D. GUTMAN,A. J. HAY,AND R. L. BELFORD
1786
Shock Wave Studies with a Quadrupole Mass Filter. I.
Experimental
Apparatus: Its Design and Performance
by David Gutman, Arthur J. Hay, and R. Lmn Belford Noyes Chemical Laboratory, Department of Chemistry and Chemical Engineering, University of Illinois, Urbana, Illinois 61803 (Received November 1 , 1966)
This series of papers reports the successful use of a Paul (quadrupole) mass filter coupled to a shock tube to study reaction kinetics. The first paper describes the construction and operation of the equipment and the results of some performance tests. Smoothness of the leak, design and alignment of the ion source, and placement and output circuitry of the ion detector proved to be critical. I n the final tests, the ion signal rose rapidly after arrival of the shock and leveled off to a value proportional to the partial pressure of the species parent to the ion peak monitored. Rise times corresponding to the time constant (-10 psec) of the electronic circuitry were observed.
Introduction The possibilities of coupling a shock tube to a mass spectrometer in order t o study high-temperature gas phase reactions were first explored by Bradley and Kistiako~sky.'-~Subsequently, Diesen*J and otherse!' have refined the technique. This technique is becoming more and more valuable to gas kineticists. Shock-heating is the only method of heating a gas sample quickly and uniformly to a predetermined high temperature, and the mass spectrometer can follow the concentration of almost any chemical species as a function of time. The advantages of this experimental technique are of such great value that thorough studies of the problems and limitations imposed by this experimental method would be extremely useful. All mass spectrometric shock tube studies to date have been made with a shock tube coupled to a time-of-flight mass spectrometer, which has the great advantage of being equipped to scan an entire mass range as fast as every 20 psec. However, it cannot remain tuned to one mass peak continuously. In order to obtain detailed continuous ion current records we built a quadrupole mass filter and coupled it to a shock tube. The absolute sensitivity is comparable with that of the timeof-flight instrument. Therefore, the continuous nature of the quadrupole trace offers a considerable improvement in accuracy, for which one sacrifices The Journal of Physic& Chemistry
the ability to observe several species simultaneously in one shock experiment. This- paper describes the first working version of the apparatus and tests of its performance. It was our purpose to explore such problems as those listed here. (1) How fast does the ion signal rise after shock arrival at the end of the shock tube, and under what conditions does it rise fastest? (2) What type of sampling leak gives ion signals most nearly proportional to gas pressure? (3) What are the important considerations in the design of the ion source for chemical kinetic studies? (4) How serious is the boundary layer cooling effect on the gas being sampled? We are now able to offer partial answers to these questions.
Experimental Apparatus The Shock Tube. The shock tube consists of a 4ft driver section and a 16-ft test section. The tube is (1) J. N. Bradley and G. B. Kistiakowsky, J. Chem. Phys., 35, 256 (1961). (2) J. N. Bradley and G. B. Kistiakowsky, ibid., 35, 264 (1961). (3) J. N. Bradley, TTan8. Faraday SOC.,57, 1750 (1961). (4) R. W. Diesen and W. J. Felmlee, J . Chem. Phys., 39, 2115 (1963). (5) R. W.Diesen, ibid., 39, 2121 (1963). (6) J. E Dove and D. McL. Moulton, Proc. Roy. SOC.(London), A283, 216 (1965). (7) A. P.Modica, J. Phys. Chem., 69,2111 (1965).
SHOCK WAVESTUDIES WITH A QUADRUPOLE MASSFILTER
made from several sections of Schedule 80 (2.900-in. i d . , 0.300-in. wall thickness) 316 stainless steel pipe and a 10-ft section of 3-in. i.d. Pyrex double-tough glasspipe. Gradual tapers cut into the two ends of stainless steel pipe adjoining the glass section give the shock tube a smooth inside wall. The last two sections of the tube, which contain the velocity probes and the leak, are of constant cross section. The diaphragms are scribed aluminum sheet, 0.005 in. thick. The velocity of the shock wave was measured by detecting its passage past four thin-film platinum resistance gauges.s The signal from the first gauge triggered a single sweep of a Tektronix 5358 oscilloscope fitted with a Type CA dual-trace input amplifier. The subsequent signals were then displayed on top of an unfolding raster time base generated by a Radionics Inc. Model TWM-2A crystal-driven triangle wave and time mark generator. The shock velocity at the end of the tube was calculated by extrapolating the velocity measured between probes to the end of the tube. The end flange on the last section of the shock tube contains a small conical or hyperbolic orifice, which allows gas from the shock tube to flow directly into the ion source of the mass filter (Figure 1). The leak is in the center of a small disk, 0.625 in. in diameter and 0.020 in. thick, which is in turn brazed into the larger end plate. The hole in the center of the end plate is step-shouldered such that when the disk is in place, the surface is again even. Another thin-film resistance gauge is mounted flush with the inside surface of the shock tube end flange. When the shock arrives at the end of the tube, the signal from this gauge triggers the single sweep of the 565A dual-beam oscilloscope. The ion current is recorded by one beam and the thin-film gauge signal is recorded by the other beam. The Mass Filter. The vacuum chamber which houses the quaclrupole mass filter connects to the last section of the shock tube through a coupling that allows necessary spatial alignment without spoiling the vacuum seal. It is made from a 150-lb 8-in. diameter stainless steel flanged cross. The cross is connected to a 150-lb %in. diameter stainless steel flanged tee through an %in. gate valve (Figure 1). Two 6-in. N.R.C. oil diffusion pumps maintain a high vacuum in the system. Before the gas is admitted into the shock tube, the chamber is pumped down to a pressure torr. With gas in the test section of the of 1 X shock tube, there is a steady flow of gas through the orifice and a resultant steady-state pressure in the vacuum chamber. With a 0.002-in. orifice and a pressure of 10 torr in the shock tube, the background pressure can be held below 5 X 10-6 torr.
1787
QUADRUPOLE POYER suppLr
ELECTRON MULTIPLIER POWER SUPPLY
TRIQQER PROBE’
TO 8 IN. OIL DIFFUSION PUMP
(L
LIQUID N, BAFFLE’
e
IN. CROSS
‘SHOCK TUBE TEST SECTION
TO 6 IN. OIL DIFFUSION PUMP (L LIQUID N, BAFFLE
I 1 Figure 1. Schematic drawing of the vacuum system housing the quadrupole mass filter and the last shock tube sections. SAPPHIRE RODS
ION
COLLIMATOR SOURCE MOUNT
SHOCK TUBE’
END
CAP ELECTRON FOCUSING LENS
Figure 2. The ion source of t,he quadrupole mass filter and the shock tube end section.
The ion source (Figure 2) consists of an electron gun assembly to ionize the gas and a series of focusing cylinders to move the ions from the ionizing region to the entry hole of the quadrupole mass filter. Most of the metal parts in the ion source were constructed from 0.018-in. thick Inconel sheet. Where thicker material was needed, 304 stainless steel was used. The assembly contains several insulating materials. ( 8 ) These gauges were prepared according t o the technique described by M. Steinberg and W. 0. Davies, A.R.L. Technical Report 60-312 (1960).
Volume 70,Number 6 June 1966
1788
Sapphire rods support and align the ion source and boron nitride bushings space the focusing cylinders. Glass-bonded mica (Mykroy 1100) was used to make electrical terminal strips and other miscellaneous parts. The filament is of tungsten wire supported on both ends by tungsten posts. Midway between the posts, the wire is wound in a three-turn helix,9 which presents a large emissive surface to the first (circular) collimating lens. No attempt is made to regulate the filament emission, because the entire recording time of one experiment is less than 1 sec. Thus, no long-term stability is required. About 2 min before a shock is fired, the emission current is checked and is adjusted, if necessary, by changing the filament-heating voltage. The electrons emitted from the filament pass through a collimating lens and a focusing cylinder before entering the ionizing region, ie., the case. The electron beam then passes transversely directly across the front of the leak parallel to the end flange of the shock tube and intersects the beam of gas molecules emerging from the leak. The center of the electron beam passes 4 mm from the leak. On the other side of the case, the electrons exit through a 0.120-in. diameter hole and are collected on a flat metal sheet held 1.5 v positive with respect to the case to suppress secondary electron emission. The case region in which the electron and gas beams cross is designed for very rapid pumping. It has a large cross section facing the gas beam (0.625 in. in the direction of the electron beam X 0.75 in. perpendicular to the beam) and is narrow (0.188 in.); it allows the un-ionized gas to escape easily into the large vacuum chamber. The ions formed by the electron bombardment are forced upward by the ion repeller into the focusing lenses, which they enter through a 0.106-in. diameter hole in the top of the case. The center of the electron beam is 0.156 in. below this hole. I n order to extract the ions from the case, the ion repeller is made 22 v positive with respect to the case. There are 11 cylindrical lenses (Figure 2) 0.089 in. in diameter and 0.150 in. long. The voltages on all 11 lenses are continuously adjustable from that of the case (75 v) to ground, the potential of the entry hole into the quadrupole. The geometry of the lens system made the computation of optimum voltages virtually impossible, so that the settings used are those found by maximizing the recorded ion current. Since this method of focusing ions is rather unconventional, especially in mass spectrometry, the reasons for making the ion optics as they are listed below. (1) Circular optics instead of slit optics are emThe Journal of Physical Chemistry
D. GUTMAN, A. J. HAY,AND R. L. BELFORD
ployed because in a quadrupole mass filter the entry hole is circular. (2) Cylindrical lenses are used because their ends may be placed arbitrarily close together. The lenses run parallel to the end cap of the shock tube, nearly touching it. To prevent the potential of the end cap from penetrating into the lens system, we placed the cylinders 0.007 in. apart. The end cap of the shock tube is floated at the voltage of the case. The rest of the shock tube is grounded. (3) The large number of lenses is necessary because the ions must travel 2 in. to clear the side of the shock tube before entering the mass filter. Also, for maximum focusing versatility, the lenses were made only a little longer than their diameter. (4) The diameter of the cylinders (0.089 in. i d . ) was a compromise between two desired properties. Their diameters should be large to admit the largest number of ions from the case, thus increasing the sensitivity. On the other hand, the diameters should be small to allow the central axis of the cylinders to be as close as possible to the end of the shock tube, which in effect moves the electron beam closer to the leak. No unique set of voltages gave an absolute maximum ion current. Several different sets which differed markedly from each other all gave approximately the same resultant ion current. However, each set gave a maximum current in the neighborhood of its voltage settings (f10 v variation of each lens, varied one a t a time). A typical set of ion source voltages is given in Table I. The ions enter the mass filter through a final collimating hole. The size of this entry hole has an effect on the resolution of the quadrupole. For this reason it was made adjustable. For any particular experiment, in order to maximize the ion current transmission, the largest hole is used consistent with the resolution required. The quadrupole mass filter consists of four stainless steel rods 0.750 in. in diameter and 17 in. long arranged so that the electric field in the space between them is essentially hyperbolic. The poles are held in place by two stainless steel flanges, each fitted with four boron nitride cups to provide elect,rical insulation, support, and location. The flanges are mutually aligned by bolting them to two ends of a cast stainless steel cylinder. The principle of operation of the quadrupole mass filter has been well documented and will not be repeated here.10.11 Opposite poles are connected elec(9) The entire filament unit was obtained from Consolidated Electrodynamics Corp., Part No. 38760, Diatron filament.
SHOCK
WAVE
S'PUDIES WITH A
QUADRUPOLE ;1'IASS FILTER
Table I : Typical Ion Source and Lens System Voltages
1789
MULTIPLIER SHIELD\
1
Potential,
Element
1. Ion repeller 2. Case 3. Ion focusing lenses x o . 11 K O . 10 KO.El
No. 6 No. 7 S o . ti No. 5 No. 4 KO. 3 No. 2 KO.I 4. Ion source mount 5. Filament and flat electron lens 6. Cylindrical electron lens 7. Electron trap (trap current 20 pa) ~
V
ION
97.0 75.0
DEFLECTINO. PLATES
Q y i s o N TRAJECTORY
75.0 49.3 72.8 28.7 65.2 31.1 65.4 0.0 57.9 11.7 50.4 0.0 -25.0
ELECTRON TRAJECTORIES
FIELD STRIP
HI FE
225.0 76.5
trically and excited with both ac and dc voltages. A radiofrequency oscillator circuit supplies a 1 Mc/sec signal to the poles, such that opposite pairs of poles have signals of the same amplitude but 180" out of phase. The same circuit supplies a positive dc voltage to one pair and a negative dc voltage to the other pair, also of the same magnitude. The ratio of ac to dc voltage is externally variable and is adjusted to establish the desired resolution. A resolution of 150 (10% valley) is possible with this instrument. The mass which is allowed to pass through the filter is controlled by the magnitude of the radiofrequency voltage.'* The beam of ions emerging from the mass filter is bent 45" by a parallel-plate deflection lens before striking the cathode of a Bendix M-306 magnetic electron multiplier (Figure 3). The output of the multiplier passes through an emitter follower arranged in the Darlington configuration according to a published circuit.13 The input resistance is 1.1 X lo6 ohms paralleled by 35 pf capacitance. The output impedance is ca. 500 ohms. The voltage thus generated is amplified by the 2A61 preamplifier of the Tektronix 565A oscilloscope. A magnetic multiplier was used because it has severaI advantages over conventional types. Since the active coating on the dynode strip is an inert oxide, the entire multiplier is easily cleaned and can be recycled to moist atmosphere without changing the gain characteristics. This feature proved very important to us because we raise the vacuum chamber to atmospheric
Figure 3. Electron multiplier and mounting for ion current detection.
pressure every time we clean the leak. There is, however, one serious disadvantage to its use. Because the magnetic multiplier has only one dynode, the last stages of amplification cannot be decoupled as in the conventional multi-dynode type. Therefore, the output, current must be kept below about loF7 amp for amplification to remain linear. This necessitated the use of the high anode resistor, which caused the overall electronic time constant to be relatively high (-10 psec), and the rather sensitive preamplifier. When this multiplier was first used, it was situated directly behind the exit hole of the mass filter and no deflection lens was used. This gave a direct line of sight from the ion source to the detector. With this geometry, the base line of the mass spectrum never lay a t zero ion current. There was always a residual output current on the multiplier, even between mass peaks and when the accelerating voltages mere turned off. Thus we suppose that the current came from hard ultraviolet or soft X-rays which were generated in the ion source and were reaching the multiplier up the direct line-of-sight path through the quadrupole mass filter. (10) W. Paul, H. P. Reinhard, and V. von Zahn, Z. Physik, 152, 143 (1958). (11) W. Paul and M.Raether, ibid., 140, 262 (1955). (12) The circuit for the oscillator was supplied by K. R. Wilson, University of California, Berkeley, Calif. The circuit has since been published with modifications. See K. R. Wilson, University of California Radiation Laboratory Report No. UCRL-11605 (1964). (13) The emitter follower output circuit was taken from K. G. P. Sulsmann, J . Quant. Spectry. Radialhe Transfer, 4, 375 (1964).
Volume 70,Number 6 June 1966
1790
These high-energy photons also cause secondary emission on the multiplier cathode and hence would be recorded as ion current. For this reason we offset the multiplier and added the deflecting plates illustrated in Figures 1 and 3. With this arrangement the residual ion current completely disappeared. The magnitude of this anomalous signal was directly proportional to the gas pressure in the ion source. Therefore, the signal was due primarily to luminescence of the bombarded gas molecules and not due to emission from the case itself brought on by electron bombardment. Experimental Results
Ion Source Perfonnance. At the end flange of the shock tube, the density of argon as a function of time looks like a step function. Before the shock arrives, the density is low and constant. On shock arrival, it rises rapidly (less than 5 psec) and remains constant until the reflected shock wave interacts with the contact surface and returns to the end wall. This period of constant density (and temperature) lasts between 500 psec and 2 msec, depending on the final temperature. Shocks were fired into pure argon and t.he Art- ion currents were recorded. In the earliest experiments, the ion current signals rose very slowly (2W400 psec) and did not level off to a constant value. This phenomenon was soon traced to the ionization region (the case); the electron beam was centered too far away from the leak. Thus, the main contribution to the ion current came from the static gas pressure’s building up in the case. A narrower source was built which allows the electron beam to be placed closer to the leak. The distance was reduced from 25 to 4 mm. The narrower source made the conductance of the case much greater so that gas from the shock tube which was not ionized could exit more easily into the larger vacuum enclosure. The new source gave faster rises (
