International Journal of Mass Spectrometty and Ion Processes, 76 (1987) 85-93 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands

85

A HIGHLY SENSITIVE MASS SPECTROMETER DETECTOR SYSTEM

DANIEL MURPHY * and KONRAD MAUERSBERGER

School of Physics and Astronomy, University of Minnesota, Minneapolis, MN 55455 (U.S.A.)

(First received 4 November 1986; in final form 28 January 1987)

ABSTRACT

A mass spectrometer detector system has been constructed and tested to measure up to 70 masses simultaneously. The detector increases substantially the sensitivity of the instrument for the analysis of trace gases. The system consists of a microchannel plate multiplier and a phosphor layer on a fiber optics bundle mounted in a Mattauch-Herzog mass spectrometer. A photodiode array and counting electronics record individual ion events. Each mass peak

can be independently measured at count rates ranging from 1000 counts s-l to much less than 1 count mir- ‘. Gases with mixing ratios of less than 1 X measured

directly in air samples. The mass spectrometer and detector have performed well in two balloon flights to sample stratospheric air. Other applications include residual gas analysis, gas purity analysis, and pyrolysis of solid samples. Such a detector could be employed in a mass spectrometer with continuous photoionization.

INTRODUCTION

The detection of very low gas concentrations present in mixtures of gases has always been a challenging task. When using a mass spectrometer, the most important parameter influencing the dynamic range of the measure- ments is the ion detector following the mass analyzer. Large currents are usually measured with an electrometer, while multipliers operating in count- ing mode cover the low range of ion detection. In the past, a combination of a grid collector attached to an electrometer and a counting multiplier has provided a dynamic range of over lo8 [l].

During the last few years, the range of ion detection has been further extended toward low ion currents by installing a microchannel plate (MCP) multiplier in a Mattauch-Herzog magnetic mass spectrometer. The Mat-

* Present address: NOAA, Aeronomy Laboratory, Boulder, CO, U.S.A.

0168-1176/87/$03.50 Q 1987 Elsevier Science Publishers B.V.

86

tauch-Herzog geometry [2] provides a straight line of focus at the exit of the magnetic field suitable for mounting a plate multiplier.

The microchannel plate detector system was initially developed for trace gas analysis in the stratosphere [3,4]. Many of the reactive gases involved in the ozone chemistry are present with mixing ratios of less than parts per million. For example, ozone at mass 48 has a maximum mixing ratio of less than 10 ppm, while its rare isotopes at 49 u and 50 u are present in parts per billion. Reactive chlorine and nitrogen compounds are present at a few parts per billion or less. Only a microchannel plate detector system would permit the simultaneous detection of those trace gases with sufficient preci- sion.

An imaging detector is also desirable, even when only one or two trace species are being measured. By monitoring a range of masses near the mass of interest, a continuous check can be made on the hydrocarbon background usually present in vacuum systems. One can also monitor long tails from major mass peaks that could interfere with the measurement of trace species. In general, when working at very low count rates, recording a large mass range provides the information needed to evaluate all possible interferences.

The MCP detector system described in this paper was designed to operate in an ion counting mode. Compared with analog measurements, ion count- ing offers the highest possible sensitivity. The gain stability and spatial uniformity of the multiplier are less critical for pulse counting than for analog measurements [5]. Finally, a fixed pattern noise must be subtracted when operating the photodiode array in an analog mode; the noise may be ignored when pulse counting.

A number of groups have developed ion detectors using MCPs [6-111. Of these, only Coplan et al. [6] used ion counting and they have not described dynamic range, gain stability, or background levels for their detector. More relevant to the work described here are MCP detectors in optical spectrome- ters, particularly those discussed by Hartig et al. [12], Lawrence and Stone [13], and Timothy [14]. The presence of a photocathode in optical detectors, however, causes some parameters, such as dark counts, to be different from those for ion detectors.

EXPERIMENTAL

The essential parts of the mass spectrometer and detector are shown in Fig. 1. The construction of the detector has been described in more detail in an earlier paper [15]. Experience has been gained with the microchannel plate detector on a number of operating parameters, including resolution, cleanliness, dynamic range, and stability. Briefly, ions dispersed by the mass spectrometer strike the surface of the MCP to release electrons which are

87

SPIRALTRON

Fig. 1. Diagram of the mass spectrometer and microchannel plate detector system.

amplified as they travel down a channel. The electron pulse is then accel- erated to a phosphor layer, coated on a fiber optic bundle, where it produces a flash of light. This flash is carried through the fibers to a photodiode array [20] located outside the mass spectrometer. The array’s 1024 elements are paired into 512 pixels, each 50 pm X 2.5 mm, providing about 10 pixels mass-’ for 50 masses simultaneously measured.

To operate in the counting mode, the array is scanned at the high frequency of 4300 Hz (450 ns per pixel), and the pulse height on each pixel is fed through a discriminator to determine the presence or absence of a count. The counts are accumulated in a multichannel scalar [16]. Every 15 s, a mass spectrum of approximately 50 masses is obtained. Longer integration times are possible by adding spectra stored in a computer.

The mass spectrometer is a Mattauch-Herzog magnetic sector instrument with a radius of curvature (for the center of the MCP) of 5.9 cm. Ions are formed by electron impact. As shown in Fig. 1, a single Spiraltron multiplier is placed at a 3.5 cm radius to measure peaks with higher count rates than are suitable for the MCP. Magnetic mass spectrometers with single multi- plier detection have been described by von Zahn and Mauersberger [l] and Nier and Schlutter [17].

Most of the results described below were obtained when the gas was admitted in the form of a molecular beam into the ion source [18]. Gases present at pressures above 1 torr are expanded into a beam using orifices and liquid helium pumps. For this paper, details of the beam system are not of importance, except that it provides a convenient method of introducing gases into the mass spectrometer ion source.

The combined spatial resolution of the MCP, fiber optics, and photodiode array is about 70 pm. For all measurements reported here, the mass resolution M/AM = 200 is limited by the mass spectrometer and not the detector. The width of the imaged peaks is very close to the theoretical

88

width, indicating that there are only minimal aberrations caused by the magnetic and electric fields in front of the MCP detector and by the detector itself.

The mass range covered by the MCP depends on the size of the mass spectrometer and the position of the detector in the focal plane and the ion accelerating voltage. For this spectrometer, the range is about 1.72 to 1, i.e. if the lowest mass on the MCP is 50 u, the range will be 50-86 u. Since the mass spectrometer employed a permanent magnet, the mass range is adjusted by changing the ion accelerating voltage. The upper limit to the mass range is set partly by the resolution and partly by low detection efficiencies for the lower energy ions at high mass ranges. Both of these limit the upper mass to about 250 u. A larger mass spectrometer could extend the range to higher masses. For analyzing air samples, the lower mass range for measurements using the MCP is set by the presence of very large mass peaks such as 40Ar, 44(C0,), or even 46(C0,), depending on the sample pressure.

The lower sensitivity limit of the mass spectrometer is given by the dark count noise in the MCP. We measured the noise at 4 X lop4 counts s-l on each pixel with 1785 V across the MCP (a normal operating voltage) when the emission current in the ion source was turned off. The noise increased to 6.5 x low4 counts s-l at 1915 V. The spatial distribution of noise counts on different pixels appeared to be completely random, following almost exactly a Poisson distribution. This matches the experience of Timothy [19] who used a similar detector. For a mass peak covering three pixels, this noise level implies that a peak of less than 0.002 counts s-l could be measured.

It is our practice to bake the MCP detector at 180 o C after every exposure to atmospheric or even to rough vacuum pressures. The mass spectrometer is operated in an oil-free vacuum system with residual pressures below 5 X 10m9 torr. An outgassing procedure is occasionally used for the MCP by increas- ing the MCP voltage until feedback sets in. The voltage is slowly raised until a voltage, several hundred volts higher than the operating voltage, can be sustained with minimal feedback.

RESULTS AND DISCUSSION

An example of small peaks measured with such a detector is shown in Fig. 2. This spectrum of residual gas present in the vicinity of the ion source covers the mass range of 78-135 u. The total pressure was less than lo-” torr, and the largest peak in this mass range represents a partial pressure of about 1 x lo-l4 torr. The smallest discernible peaks are limited by statistical limits of the integration time (12 min) and not by the noise in the multiplier. Using a single electron multiplier, a scan over this mass range at a compara- ble resolution and sensitivity would have taken many hours.

89

0.10

1

0.08

i

am 0 2

PIXEL

Fig. 2. Part of a mass spectrum of residual gas in the ion source. Total pressure is 610-‘” torr. The partial pressure of mass 91, the largest peak in this mass range, is about lo-l4 torr. Total integration time was 12 min; data are plotted in counts s-l.

The dynamic range of the MCP detector is greater than 104. A spectrum illustrating this range is shown in Fig. 3, taken with a gas beam of air travelling through the ion source [18]. The ratio between 84Kr and 126Xe in air is 8.4 X 104. Note that 84Kr, the largest peak, has, in air, a mixing ratio of only 649 ppb. The highest dynamic range attainable would be from 1000 counts s-l to less than 0.01 counts s-l, a range of over 105.

The mass spectrometer itself has a dynamic range of over 10”. An electrometer in the proper position could have measured the N, peak present in the ion beam at the same time as the spectrum in Fig. 3 was obtained. A “finger” (shown in Fig. 1) was added inside the magnet gap to separate high and low mass ranges early in the passage through the magnetic field. An early separation greatly reduces background on high mass ranges due to ions scattered off neutral molecules or walls. This is especially effective for air samples, in which 99% of the gas has a mass less than 35 u and over 99.99% less than 45 u. The finger reduced the background from ion-neutral scattering to less than lo-” of the major N2 and 0, peaks.

The stability of the MCP detector has been excellent over approximately 2 years of use. The operating voltage has been increased from 1720 to 1785 V to maintain the same gain. The system has survived two parachute landings after balloon flights with no damage. On one occasion, we left the detector open to the atmosphere for several months. After pumpdown and a

90

0 128 256

PIXEL

384 512

Fig. 3. Spectrum of stratospheric air, 12 July 1985, from 78 to 135 u (‘36Xe is partially cut off). The integration time was 6 min. The peak heights between Kr and Xe should not be intepreted as upper limits on stratospheric gases, since a malfunction in the sampling inlet prevented some highly reactive gases from reaching the mass spectrometer. Kr and Xe were unaffected.

single 180°C bake, the gain vs. voltage curve was identical to one taken before the system was opened to the atmosphere.

Many applications of mass spectrometry require precise measurements of isotopic ratios. An example of the repeatability and precision of data from the MCP detector is given in Table 1, showing the ratios of count rates on several Kr peaks and ratios *°Kr/86Kr in air as measured over several weeks with the same tuning of the mass spectrometer (the same magnet strength,

TABLE 1

Ratios of count rates for krypton peaks measured in air as an example of the precision and repeatability of the MCP detector system

The errors quoted are la, based solely on sampling statistics. The ratios are not corrected for mass discrimination and do not represent true krypton isotopic ratios.

Date 80 u/86 u 82 u/86 u Sample source

21 May 85 0.114 f 0.004 0.613 f 0.011 Bottled, zero air 30 May 85 0.103 f 0.003 0.591 f 0.009 Bottled, zero air 19 June 85 0.107 f 0.004 0.610 f 0.012 Bottled, zero air 12 July 85 0.112*0.002 0.612 f 0.005 Stratospheric air

91

100 _

10 - . J32u

3 1

?

5

:: 0.1

0.01

0 128 258 304 512

PIXEL

Fig. 4. Spectrum covering 98-169 u for a sample of ultrahigh purity 0, after passage through a liquid nitrogen trap. The 0, obviously contained some Xe. The three most abundant Xe isotopes are each present at about 100 ppbv. The integration time was 6 min.

ion source extraction potential, etc.). Without the use of standards, the accuracy of isotope ratios is about lo%, in part because of mass discrimina- tion in the molecular beam system and ion source, but also because of non-uniformities in the MCP. For example, the isotope ratios taken directly from Fig. 2 agree with standard values to 9% plus statistical errors. Table 1 shows that repeated measurements of the same isotopic ratios are stable to within 2% over a period of many weeks. If internal standards are used, the accuracy would be determined by statistical errors.

One application of such a detector is the analysis of the purity of gas samples, as illustrated in Fig. 4. This spectrum covers 98-168 u for a sample of ultrahigh purity 0, [21]. The 0, was exposed to a liquid nitrogen trap to remove hydrocarbons. There is nearly 0.5 ppm of Xe in the 0,, while other contaminants in this mass range are less than 5 ppb. The entire spectrum was obtained in 6 min.

A more complicated spectrum is shown in Fig. 5. This spectrum is of stratospheric air near an altitude of 32 km taken during a balloon flight from Palestine, TX. The relative heights of 47(C0,), ‘*Kr, *‘Kr, ‘29Xe2+ all closely match laboratory calibrations. The ozone mixing ratio and informa- tion on the isotopic composition of ozone can be obtained from the peaks at 48, 49, and 50 u. An example of the usefulness of the imaging detector is the evidence of the long tail to lower masses from the 56 u peak, with no corresponding tail either to higher masses or on the larger 80 u peak. This is

92

moo r

0.1

0 12e

Xe++

258 384 512

PIXEL

Fig. 5. Spectrum of stratospheric air taken near 32 km on 30 October 1984 east of Palestine, TX. The spectrum is dominated by ozone. The CO,, Kr, and Xe peaks closely match laboratory calibrations.

due to 56(N2 a N2) dinners formed in the molecular beam system dissociating inside the mass analyzer.

A thud application of the MCP detector being pursued in this laboratory is the analysis of gas samples in a pyrolysis cell. The simultaneous detection of many masses is a clear advantage, since different species will be released from the solid sample for a limited time as the cell is heated. A mass stepping program for a single multiplier would either have to use very short integration tunes or risk missing some outgassing species entirely.

CONCLUSIONS

The MCP detector has proven itself to be a reliable and highly sensitive way of measuring a wide range of mass peaks. There are many applications for such a detector, some are stated above, but some limitations as well. The most important limitation is the time scale of the measurement. With a maximum count rate of 1000 counts s-l, the MCP is not well suited for measuring events on a time scale of 1 s or less. Either the peaks will be saturated or too few counts will be recorded to be statistically useful. CC-MS is another candidate for an imaging detector, but the time period during which a compound is eluted must be long enough for mass measure- ments to utilize the dynamic range of the MCP detector. Low repetition rate pulsed ionization methods, such as photoionization with a pulsed laser, may

93

produce low average count rates, the instantaneous count rates, however, may be too high during a pulse. The detector, however, is well suited for high repetition rate spark ion source mass spectrometry. Another promising application is the use of a MCP detector in a continuous laser photoioniza- tion mass spectrometer. The sensitivity and mass range of the detector would help compensate for the low efficiencies of the laser ionization. The selectivity of the laser ionization would avoid bombarding the MCP with huge peaks, allowing it to work near its noise limit.

ACKNOWLEDGEMENT

Financial support was provided by NASA’s Upper Atmosphere Research Program.

REFERENCES

1 U. von Zahn and K. Mauersberger, Rev. Sci. Instrum., 49 (1978) 1539. 2 J. Mattauch and R. Herzog, Z. Phys., 89 (1934) 786. 3 K. Mauersberger and R. Finstad, Rev. Sci. Instrum., 50 (1979) 1612. 4 K. Mauersberger, Adv. Space Res., 2 (1983) 287. 5 D.M. Murphy, PhD Thesis, University of Minnesota, 1985. 6 M.A. Coplan, J.H. Moore and R.A. Hoffman, Rev. Sci. Instrum., 55 (1984) 537. 7 D.L. Donohue and J.A. Carter, Int. J. Mass Spectrom. Ion Phys., 33 (1980) 45. 8 C.E. Griffin, H.G. Boettger and D.D. Norris, Int. J. Mass Spectrom. Ion Phys., 15 (1974)

437. 9 G.J. Louter and A.N. Buijserd, Int. J. Mass Spectrom. Ion Phys., 50 (1983) 245.

10 P.J.C.M. Nowak, H.H. Holsboer, W. Huegers, R.W. Wijnaendts van Resandt and J. Los, Int. J. Mass Spectrom. Ion Phys., 34 (1980) 375.

11 B. HedjfaII and R. Ryhage, Anal. Chem., 53 (1981) 1641. 12 G.F. Hartig, H.W. Moos, R. Pembroke and C. Bowers, SPIE J., 331 (1982) 44. 13 G.M. Lawrence and E.J. Stone, Rev. Sci. Instrum., 46 (1975) 432. 14 J.G. Timothy, Publ. Astron. Sot. Pac., 95 (1983) 810. 15 D.M. Mushy and K. Mauersberger, Rev. Sci. Instrum., 56 (1985) 220. 16 D.M. Murphy, Nucl. Inst. Methods Phys. Res. A, 236 (1985) 349. 17 A.O. Nier and D.J. S&hitter, Rev. Sci. Instrum., 56 (1985) 214. 18 K. Mauersberger, Rev. Sci. Instrum., 48 (1977) 1169. 19 J.G. Timothy, Rev. Sci. Instrum., 52 (1981) 1131. 20 EG&G Reticon, Sunnyvale, CA 94086, U.S.A., Model Reticon 1024-S. 21 Air Products, Hillside, IL 60122, U.S.A., Ultrapure Carrier Grade.