Instrumental drift

To find explosives Gas analyzers, chromatography instruments, drift-spectrometers, neutron defectosopes, nuclear-magnetic and nuclear-quadrupole resonant instruments... 

It is usually difficult, if not impossible, to quantify all of the components in our samples. This is expecially true when we consider the meaning of the word "components" in the broadest sense. Even if we have accurate values for all of the constituents in our samples, how do we quantify the contribution to the spectral absorbance due to instrument drift, operator effect, instrument aging, sample cell alignment, etc. The simple answer is that, generally, we can t. To the extent that we do not provide CLS with the concentration of all of the components in our samples, we might expect CLS to have problems. In the case of our simulated data, we have samples that contain 4 components, but we only have concentration values for 3 of the components. Each sample also contains a random baseline for which "concentration values are not available. Let s see how CLS handles these data. 

As was the case for PCR, we see that the PLS spectral residuals for a sample will be higher whenever there is something in the data that introduces a mode of variation into the spectrum that was not present in any of the training samples used to develop the basis space. The anomolous variation could be caused by instrument drift, an unexpected interfering component, a misaligned sample cell, or whatever. We can use this property of residuals as an indicator that can signal... 

Various calibration schemes similar to those given in Section 2.2.8 were simulated. The major differences were (1) the assumption of an additional 100% calibration sample after every fifth determination (including replications) to detect instrument drift, and (2) the cost structure outlined in Table 4.6, which is sununarized in Eq. (4.2) below. The results are depicted graphically in Figure 4.5, where the total cost per batch is plotted against the estimated confidence interval CI(X). This allows a compromise involving acceptable costs and error levels to be found. 

The major causes of spectral variation were (1) instrumental drift, as Goodacre and Kell realized, but also (2) sample history, as discussed above. In particular, variations in the supplier or even the batch of tryptic soy agar (TSA) used for cell culturing led to spectral variations that differed in degree among disparate species. This phenomenon was attributed to the differential metabolic capabilities of the species with respect to the changed nutrients. 

Subsequently 36 strains of aerobic endospore-forming bacteria, consisting of six Bacillus species and one Brevibacillus species could be discriminated using cluster analysis of ESMS spectra acquired in the positive ion mode (m/z 200-2000).57 The analysis was carried out on harvested, washed bacterial cells suspended in aqueous acidic acetonitrile. The cell suspensions were infused directly into the ionization chamber of the mass spectrometer (LCT, Micromass) using a syringe pump. Replicates of the experiment were performed over a period of six months to randomize variations in the measurements due to possible confounding factors such as instrumental drift. Principal components analysis (PCA) was used to reduce the dimensionality of the data, fol-... 

Results of pyrolysis mass spectrometric analyses can be influenced by both phenotypic drift and instrument drift. Phenotypic drift can result from variations in culture growth immediately prior to analysis, and from variations during serial subculturing before analysis.125,126 However, this type of drift is not perceived as an obstacle in microbiological work because it can be largely overcome by standardizing culture conditions and by analyzing more than one sample from each culture. 

Instrumental drift results from variations in the physical conditions of a pyrolysis mass spectrometer over time.127 It leads to variation in spectral fingerprints taken from the same material on different occasions. Short-term (<30 days) instrument reproducibility was examined by Manchester et al.57 who used PyMS to differentiate strains of Carnobacterium over a four-week period. Excellent reproducibility was obtained as separation of the five type strains was sustained and spectra did not change significantly over the four weeks. 

The first strategy to compensate for mass spectral drift is to tune the instrument. This is typically achieved with the volatile standard, perfluorokerosene, and tuning so that mlz 181 is one-tenth of m/z 69. Unfortunately, this procedure is insufficient to compensate for all the instrumental drift and additional methods are required. 

For PyMS to be used for (1) routine identification of microorganisms and (2) in combination with ANNs for quantitative microbiological applications, new spectra must be comparable with those previously collected and held in a data base.127 Recent work within our laboratory has demonstrated that this problem may be overcome by the use of ANNs to correct for instrumental drift. By calibrating with standards common to both data sets, ANN models created using previously collected data gave accurate estimates of determi-nand concentrations, or bacterial identities, from newly acquired spectra.127 In this approach calibration samples were included in each of the two runs, and ANNs were set up in which the inputs were the 150 new calibration masses while the outputs were the 150 old calibration masses. These associative nets could then by used to transform data acquired on that one day to data acquired at an earlier data. For the first time PyMS was used to acquire spectra that were comparable with those previously collected and held in a database. In a further study this neural network transformation procedure was extended to allow comparison between spectra, previously collected on one machine, with spectra later collected on a different machine 129 thus calibration transfer by ANNs was affected. Wilkes and colleagues130 have also used this strategy to compensate for differences in culture conditions to construct robust microbial mass spectral databases. 

To optimize resolution in lifetime-based assays, a comparison of relative estimates is always favorable. If the FLIM experiment is carried out in an environment where temperature cannot be tightly controlled, it is also convenient to cycle between different samples during the same experimental session, in order to average out thermal and other instrumental drifts. When applicable, this practice may be useful to suppress any nonrandom variation in the detection. 

Corrigan and Weaver employed the PDIR approach to study the potential-dependent adsorption of azide, N , at a silver electrode. The potential was switched between the reference value, —0.97 V vs. SCE (where adsorption is known to be limited) and the working potential every 30-60 scans, i.e. up to a minute per step, to a total of c. 1000 scans. The high number of scans was required in order to obtain the required S/N ratio hence the PDIR technique was employed to minimise instrumental drift. Since the electrochemical process under study was totally reversible on the timescale of the experiment, the PDIR technique was a viable option. 

Sample-standard comparison is more applicable in MC-ICP-MS, in which instrument mass fractionation is fundamentally a steady state phenomenon (Marechal et al. 1999). This method has been used successfully for some non-traditional stable isotopes, particularly involving Fe, in which analyses of samples are bracketed by standards to cope with systematic instrumental drift (e.g., Zhu et al. 2002 Beard et al. 2003). However, other methods have been used for Mo stable isotope work published to date because of concerns about non-systematic changes in instrument mass fractionation, particularly arising from differences in matrices, between samples and standards. Such concerns are more acute for Mo than for Fe and many other elements because Mo is a trace constituent of most samples, increasing the challenge of rigorous, high-yield sample purification. 

Selecting the placement of Q.C. samples within the anaytical run depends upon the purpose of the Q.C. program. While random placement is statistically justified, it may not provide sufficient diagnostic information. If instrumental drift is an important concern (as it is in many automated, operator unattended techniques) the two Q.C. samples should be spaced at intervals that are appropriate to detect the anticipated drift. Placement near the beginning and end of the analytical run has been been beneficial in detecting instrumental drift. By bracketing groups of routine samples with Q.C. samples it is easy to identify specific samples that require re-analysis. 

It is also possible to use an internal standard to correct for sample transport effects, instrumental drift and short-term noise, if a simultaneous multi-element detector is used. Simultaneous detection is necessary because the analyte and internal standard signals must be in-phase for effective correction. If a sequential instrument is used there will be a time lag between acquisition of the analyte signal and the internal standard signal, during which time short-term fluctuations in the signals will render the correction inaccurate, and could even lead to a degradation in precision. The element used as the internal standard should have similar chemical behaviour as the analyte of interest and the emission line should have similar excitation energy and should be the same species, i.e. ion or atom line, as the analyte emission line. 

A better method, which accounts for any instrumental drift, is to measure the isotope ratio of two isotopes of an element with different abundances, such as In, Pb or Rb, and use the following expression derived from Eqn. 5.5 ... 

Variations in lamp intensity and electronic output between the measurements of the reference and the sample result in instrument drift. The lamp intensity is a function of the age of the lamp, temperature fluctuation, and wavelength of the measurement. These changes can lead to errors in the value of the measurements, especially over an extended period of time. The resulting error in the measurement may be positive or negative. The stability test checks the ability of the instrument to maintain a steady state over time so that the effect of the drift on the accuracy of the measurements is insignificant. 

Another important concept is parsimony. This means that if you have to select among several models that perform more or less equally, the preferred model is the one that has fewer factors, because the fewer factors are used, the less sensitive its predictions will be to non-relevant spectral phenomena. Hence the PLS model will still be useful for future unknowns even when slight dilfe-rences appear (e.g. some smooth instrumental drift, slightly higher noise, etc.). Such a model is said to be more parsimonious than the others. 

This discussion deals with random errors and their propagation in reported HO concentrations. Equal attention should be given, of course, to systematic errors of calibration or instrument drift. 

The CD spectrometer is usually required to work near the limits of sensitivity—e.g., reading AA values of <10 4 at a total absorbance of 1. Thus, particular care needs to be taken with cleanliness and orientation of cells and with settings of scan rate, time constant, and bandwidth. It is also important, especially when recording far-UV spectra, that the lamp is not old and that the mirrors are not clouded from radiation and traces of ozone. Because the spectrometer is a single-beam instrument, it is essential always to watch carefully for evidence of instrumental drift during measurements of sample and baseline. 

---