sode begins with a rapid lowering of pressure, which is reflected in the graph by a rapid increase of percent transmission. If the pressure drop is maintained for one to two seconds then the pattern expected from Le Chitelier's Principle will be observed, but a longer time may result in leakage of air into the system. Discussion Practical Aspects
The exueriments chosen for this article were such that interfaciig added greatly to the convenience of the procedure. Indeed, in those examples where the time frame was short (nitrogen dioxide equilibrium) or long (rusting of steel wool) the exueriment would have been difficult to accomplish by other means. Expense was not a serious issue as the package of interface, software, and temperature and light probes cost about $350US. While the pressure probe as supplied by Tain Electronics was relatively expensive (about $135US), other probes cost only tens of dollars or consisted of equipment not originally purchased for interfacing. The TCS2 interface can read the voltage at the amiliary outputs of pH meters and similar equipment. We used the TCS2 because it was cheap and readily available in Australia, but the experiments could be run easily on any multi-functional interface. The TCS2 interfaces quite satisfactorily with old XT microcomputers that are obsolete for most other uuruoses. Experiments sueh as those described in this work could also be run with a chart recorder, but the ease of entry of data into a spreadsheet is lost. Spreadsheets may be difficult to learn. but thev can be useful tools in science education (13,18222). In this work they have been employed to present experimental results, but they also can be used to make models from theory to compare with these results. Learning Aspects
experience with the equipment and software, students could be asked to do some "real science" by designing their own experiments, such as to devise a procedure to analyze a commercial product or to choose a method (conductance, pH, color of indicator) to follow the hydrolysis reaction (2425). Then there is the field of computer control. The TCS2 has outputs that can be govrrned-by inputs, so that some preset level of, say, pressure would cause a light or motor or other device to be switched. The TCS2 also can be owrated by a library ofpascal and QuickBASIC routines, leaving open the possibility of quite sophisticated programming. Literature Cited
4. Flauell, J. H. Dola Caplare Ezperimnfn in the Scirnco Naflond Cum~culum:Borough ofDudley: hdley, U.K. 1990. 5. Jaffar,M.; Zahid, Q. J. Cham. Educ 1988,65, 1099-1100. 6. M&, W.C.: ?ge, R. S. J. Chem Edue. 1991,68,A95. 7. Curtin.T.A.: Wahlatmm.D.:MeCmick.J. J. Cham.Edu 1991.68.781. 8. Amend. J. R.:Furstenau, R. P;Howald, R. A,; Ivey, B. E.; lkk& K A. J. Ckm. Educ 1990,67,333336. 9. Garret& D. D.; Banta,M C.;Amey. B. E. J. Ckm. Edve 1991.68,661-668. 10. Berka, h H.; Clark. W. J.; White,D. C. J Chem. Edur 1992,69,891497. 11. SpmIgin, C. B. Sch. Sci. RPU.1999,71(2541,47M. 12. Btimimmbe,M. W.Sch. Sci. Re". 1990.71(256). 151. 13. Amend, J. R.;Tucker,K. A.; Furstenau,R.A.J Chem Edue. 1991,68,857-860. 14. Lvnch. J.A.:Namamom. J . D . J. Ckem. Educ. 1990.67,53&335. 15. Skphrn8.J.C.H.Sch. Sci. &u. 1989.7112541,92~6. 16. Solorzs, O.;OPvsres,L.J Chem Edue 1991.68,17&177. 17. Yang.2. J. Ckm. Educ 1#98,70,9&95. 18. Bmmsn,TSch. Sc;. Rou. 1989,7M2521,3%i7. 19. Bmman,T.Sch. Sci. Reu 1990.7112561,5&39. 20. de h w e , R.J. Ckm. Educ 1998,70,209-217. 21. Goodfellow,T Sch. Sci. Re". 1990,71(2571,4745. 22. Webb, h Phydes Educ 1993.28.77-82 23. Bauer. S. J. Chrm Edue. 1990.67.692493. 24. Amend, J. R.;Furrtenau. R. P:Tueker,K.A. J Ckm. Edm. 1990,67,593495. 25. Lieu, V. T.:K4bus.G. E. J . Chem. Educ 1988.65, 18G185.
Using Computers To Replace
One thing the computer does not (or should not) do is tell Some HPLC Laboratory Work students the "correct" result. In order not to waste their time with this technolorn students must know enouph to Ian C. Bowater and Ian G. McWilliam recognize what is a reas-&able result. What the corncuter Swinburne University of Technology should do is to take measurements in quantities or at rates John Street, Hawthorn, Victoria 3122, Australia inconvenient to do by hand, and to pe;form routine calculations on the raw data to produce a set of results from The most expensive component of a tertiary course in which a conclusion could be drawn. By removing much of chemistrv is the laboratorv work. Cost-effective imuerathe technical difficulty and computational tedium the comtives are forcing us to reevaiuate our laboratory outer should enable students to concentrate on the conce~t Some experiments or parts of experiments must be rebf the experiment. This does not mean that all mathemaktained because they teach and reinforce essential techcal operations must be removed. In the hydrolysis experiniques. However, experiments that are expensive to run, ment the computer generates a plot of conductance versus that contain a lot of waiting, that involve a large number time. If students are asked to eet the rate constant for the of repetitive experiments, or are now considered unsafe reaction then they must know in what way to replot the are prime candidates for change. data and how to use the spreadsheet to do this. Aprogram Computer experiments are cheaper, quicker, and safer, that performed the experiment and presented the rate conand require less supervision. A computer exercise can be stant as its only result would have limited educational used in coniunction with a laboratorv exueriment or it can value. Then there is the "Black Box" problem that is inherbe a stand-"alone activity. Sometimes deeper learning can ent in this sort of work (23). Unless you also are teachim electronics there is not much to gain hy r e q ~ i n n ~ s t u d e n t ~ be achieved from a well-desimed comuuter exercise than the equivalent laboratory experiment. to understand what is going on inside the interface, or, for All of our maior instruments are now either comouterthat matter, the computer. Rut it is important that they ~ ~ ~ ~- -~ driven or use a computer for data storage and processing. comprehend what is being measured, and how. With some In some instances it should be uossible for undereraduates of the sensors there is the literal oossibilitv of "hands on" to become familiar with iustr&nental software Ybefore usexperience. Students could (with approp"riate precauing it in the chemical laboratory. In other situations, they tions!) blow into the Dressure transducer to see the readshould be able to use the software to process their laboraing change on the moktor, and could use the thermistor to tory results elsewhere. measure their skin temperature, while the conductivity probe could be demonstrated with some distilled water to One of the most widely used instrumental techniques is which a little acid was added. high performance liquid chromatography (HPLC).This paExperiments in this article have been presented in "recper will describe three exercises where computers are beipe" form, but this need not be the case. ARer some initial ing used to help students learn about HF'LC.
" . ~
674
Journal of Chemical Education
~~
L~
~
-
-
Two of these are stand-alone computer exercises that are done using a spreadsheet in subjects called Computers in Chemistry ( I ) . The other provides an appropriate hackground for an associated laboratory experiment. These exercises will be presented in the order that students meet them in the course. Simulation of Chromatographic Resolution In the first exercise the students study the parameters in a spreadsheet that generates a chromatogram containing two Gaussian peaks with identical areas. The main theme of this exercise is to simulate how the resolution of two peaks changes when the values of the number of theoretical plates (n), the capacity factor (k'),and the selectivity factor (a) are altered as would be done by changing the experimental conditions. A spreadsheet which contains two peaks a t 1.5 and 1.6 min and the corresponding chromatogram are shown in Table 1and Figure 1, respectively. The chromatogram is generated by summing they coordinates of both peaks. The dead time ( t ~is) marked on the chromatogram at 1minute. Different chromatograms can be generated by altering the values of the initial parameters column length (L),average linear velocity (GI, capacity factor of the first peak (kl') and a. The values of the parameters n, t ~kz', , the retention times ofthe two peaks ( ~ R I and tm) and their standard deviations (al and 02)are calculated using the following equations. (1) n=LIH
tM=LIE
(2)
tE = tM(l+ k')
(4)
When the students initially load the worksheet, Kl and a are set to 1and 3, respectively. This generates two wellseparated peaks with retention times at 2 and 4 min. Initially the students 1. measure tM,tR1,and tm and hence calculate the values of Kl, k'z, and a 2. estimaten fmm the width at half height of the second ~ e a k 3. studv the formulae in the snreadsi&t and the a r r k of numbers that generate the chromatagram 4. enter two formulae to calculate the resolution (R)using 2r.m
(a) 3 a
1p.m
o.m
4 R=2@+oz) n%
(6)
a-1
r=T [T][&]
(7)
When using an HPLC, the easiest experimental condition to change in order to improve the chromatagram is the percent of organic modifier in the mobile phase. To a first approximation this alters k' but not n or a. Therefore we would like students to appreciate that eq 7 states that R is proportional to k'l(k' + 1)and hence there is no resolution when k' = 0 (no time is spent in the stationary phase), half of the maximum resolution is achieved when k' = 1,O.S of the maximum resolution is achieved when k' = 4, etc. The students study two series of chromatograms a t a ranee of different values of k' (n is fixed at 5000). In the frstseries k' can be as low as 0.05 because the peaks are easv to s e ~ a r a t e(a = 3). However in the second series k' mu& be &eater than 1to get baseline resolution because the peaks are much harder to separate (a = 1.2). The Table 1. Spreadsheet Data for Chmmatogram with n = 5000, K = 0.5 and a 11.2. HPLC simulation
mwpmmms Column length (cm) Height equivalent of a theoretical plate (cm) Average linear velocity (cmis) Capacity factor of peak 1 (k') Selectivity factor(alpha) Calculated Darameters Number of theoretical plates (n) Dead time (min) Capacity factor of peak 2 (K) Resolution calc (exact): t and std devn Resolution calc (approx):n,kand alpha Peak I Retention time (min) Standard deviation (min) . . x Y1 Y2
15 0.003 0.25 0.50 1.20
5000 1 .O
0.60
Peak 2
Sum
Resolution with n=5000, a = 1.2. k' = 0.5
1.m
i.m
r.m
Fmtcntim ti-
r.m
r.m
I
r.m
Figure 1. (a) Graphical display of data shown in Table 1. (b)Same display with K = 4.0. Volume 71 Number 8 August 1994
675
..
selecting the % organic modifier and flow rate selecting an injection volume and injecting the sample viewing the chromatogram an a plotter using a data system to display, save, and process the chromatogram.
IS.@ I0.m -
(a)
Capillary GC column with n=125000 k'=4.a=l.l
IS,@1o.m -
rsa . -
(b)
When using the simulator, the time spent waiting for a sample to pass through a column is eliminated. Hence more samples can be ~n in a given time compared with the laboratory and hence a greater percentage of this time can be spent analyzing and t h i n g about results. In 3 hours students use the simulator to
Typical HPLC column with n-5000
-Figure 2. Comparison of separation achieved by a capillary GC column and a typical HPLC column.
prepare a standard of four known mmpounds run the standard using 80%methanol use reference W spectra to assign which compound in the standard causes each peak view a 3D plot of absorbance versus time versus wavelength add another 5 mg of one of the compounds to the standard and run the spiked standard use the retention time of the enhanced peak to contirm the previous assignment A n thcstandGrd using65'h methanol wherctheorderafelutmn is different
spreadsheet is then used to calculate the lowest value of a that will give a resolution greater than 1.5 (valley to peak ratio less than 2%) when n = 5000. In order to appreciate the changes in the series, the screen is split to display both the new chromatogram and a reference chromatogram. The change when k' i s increased from 0.5 to 4 with a = 1.2 and n = 5000 is shown in Figure 1. The spreadsheet is also used to illustrate that the increase in resolution when a good HF'LC column (n = 5000) is replaced by an outstandine one (n = 10000) is onlv 40% becaise R i; proportional to-nLI. However the resoiution can be increased bv a factor of 5 for analvses that can be done using capilla& GC with n = 125,000 rnstead of HF'LC with n = 5000. The chanw when n is increased from 5000 to 125,000 with k' = 4 an2 a = 1.1 is shown in Figure 2. For the capillary GC simulation
generate an unknown mixture of seven compounds find the % methanol that will give baseline resolution of all seven peaks use referenceW spectra to identify which of 10 known eompounds causes each peak in the unknown mixture find the % acetanitrile and % tetrahydrofuran that gives the same retention time for the final peak observe how the chromatogram and the column pressure alters when the column length and flow rate are altered When the percent methanol is changed from 80% to 65% in this simulation, the order in which benzene and anisole elutechanges. This illustrates that u canalsochange when the organic modifier is changed.
1. L and H are altered to 2500 and 0.02 cm, respectively, to make n equal to 125,000 2. E is altered to 41.7 cmls to keep tMequal to 1min. Finally the spreadsheet is used to show that by increasing the average linear velocity a shorter analysis time can he achieved provided that the degradation in the resolution (n is smaller) is acceptable. A Combined ComputerIChemical Laboratory HPLC Experiment In the second exercise a software package HPLC: An Instrument Simulator, which was produced by Rittenhouse and is marketed by JCE Software (2),is used to simulate the conditions reauired to use a reverse ~ h a s ecolumn to isocratically sepa;ate and identify a raniomly generated mixture of seven com~oundsfrom a librarv of 10 aromatic compounds. The students later attempt toseparate a mixture of the same seven compounds in the laboratory using a Waters CIS Radial Pak column. The decisions made in generating a chromatogram using HF'LC: An Instrument Simulator are as close a s possible to those made in the laboratory using a n HPLC with a diode array detector. These include
turning on the power to the instrument mnsole, the detector and the . oumos . choosing the column and organic modifier ehoasina the detector conditions (diode arrav or selected wavelength, and absorbance range) preparing a standard solution
676
Journal of Chemical Education
Figure 3. Retention timesof the 5 carboxylicacidsversus pH showing changes in the order of elution.
Table 4. Separation Factors for TPA, PABA, p 4 B A and BA
100 90 80 70 60 50 40 30 20
Figure 4. Separation factor versus pH for critical pairs of carboxylic acids.
10
no no no no no no no no no no no
0 %ACNO
no no no no no no no no no no
no no no no no no no no no
10 20
YES is all three > 0.2 no is at least on c 0.2 no no no no no no no no
no no no no no no no
30 40
no YES YES YES no no
YES YES no no no
YES no no no no no no no no no
50
60
70
80
90
100
Optimization of HPLC Separation Parameters
In the third exercise, students use a spreadsheet to reprocess two related sets of published chromatographic data on carboxylic acids. All the analyses were done isocratically using reverse phase separations on a C18 column. Table 2. Compositions of MeOH in isocratic Solvent Mixtures
MeOH is 20% in water ACN is 16% in water THF is 10% in water
%THF
100
0
90
10
80
20
10
70
30
20
10
0
60
40
30
20
10
0 0 0
50
50
40
30
20
10
40
60
50
40
30
20
10
0
30
70
60
50
40
30
20
10
20
80
70
60
50
40
30
20
10
10
90 80
70
60
50
40
30
20
10
0
0
100 90
80
70
60
50
40
30
20
10
0
20
30
40
50
60
70
80
90
1W
%ACN 0
10
Deming and Turoff (3) measured 12 retention times in the pH range 3.76-5.66 for five aromatic carboxylic acids using an aqueous buffer (acetic acid, pK. = 4.7) as the mobile phase. The order of elution of the five acids changed a number of times over this pH range as shown in Figure 4. They used a nonlinear least squares fit to calculate values of tR(HA),~R(A-) and pK, for each acid. After studying the shape of the retention time versus pH curve and comparing the raw data with that predicted by the curve of best fit, students identify the pH ranges where pairs of these carboxylic acids would be difficult to separate. For each ofthese pairs they calculate a separation factor ( S )where
0
The parameter S was chosen because it is easier to measure than R. These parameters are related by
0 0
-
Table 3. Separation Factors for TPA and p-ABA
%THF 100
0.14
90
0.15
0.15
80
0.15
0.15
0.16
70
0.15
0.16
0.16
0.17
20
0.14
0.15
0.16
0.18
0.19
0.20
0.22
0.23
0.25
10
0.13
0.14
0.16
0.17
0.19
0.21
0.22
0.24
0.25
0
0.12
0.14
0.16 20
0.17 30
0.19 40
0.21 50
0.22 60
0.24 70
0.25 80
A plot of S versus pH for each of these pairs was used to find a DHwhere I S I > 0.08 for all airs. This corresuonds to R > ?2 (valley to peak ratio less thin 0.1%)if n = 2560. A n example of this plot is shown in Figure 4 for two pairs of acids. Harvey et al. ( 4 ) based an experiment on the work of Deming and Turoffusingfour of the five carboxylic acids. They found that the analysis time could be reduced from about 25 min to 6 min with a buffer (pH = 4.1) containing either 20% methanol (MI, 16%acetonitrile (A) or 10% tetrahydrofuran (T). While an adequate separation of all four acids could not be obtained using only one of these three modified buffers, it a 6 peared that a mixture of two or three would give a suitable separation. For each pair of adjacent peaks, they used measurements of the se~arationfactor for the solvent mixtures M, A, T~MA, MT,AT and MAT to predict the separation factor for any solvent 0.27 mixture containing M, A andlor T. They then 0.27 0.28 used a triangular response surface to fmd the region wherethe separation factor was above a predetermined critical value. Volume 71
Number 8 August 1994
677
Students use a spreadsheet to repeat the calculations and create a triangular response for each of the three pairs of adjacent peaks (the order of elution of the four peaks does not change). The traditional equilateral triangle is displayed in the spreadsheet as a right-angled triangle. Table 2 shows the composition of the modified buffers in the triangle and Table 3 shows the corresponding separation factors for one pair of acids. ARer converting each triangle to regions of good resolution (S > 0.2 for this column
678
Journal of Chemical Education
because n is smaller) and bad resolution (S < 0.21,a composite triangle is formed which shows the region where good resolution is obtained for all three pairs of adjacent peaks. This is shown in Tahle4. Literature Cited 1. Bowster, I. C.; MeWIIBam, I. G.; Wong, M . G. J.Cham. Educ,in press. 2. Rithhouse. R. C."HPLCSimulator? J Chem. Edue: SoRwore 1988,Ib , S.N.: mmR,M. L.n.nnnl. c b m . l m ,50.546548. 4. narvey, D. T.; ~~~~l~8.:BO-W, A ; lbmlin,J.J chm.E ~ U Cissl,68,162168.