54 ANALYTICAL METHODS/Geochemical Analysis (Including X-ray) Geochemical Analysis (Including X-ray) R H Worden, University of Liverpool, Liverpool, UK ß 2005, Elsevier Ltd All Rights Reserved Introduction Geochemistry is the study of the occurrence and distribution of elements, isotopes, minerals, and compounds in the natural environment Geochemical analysis, the measurement of the quantities of elements, isotopes, minerals, and compounds in a rock, natural liquid, or naturally occurring gas, is a colossal topic The applications of geochemistry stretch from the core to the outer atmosphere and beyond The objects of study include minerals, metals, and salts, organic biomolecules, coal, bitumen, kerogen, and petroleum, atmospheric-, river-, ground-, and formationwater, and gases in the crust, sediments, and atmosphere Analysis is also a huge topic, with subjects ranging from rocks, minerals, compounds, and species to isotopes, elements, and atoms The Role of Analysis: Hypotheses, Questions, Problems, and Theories In the context of Earth sciences, geochemical analysis commonly implies quantitative work with a numerical output Geochemical analysis is thus the quantitative examination of the composition of natural Earth materials Most geochemical work is undertaken to address a hypothesis, answer a question, or solve a problem Even routine environmental geochemical monitoring is done in order to address the question of the continued safety of natural materials over time It is typically important to formulate a hypothesis, question, or problem carefully before embarking on an analytical programme Such an approach is best since the outcome should be cost-effective while being statistically rigorous and ultimately defensible signal over a range of concentrations (Figure 1) The detector outputs from the standards must be converted into some sort of equation that permits the eventual back-conversion of output signals into concentrations for unknown samples In the example in Figure 1, the output must be determined as a function of concentration (equation 1) and then inverted to allow concentration to be determined from the output signal for unknown samples (using equation 2) The quality of the analytical output from a device is fundamentally a function of the quality of the calibration curve No calibration is perfect, and the degree of imperfection can be described in terms of accuracy and precision These discrete characteristics are best described for a dataset composed of repeated analyses of the same standard sample (Figure 2) When the results are widely dispersed about the known answer, the output is said to be imprecise When the results are tightly clustered about the known answer, the output is said to be precise If the results cluster around the correct figure, they are said to be accurate, whereas if they cluster about a figure other than the correct output, they are said to be inaccurate (Figure 2) Accuracy and precision are the combined result of the quality of calibration, the sensitivity of the device, Producing Geochemical Data When any analytical geochemical device is used to quantify the composition of a material, the detector output is usually in the form of some sort of electrical signal The electrical signal must be converted into a meaningful geochemical parameter, such as a unit of concentration This conversion is usually achieved via a calibration curve For any technique, a series of previously characterized standards must be analysed, in exactly the same way as an unknown sample, and the output signals measured In essence, the calibration is a plot of known concentration versus the output Figure Example of a calibration curve using standards of known concentration and their output signal strengths The small black dots represent the calibration Equation describes how concentration could be converted to signal strength The more useful equation describes how signal strength from an unknown sample can be converted into concentration Unknown sample A yields a valid result since it falls within the calibration range The analysis of sample B is not valid since the output signal falls outside the calibration range