Spatial Emission Profiles for Flagging Matrix Interferences in Axial-Viewing Inductively Coupled Plasma-Atomic Emission Spectrometry: 2. Statistical Protocol

A statistical protocol was developed and verified for automated signaling of matrix interferences in inductively coupled plasma-atomic emission spectrometry (ICP-AES). Spatial emission profiles in ICP-AES are versatile indicators for flagging matrix interference. A family of calibration curves is first generated by measurements of standard solutions at different spatial locations in the plasma. The determined-concentration profile of the analyte along a spatial measurement axis of the plasma is then obtained by analyzing the sample at each spatial location by reference to the respective calibration curve. The absence or presence of a matrix interference is gauged from the shape of the determined-concentration profile of the analyte. A flat determined-concentration profile indicates absence of matrix interference, whereas a dissimilar (i.e., curved) concentration profile offers a clear warning signal that the analytical results are compromised by interferences. The developed protocol automatically classifies a spatial profile as flat or curved; it involves the computation of three statistical parameters: relative range(0.05-0.95), σ(sample), and σ(successive). The term relative range(0.05-0.95) refers to the ratio of the range to the mean of the relative-intensity (or determined concentration) values between the 5th and 95th percentiles in a spatial profile, whereas σ(sample) and σ(successive) refer to the sample standard deviation and the standard deviation of successive values, respectively, of all values in a spatial profile. It was found that whenever the relative range(0.05-0.95) of a spatial profile is below 1.5%, the profile can be considered to be flat and no further statistical testing is needed. If relative range(0.05-0.95) > 1.5%, the σ(successive)/σ(sample) ratio provides useful information on the flatness of the profile. If the profile is flat, σ(successive) will be statistically equivalent to σ(sample) (i.e., σ(successive)/σ(sample) = 1). In contrast, if curvature is present in the profile, σ(successive) will be statistically smaller than σ(sample) (i.e., σ(successive)/σ(sample)< 1). A statistical test, based on the transformation of the experimental σ(successive)/σ(sample) ratio to the z value of a standard normal distribution, was used to judge if the difference between σ(successive) and σ(sample) is statistically significant. This statistical protocol for characterization of flatness in a spatial profile was verified in experiments carried out under the influence of various matrix interferences and different plasma operating conditions.