In the Laboratory
Modular Spectrometers in the Undergraduate Chemistry Laboratory Paul Bernazzani and Francine Paquin* Département de chimie, Université du Québec à Montréal, C.P. 8888, succ. Centre-Ville, Montréal, PQ H3C 3P8, Canada; *
[email protected] At the Université du Québec à Montréal, we are committed to offer a B.Sc. chemistry program that has a strong experimental component. Our program unfolds over a three-year period. As part of the 90-credit chemistry degree, students are required to complete a number of 3-credit laboratory courses. Among the laboratory courses offered is a 3rdyear instrumental course on FTIR, NMR, AA, fluorescence, phosphorescence, and UV–vis spectroscopy. The lab class is divided into groups of two students and each group is required to complete 11 experiments during the semester. The labs run on a 6-hour period. New technology has modified classroom academics significantly in recent years. Computers and a wide range of software facilitate data acquisition, allowing us to integrate signal-processing discussions into our course. We wanted to include such a discussion within an applied frame to show the limits of these treatments when applied to real problems. To implement this concept, our lab needed instrumentation that was both relatively inexpensive and capable of various signal-processing treatments. As these treatments are not offered on simpler and less costly instruments such as the Spectronic 20, such spectrophotometers were not a valid option. Most spectrometers offer a variety of mathematical treatments but few show their effect in a dynamic way. In this paper, we discuss the effect of integrating optic modules into our undergraduate-level laboratory classroom. Two miniature 2048-element CCD-array fiber-optic spectrometers were purchased from Ocean Optics Inc., the CHEM2000 VIS and S2000 UV-VIS. They are briefly described below. However, the object of this paper is not to compare different spectrometers but to show that implementation of the Ocean Optics spectrometers has improved the quality of our students’ learning. Equipment Components Like all such equipment, each instrument comprises an IBM-compatible computer and a spectrometer. The spectrometer has five basic components: 1. Source 2. Sampling optics 3. Optic fibers 4. Micro-machined diffraction grating and detector (CCD-array) 5. A/D converter
The light sources vary according to the type of analysis. For the S2000, the light source, a deuterium tungsten-halogen lamp, combines a continuous spectrum in the UV and visible regions. Its output ranges approximately from 200 to 1100 nm. The tungsten-halogen light source, on the other hand is versatile for the visible region, between 360 and 1700 nm. 796
The sampling chamber, capable of holding standard 1-cm2 cuvettes, is directly connected to the source and the detector via optical fiber cables. The cuvette holder has a 5-mm-diameter f/2-collimating lens to collect the light and funnel it to the exit solarization-resistant optical fiber. Adjustment screws allow optimization of angles of incidence. The optical fiber cables used limit the range of the source. Attenuation is fairly flat in the visible but increases strongly in the UV. In the NIR, water absorption bands at 750 and 900 nm affect fiber attenuation (1). The detector comprises a 2048-element linear silicon CCD-array and has an effective range of 200–850 nm, thereby limiting the spectral output range of both light sources. The diffraction grating, the detector, and the A/D converter are housed on an ISA-bus card for PC. This assembly allows integration times of 5 ms to 60 s. Finally, a Windows95-based program provides rapid acquisition and various processing functions, such as delay time of acquisition and point-to-point averaging, both of which are discussed below. This software is user friendly and requires no prior experience. The basic concept of this software permits immediate time display of data, allowing the students to verify the effectiveness of their experimental setup. Compared to more conventional spectrometers, these modular instruments offer another major advantage: their low cost. The S2000, the more versatile system, is 5 to 6 times less expensive than its conventional competitors. However, Yee et al. have demonstrated that the dispersion and resolution of modular spectrometers and a conventional low-end commercial instrument are comparable (2). The Experiments Knight and Farnsworth have successfully used the Ocean Optics modules in their university classroom to provide live visible spectra to a large audience (3). However, we know of no published accounts of direct implementation of miniature spectrometers in lab courses. Two experiments, previously performed with conventional spectrophotometers such as a Cary 1E (Varian) and a 8800 (Pye-Unicam) model, were constructed with miniature optical fiber Ocean Optics spectrometers. We describe the experiments that were successfully adapted using the modular spectrometers. Data obtained with all three instruments were comparable.
Experiment 1 In the first experiment, the CHEM2000 is used to determine (i) the ligand/metal ratio in a Fe(II)/phenanthroline complex using the standard method of Job and (ii) the dissociation coefficient of bromocresol green. Job’s method establishes that the intensity at a fixed wavelength of a ligand/metal solution of constant molarity is greatest when the molar fraction of the
Journal of Chemical Education • Vol. 78 No. 6 June 2001 • JChemEd.chem.wisc.edu
In the Laboratory
ligand in solution is equal to that in the complex. Students should conclude that three molecules of phenanthroline are found in the coordination sphere of the Fe. Equation 1 shows the formation of the complex. Fe2+ + nPhenanthroline → (Fe(Phenanthroline)n)2+ (1) Each of these components absorbs in a different region and the students are asked to find the value of n. In the second part of the experiment, students determine the dissociation coefficient of bromocresol green. This can easily be accomplished during the hour when the ligand/metal complex is forming. (Since the students have a limited amount of time to make a great number of solutions and measurements, they are forced to use their instrumentation time efficiently.) Students should find that the dissociation constant of bromocresol green is close to 4.72.
Experiment 2 In the second experiment students use the S2000 spectrometer to construct calibration curves for three aromatic acids based on the UV intensities of benzoic, salicylic, and p-hydroxybenzoic acid solutions. Using these curves they evaluate relative fractions of unknown solutions containing a mixture of these acids. The correlation coefficients of the acids ranged from .784 to .999. The relative errors for the determination of the concentration of the acids in the unknowns are of the order of 0.2–4.5%. Both the correlation coefficients and the error factors of the mixtures studied were of the same magnitude as those obtained using conventional UV–vis spectrophotometers. Discussion: Signal Processing These two experiments were chosen because the time factor is critical in the first, whereas in the second, precision is important. Both experiments also provide an introduction to signal processing. The students extensively vary two parameters (curve smoothing and scanning average) and observe the effects on their calibration curves. They must evaluate the influence of each parameter with respect to the objective they must achieve. They are shown that curve smoothing by the running average method (dynamic averaging) is useful if the fluctuation of the noise is significantly greater than that of the signal. The method has two distinct drawbacks: (i) it eliminates some data points at the extremities of the spectra, and (ii) oversmoothing can easily lead to loss of data. However, although the Ocean Optics software will run the average over any number of points, we seldom encountered over-smoothing because the apparatus has a high resolution (0.6 nm with a pixel resolution of 0.15 nm [4 ]). The students must also verify that static averaging, which requires either taking multiple scans and averaging them or varying the integration time, can improve the data as a function of the square root of the number of scans (or more accurately, the time). Students can thus observe whether this is correct, depending on the limiting types of noise and their nature, and whether the number of scans or the time required is low enough to be of any value. If the square root relationship between time and signal-to-noise ratio is valid, then white noise, defined as shot noise in the background or dark signal, is the limiting noise factor. Since we found that the signalto-noise ratio varied proportionally with time, 1/f noise is the limiting noise.
Because the Ocean Optics spectrometers are capable of rapid measurements over the complete region of interest, students have sufficient time to evaluate the effects of parameter settings and optimize them. They can complete the experiment in less than the allotted 6 hours. However, two drawbacks of these instruments are noteworthy. First, at the end of the course, students take home their results on floppy disk for subsequent evaluation. However, some students found exporting the data to be cumbersome because of difficulties with software incompatibility. This part of the software could be improved. Second, the absolute values of intensity where found to fluctuate. Pokorny et al. used an S2000 to evaluate the color of white wines (5). They found that using 30–36 responses yielded an average standard deviation of the mean of the order of 3%. In the best experimental conditions, we observe a 2% standard deviation. Compared to the 0.3% variation observed on the much slower Cary 1E from Varian, the Ocean Optics spectrometer’s accuracy fares poorly. In the scope of our teaching goals however, this is not a problem because the precision (reproducibility) was comparable for the two apparatuses. Previous work also demonstrates the overall reliability of the instruments (6 ). Since a major objective of the course is to introduce signal processing, the Ocean Optics spectrometers are still advantageous. Evaluation Judging by the students’ evaluation, these experiments are a success. Students are eager to operate these miniature spectrometers. They are fascinated by the speed of data acquisition with the Ocean Optics spectrometer and its ease of use. They develop a curiosity about spectrometry and a basic understanding of signal processing methods. Our evaluation of these modular spectrometers is summarized in the box. As this list shows, the advantages outweigh the disadvantages.
Advantages and Disadvantages of Miniature Fiber-Optic Spectrometers Advantages Low cost Useful for introducing signal processing Occupies minimal counter-top space No moving parts to break or misalign Easy to install User-friendly software Entire spectrum viewed in a few seconds Raw data compatible with most standard spreadsheets
Disadvantages Computer needed Blooming a Single beam Non-selectable excitation frequencies b Absolute reproducibility a
Occurs when a strong signal spreads to adjacent sensor elements (7). We have not quantified this effect. b This causes a broadening of fluorescence peaks and complicates the teaching of flourescence phenomena because all chromophores are excited at the same time. No fluorescence experiment has been implemented yet.
JChemEd.chem.wisc.edu • Vol. 78 No. 6 June 2001 • Journal of Chemical Education
797
In the Laboratory
Conclusion The acquisition of these modular spectrometer units enhanced our instrumental laboratory course. In general, we are very satisfied with the effect these modules have had on our curriculum. Their impact is twofold. First, they are less costly than the conventional equipment currently available. Second, their speed, ease of use, and capacity to vary signalprocessing parameters facilitate the understanding of spectroscopic principles. These instruments are a favorite among both students and teaching staff. Literature Cited 1. Ocean Optics. Optical Fiber Performance Characteristics; Technical document; http://www.oceanoptics.com/productsheets/
798
2. 3. 4.
5. 6. 7.
optical%5Ffibers%5Fperformance%5Fcharacteristics.asp (accessed Feb 2001). Yee, G. M.; Maluf, N. I.; Hing, P. A.; Albin, M.; Kovacs, G. T. A. Sens. Actuators, A 1997, 58 (1), 61–66. Knight, J. B.; Farnsworth, P. B.; Spectrochim. Acta, B 1998, 53, 1889–1893. Ocean Optics. Selected Components: Optical Resolution: http:// www.oceanoptics.com/specifications/optical%5Fresolution.asp (accessed Feb 2001). Pokorny, J.; Filipu, M.; Pudil, F. Nahrung 1998, 42, 412– 415. Waterbury, R. D.; Yao, W.; Byrne, R. H. Anal. Chim. Acta 1997, 357, 99–102. Ingle, J. D. Jr.; Crouch, S. R. Spectrochemical Analysis; PrenticeHall Canada: Toronto, 1988.
Journal of Chemical Education • Vol. 78 No. 6 June 2001 • JChemEd.chem.wisc.edu
