity, and the methodology of conductivity measurements in the GE Sievers M9 TOC Analyzer. Electrical conductivity is a measure of a substanceβs ability to transmit electric charge. For purposes of water quality, conductivity is a result of dissolved ionic material. This is typically indicative of some sort of contamination but can also be a gauge of composition, chemistry, and consistency. Because of the relative speed and ease of measuring conductivity, it has long been a regulated measurement in many applications and is used as a first indicator of water quality changes in many others. While the measurement of conductivity, itself, is typically a stable, accurate, and reliable one, it is important to consider the effects that the environment can have. These effects can change the measured conductivity, even though the composition of the solution has not changed. Of the many variables that can affect a solutionβs conductivity, temperature is the most studied and modeled. To better compare the conductivity measurements of aqueous samples, the concept of Temperature-Compensated Conductivity was created. This allows for the comparison of conductivity measurements by using a reference temperature (typically 25 °C). To properly compensate for temperature, the ionic species must be known or approximated. In the latter case, the compensated value of conductivity to the reference temperature would also represent an approximated value. Different conductivity measuring apparatus handle compensation differently. In all cases, however, the temperature compensating algorithm is only as accurate as the constant used. In the GE Sievers M9 TOC Analyzer, the constants are fixed, which requires the operator to choose appropriate standards, control the samples to 25 °C, or use raw conductivity and temperature data to manually calculate an accurate temperature-compensated conductivity value. Cell Constants and Measuring Conductivity Electrical conductance is the inverse of electrical resistance, which is, in turn, defined by Ohmβs Law: π = π πΌ * Trademark of General Electric Company; may be registered in one or more countries. ©2016, General Electric Company. All rights reserved 300 00322 EN Rev. B Where R is resistance, V is an applied potential, and I is a measured current. Whereas resistance is measured in Ohms (ο), the SI unit for conductance is Siemens (S). Again, the reciprocal nature exists; if a sample has a resistance of 5.0 ο, it will have conductance of 0.2 S. To measure conductivity, a probe or cell is used to measure the dynamics of current and potential. Though there are many designs, most employ two or more electrodes of known dimensions across which a known potential is applied. The current is measured and conductance (resistance) is calculated per Ohmβs Law. In order to convert conductance to sample conductivity (or specific conductivity), the conductance must be normalized by the cell constant, K, πΎ = π/π where d is the distance between the electrodes and a is the area of the electrodes. Note that the units of K are [length]-1, typically given in cm-1. Therefore, whereas conductance is given as S, sample conductivity is generally expressed as S/cm. Conductivity Measurement Theory & Ions In the context of aqueous solutions, the concept of electrical conductance is simply the ability of the water to transmit an electrical charge. Empirically, it is understood that as the purity of water increases, the ability to conduct electricity is reduced. Furthermore, it is generally well understood that the presence (or absence) of ions (charged molecules) affects a solutionβs conductivity. This dependence on ions for conductivity is most commonly observed with salts.1 For example, sodium chloride is a common salt that is frequently found in aqueous solutions. The behavior of sodium chloride in water is fairly easy to describe: πππΆπ β ππ+ + πΆπ β This clearly shows how a salt dissociates into ions which, in turn, are free to carry electrical charge. The degree to which these ions carry charges is highly variable, however. It is the nature of these variables that is fundamental to understanding a sampleβs conductivity. For example, magnesium sulphate (MgSO4) is a salt that is highly soluble in water but will impart different conductivity behaviors due to the anions and cations produced upon dissociation. Without delving into the atomic and sub-atomic theories of physics and physical chemistry, the role ions play in electrical conductivity can be summarized as one of mobile facilitators. That is, the presence of ions facilitates the transfer of electrical charge through their mobility in solution. Ions are carriers of electrical charge; when a potential is applied to solution, ions increase the speed at which the electrical charge is carried. This is what we measure as conductivity. 1 A salt is an ionic substance that results from the neutralization reaction of an acid and base. Because salts are ionic solids (typically), most have some degree of solubility in water. Page 2 300 00322 EN Rev. B Once it is understood that ions, through their mobility, facilitate the carrying of electrical charge in the solution, it is imperative to acknowledge that not all ions are created equal with respect to carrying a charge. Mathematically, this can be demonstrated with the following equation: S ο½ ο¬1c1ο§ 1 ο« ο¬2 c2ο§ 2 ο« ο¬3 c3ο§ 3 ο« ο¬4 c4ο§ 4 ο« .... where S ο½ conductivity ο¬i ο½ specific conductance of the ion c i ο½ concentrat ion of ion ο§ ο½ charge of the ion The specific conductance are studied and determined empirically, rather than by theoretic modeling. The values can often be found in the literature to complete an appropriate model. Affinity, Solubility, Concentration, Mobility, Viscosity, and Temperature There are a significant number of variables that can affect a solutionβs conductivity. Detailed analyses of these variables can be found in the literature and are the subject of many peer reviewed journal articles written by esteemed researchers at major universities. The objectives here are not academic; therefore, a summarized approach is given. Though, perhaps obvious, it is important to understand that ions can be quite different from one another. πΆπ β , ππ42β , ππ+ , ππ»4+ , and others all have different properties that affect their respective abilities to carry charge and, therefore, influence sample conductivity. Different molecules have different affinities, or attraction to electrons. How tightly a molecule governs its electrons affects its influence on conductivity. Solubility and concentration also are important to conductivity. For example, sodium-based salts are typically readily soluble in water and lead to larger concentrations of ions than less soluble substances, e.g., carbonate (πΆπ32β ) salts. As concentration increases, there are more ions in solution to facilitate the carrying of charge and conductivity increases. As a general rule the smaller the ion, the greater the electrostatic field around the ion, and the slower the ionic mobility of that ion becomes. This is called the Hydration Effect and influences sample conductivity. Note that the hydration effect can be minimized by the conjugate ions a substance contributes. Potassium salts (e.g., KCl) have a much different conductivity behavior than potassium hydroxide (KOH). Sample viscosity has an inverse relationship with sample conductivity. As viscosity increases, conductivity decreases. To the extent that a substance increases the waterβs viscosity, this will decrease the conductivity that might otherwise have been expected due to that substanceβs contribution of charge-carrying ions. Temperature has a dynamic effect. Generally speaking, as temperature increases, viscosity decreases, solubility and concentration increase, and mobility increases. Nearly all of these dynamics have an overall net increase on sample conductivity. Page 3 300 00322 EN Rev. B Because temperature is most likely to change significantly, it has the greatest effect on sample conductivity measurements. For this reason, it is the most commonly studied, modeled, and discussed variable on conductivity. Temperature Correction Methodologies Modeling the temperature dependence of conductivity requires careful studies of ion-specific behaviors across a range of temperatures and concentrations. Common conductivity values for various types of water can be classified as follows: Solution Type Theoretical Limit Online DI & WFI Distilled Water Drinking Water Sea Water Typical Conductivity Value µS/cm mS/cm 0.055 0.1 - 1.0 0.5 - 3.0 500 - 2,000 0.5 - 2.0 50,000 50 Temperature affects conductivity of samples, changing by 2% per ο°C.2 For example, a 0.05 ο°C fluctuation will cause a ±0.1% change in conductivity. To account for this variation, most laboratory-based measurements are performed on samples that have been equilibrated to 25 ο°C in a water bath, or equivalent. Similarly, other instruments, including the M9, have the ability to apply an algorithm to calculate a βtemperature-correctedβ or βtemperaturecompensatedβ conductivity value.3 Typical conductivity instruments will show three measurements: ο· Raw conductivity β the actual measured conductivity value ο· Temperature ο· Temperature-compensated conductivity β the actual measured conductivity value normalized to the reference temperature, usually 25 ο°C An additional dynamic is observed with the self-ionization of water, governed by the following equilibrium: 2π»2 π β π»3 π+ + ππ» β The conductivity of pure water is given by: S ο½( ο¬H ο« ο¬OH ) K w where K w ο½ ionizationproduct water While the specific conductance of H and OH are fairly linear with temperature, the ionization product is very nonlinear with temperature. This is the main reason the conductivity of water as a function of temperature is so nonlinear. 2 Ibid. Virtually all conductivity standards are specified at 25 ο°C. For example, a calibration standard wonβt be listed as 1,409 µS/cm, such a standard will be described, marketed, and sold as 1,409 µS/cm at 25 ο°C. 3 Page 4 300 00322 EN Rev. B It also represents a temperature-dependent phenomenon that becomes significant at ultrapure levels. ASTM D11254 shows one study to demonstrate the nonlinear conductivity of pure water. The chart below shows a subset of Table 3 from the ASTM standard: ASTM D1125 Table 3: Conductivity Value of Pure Water Conductivity, µS/cm 0.20 0.15 0.10 0.05 0.00 0 10 20 30 40 50 60 Temperature, ο°C In aggregate, the dynamics discussed above (and others that are not discussed5) lead to a nonlinear relationship between temperature and conductivity. But, it can also be seen that most of the discussed dynamics are significant at relatively low conductivity values. For example, if drinking water is being analyzed at 1,000 µS/cm, then the self-ionization of water is probably not of any concern. And, like many scientific laws and governing equations, there is a linear approximation for temperature compensation. While there is no definitive criteria in the literature to guide when it is appropriate to use a linear approximation to model the temperature dependence of conductivity, the general guideline employed by GE Analytical Instruments is 10 µS/cm. However, when the substance is known (such as a calibration or verification standard) and the nonlinear compensation algorithm is available, it is always recommended to use the nonlinear model. Each model is examined in further detail below. Linear Temperature Correction and Conductivity Standards As discussed above, conductivity is dependent on temperature, which can make it difficult to compare readings of the same sample on different instruments. It complicates the calibration and verification of conductivity probes, meters, and instruments. It also explains 4 ASTM D1125 β 14 (2014), Standard Test Methods for Electrical Conductivity and Resistivity of Water, Table 3 Conductivity Values of Pure Water and Increases Due to Sodium Chloride, β2015 Annual Book of ASTM Standards, Section 11: Water and Environmental Technology, Volume 11.01: Water (I)β, pg 108. 5 For example, slightly acidic or basic salts (e.g., ammonium chloride) create a dynamic relationship between the conductivity of the water and the conductivity of the contaminating salt. NaCl is a good example of a neutral salt. This leads to a model that assumes the water conductivity and the saltβs contribution to conductivity are completely independent of one another. This assumption begins to fail with non-neutral salts and contaminants. Page 5 300 00322 EN Rev. B the temperature rating on commercially available conductivity standards. A given standard isnβt simply expressed as a conductivity value, such as 1.409 mS/cm. Rather, it is given as a specific substance at a specific temperature: 1.409 mS/cm KCl at 25.0 ο°C. The concern naturally comes from the need to verify instruments against a standard of known concentration. In practice, laboratories, instruments, and the standards themselves are typically at some temperature other than 25.0 ο°C. Unfortunately, that 1.409 mS/cm standard will read ~1.460 mS/cm at 26.8 ο°C (using an average of 2% per ο°C). This is not due to contamination or some other alteration of the standard, but due to the increase in temperature. Fortunately, the behaviors of common ions have been studied enough to be modeled.6 Though the relationship between conductivity and temperature is typically nonlinear, it may be acceptable to use a linear approximation given by the following equation: πΆπ‘ = πΆπ 1 + πΌπ (π β π‘) Where: π = π‘βπ πππ‘π’ππ π‘πππππππ‘π’ππ ππ π‘βπ ππππ π’πππ π πππππ π‘ = π‘βπ πππππππππ π‘πππππππ‘π’ππ, π‘π¦πππππππ¦ 25 β πΆπ = π‘βπ πππ‘π’ππ ππππ π’πππ πππ€ πππππ’ππ‘ππ£ππ‘π¦ ππ‘ π‘πππππππ‘π’ππ π πΌπ = π‘βπ π‘πππππππ‘π’ππ πππππππ ππ‘πππ πππππππππππ‘ πΆπ‘ = π‘βπ π‘βπππππ‘ππππ πππππ’ππ‘ππ£ππ‘π¦ πππππππ ππ‘ππ π‘π π‘βπ πππππππππ π‘πππππππ‘π’ππ π‘ Note that the temperature compensation coefficient πΌπ (the constant) is an empirically measured value that is available in literature, user manuals, and other sources. It is also important to understand that, for myriad reasons explained above, this constant is unique to every substance. Thus, there is no compensating algorithm that would be appropriate for all samples of various ions. For calibrations and verifications, the substances are known. Typically KCl is used as the calibration media, for reasons discussed in the next section. The relative purity of this standard allows the user to choose the appropriate constant for KCl and use the appropriate equation to account for temperature variations. Similarly, as instruments are challenged with known substances to calibrate or verify the instrumentβs performance, it is imperative that the proper constant be used to model the temperature dependence. For unknown samples, it becomes a problem to find an appropriate algorithm to model the temperature dependence of the compounds and ions in solution. Typically, the constant for sodium chloride is used to approximate the temperature dependence of the unknown sampleβs conductivity. Sodium chloride is a representative salt that mimics an average constant value; the constant for sodium chloride falls near the middle point of the range of common ionic constants, πΌπ . 6 Barron, J. J. and C. Ashton, The Effect of Temperature on Conductivity Measurement, Reagecon Diagnostics, Ltd. Page 6 300 00322 EN Rev. B It is due to the temperature dependence of conductivity and the ion-specific nature of this dependence that compendial methods (USP, EP, JP, etc.) typically require raw conductivity values be used or that the sample be maintained to 25.0 ο°C (thereby eliminating the need to compensate for temperature). In either case, the compendia eliminate the potential inaccuracies of the temperature compensating constants and equations by instead using the raw conductivity measured by the instrument. Nonlinear Temperature Correction Complimentary to the above chart based on ASTM D1125 is the following taken from Light et al.7: This again shows the nonlinear relationship between temperature and conductivity. Similarly, it demonstrates that below 30 ο°C, the percent change of the conductivity reading is between 5-7% per ο°C of change. Thus, a sample with 1.0 µS/cm conductivity reading at 30 ο°C will read β₯0.75 µS/cm at 25 ο°C. This is significant and the nonlinear behavior becomes important to accurately model in ultrapure water. The following chart for sodium chloride demonstrates that this general behavior is consistent for other ionic species, but that the specific model varies: 7 Light, T.S., Licht, S., Bevilacqua, A.C., and Morash, K.R., βThe Fundamental Conductivity and Resistivity of Water,β Electrochemical and Solid-State Letters, 8 (1), E16-E19, 2005. Page 7 300 00322 EN Rev. B The discussion to this point clearly demonstrates the nonlinear relationship between temperature and conductivity. Furthermore, the ion-specific nature of this relationship can clearly be seen. Thus, it is incumbent upon the user to ensure that he appropriately models the solution. For samples near the reference temperature and whose expected values are larger than 10 µS/cm, the linear relationships may be an appropriate approximation. If a nonlinear equation is necessary, this can be determined experimentally using the following methodology. For a dilute solution of a salt, the relationship between temperature, T, and conductivity, C, may be modeled as such: πΆ = (ππ (π) + ππ (π))π + πΆπ»2 π , where ππ and ππ are limiting ionic conductance of the individual ions m is the moles of salt in solution πΆπ»2 π is the conductivity of pure water By substituting 25 ο°C as the reference temperature of interest and solving the equation for conductivity at this point, the governing equation is used to define the temperaturecompensated conductivity: πΆ25 = where: ππ (25)+ππ (25) (πΆπ ππ (π)+ππ (π) β πΆπ»2 π(π) ) + πΆπ»2 π(25), 25 represents that temperature is the reference 25 ο°C T is the actual temperature of the measurement Additional detail on how to set up and conduct experiments to model a specific substanceβs temperature dependence can be found in the literature and various guides. For the purposes of this discussion and the methodology found in the GE Sievers M9 TOC Analyzer, the following were used: Page 8 300 00322 EN Rev. B ο· ο· ο· ο· Pure water: ASTM D1125-14. Other common guides include ISO 7888. Sodium chloride (NaCl): Data included in Langeβs Handbook of Chemistry8 and Light et al.9 Potassium chloride (KCl): Data included in the NIST guidance10. Note that the model here is not for dilute solutions. Hydrochloric Acid (HCl): Data included in Strongβs research.11 Conductivity Calibration, Verification, & USP <645> Per pharmacopoeia standards the cell constant of the cell must be known to within 2%.12 This is typically achieved by calibrating the instrument and then verifying against a known standard. The first step is to choose a suitable calibration point. This is the first area that can cause confusion as there are two seemingly contradictory guidelines13: 1. Calibrate at a nominal value of 1.409 mS/cm for βgeneral useβ 2. Choose a calibration point near the point of interest To understand the nature of the seemingly odd 1.409 mS/cm standard, an analysis of the NIST primary standard is required.14 Potassium chloride (KCl) is the standard solute for conductivity standards due to its stability and ease of purification. Standard concentrations are generated in terms of molality15, wherein 1.0, 0.1, and 0.01 m solutions are generally used. The 0.01 m standard yields a 1.40933 mS/cm response. By adhering to the guidance in NIST, calibration standards at 1.409 mS/cm (or 1,409 µS/cm) provide traceability.16 As noted above, the regulatory limits are in the 1.0 β 5.0 µS/cm range and the points of interest for most life science customers are in the 0.5 β 3.0 µS/cm range. This is three orders of magnitude lower than the 1.409 mS/cm calibration point discussed in the previous paragraph. Common sense prevents us from reconciling those two guidelines, unless one takes a liberal view of βnearβ; most would conclude that three orders of magnitude represents a wide gap 8 Langeβs Hanbook of Chemistry, 16th Edition, βConductance.β Light, T.S., S. Licht, A.C. Bevilacqua, and K.R. Morash, βThe Fundamental Conductivity and Resistivity of Water,β Electrochemical and Solid-State Letters, 8 (1), E16-E19, 2005. 10 NIST Special Publication 260-142, 2004 Ed. Standard Reference Materials: Primary Standards and Standard Reference Materials for Electrolytic Conductivity, Shreiner, R. H. and Pratt, K. W. 11 Strong, L.E., βAqueous Hydrochloric Acid Conductance from 0 to 100 ο°C,β J. Chem. Eng. Data, 1980, 25, pg 104106. 12 Per USP and EP. Other pharmacopoeias may have different criteria, such as CP at 5%. 13 Guidelines are suggestions and appear in myriad publications, including regulatory documents such as USP. Guidelines are not prescriptive, but, rather, suggest generally accepted good practices. Adherence is typically not mandatory or enforced. 14 Ibid. 3, pg 13-24. 15 Molality is defined by the number of moles of a solute divided by the mass of the solvent. In this case, the moles of KCl divided by the mass of the water in which the KCl was dissolved. 16 Some offer standards that are 1.408 mS/cm, but this value discounts the conductivity contribution from the solvent (water). 1.409 mS/cm is a better descriptor of the conductivity of a 0.01 m KCl in water solution, per Table 6 in the NIST document. 9 Page 9 300 00322 EN Rev. B between two points. Rather, it is necessary to revisit the effects of creating calibration standards near these low points of interest. NIST specifically addresses this in §6.3 Low Conductivity Standards.17 In this section, NIST addresses inherent variation in a 5.0 µS/cm standard produced with KCl and water. Its solution is to create a solvent that is 30% n- propanol. This alcohol-water solution lowers the effect of CO2 on the stability such that a ±0.1 µS/cm (2%) variability is observed theoretically.18 However, before using a 5.0 µS/cm calibration standard, itβs important to understand the effectof the standardβs variability on the resulting calibration curve. Whereas 5 µS/cm standards have uncertainties in the 2 β 10% range, typically the 1.409 mS/cm have variability in the 0.01 β 0.25% range. Gingerella and Jacanin19 explore the real-world performance of βlow-conductivity standards,β defined as 10 µS/cm. While advances have been made in the years since this study, the conclusions are still generally valid today. Specifically, the authors cite several sources of variability: ο· Atmospheric CO2 via the following reaction describing the dissociation of the carbon dioxide molecule: πΆπ2 + π»2 π β π»2 πΆπ3 β π» + + π»πΆπ3β The presence of the proton and bicarbonate ions raises sample conductivity by as much as 1.05 µS/cm.20 ο· ο· ο· ο· ο· ο· Container surface contamination (both the vessel in which the standard is packaged and the vessel from which the measurement is made) Sample handling Manufacturer quality control Container head-space Tolerances that exceed scientific theoretical limits Tolerances that exceed the ability of the instruments purported to measure them Ultimately, many of these standards show the theoretical tolerances at the time of manufacture and do not correlate to tolerances the user can expect to see when actually measuring them. This leads to difficult-to-use and difficult-to-validate standards. 17 Ibid. 3, pg.21. Ibid. 19 Gingerella, M. and Jacanin, J.A., Is There an Accurate Low-Conductivity Standard Solution? The International Journal of Metrology, July-August 2000, pg 29-36. 20 Ibid. 18 Page 10 300 00322 EN Rev. B From authors of papers, to conductivity measurement equipment manufacturers21, to regulatory bodies22, there is overwhelming caution against using low-level standards. With that said, and as mentioned above, there continue to be advances made in all technological fields, including low-level conductivity standards. There are a number of manufacturers who develop novel and oft-patented systems that purport to maintain the integrity of low-level standards. Though not endorsed or used by GE Analytical Instruments, Aurical Company is one such manufacturer. The table below shows how a 1.409 mS/cm standard compares to three standards by the Aurical standards and the theoretical NIST reference material of 5 µS/cm ± 2%. Standard Nominal 1409 10.0 5.0 1.0 5.0 Tolerance per Actual Values Reading at 1.0 mS/cm CoA Lower Upper Lower Upper 0.25% 1405.478 1412.523 0.9975 1.0025 5% 9.5 10.5 0.9500 1.0500 10% 4.5 5.5 0.9000 1.1000 20% 0.8 1.2 0.8000 1.2000 2% 4.9 5.1 0.9800 1.0200 23,24 As is easily seen in the table, using an accurate calibration standard three orders of magnitude above the point of interest yields a much more accurate measurement than do the best low-conductivity standards. This may seem counterintuitive, but it illustrates two points that govern nearly all analytical instruments: ο· Calibration errors propagate to every additional measurement; the accuracy, stability, and precision of calibration standards is critical in defining instrument performance. ο· Calibrating near the point of interest is not necessarily as important as maximizing the calibration stability itself. This second bullet-point must be explored a bit further, as it assumes linear behavior. In some instruments, the linearity of the measurement technology would not support interpolation between the origin and a point three orders of magnitude above the point of interest. Thus, the language was chosen to provide provisions where calibrating at or near the point of interest might be of paramount importance. The operator can quickly ascertain whether a given instrument is linear in nature by verifying that instruments performance. It 21 E.g., Thornton (Mettler Toledo) βTherefore more accurate results are normally achieved through verification and/or calibration using 25 or 100 ΞΌS/cm standards than could be obtained using standards with lower values.β 22 E.g., ASTM Standard Test Method D5391 β 14 (2014), Standard Test Method for Electrical Conductivity and Resistivity of a Flowing High Purity Water Sample, section 10.4, states that βcell calibration with standard solutions below 100 µS/cm is not recommended.β 23 The βTolerance per CoAβ is the stated variation or uncertainty on a standardβs certificate of analysis. The βActual Valuesβ define the expected range of values based on the standardβs nominal value and the tolerance specification. 24 The bottom shaded row is the theoretical NIST standard described in the literature. Page 11 300 00322 EN Rev. B will be demonstrated below that, by employing patented technology,25 the M9 possesses such linearity, lending validity to the conclusions made above. Conductivity Verification & Linearity While the above section concludes that a calibration with highly stable and accurate 1.409 mS/cm can be advantageous, this is only true if the linearity of the instrument can be demonstrated. For the purposes of verification, lower level standards are often sought. Verification standards can accommodate more variability because that variability is specific to the measurement of that a single sample. Thus, the variability does not propagate through successive measurements the way a calibration standardβs error will. Perhaps the most common verification standard is the 25 µS/cm HCl standard. By creating an acidic standard, the pH is lowered and the chemical reaction below demonstrating the dissociation of CO2 to bicarbonate and hydronium ions is affected. πΆπ2 + π»2 π β π»2 πΆπ3 β π» + + π»πΆπ3β pH < 4 pH > 4 This invokes Le Chatelierβs principle and shifts the equilibrium in favor of gaseous or dissolved CO2 molecules. Without this dissociation, atmospheric CO2 no longer poses a contaminant and the relative accuracy of the 25 µS/cm HCl standard is maintained. Unfortunately, at lower concentrations, the acidity of the standard and the resulting pH reduction are not enough to prevent the dissociation of CO2 and the contamination of the standard. Thus, the reliability of low HCl standards (β€ 10 µS/cm) is no better than KCl-based standards. Nevertheless, the 25 µS/cm HCl standard is quite effective in demonstrating linearity of the calibration to two orders of magnitude lower than the calibration point of 1.409 mS/cm. This accommodates the need to show linearity and the natural conclusion is to expect that, provided the calibration is linear for the first two order of magnitude (1,409 to 25 µS/cm), then it should maintain this linear behavior for the final order of magnitude (25 to 1 µS/cm). The table below shows the verification results against GE Analytical Instrumentsβ 25 µS/cm verification standard, as well as three Aurical standards. These data were collected at a customer facility in an open environment. The Theoretical Range is defined as the specified variability of the standard itself plus the 2% tolerance in the calibration, per USP and EP. There are no allowances in the Theoretical Range for sources of contamination discussed above or any additional influences in the measurement. 25 Patent US 9116099 B2 βWide Dynamic Range Conductivity Measurement in Waterβ Page 12 300 00322 EN Rev. B Theoretical Measured Standard Range Value 25 24.5 - 25.5 24.62 10 9.31 - 10.71 10.72 5 4.41 - 5.61 5.45 1 0.784 - 1.53 0.92 The data clearly show that the measurement response of the M9 with a 1.409 mS/cm calibration is quite linear through the point of interest, 1.0 µS/cm.26 Complementing the above data is a controlled study performed by GE Analytical Instruments. Using commercially available 100 µS/cm sodium chloride (NaCl) standards with carefully controlled serial dilutions, the following data were collected: Non-Blank Adjusted, Non-Temperature Corrected Actual Measurement, µS/cm 120 100 80 60 y = 1.0911x + 0.6508 R² = 1.0000 40 20 0 0 20 40 60 80 100 120 Theoretical Standard Value, µS/cm This shows a high degree of linearity down to 1.00 µS/cm based on a calibration point of 1.409 mS/cm. Note that the slope of this curve suggests that the data are not accurate, showing in excess of 9% variation between theoretical and actual. This can easily be explained by the serial dilutions and the temperature effects. The dilution water used contributed some amount of conductivity to the measured standard. Furthermore, the standard is certified to a reference temperature of 25 ο°C and the temperature of measurement was elevated. If the data are corrected for the dilution water contribution and the nonlinear temperature compensation for NaCl, the chart becomes more reflective of the accuracy capability of the M9: 26 Note that response associated with the 10.0 µS/cm standard was 0.01 µS/cm out of the Theoretical Range, but this can be explained by any number of contamination sources described above. Page 13 300 00322 EN Rev. B Blank Adjusted, Temperature Compensated Actual Measurement, µS/cm 120 100 80 60 y = 1.0102x + 0.0475 R² = 1.0000 40 20 0 0 20 40 60 80 100 120 Theoretical Standard Value, µS/cm The primary purpose of these data is to demonstrate linearity between the calibration point and the expected point of use for ultrapure water analysis. This linearity is demonstrated in both plots. Furthermore, the accuracy of the GE Sievers M9 Analyzer is also demonstrated after correcting for the dilution water (blank) and the temperature. Temperature Compensation in the GE Sievers M9 For many instruments in the commercial marketplace, the user must select and enter an appropriate constant. These constants are sometimes provided by the manufacturer, but can also be also be found in the literature and using online search tools. The M9 currently does not have a user-selectable temperature compensation feature. The reference temperature is fixed to 25 ο°C and the temperature-compensating algorithms are nonlinear and fixed to the following substance type: Operating Mode Sample Conductivity Calibration Sample Conductivity Verification Measurement Mode Constant Used KCl HCl NaCl For this reason, the M9 cannot be used to verify samples with known quantities of alternative ions. When the M9 is in normal measurement mode, it will apply a temperature compensation algorithm appropriate for NaCl. If, for example, an HCl standard were being used and the temperature were not 25 ο°C, the correction algorithm applied would not yield an accurate result. A graphical representation of the linear temperature-compensated conductivity readings are shown below: Page 14 300 00322 EN Rev. B As noted, as the actual temperature moves away from the reference temperature, the measured value of conductivity will be quite different. Therefore, if a conductivity value is measured and an improper constant used to compensate that value to a reference temperature, it is easy to see how this will lead to an inaccurate compensated value. USP <645> and Equivalent Standards Though there are many global standards (such as USP, EP, JP, other pharmacopoeias, ASTM, etc.), many have standardized on the methodology documented in USP <645> Water Conductivity. Many of these standards and compendia have been harmonized, an effort to ensure the same procedures and criteria are present in each document. While this has helped reduce complexity for customers, some of these standards and compendia may have slight differences between them. A focus on USP <645> follows, so be certain to check with your local regulations before proceeding with your testing regime. USP <645> contains three types of conductivity testing: Stage 1, Stage 2, and Stage 3. Stage 1 is the simplest method, but contains the tightest pass/fail levels. The user need only to measure temperature and conductivity online or in βa suitable container.β Note that the definition of βsuitable containerβ is left unspecified and, therefore, may be defined by the customer. There are no other requirements or prescribed actions for Stage 1 measurement, save that the reported value of conductivity be the raw measurement and not a temperaturecompensated value. The pass/fail criteria are then dictated by the abridged table below: Page 15 300 00322 EN Rev. B (The table continues to temperatures of 100 ο°C in the USP monograph.) The pass/fail criteria for nominal sample temperatures are in the 1.0 β 1.4 µS/cm range, which again underscores the importance of minimizing contamination, including atmospheric CO2. If the customer fails Stage 1 conductivity, he must proceed to Stage 2. Stage 2 is much more prescriptive about the testing protocol. The sample must be β₯ 100 mL, be kept at 25 ± 1 ο°C, be agitated, and the conductivity measurement must be stable within 0.1 µS/cm over a five minute period. None of these requirements or restrictions existed in any form for Stage 1. The pass/fail limit for Stage 2 is 2.1 µS/cm. Note that this is nearly double the pass/fail limit at 25 ο°C for Stage 1. The customer must devote quite a bit more time and effort to measure with Stage 2, but there is a wider acceptance criterion. If Stage 2 conductivity fails, the customer adds KCl to the sample, measures pH, and compares conductivity to a new pH vs. conductivity table, per Stage 3. Pass/fail criteria are as high as 4.7 µS/cm. Note that while conductivity calibration standards are specified (and discussed below), temperature must only be verified and, as a limit test, there is no required operational and performance qualifications required. Mechanics of Measurement in the M9 The measurement cell in the M9 comprises gold electrodes mounted in a quartz cell with a thermistor at the exit end of the cell. There are a number of proprietary electronic algorithms and methodologies applied to the measurement module that yield the linear behavior characterized above. The conductivity measurement is performed upstream of the TOC measurement and prior to the addition of any reagents for TOC measurement; the conductivity cell is the first component in the sample flow path. During the normal mode, a measurement cycle lasts two minutes. For the first 30 seconds, a conductivity measurement is made ten times per second; thus, 300 measurements are Page 16 300 00322 EN Rev. B made.27 The reported value is the average, or mean, of these data. After careful analysis of many data sets, the mean was selected as the best indicator of the central tendency. Suitable Sampling Vessels The primary value proposition behind the sample conductivity option is the ability measure both TOC and conductivity simultaneously. This creates a unique challenge in that TOC needs to be measured from a glass container, but glass containers typically leach ions into water and are unsuitable for conductivity measurement. To overcome this challenge, GE Analytical Instruments has developed Dual Use Conductivity and TOC (DUCT) Vials. These vials have a proprietary coating that seals the glass and protects the integrity of the water samples with respect to both TOC and conductivity. 28 When using the M9 for sample conductivity, the DUCT Vials must be used. Summary and Conclusions Conductivity is a complex measurement that is dependent on a number of variables. Though its relationship with temperature is understood, it is a unique relationship for each ionic species. For compendial reporting purposes, temperature-compensated conductivity is either prohibited or effectively eliminated by controlling the samples to the reference temperature of 25°C. For calibration and verification, the standards contain a known, purified substance that can be adequately modeled to provide accurate and precise results. However, it is imperative that the proper constants be used when modeling these substances. The GE Sievers M9 TOC Analyzer does not yet allow the user to enter his own constant and instead relies on fixed constants that vary per operating mode. Users must either conform to the proper ionic species that match the constants, control their samples to the reference temperature, or conduct manual calculations based on the given equation and the appropriate constants. 27 In the Turbo mode, the measurements are made over a three second period; 30 measurements are taken, but the remainder of the methodology is unchanged. 28 See GE AI application note on Dual Use Conductivity and TOC Vials for more information. Page 17 300 00322 EN Rev. B
© Copyright 2026 Paperzz