Performance Characterization of an Instrument Eric D. Salin McGill University, Montreal, P.Q., Canada H3A 2K6 There is a tendency among certain students to have nnquestioning faith in instrumentation. This is particularly true of modern instrumentation with digital readout. These stufiadents reDort data to a phenomenal number of sienificant . . ures while expecting a corresponding accuracy. An appreciation of the power of modern instrumentation must he tempered with an awareness of its limitations. Students should apply to instrumentation the same statistical awareness that is so conveniently and commonly applied to classical techniques. Propagation of error techniques can he used to pinpoint instrumental limitations and breakdowns as well as they can be nsed to demonstrate the capabilities and ultimate limitations of volumetric and gravimetric methods. With this in mind, an experiment has been developed for use in our second analytical chemistry course. We have used an atomic emission apparatus as the experimental vehicle; however, the techniques are generally applicable. All instruments have limitations over their entire performance ranae. The experiment to he discussed introduces the student tothe terms 'detection limit," "signal-to-noise ratio," "dynamic range" and "linear dynamic range" as well as providing the student with a step-by-step procedure for the determination of experimental noise sources in the instrumental operating ranges. Flame atomic emission (FAE) was nsed due to the simplicity of the apparatus and data handling. Experimental Procedures and Equipment A FAE apparatus with a digital readout is most convenient for this experiment. A meter "needle" type of readout usually cannot provide the readout resolution required to quantize the experimental noise. A strip chart recorder is somewhat better if it has multinle ranees: however. we have found that an inexpensive 3112-digitdigital multimeter (DMM) is excellent, because it usually provides voltage ranges so that a change in amplifier settings is unnecessary throughout the entire experiment. We have found it convenient to use calcium (Ca) as an analyte for this experiment. It can he seen visually in the flame. has both elemental and molecular hands, is subject to the classic phosphate chemical interference, and ~ r o v i d e aood s oerformance. The ;it&nt'i iirit t8;k is to adjust the specfrophotometer so that the 4'22.7.nrn C~atnmicemissionline is obirrved.Thr wavelength dial usually does not readout the wavelength precisely thus providing a convenient first contact with reality. While aspirating a high concentration (around lo4 ppm) of Ca the student must set the gain of the instrument to provide a readout of 5.0 V on the DMM with appropriate initial operating parameters (gas flows, slit, etc.). This is the highest concentration that will he aspirated and the gain must he adjusted on our system to avoid saturation of our external onerational amolifiers. The student is expected to determine the electronic time bv blockine.. a hieh-inconstant (. 1.) of the entire aooaratus .. .. tensity signid and thvn quickly r(:mwing the I h k . Studenti usc the ruideline that 99'10 of the final value will he reached in 5 timeonstants (TC's). The students are then encouraged to use the eeneral rule that dieital observations should he made every5 .IT or farther apart so fnaf the values will Ile statisticalls indrl~cndrnt.It is la^ ifthr time constant oftht: system is from 0.i to 1s. This range normally will he available on both modern and old eaui~ment.If the time constant exceeds 1s then the studentiwill spend inordinate amounts of 70
Journal of Chemical Education
Svmbols and Terms Symbol
d m
n s t
B C cd,
E
N.z S AFE AS BA BE BFE
Definition dark signal (no light falling on detector) slope of the analytical calibration curve in units of volts/concentrationunits number of measurements analytical signal (E, = EAEF)used in figures tstatistic value from Student's t Test background concentration the concentration at the limit of detection voltage noise (RMS) signal analyte flame emission total analytical emission signal blank aspiration total blank emission signal blank flame emission
time recording their data and may introduce unnecessary long term drift (low-frequency noise) into their experiment. Time constants that are too small will allow unnecessary noise through the electronic handpass. Students are now ready to begin determination of instrumental noises. All noises are determined bv. acauirine 10 . readings a t 5 TC or greater intervals and taking the stanzard deviation, z, of those values. While it might he argued that 32 or some intermediate value is more appropriate, the inclusion of more values may introduce low-frequency noise contributions as well as student fatigue. For purposes of later reference we will refer to the followina of the experiment as -portions . steps. Step 1. The first noise to he determined is the "dark" noise, zd. E d. is measured by blocking the cntrsnce I> B. Under these conditions the optimization procedure may not be valid.
concentration range. The students are instructed to examine closely Plot C (Fig. 2). It isalogfiog plot of a relationship whichshould, ideally, be linear and follow the general formula Volume 61
Number 1
January 1984
71
