than 10 /s. In the Py-MS configurations shown in Figures 18 and 20, the pressure rise in the ion source after each pyrolysis lasts for about 5-10 s. Thus the total
maximum number of ions measured per peak is well below 5.10 . Since at least 10 ions are required for a reasonably well defined (10% level) peak amplitude measurement, the useful dynamic range (ratio between smallest and largest accurately measured peak in a spectrum) is less than 50:1!.
For this reason,. Meuzelaar et al. used a specially built high-speed ion counting system, capable of counting up to 2.107 ions/s (ref. 43), thus obtaining a 20-fold improvement in the useful dynamic range. Construction of the high-speed ion counting system required special precautions to be taken against interference from outside noise signals, e.g. triaxial feedthrough and cable connections from multiplier to amplifier. The multiplier used was a Bendix 4700 channeltron multiplier. When using amounts of sample below 20 yg and ionizing at 14 eV, only a relatively small number of mass signals encountered in biomaterials exceeded the maximum count rate in this system.
Recording of pyrolysis mass spectra obtained by quadrupole instruments can be done either by a signal averager or by computer. Signal averagers are often suitable for interfacing with quadrupole mass spectrometers (ref. 44). Most signal averagers will supply a voltage ramp output which can be used to direct the mass scan of the quadrupole. Alternatively, the quadrupole may use its internal scan and the signal averager may follow the quadrupole scan in a trigger mode. Some signal averagers are even equipped with pulse count inputs allowing direct recording of signals generated by ion counting equipment. However, as with commercially available ion-counting equipment, the maximum count rate of pulse inputs on signal averagers generally does not exceed 2 - 3.10 pulses/s. The advantages of spectrum recording by signal averagers are simplicity of operation, ease of data inspection and high scan speed. Disadvantages, in comparison with signal recording by computer, are limited memory size, unsuitability for further data processing and inability to perform more complex scanning routines, e.g. jumping to selected peaks. The limited memory size will generally preclude the recording of time-resolved pyrolysis patterns. To allow further data processing, on- or off-line transfer of spectral data to a suitable computer system will be necessary.
Various computerized quadrupole systems are now commercially available. However, when selecting a computerized system for Py-MS applications, attention has to be paid to some special requirements not fulfilled by every system. In fact, none of the presently available mass spectrometer/computer systems is ideally suited for Py-MS applications. First, the computer should be readily programmable to perform signal averaging tasks for the purpose of obtaining a single integrated mass spectrum, especially if the available memory size does not permit sequential storage of all spectra for later summation. Secondly, if time-resolved pyrolysis patterns are to be recorded, the memory size of the computer should be large enough to accomodate sequentially recorded spectra or, alternatively, fast dumping of recorded spectra to disk should be possible during the measurements. Thirdly, the computer/mass spectrometer interface should be capable of handling ion counting signals instead of analogue signals. At present, this capability is available on one of the commercial c
Py-MS systems (ref. 120a) but only for count rates up to 2.10 ions/second. Finally, the computer system should be evaluated with regard to its capability for numerical evaluation of the pyrolysis mass spectra. None of the commercially available mass spectrometer/computer systems provides software packages for multivariate statistical analysis such as described in Chapter 6. Therefore, the user either has to develop his own software by translating available software packages from larger computer systems, or may simply want to make an on- or off-line connection to a large system. A minimum "minicomputer" configuration suitable for developing multivariate statistical analysis programs will probably have at least 64K core memory, a floating point processor, two disks (or one disk and one magnetic tape), a printer and a graphic display terminal with hard copy unit or X/Y plotter.
Was this article helpful?