Initial instrument calibration requirements

Initial instrument calibration requirements. (1) General provisions. (a) The laboratory shall calibrate or verify the calibration of all analytical instruments before the instruments are used to provide any quantitative results. (b) Once a calibration model is selected, a calibration function is established, and an initial calibration is finalized, a laboratory may not change the model or calibration function after samples have been analyzed without performing another initial calibration. (c) The laboratory shall perform an initial calibration if any of the following apply: 1. After instruments undergo non-routine maintenance. 2. Conditions change the expected behavior of the instrument. 3. When a CCV standard fails and any of the following occur: a. Corrective action taken does not result in a passing CCV standard. b. A second consecutive (immediate) CCV standard is performed under the same conditions and it also fails and the corrective action taken does not result in two consecutive passing CCV standards. (d) The laboratory shall retain all the raw data necessary to reconstruct or reproduce calibration functions associated with initial calibrations. (e) For colorimetric technologies, the laboratory may not use a method blank to zero the instrument. (f) The laboratory may not utilize pre-programmed initial calibrations, provided by the instrument manufacturer, for compliance testing. (g) The laboratory shall include or reference the details of initial instrument calibration procedures including algorithms, any required equations, and acceptance criteria in the method standard operating procedure. (h) When required by method, the laboratory shall process each calibration standard in the same manner as samples. (i) Point-to-point calibrations are not allowed unless otherwise specified in this chapter. (2) Minimum number of standards. To establish calibration, the laboratory shall select the number of nonzero standard concentrations that is appropriate for the calibration model selected and the expected range of concentrations. If a method requires analyzing more than three standards to establish a linear calibration, and the laboratory chooses to narrow the calibration range of the determination to no more than two orders of magnitude, the laboratory may use 3 standards to generate the initial calibration. The minimum number of nonzero standard concentrations selected to establish calibration shall be three except for all the following: (a) Dissolved oxygen meters, for which the minimum shall be one. Dissolved oxygen meters shall be calibrated against water-saturated air or air-saturated water at a known temperature and pressure. Alternatively, calibration may be performed using an iodometric method. (b) Conductivity meters, for which the minimum shall be one. Conductivity meters shall be calibrated by verifying the cell constant or adjusting the meter based on the analysis of a potassium chloride standard solution. (c) Inductively coupled plasma emission spectrophotometers and inductively coupled plasma mass spectrometers, for which the minimum number shall be one. (d) pH meters, for which the minimum number shall be two. (e) Quadratic calibration models, for which the minimum shall be five. (f) Cubic calibration models, for which the minimum shall be seven. (3) Concentration levels of standards. The concentration of the standards chosen to establish a calibration function shall be within the same orders of magnitude as the expected concentration of samples. (4) Calibration models. The laboratory shall select a calibration model that is appropriate for the expected behavior of the analytical instrument to be calibrated. To generate a calibration model, the laboratory shall select a reduction technique or algorithm that is appropriate for the calibration model and the number of nonzero standards used, subject to all the following: (a) The selected algorithm or reduction technique shall be describable mathematically and shall provide equations, coefficients, or parameters necessary to characterize the calibration function uniquely, unless an analytical instrument is tuned to conform to a universally accepted scientific law or scale. (b) Non-linear functions may not be used to compensate for instrument saturation, insensitivity, or malfunction. (c) The laboratory may use weighted algorithms, unless the weighted algorithms are chosen to compensate for deviations from the expected behavior of a detector of an analytical instrument resulting from saturation, insensitivity, or malfunction. (d) Except for methods that allow average response factors and average calibration factors, the laboratory may not use reiterative reduction techniques or algorithms that force calibration functions through zero. (5) Excluding calibration points. If one or more calibration standards are excluded from the calibration, all the following criteria shall be met: (a) The rationale for the exclusion is documented. (b) Any required regulatory limits can still be met. (c) Except for ICP, ICP/MS, and HRGC/MS, if the highest calibration standard is removed, the linear range shall be limited to the remaining high standard concentration. (6) Evaluating algorithm validity. The laboratory shall establish acceptability criteria for initial calibrations. The type of criteria chosen, and the acceptance range shall be appropriate for the type of analytes to be quantitated, the calibration model selected, and reduction technique or algorithm chosen. Acceptability criteria shall be established using any of the following: (a) When the x-intercept is used to evaluate the calibration, then the value of the x-intercept of the calibration function for each analyte may not exceed its LOD. (b) Unless otherwise specified by the method, when RSE is used to evaluate the calibration, the relative standard deviation may not exceed 15% for inorganic analytes or 20% for organic analytes. (c) Unless otherwise specified by the method, when residuals of each calibration standard are used to evaluate the calibration, the standard recovery for all but the lowest calibration point shall fall within 90% to 110% for inorganic analytes or within 70% to 130% for organic analytes. Recovery for the lowest calibration point shall be within 80% to 120% for inorganic analytes or 50% to 150% for organic analytes. (d) When average response factors are used to reduce calibration data, the relative standard deviation of the response factors may not exceed 20% unless the method allows a larger percentage. (e) When linear regression or least squares analysis is used to reduce calibration data, the correlation coefficient (r) of the resultant calibration shall be at least 0.995 for inorganic analytes or 0.99 for organic analytes. (f) When quadratic (2 nd order) or cubic (3 rd order) analysis is used to reduce calibration data, the coefficient of determination (r 2 ) of the resultant calibration shall be at least 0.995 for inorganic analytes or 0.99 for organic analytes. (7) Verifying accuracy. Except for calibrations generated using dissolved oxygen meters, pH meters, or conductivity meters, the laboratory shall verify all initial instrument calibrations after the calibrations are generated, but before the calibrations are used to quantitate any samples, with a second source standard, referred to as an ICV standard. ICV standards shall be treated in the same manner as the standards analyzed for the initial calibration. Unless otherwise required by method, regulation, or covered program, the acceptance criteria for the ICV standard shall be all the following: (a) Obtaining concentrations within 10% of the theoretical concentrations of all reportable inorganic analytes. (b) Obtaining concentrations within 20% of the theoretical concentrations of all reportable organic analytes. (8) Evaluating sensitivity. When methods require an ICB be analyzed after the initial calibration, the ICB shall be treated in the same manner as the initial calibration standards. The concentration of an analyte in an ICB may not exceed its LOD.