Calibration of a Reference Standard is "the determination of one or more physical, chemical, biological, or engineering properties related to the intended use of the Reference Standard.
As a measuring instrument, a Reference Standard has the function of preserving, reproducing and transmitting quantitative values. Whether the characteristic value of the standard substance is comparable and reliable in different time and space depends on whether the fixed value measurement of this standard substance establishes traceability. Validation is the process of assigning a value to a Reference Standard, which is also a key link in the process of identifying a Reference Standard. The process of Reference Standard valuation requires more stringent quality assurance measures and requirements (see ISO/REMcO Guide 34 and ISO/REMCO Guide 35).
Accuracy/uncertainty, comparability, traceability
When applied to a range of measurements. The term "accuracy" is a combination of a random component (reproducibility) and a systematic component (correctness). In ISO/IEC Guide 99 (2007), "International Vocabulary of Weights and Measures - Basic common concepts and related terms", accuracy is defined as the degree of agreement between the measured value of a measurement and its true value. Since accuracy is not a quantity, it cannot be given as a numerical value. A measurement is said to be accurate when it provides a small measurement error. In addition, with the widespread international acceptance of the "Guidelines for the Expression of Uncertainty in Measurement (GUM)", the positive term "accuracy" is being replaced by the negative, but clearer, term "uncertainty".
The positive term "accuracy" is being replaced by the negative, but clearer, term "uncertainty.
Uncertainty is defined as "a parameter that characterizes the dispersion of values reasonably assigned to a measurement, and is associated with the result of the measurement". If the above definition is linked to the definition of measurement traceability to analyze the meaning of the definition, the following information can be derived: dispersion is a property of the measurement result, uncertainty is the quantification of dispersion; the measurement result is incomplete without uncertainty; and traceability cannot be achieved without uncertainty.
Both the procedure described in GUM and the state-of-the-art determination procedure emphasize the word "reasonably". There is no point in characterizing a quantity with a very small uncertainty when there is a greater risk that the true value will exceed the uncertainty range with a given probability; on the other hand, there is no point in artificially expanding the uncertainty range for safety reasons alone. If a confidence level of 95% is given for uncertainty, but 99.9% is actually represented, the user of a CRM will introduce an unreasonable uncertainty component into his or her uncertainty calculation.
One of the main purposes of using CRMs is to improve the comparability of measurements made in different laboratories. Reference to a recognized standard for chemical composition or to a measurement standard traceable to a higher level of recognized standards is a powerful tool for obtaining comparability within the framework of international guidelines or among trading partners.
The definition of comparability is inconclusive and ranges between the following statements: "The ability of measurements to be compared for the purpose of determining whether they are equal or different (greater or lesser). This is obtained by expressing measurements on the same, preferably internationally recognized, measurement scale."
And the characterization of measurement results, obtained from subsamples of the same material, when the results are measured on the same scale of measurement (i.e., expressed in the same units), agree within their uncertainties.
In a qualitative sense, the latter definition is preferred by most when considering bilateral mutual recognition of analytical measurement results. The basis for comparability of measurement results is to achieve traceability of measurements. The term is defined as follows, "The property that enables the results of a measurement or the value of a measurement standard to be related to a defined reference standard, usually a national or international measurement standard, by means of an unbroken chain of comparisons with specified uncertainties."
The traceability of a recognized value is mentioned in the Certified Reference Material (CRM) as an important requirement for the identity of the substance. But measurement traceability is largely not just a matter of value, but a fundamental prerequisite for establishing comparability of measurement results.
For high-level certified reference materials (CRMs, or "Reference Standards"), it is common for the process of determining the value of a measurement to be made without reference to other CRMs, but rather to be traceable directly to an sI unit such as a kilogram (kg) or mole (moL), or to a written standard describing the method of measurement, i.e., a method that has been systematically researched and accurately described. systematically studied and accurately described. Direct traceability to sI units of kilograms (kg) or moles (moL) is achieved by introducing a weighed amount of a sufficiently pure element or compound as a calibrator in the analytical procedure. Thus, calibration is an important part of the determination procedure, but by no means the only important part.
The question of how much of a pure element or compound can be weighed before it can be considered sufficient to "accurately reproduce the unit of identity" continues to be discussed among metrologists and analytical chemists.
The use of CRMs is the most important tool for users to make their measurements traceable, and they can even extend the chain of traceability down the chain of comparisons by referring to a CRM when assigning a value to a secondary standard, a working standard, or a quality control material (QCM).
Models of Reference Standard Valuation Measurement Methods
The valuation measurement methods used in the determination process shall be those that have been tested and confirmed to be accurate and reliable, both theoretically and practically. Before starting the fixed-value measurement, the measurement method, the measurement process and the sample handling process inherent and systematic errors and random errors, such as dissolution, digestion, separation, enrichment and other processes of the measured sample staining and loss, the measurement process of the matrix effect, etc., the measurement instrument should be calibrated regularly, the use of traceable reference reagents, there should be a feasible quality assurance system to ensure that the measurement results of the Traceability.
Based on the type of certified reference material (CRM) and its potential use, the capability of the fixed-value measurement laboratory and the quality of the measurement method used, one of the following method modes can be selected for fixed-value measurement.
(1) Benchmark (Authoritative) Measurement Method ModeWhen the uncertainty of a fixed-value measurement must be narrowed, a single, so-called benchmark analytical measurement method may be used to perform the fixed-value measurement. If the fixed-value measurement is undertaken by only one laboratory, it is recommended that another non-benchmark method be used for error checking.
The systematic error of a benchmark (absolute or authoritative) measurement method is estimable, and the level of relative random error is sometimes negligible. Measurements require two or more analytical workers to operate independently and to use different experimental setups wherever possible, with comparisons of quantities where available.
But the definition of the term "primary method of measurement" (PMM) and the identification of measurement methods that meet the requirements of the definition are far from complete.
As mentioned earlier, today, the International Committee for Weights and Measures/Consultative Committee for the Quantity of Matter (CIPM/CcQM) publishes the following definition: "A primary method of measurement is a method of measurement of the highest metrological quality, whose operation can be fully described and understood, whose final uncertainty can be expressed in sI units, and whose results do not depend on the measured measurement standard.
Explanatory notes:
1) Benchmark measurement methods also require the establishment of equations between the (direct) object to be measured and the (indirect) target object to be measured that do not contain any meaningful empirical correction factors.
② For a measurement of a quantity of a substance to be benchmarked, a method must be used that is specific to the specified substance, as well as relying on the fact that all parameter values or corrections for other substances or matrices are known or can be calculated with approximate uncertainty.
The most important analytical measurements considered by the International Advisory Committee on the Quantity of Substances (CCQM) for the measurement of quantities of chemical constituents are: isotope dilution mass spectrometry (IDMS); gravimetry; titrimetry; freezing point depression; and electrometric (coulometric) methods.
Also proposed in 2013 as potential benchmark measurements were: Cavity Ring Down Spectrometry, Instrumental Neutron Activation Analyses (INAA - Instrumental Neutron ActiVation Analysls).
The International Advisory Committee for the Quantity of Matter (CCQM) also defines the new term "Reference Standard": "A Reference Standard is a standard of the highest metrological quality for the determination of quantities by means of reference measurements."
The use of reference (authoritative) measurements for the characterization of chemical composition is metrologically sound, but very few such measurements meet the requirements of the definition. When benchmark measurements are used, the valuation of a standard is often done independently by a single laboratory. Today, the weight, volumetric, electrical (coulometric), freezing point depression, and isotope dilution mass spectrometry (IDMS) methods mentioned above meet the definitional requirements and can be used for the measurement of quantities of chemical constituents, and are considered to be potential reference measurements. The accuracy of these methods is very high, while the systematic error is very low or even close to zero. Therefore, they are able to measure the value of a parameter within a limited range of uncertainty. For example, isotope dilution mass spectrometry is able to overcome many problems in the accurate determination of trace elements. It is able to compare the number ratios of isotope atoms of different masses while not requiring quantitative separation of the sample. Theoretically, the results obtained are directly traceable to the basic unit mole. Labeling the sample with isotopically labeled elements or compounds, and then trying to homogenize the isotopes, enables mass spectrometry to be used to determine the ratio of analytes in the sample without any matrix action.
In practice, however, for chemical composition measurements, most baseline measurement methods are dependent on certain measurement standards and they can be used in relatively few situations. In addition, they are quite time-consuming and costly. As a result, the scope of use of benchmark measurement methods is somewhat limited. However, if they are implemented correctly, they can produce measurements of fairly high quality.
In the development of Reference Standards, the gravimetric method is generally preferred, if feasible, for the determination of chemical composition, but unfortunately this is not often the case. Typical examples of gravimetric applications are the preparation of gas mixtures by gravimetric methods, the use of pure material certified standards or analyte certified standards by the quantitative method to accurately formulate a well-dispersed matrix certified standards or calibration solutions. Traceability of gravimetrically prepared CRMs can be established by tracing the weighing measurements to national quality standards, to the atomic/molecular weights of the weighed components and to their purity. Traceability to national quality standards is theoretically relatively easy to achieve. It should be of good stoichiometric quality, but in practice, due to unknown instabilities, segregation, adsorption or other unknown factors, the prepared mixture may still differ significantly from the expected composition. Therefore, when using the gravimetric method to perform a calibration measurement on a standard substance, another method is usually chosen to further validate the resulting calibration results.
(2) Multiple method models that are independent of each other
When one or several organizations have many different analytical measurement methods at their disposal, and the following conditions are met, it is possible to value a Reference Standard on the basis of the results of only two to four laboratories/methods: the analytical measurement method has been fully validated; the validation covers a range of values for sample constituents similar to those of the CRM candidate; and the validation covers a range of values for sample constituents that are similar to those of the CRM candidate; and the validation covers the range of values for sample constituents similar to the CRM candidate. range of sample constituents similar to the CRM candidate constituents; the personnel applying the methods are well trained; the methods used were selected with an eye to minimizing ***identical*** sources of uncertainty; and there is good agreement between the various method measurements.
The implementation of this model requires a wealth of measurement experience accumulated over a long period of time by the participants in fixed-value analytical measurements, good technical reserves and documented technical experience, as well as an in-depth knowledge of the condition of the measuring instruments involved to ensure it.
Individual laboratories may use independent measurement methods to characterize pure substances of a single composition, such as certified standards of pure pesticides or solutions requiring gravimetric preparation for analysis. In such cases, it is desirable to use two or more well-recognized analytical measurements based on different fundamentals, and if the results agree within the total uncertainty, either method used can be considered to be unbiased. The mean value can then be considered a reliable estimate of the true value. For example, a deemed fixed value measurement for a pure pesticide using gas chromatography, high-pressure liquid chromatography, and differential scanning calorimetry was performed at the same time. In this case, the method used should have a small measurement uncertainty relative to the intended use of the certified standard substance.
Interlaboratory comparison studies
This is a widely used concept in which multiple laboratories independently perform a series of measurements of one or more quantities on a given sample. It is sometimes referred to as a "round robin", "cooperative research program", or "collaborative analytical study". Such "studies" can be used for many purposes other than standardization.
In some analytical testing applications, inter-laboratory comparison (collaborative) studies (laboratory consensus methods) are the most commonly used method of Reference Standard valuation measurements, especially those using natural matrices. When using this approach for the valuation of a CRM, the Accreditation Body must first find a group of qualified laboratories with expertise in the measurement of similar analytical materials to the CRM candidate.
Secondly, if possible, the body must also encourage the laboratories involved in the interlaboratory comparison (collaborative) study to use as many independent analytical methods as possible. These measurement methods should not only differ in operational details, but should be based on different principles. Doing so puts unknown potential systematic errors causing positive or negative biases on the same level. Another advantage of this approach is that when there are 10 to 18 participating laboratories, outliers in the measurement results can be identified and rejected for treatment through technical review.
Thirdly, it is advisable for the accreditation body to conduct a round of inter-laboratory pre-comparison studies with another sample prior to the formal calibration measurements, in order to ascertain that the selected laboratories are indeed at the same level of measurement capability. It is important to note that, in order to obtain high quality calibration measurements, the sample matrix material used in the pre-comparison study should be as consistent as possible with the target Reference Standard candidate. The measurements from the pre-comparison study are used for comparison to verify that all results obtained agree within their measurement uncertainties. If the dispersion of measurement results from this round of inter-laboratory comparison studies is large, it indicates that a large systematic error has occurred, which does not satisfy the requirement to assign a value to the certified Reference Material (CRM), and the necessary measures should be taken to eliminate the systematic error prior to performing the formal assignment measurements. Usually, in order to eliminate systematic errors, all laboratories should participate in the seminar to find out the causes of errors and solutions, and organize another round of laboratory comparison studies, until the consistency of the measurement results of the comparison studies can prove that the measurement capability of the selected laboratories is at the same level and meets the requirements for the calibration measurement of the target CRMs, then the calibration measurements can be organized in the same form.
Additionally, it should be noted that, in the case of sufficient number of laboratories to meet the requirements, if in the first round of laboratory comparison study, only a few laboratories out of the measurement results, the consistency of the results of the other laboratories is very good, and can meet the target certified standard substances of the fixed value of the requirements, but also the results of the outlier laboratories can also be eliminated in the remaining laboratories to carry out the fixed-value measurements among the laboratories.
When the results of the different methods used by the participating laboratories are in perfect agreement within their measurement uncertainties, the chances of systematic error are small. This is because the source of the systematic error in one laboratory is usually not related to the systematic errors in other laboratories, and thus the systematic error is usually retained to calculate a reasonable estimate of the true value later. The end result of a fixed-value measurement is that the values from each laboratory are treated as a whole of equal quality, and the final mean and standard deviation of the measurements are calculated accordingly.
Laboratory comparisons presuppose the existence of a certain number of laboratories of equal competence. The methods they use are each independently validated. This means that the differences between the different measurements are purely statistical and can therefore be dealt with using purely statistical methods. Although this method of identifying measurements is sometimes a no-win situation, strictly speaking, the results of this method of identification are only comparable to those asked by the laboratories involved in the identified measurements. Here even erroneous values have clear authority, especially when relying on purely statistical treatments for decision-making and judgment.
The laboratories participating in the collaborative study - should have the necessary conditions for the standardized measurement of the substance and have some technical authority. Each laboratory can use a uniform measurement method, or can choose the laboratory recognized as the most effective method. The number of cooperating laboratories or the number of independent constant value measurement groups should meet the statistical requirements (when the same method is used, the number of independent constant value measurement groups is generally not less than 8; when multiple methods are used, the number of independent constant value measurement groups is generally not less than 6). The organization responsible for the accreditation must establish clear guidelines for quality control of laboratories participating in the collaborative valuation of comparative studies.
Cooperative comparative study calibration becomes a method of calibration that specifies the measurement method if the participating laboratories are essentially using the same measurement method to establish the characteristic value. In this case, the resulting characteristic value is method-dependent, which must be stated in the corresponding Reference Standard certificate. In many areas, inter-laboratory comparative studies are the only valid way to determine values, especially in areas of mandatory control, such as the content of leachable toxic metals in paints, the flash point of flammable solvents, etc. Many of the certified standards used in clinical chemistry are determined by the reference method. The catalytic activity of an enzyme can be derived by evaluating its ability to increase the rate of a particular chemical reaction at a specified pH, temperature, and concentration.Over the course of 2013, the European Union has come to recognize the importance of calibrating routine medical devices with certified reference materials (CRMS) that are rigorously traced back to a reference method, and has included a requirement to do so in the relevant regulations.
There is some difficulty in estimating the measurement uncertainty of the final recognized value when using inter-laboratory comparison studies as a way of making fixed-value measurements. This has led some metrologists to refer to the method as one that produces non-traceable "consensus values". As a result, further research work is now being done in the metrology community to improve the statistical tools applicable to this approach. On the other hand, a number of internationally renowned metrological institutes have compared the results of a series of collaborative inter-laboratory calibration measurements in the field of certified reference materials (CRMS) for metal alloys with the results of measurements made using the benchmarking method, isotope dilution mass spectrometry (IDMS), and the results of the comparison show that there is no inherent contradiction between the results of the two types of methods.
①The influence parameters of the characteristic value must be paid attention to when the standard substance is measured at a fixed value. Measurement operators must experimentally determine the various operating conditions on the characteristic value and its uncertainty of the impact of the size, that is, to determine the value of the influencing factors, which can be expressed in numerical terms or numerical factors. For example, the melting point standard substance for standard capillary melting point apparatus, its capillary melting point and its uncertainty by the rate of temperature rise. Fixed value measurement to give different heating rate of the melting point and its uncertainty.
② Characteristic value of the influence of the function of some standard substances may be measured by the environmental conditions of the characteristic value. Impact function is the value of its characteristics and the influence of the quantities (temperature, humidity, pressure and other factors) the relationship between the mathematical expression. For example: pH-A/y+N+C-Din+D-Din. Therefore, standard substances must be determined when the value of its influence function.
Statistical processing of the fixed value data
When the use of high-accuracy absolute measurement method or authoritative measurement method (now generally referred to as the benchmark measurement method) of the standard material for the fixed value of the measurement, the measurement data can be processed in accordance with the following procedures:
On each operator's set of independent measurements, the technical explanation of doubtful values and be eliminated, can be used to Grubbs (Grubbs) method (Grubbs method) (Grubbs method). （The Grubbs method (Grubbs test I critical value) or the DiKon method (DiKon test critical value) can be used to statistically eliminate the suspicious values again.
The means and standard deviations of the data measured by two (or more) operators were tested separately for significant differences. If the test result is that there is no significant difference, the two (or more) sets of data can be combined to give the total mean and standard deviation. If the test results that there is a significant difference. Difference, should check the measurement method, measurement conditions and operation process, and re-measurement.
When using more than two different principles of known accuracy of reliable measurement methods for fixed value measurement. Measurement data processing is approximately the same as using an absolute measurement method with high accuracy or an authoritative measurement method. If the test results are not considered to be significantly different, the two (or more) averages can be combined to find the total average, and the standard deviation can be found by dividing the sum of the squares of the two (or more) standard deviations by the number of methods, and then squaring.
When more than one laboratory cooperates in the determination of the value, the measurements of each laboratory can be used to statistically remove the suspicious value again by the Grubbs method (Grubbs test critical value) or the Dixon method (Dixon test critical value). When the data are scattered or there are many suspicious values, the measurement method, the measurement conditions and the operation process should be carefully examined. List the results of each operator's measurements: raw data, mean, standard deviation, number of measurements, and then examine the normality of the distribution of all measurement data. In the case of data obeying a normal or approximately normal distribution, the average of the measured data from each laboratory is considered as a single measurement value, constituting a new set of measurement data. Suspicious values were statistically eliminated by the Grubbs or Dixon method, and when the data were more dispersed or when there were more suspicious values, the measurement methods, measurement conditions, and operating procedures used in each laboratory should be carefully examined. The Cochrane (coch-ran) method is used to check whether there is equal precision between groups of data, and if the data are of equal precision, then the total mean and standard deviation are calculated. In the case of all the original data obeys normal distribution or nearly normal distribution, it can also be regarded as a new set of measurement data, according to the Grabs method or Dixon's method from the statistical elimination of doubtful values, and then calculate the total mean and standard deviation of all the original data. When the data do not obey the normal distribution, the measurement method should be checked to find out the possible systematic errors in each laboratory, and the treatment of fixed-value measurement results should be cautious.