Mass Soectrometric Analysis of Germanium - Analytical Chemistry

---

Spectrometric Analysis of Germanium RICHARD E. HONIG, RCA Laboratories, Princeton, S. J . A mass spectrometric method, employed in conjunction with a physical process of impurity-enrichment, has yielded semiquantitative analytical results on germanium samples used in transistor work. After a two-stage treatment in a gradient furnace, the enriched sample is vaporized thermally in a mass spectrometer, at temperatures that are raised in suitable steps. As a result, most impurities are separated, in temperature and time, from the bulk germanium. In this fashion, it seems possible to extend instrumental sensitivities from a present mole fraction limit of 10W6 to perhaps lo-*. After

T

HE electrical properties of a semiconductor are known to

depend i n a critical manner on the nature and concentration of impurities in the lattice. For this reason, the qualitative and quantitative analysis of germanium is of considerable interest. Furthermore, such analytical data are helpful in improving the chemical or physical purification procedures. The range of imand 10-5 mole purity concentrations of interest-between fraction-makes it difficult, if not impossible, to apply analytical methods based on chemical means, particularly because a chemical method is, of necessity, specific rather than general. Optiral spectroscopy, with a maximum claimed sensitivity of 10-6, shows considerably more promise, but is a t best a semiquantitative method that does not lend itself readily to the detection of the important Group V impurities that are encountered. I n contrast to the methods mentioned above, use of the mass spectrometer as an analytical tool offers advantages. I t is capable of determining quantitatively all impurities that have a sufficient vapor pressure (about mm. of mercury) a t experimentally realizable temperatures. The only element of interest excluded by this specification is carbon. The useful sensitivity of a mass spectrometer not specifically constructed for germanium analysis is about 10-5. Thus, taken by itself, this method would apply only to samples with relatively large impurity concentrations. However, the mass spectrometer, when used i n conjunction with a physical method of impurity enrichment, will yield meaningful results. Furthermore, it is believed that in an improved mass spectrometer tube the method of differential evaporation described below may extend the instrumental sensitivity limit to 10-6. METHOD

A typical mass spectrometer (see Figure 1) consists essentially of source region, analyzer, and collector. Vaporization and ionization of the sample material are accomplished in the source region. The resulting ion beam passes through electrostatic fields, where it acquires a suitable energy. In the transverse magnetic field of the analyzer, the ions are separated out according to their mass-to-charge ratio. Finally, the mass-resolved portions of the ion beam are made to fall, in succession, onto the collector, where their relative intensities are recorded. 41though mass spectrometric analyses of gases and liquids are, by now, largely a routine matter, only one detailed paper (6) has been published in recent years on the analysis of solids. This may be due to the fact that i t is difficult to produce steady ion beams that are representative of the impurity concentrations present in a solid. In the following paragraphs, various possible forms of ion sources suitable for this type of work are discussed, and the problems associated with quantitative analyses are investigated.

the sample is completely evaporated, a time integration of ion currents yields concentration estimates, whose accuracy is mainly limited by uncertainties in the ionization cross sections. In analyses made on Eagle-Picher germanium, small concentrations of the following impurities were found: arsenic, phosphorus, antimony, tin, lead, iron, and cadmium. Calibration runs that have been made on germanium, tin, and lead show that these Group IVB elements evaporate partially in molecular form, while in the case of arsenic the molecular species As4 and As2 predominate.

Suitable Ion Sources. I n their analyses of stainless steel samples, Gorman ef al. (6) used a modified spark source of the Dempster ( 2 ) type. The unavoidable fluctuations in source intensity were overcome by measuring the ratios of individual ion currents to total ion current, rather than the individual currents themselves. In this fashion, analyses reproducible t o 0.1 % were obtained, but no statement is available concerning the ultimate sensitivity of this method. Apart from the problem of adequate sensitivitj; equipment of a very specialized nature (a Denipstertype double-focusing mass spectrometer) has to be constructed to permit the use of a spark source. E

SOURCE

\H\

COLLECTOR

/

@

MAGNETIC FIELD

I

Figure 1.

Diagram of XIass Spectrometer Tube

I t has been suggested ( 1 8 ) that the so-called exploding wire technique ( I ) ,used as a light source in optical spectroscopy, be employed to evaporate the sample completely in a short time (about 10-6 second). By this means it should be possible to increase greatly the intensities of small ion currents without a similar increase in the background, thus improving the useful sensitivity of the system. However, because of the difficulties involved in recording ion currents of such short duration, this suggestion has not yet been explored. The method of complete thermal evaporation has been briefly in mentioned by Hickam ( 7 ) , \Tho reported sensitivities of the analysis of copper samples. For the present study, it was concluded that a thermal evaporation source would involve the simplest experimental setup, permitting the use of existing equipment with only minor modifications. I t was realized that, in order to eliminate the effects of fractional distillation, samples would have to be evaporated completely and ion currents integrated over time. Upon completion of the study described below, details of Hickam’s work were received (8) which indirate 1530

V O L U M E 2 5 , NO. 10, O C T O B E R 1 9 5 3

1531

that he introduced his samples into a furnace kept at a fixed temperature such that the impurities evaporated from the molten copper sample in less than a minute. Quantitative Analysis. I n the quantitative analysis of a solid sample by complete thermal evaporation, the current representing the number of ions of a given species collected per unit time a t the receiver of a mass spectrometer is given by

I T = ~QI-njl

(1)

where efficiency of positive ion collection-i.e., the ratio of ions collected to ions formed in the ionizing region Q = ionization cross section, sq. cm. I - = electron beam current, inunits consistent Jviththoseof I + ni = concentration of atomic species in ionizing region, cc.-1 1 = active path length of electrons, em. 7~

=

In order to simplify the mathematical treatment, two approximations are made. First, it is assumed that the particles evaporating from the sample leave the evaporator as a parallel beam, whose cross section is defined by the crucible opening (see Figure 2 j. Second, the condensation coefficient is taken to be unity for all species-Le., all particles will cross the ionizing space but once. Thus, the rate of evaporation of particles from a source kept at temperature 1' is given by

.\-/A

=

3.63

x

103n~(~/111)~~

(2)

where

3- = cZS: dt = number of payticles e v a p o r a h g , per second A = particle beam cross section, sq. cm. ni

=

r ,

= =

1 111

particle concentration at source and in ionizing space, cr.-1 sample temperature. a I