1S-of P~HYDROXYPHENYL ACETIC ACID FIGURE & GAS CHROMATOGRAPH OF URINARY ACID PROFILE A, PATIENT ACIDOTIC (2-17-73) BL OPATTENT mopery, (10278) oO = >» OXYPHEN Sh mass 149, amp hohe mass 186, amn £9,953 Amtna Actd ALA Hasses mass 168, amn 7.23 mass 178, am hhh Amtinn Aete VAL Passes mass 126, ann 22,75 cmass 178, amy 227.76 Aminn Acie CLY Masses mass 1£2, arn 1.89 mass 192, ann 1h,25 Amino Actd thr Masses mess 182, amp 3.59 mass 197, arn 26,32 Asatno Acta Leu Masses mass 166, ann F954 mass 173, amo 199,99 Antnan Actd epo MASSPS mass 153, amn h.59 mass 158, amn 19,31 Anina Acted TUR Masses mass 139, amn 3.10 mess 182, amp 12,32 Amtng Acid S&P Masses mass Tho, amo 9.85 mass 155, amn 32,27h Amting Acfd Pre Masses mass 280, ann S.f1 mass 2h3, ama 20,71 Amtna Actd ASP asses mass 198, ann 58,15 mass 293, ann f7.5F Amino Acted Chi Masses Amino Actel 4 12 Tacatton Fyernt VhO/1%h URE /17A 126/128 122/192 122/192 THE/175 153/158 139/182 Thasyss IhN/2hS 198/205 errer ANALYSES ( me/198nmt) 2 2 1 3 2 4 6 49 N APEC ACHE Faure ALA 1,8 VAT. 1,5 nv 3.5 Ihe n,2 te Ah Pra 29% Ten ..7 ener 2.7 ror n.7 asr 7,0) mae o 0 Areas Areas Areas Areas Areas Aroas Areas Areas Areas Areas Areas 11349,2/ 21382,5 rode .n/ 10987.3 6222.h/ 65H,7 393.27 3653.7 R36,1/ 6769.0 7603,1/+33028.8 1162,.7/ 2722.9 3293,8/ 3104.7 2929.1/ 8971.8 2281,8/ 5751.9 16883,6/ 19353,6 FIGURE 5 PATI PATIO RATIO RATIO PATIO PATIC PATIN PATIO PATIO PATIO RATIO 0.5308 ** 0.1771 1.0277 N.1076 0.1235 “0.2298 0.4270 LONE 1.3255 9.3968 N,872h ANALYSIS OF 12 AMINO ACIDS IN URINE USING MASS FRAGMENTOGRAPHY Lacatton Location Location Lecation location Locatton Lacatton Location Location Locatton Locatton Error frror Error Error Frror Error Error Error frror frror Error Let. Diet hem... tn fp xss a. a IE NE WF THE SIMULTANEOUS QUANTITATION OF TEN AMINO ACIDS IN SOIL EXTRACTS e BY MASS FRAGMENTOGRAPHY W. E. Pereira, Y. Hoyano, W. E. Reynolds, R. E. Summons and A. M. Duffield Department of Genetics, Stanford University Medical Center Stanford, Califormia Running Title: Mass Fragmentography of Amino Acids Address for Proofs: Dr. A. M. Duffield Department of Genetics Stanford University School of Medicine Stanford, Califomia 94305 Received P-S3 The analysis of amino acids from terrestrial and extraterrestrial sources is becoming increasingly important (1-5). The need for a specific, sensitive and rapid method of quantitation is desirable. The methods currently employed for amino acid analysis involve ion exchange procedures (6,7) or gas chromatography (8-10). These techniques, although of immense value, are limited by their non-specificity for the absolute identification of any substance responsible for a gas chromatographic peak. In the present communication we report an absolute, unambiguous method for the positive identification end quantitation of ten amino acids present in soil extracts using GLC-mass fragmentography. In mass fragmentography the mass spectrometer is used only to detect certain preselected ions known to be characteristic for each compound being quantitated, and the internal standard. The technique of mass fragmentography using sector mass spectrometers is usually restricted to the simultaneous monitoring of up to three integer mass values (11, 12), although with one instrument five ions were used (13). Using a quadrupole mass spectrometer up to eight ions have been selected and their respective analog signals monitored (14). We now wish to report the modification of the gas chromatography-quadrupole mass spectrometer-computer system previously described (15) for the simultaneous monitoring under computer control of the ion currents from 25 pre-selected integer mass values. | hese valves Com range betwaw masser Cond 750 iw conftest & the lini ed Poange uveilable for mess fragmeniogqraphy ving S¢! ometes, Tf required this number could be increased by suitable alteration of the computer control programs. Specifically we wish to report the application of this system to the quantitation of ten of the amino acids present in soil extracts, p-s7 METHODS Reagents: A deuterated amino acid mixture was supplied by Merck Laboratory Chemicals (New Jersey). 1.25N HCl in n-butanol, \ 25% (v/v) trifluoroacetic anhydride in methylene chloride and Tabsorb(EGA on Chrome column packing were obtained from Regis Chemical Co., Illinois. A standard amino acid solution was purchased from Pierce Chemical Co., Illinois. Equipment: A Varian model 1200 gas chromatograph was coupled by an all glass membrane separator (16) to a Finnigan 1015 Quadrupole mass spectrometer which was interfaced to the ACME computer system of the Stanford University Medical School (15). GLC separations were conducted using a 6 foot by 4 mm. (1.D.) coiled glass column packed with Tabsorb (Regis Chemical Co.). The flow rate of the carrier gas (helium) was 60 ml/minute. | The uniqueness of the mass spectrometer instrumentation lies in the modified computer software (program) used. The hardware is the system | previously described (15) and ae an operating cycle of: (a) transmission of a control number, N, from the computer to an interface controller which sets the quadrupole mass analyser to a particular mass point in the n/e continuum. (b) an integration of the ion signal for a pre-set period, T (integration time = 8 milliseconds in our work), and (c) computer reading of the integration value with a twelve bit A+ D conversion. e For the recording of normal mass spectra N is selected such that successive cycles result in m/e values of 1,2, .....750. At the beginning of each day the instrument is calibrated using a reference compound. Idiosyncracies of the IBM 360/50 to IBM 1800 computer data paths dictate that the mass ’ values be buffered into groups of 250." For normal g.c.-m.s. procedures the operator is allowed to select a mass range of 1 to n x 250 (n = 1,2, or 3 buffers). For mass fragmento- graphy n is set to zero and instead a “precision collect" buffer of 250 control-data acquisition cycles is employed. The operator must then enter the pre-selected m/e values he wishes to scan. When the precision collect buffer is constructed, 10 cycles are allocated to each m/e value selected. The first of the 10 cycles sets N to Nova. The returned integrated ion measurement is discarded; this cycle serves only to slew the quadrupole electronics from anywhere in the m/e continuum to the mass region of interest. The additional 9 cycles are used with N = Nae Nee ee NOTA. The returned values represent a set of readings about the m/e value of interest + 0.5 amu. The center three points are then smoothed with a five point qudratic function (17). The highest value of these three smoothed points is then selected ,as the precision collect value. Thus small drifts in calibration are corrected and a signal average obtained. Finally, the abbreviated "spectrum" of 25 precision intensities for each n/e are filed on disc. Such a "spectrum" is recorded every 2 seconds and a summation of all the fon intensities is used to construct the ion chromatogram shown in Fig. 2. Individual ion chromatograms can also be constructed if required (Fig. 3). A threshhold is established from the ion currents before and after each gas chromatographic peak and a computer program performs integration of the ion currents under each peak. P-£% Procedure 1 g of sieved, air-dried soil (Stanford University garden soil) was refluxed with 6N HCl (10 ml) for 20 hrs. The mixture was filtered and the residue washed with IN HCl] (5 ml). The combined filtrate _and washings were extracted with chloroform (4 x 10 ml) and the aqueous phase evaporated to dryness. The residue is dissolved in water (5 ml) and passed through a column of "Ion Retardation Resin" AG 11-A8 (50-100 mesh, 1 x 21 cm). The amino acids were eluted with water (50 ml) and the eluate evaporated in vacuo to dryness. The residue is dissolved in water (5 ml) and placed on a colum of cation exchange resin (AG 50W-X12, 50-100 mesh, 1 x 21 cm) and washed with water (50 ml) to remove neutral and anion contaminants. The amino acids were eluted with 4N NH, OH (80 ml) and the eluate evaporated to dryness. The residue was dissolved in water and made up to a volume of 4 ml. A portion of this solution (1 ml) was used for the amino acid analysis using an amino acid analyser. To another 2 ml of the processed solution was added 2 ml of the deuterated amino acid standard solution (100 mg in 100 ml of 0.1N HCl) and the mixture evaporated to dryness. The residue was refluxed with 1.2 N HCl in n-butanol (1 ml) for 30 min. and evaporated to dryness in vacuo. To the residue trifluoroacetic anhydride in methylene chloride (0.72 ml) was added and refluxed for 10 min. The solution was evaporated to dryness at room temperature and the residue dissolved in ethyl acetate (100 1). An aliquot (1 ul) was injected into the injector port of the gas chromatograph and the oven kept at 100° for 1 min. when it was programmed at 4°/min. to 220°. To each of 4 tubes containing 2 ml of the deuterated amino acid standard solution (100 mg in 100 ml of 0.1N HCl) was added 150, 200, 250 and 300 yl respectively of a standard amino acid solution (2.5 ,moles - of each amino acid per ml). The solutions were mixed and evaporated to dryness. Each residue was derivatized by the above method and an aliquot of each (1 ul) injected into the gas chromatograph which was operated under the conditions described above. This procedure was used to construct a standard curve for the quantitation of each amino acid, A typical standard curve is shown (Figure 1) for glutamic. acid, f- 82 RESULTS The N-TFA, O-n-butyl derivative was chosen for the derivatization , of amino acids for two reasons. Firstly, these derivatives have excellent glc separation characteristics ex?) and secondly the selected characteristic fragment ions of the deuterated and non-deuterated derivatives do not interfere with each other, nor with other a-amino acids, Table I records the individual ions monitored for quantitation in the mass spectra of each of the deuterated and non-deuterated amino acids. The -computer-integrates-the-intensity of the deuterated and non-deuterated.-ion-currents-with-time—and-quantitation—is—achieved- vy- calculation of the ratio-of—their-respect ive-peak-areas+ Our results of a typical soil analysis are compared with those from an amino acid analyser in Table II. The higher value obtained with lysine by the amino acid analyser is due to a ninhydrin positive substance in soil interfering with the quantitation of lysine. In this respect mass fragmentography is superior to the amino acid analyser in that using a mass spectrometer as detector only characteristic pre-selected ions are detected and quantitated and any impurity present’ under the same gas chromatographic peak is not measured. A summation of 20 such characteristic ions was plotted as an ion chromatogram of a derivatized soil sample and is shown in Fig. 2. Preliminary experiments showed that when the deuterated amino acid mixture was added directly to the soil sample extensive hydrogen- deuterium exchange occurred during acid hydrolysis of the soil extract. The removal of the isotopic label was catalysed by the hot mineral acid in presence of excess mineral used in the soil hydrolysis step. Fox feo} and collaborators have reported (4) a similar finding concerning the decomposition of amino acids in soil upon direct acid hydrolysis. In the present work the deuterated amino acid mixture was added just before derivatization (i.e. after hydrolytic extraction of the soil) in order to avoid this problem. Howevér, in cases where it is necessary to quantitate the free amino acid content of complex mixtures, such as in serum or urine samples, the deuterated amino acid mixture may be added directly to ,the sample before processing without any deleterious effects sh. Although only ten amino acids present in soil were quantitated the method can be extended to all the normal amino acids found in protein. The deuterated analogs of arginine, histidine, serine, threonine and tyrosine are commercially available. Appropriate deuterated analogs of methionine, tryptophane, cysteine and cystine would have to be chemically synthesized from the appropriate precursors. In these instances at least two deuterium atoms should be incorporated in non-exchangeable positions so that for the characteristic ion chosen the P + 2 peak is separate from the 13, isotope contribution of the unlabeled amino acid. Furthermore, the deuterium substitution need not be quantitative (>90%) provided the same characteristic ion of that deuterated analog is used for the construction of a standard curve such as Figure 1. Tn our Aupevience , Wr Use cf A single wase value bow 2a, CAunino ocacd beoe trey wi the a C. petoticn 4 ume, , ig aufheent. Coy accurate dh euti{ carton AnA quantikarren | he Chemrcok Worle - wp specifically relels a asc vache Trere by elimaurctm Aorck and “Yeu tral comboumndc<, Wreck D ootal pees bly +uwe A ~ [osrasre } A ((-2rsr] TT FA ° ‘ iz F/° « aw > t a Qic a, & ' lzu aig eoze er VY mw = 4 ou qu ty ne Wow ao ef et aot wo ole 7 7 = me Lied Ww 5 g§ se, Se ——__~-3 7 aa £ Sin = w 4 a Uta ‘ j 3S z tre « < ifm &§ 7 = Ss : 34 iy zZ Lif __ w gu : a S 43181938) a <2 rs w 2 ies ce ty ee 5 w oF ‘| ws = a = = Ea wa) he yw 22 i if a wy m of Y ne . a 23 , Se =38 za 2 “sz © m “ar, 2 ra x of a w 2) || eA oh & FSG: oo Ww pep)! mi 2% « g59l/! 3 5 52 905 5 eas Ssailt 213 =z v Aa ¥ 8 joa x - STIGVD ONILSSNNOD LHOHS A ce ~ [4 au ot 3 Sis g g ~ si lz ote 215 a & $ oO Sees Slaw a 3.2 gto wu sé zi3| 35 ex = Fuse el< = Feo 3 els wy} a z be Zao E =22 ale S Sse 2 fe0 o}e 3 oa = 6 3 2 o = ” < & we ¢Y & } = o| oO « « =< 2 S Son Were | 5 nu ‘ wv ° w bE ca> wes I = a ot w z5= | 2 au Oo cos z < 5N z=D | = E> w 4 be os oa Poor | a i wo Yawe | w ns2 oz 2320 a 4 & zo< aa a } = < : 5 = 3s 25 eo Sa see S = ga 2 wie’ 3531 oy]: wie w x= = a}: z wo 2 = 5 < a x 5 2 : ok 2 6 ez 8 Zw ad wt ww ww = uy - Me ul « 2 3u ge So Figure 1. The GLC-MS instrumentation and the electronics interface to enable computer systems integration control and checkout functions, are controlled by the “C' register. There are no manual operator control functions in any o' the above steps. The control is accomplished at the telen pe writer keyboard. This keeps the system flexible and mane it independent of the idiosyncrasies of individual campute ae ANALYTICAL CHEMISTRY, VOL. 42, NO. 11, SEPTEMBER 1970 ¢« 1123 COMPUTER DIALOGUE (The user’s response is underlined) OPTION = collect EXPERIMENT# = test T = 35 MODE = single PLOT, TYPE, or FILE = plot BY MASS, AMP, or ENTIRE = entire FROM MASS = 1 TO = 510 QUICK = n PLOT, TYPE, or FILE = / EXPERIMENT# = MS26x15 CONTINUATION = yes MODE = continuous # OF SPECTRA = 130 FILED IN POSITIONS 940 to 1070 EXPERIMENT? = / OPTION = sum ~ EXP; = MS26xi5 FROM SPECTRUM = 9 TO = 1020 ~— FROM MASS = 40 TO MASS = 500 PLOT = yes EXP = / OPTION = plot EXPERIMENT? = MS26x15 FROM SPECTRUM = 957 TO = 957 a FROM MASS = 1 TO MASS = 340 COMMENT The user requests the data collection phase. A catch-all name, “‘test’’, is given; the spectrum will be used simply for a systems check. An integration time of 35 milliseconds per peak is requested. Only one spectrum will be taken (single mode). The data is acquired after this answer. The user can plot, type. or file the data col- lected; here a plot is requested. The user can plot selected masses, the highest intensities, or the entire spectrum within requested limits. The user indicates the limits. A QUICK plot omits annotation, etc. Note that “ty” and “‘n’’ mean “yes” and “‘no”. The user completed the checkout and now wishes to proceed with the experiment. The ‘‘/” is used to backup through the conversation. An existing experiment name is given here. The user confirms that the spectra are to be added to the existing experiment file. The user requests that spectra be collected continuously until 130 are taken. Data collection is complete. The user then wishes to sum the elements of each spectrum to produce a “‘total ion” plot, analogous to a GLC trace. Mass position 40 to 500 are summed for each of the collected spectra. The “total ion” curve is now plotted. (Figures 6 and 7 are illustrative of this sample dialogue.) Guided by the total ion plot, the user will plot interesting mass spectra. Only the spectrum filed in position 957 is chosen. The plot (Figure 17) is drawn and normalized to the base peak, m/e = 31. Figure 2. An example of the user-computer dialogue during operation interrupt lines and/or individual computer characteristics. This straightforward system definition makes the software design much like conventional computer programming rather than encouraging intricate techniques highly dependent upon the specific hardware. Thus the system is not oriented specifically to any given computer. It has operated on an early model LINC (/7) com- puter with 2K words of 12 bits. memory, and ona time-shared, locally programmed, IBM 360/50, buffered with an IBM 1800 (48). In all cases the computer was somewhat remote, separated by some 500 ft of cable from the rest of the in- Strumentation. The system is very economical of computer resources. Most of today’s small general purpose computers would be able to operate the described functions if it were desired to avoid time-shared computer dependency. Some (17) R. W. Stacy and B. Waxman, “Computers in Biomedical Re- search,” Vol. II, Academic Press, New York, N. Y., 1965, pp 35-66. (18) W. J. Sanders, G. Breitbard, G. Wiederhold. ef af.. “An Ad- vanced Computer for Medical Research.” Fall Joint Computer Conf. Proc., ACM, Anaheim, Calif., 1967, p 497. sort of magnetic storage for object code programs and data storage is most desirable. DEC-type tapes have been used on the LINC system and disc packs on the IBM system. THE SOFTWARE STRUCTURE The objectives of the software are to operate and control the MS, acquire data from the MS, process and present this data in a manner useful to the chemist, and provide certain control and information to aid in maintaining and servicing the instrument. With the program loaded into the computer, the user requests any one of several functions (see Table DP) by tvning the name of that function. The computer responds wih a series of prompts (see Figure 2) to elicit user macrocommands. The computer then generates the detailed control funcuons to perform the assigned task. At the completion of the tisk. requests are made for new parameters. By striking the slash (‘/"), the user can ‘“tbackup” through any conversation to correct errors or to go toa different function. This conversa- tion technique makes the system both flexible and reasonadly self-instructing. P-72 1124 « ANALYTICAL CHEMISTRY, VOL. 42, NO. 11, SEPTEMBER 1970 Table I. A List of Program Options (1) CALIBRATE: Creates an accurate N Table. The N num- bers which correspond to the peaks in the reference gas are used as the end points of a piecewise linear interpolation pro- cedure for calculating a complete N Table. (2) COLLECT: Is the primary data collection step. It is here that the 750 N Table values are sent to the MS and the 750 mie intensities recorded. This operation can be repeated at five-second intervals as the data are filed on disk under an ex- periment name. (3) TYPE: Allows the user to print out spectral data by indicating what spectra in a given file are to be reviewed. The user can request that the amplitudes of particular mje positions be typed; that a given number of the highest amplitudes be typed, or that a consecutive number of them over a given range be typed. (4) PLOT: Enables the user to have bar graphs produced by the computer controlled digitai plotter. The amplitudes to be plotted can be selected with the same flexibility as described in TYPE. (5) SUM: Produces a plot of the total ion current over a series of gathered spectra. All responses of a spectrum are summed to produce one datum point on the plot. This plot corre- sponds closely with the GLC output when running withthe GLC. (6) TRACE: Produces a record of a spectrum similar to the normal chart recording cutput. The analyzer is sampled at all N values (about 10 per amu) over a given range and the result is plotted as a “broken line.” (Used for system check out) (7) MONITOR: Provides for inspecting the peak profiles by sampling the .jectrum around a given nije position. The gathered data are then typed out. (Normally used for system service or service log) (8) DISPLAY: Enables the user to display a given mass position (or N number) in the center of the console oscilloscope. (Used in the adjustment of the mass spectrometer) (9) GAS: Allows the user to remotely turn the reference gas on or off. This is helpful when operating the system from a remote position. The example of a user-computer conversation given in Figure 2 represents the day-to-day computer-researcher dialogue given to direct the system’s operation. eeper level programming may be done at the terminal to redetine these functions or to add new modes. Additional system development may be done by the chemist-user, or his program- mer in a manner typical of general purpose computer soft- ware. In normal daily practice, the user first requests the calibrate function and then proceeds to data acquisition, analysis, and presentation. Usually the calibration is done once every four hours. It is this calibration subsection of the program that assigns to each integer mass position a value N which when sent to the D-to-A converter in the interface, will set the mass analyzer to pass that particular species of w/e. During this calibration phase a reference compound (perfluorotributylamine. FC-43), is introduced into the MS. The calibration procedure in addition to determining the N values, makes data avaiable that will aid the operator in making qualitative judgments about the stability, sensitivity, and resolution of the MES. Also a service or maintenance reccrd plot is available (see Figure 3), that, at least indirectly, shows these and cther im- portant instrument conditions. Figure 3 is actuaily 12 traced segments of a complete spectra, each segment covering a span of about 4 amu and each taken at a different integration time (gain). The m/e value, its position, and a parameter incicat- ing the gain is automatically printed below each peak. The date and time is printed by the computer, but at present the operator must insert the sample pressure and ionization pa- rameters. The file of these plots represents an excellent rec- ord of the instrument’s serviceability. The calibration ts automatic and its use less complicated than the description. It takes about 5 minutes, after which the reference gas is pumped out of the system. In the IBM 360-1800 system, the time is used to compile the main program. l I ity ity iii ri, rity rity Lily ily rly rio M 31 63 ice 131 169 219 284 414 464 Sc2 S75 Bid N 41S 1272 2293 2887 3327 434s S381 10362 114282 12223 13637 14457 T 4 -8680 7880 «2888 1.3388 «1280 5880 2.880 2.780 2 9 8 PRES _3x/0"7 torr T FACTOR MICA -_ 40% 123351 Lime IONV __20 W025. 691218 date Figure 3. A monitor plot indication of instrument serviceability 2p - 7: ANALYTICAL CHEMISTRY, VOL. 42, NO. 11, SEPTEMBER 1970 ¢ 112 Total spectral data acquisition time depends upon the number of m/e positions measured and the integration time allowed per position. It may be calculated: Spectrum acquisition time = P X (JT, + T) qd) where P is the number of m/e positions to be measured (they do not have to be contiguous or sequential), 7; is a transition time (2 msec in our system), and T is the integration time per peak (nominally 6 to 17 msec, but we have usefully used 1 to 1000 msec). Normally data are collected at each integer m/e position 1 through 750. The 750 N values are sent through the D-to-A converter to the MS and the 750 responses (a full spectrum) are recorded by the computer cia the A-to-D converter. This process can be repeated approximately every 5 to 10 seconds for an arbitrary number of times. The spectra thus gathered are stored by the computer on magnetic disks or tapes. Pro- gram changes may be made to measure any subset of the 750 nije positions and thus achieve faster repetitive spectra. Con- versely more measurements may be made at any specific peak position, a technique which may be used for accurate isotopic ratio measurements. Since many spectra are taken and stored during a GLC run or a solid probe experiment, the user requires fast methods to evaluate the data. The more useful data abstracting pro- grams we use are: ‘THe Matrix SEAnCH. The user specifies which group of spectra, what range of mass values in each spectra, and how many large peaks he wants abstracted from each spectrum. An abstract of these highest peaks is then typed out and in many cases this abstract contains useful chemical information or at least indicates the spectrum of interest. Tue Time PRESENTATION (PLOT). This is a computer drawn plot of certain peak intensities or a sum of all peak intensities of each spectrum (total ion current) plotted against time. The latter gives a good reproduction of the GLC curve and also indexes the spectra of interest (/9). NorMALizeD SPECTRUM Plots. Conventional bar graphs of mass vs. time, normalized and annotated, are routinely available. All data outputs are in the laboratory and are available immediately after data acquisition. All spectra are filed and may be recalled at a later time or date and reprocessed in any way desired. Involved programs of these magnitudes are specifically dependent upon the language of a given computer. The logic may be easily transferred, but in general the specific pro- gram may not. We have about 4 man-years of programming invested in this system. THEORY OF OPERATION During spectrum data acquisition, the computer directs the mass analyzer to a program-selected mass position and reads the output intensity of the MS. The mass analyzer is not swept in a conventional sense. As indicated in Figure 1, it is controlled by a voltage (V,) such that mle = fo) . (2) where m/e is the mass/charge ratio and f(V,)} is a monotonic function characterized by the MS. For every M;, (M = mie), to be passed by the mass analyzer, the computer has (according to the prior run calibration program) a digital number N; (19) R. A. Hites and K. Biemann, “Advances in Mass Spectrometry IV," Elsevier Publishers, Amsterdam, 1968, p 37. 1126 « ANALYTICAL CHEMISTRY, VOL. 42, NO. 11, SEPTEMBER 1970 which is transformed by the D-to-A converter to the voltage Ve. The determination of these values, N;, is accomplished by the calibration program. The value of N for 12 key peaks of the reference compound are known to approximately 1 amu from prior calibrations. The actual N value for the centroid of each of these peaks is then determined by detailed examina- tion of the m/e continuum in each of these areas. Sufficient detail is cbtained by designing the D-to-A resolution to be 10 or more values per peak width. After determining these exact 12 N values, linear interrola- tion. superimposed upon the analyticai function, m/e = (V2), is used to expand the list of 12 experimentally determined values to a full table of 750 entries. (The analytical func- tion of m/e to control voltage is linear for the quad- rupole and parabolic, m/e = &(V.?), in the case of the T-o-F MS.) Thus the procedure to measure the intensity at any WW, is as follows: a@. The number N, which corresponds to the se- lected m/e ratio (M,) is loaded from the computer into the D- to-A converter. This sets the control voltage, /,, to the mass analyzer. The output of the mass spectrometer is proportional to the quantity of ions, M;, passed from the sample. b. An analog circuit, reset and released by the computer, integrates the output of the MS. ce. Several milliseconds after the integrator is released. (the choice of integration time was initially supplied by the user upon program request), the computer samples the output of the integrator by means of an A-to-D converter. This digital value is stored as the intensity of /;. Steps (a) through (c) are repeated to acquire a complete Spectrum. The fundamental restraint upon this system is the drift of the function m = /(V,) following calibration. Our experience with the Finnigan 1015 arid a Bendix T-o-F instrument and our interface, is that this drift causes an error in N of less than 1/, the value from one N entry to the next in a 1-hour period. This is sufficiently small to allow an unambiguous mass identi- fication. Table II contains comparisons of signal-to-noise ratios and the following defined figures of merit. The comparisons are made between the described control system, a linear scan in time, a parabolic scan in time such as the T-o-F, and the ex- ponential time scan characteristic of magnetic instruments. Uniform conditions are used to give realistic values for comparison; it is assumed that in each case the peak shapes are uniform if scanned in time, and that they are gaussian, and that the resolution is commensurate with the 10% valley ($75 points on a single peak side) definition (3). In order to give typical comparison figures. it is further assumed that a spec- trum will be taken from mass 50 to 500 in 4500 milliseconds. The first column in Table If is the time the mass analyzer is on or about the mass position. In the case of the computer control system, the 4500 milliseconds is divided equally into 450 periods of 10 milliseconds each. Two milliseconds are allowed for each transition, and the mass analyzer will dwell on the peak position for 8 milliseconds. In the case of 4 con- ventional linear scan, the analyzer will enter a peak area and leave it 10 milliseconds later. By the 109% valley convention, this means the time from the beginning 5% level to the end 3°75 level of a single peak. However for the parabolic case (the T-o-F) it will be found that the resolution of the instrument will have to be set for the work case, peaks 499 and 500. Tt will be found that there is 6.6 milliseconds between these peak P-74 Table II. A Comparison of Attributes Affecting Signal-to-Noise Efficiencies Typical Operation Condition: Scan from m/e 50 to 500 in 4500 Milliseconds Time on, or between Ions detected: Figure of merit: 5% points, Time constant (Peak intensities Effective noise 1000/7 X detected Type of scan of a peak, msec of amplifiers, msec of n ions/msec) bandwidth, Hz ions; bandwidth Control and integrate 8 N/A 8.027 40 200 Linear scan m= kt 10 2 5.1 80 64 Parabolic scan m= kr 6.6 1.3 3.47 120 28 Exponential scan _ m= mek 3.9 0.8 2.0” 200 10 centers, and the theoretical 5% heights must be at the mid- point. Since all peaks are similar, all peaks of the scan will be just 6.6 milliseconds from 5% point to 5% point. During 34% of the scan time, the analyzer is not in the area of any peak position at all; but is mostly between peaks in the low mass range. The case of the exponential scans is similar. In this case it will be found that the peaks are 3.9 milliseconds wide and 61% of the time no information can get through the instrument. The next column indicates the time constant (7) of the ampli- fier channel appropriate to the scan parameters. The control “system uses a full iutegrator, so the entry is not applicable. In the conventional scanning system, the time constant is usually chosen as large as skewing permits to integrate signal and discriminate against noise. The relationship between 7 and the 3-db bandwidth (/..) of an amplifier is simply r = 1/Qrf.). If 7 is chosen to be large, peak skewing and broadening as illustrated in Figure 4 will occur. If 7 is chosen small, the bandwidth with its attendant noise is excessive and there is little integration of the signal. This is the dilemma always faced by the user of linear ampli- fier circuits: the desire to limit amplifier bandpass to smooth the signal, as opposed to the need for a wide bandpass to pass GAUSSION PEAK INPUT T [milliseconds OUTPUT OF A UNITY GAIN AMPLIFIER the signal without distortion. Since the purpose here is to compare our described amplifier and integrator system with conventional linear amplifiers, a + of 0.2 is assumed for the conventional case. This 7 is still large enough to cause deg- radation of resolution in the conventional output signal (25 to 35% depending upon the definition used). It is felt that this choice represents a fairly typical operational parameter. The assumption of a rigorous lower value would result in an unnecessary, and perhaps unrealistic, comparison advantage for the described control and integrate signal system. The column “Effective Noise Bandwidth” is the /.. for the time dependent scans. However an equivalent 3-db band- width is not as well defined for the integrator. It can be shown that for an integration interval, 7, (8 milliseconds in this example) an f,, may be determined such that a linear amplifier of bandwidth £, would pass the same amount of “white” noise as the integrator. The actual bandpass of an integrator is a sin(x)’x type function. The white noise power passed by either system may te ex- pressed as an integration of the white noise model, e“*, (29) (20) W. B. Davenport, Jr., and W. L. Root, “Random Signals and Noise,” McGraw-Hill, New York, 1958, p 88. Vin® x 2 |\6| AMPLIFIER | | TIME CONSTANTS 2 . Ri |t L. 1.35 T Figure 4. Peak broadening and skewing effect of narrow bandwidth amplifiers an ANALYTICAL CHEMISTRY, VOL. 42, NO. 11, SEPTEMBER 1970 © 1127