Hoffman D.M., Singh B., Thomas J.H. (Eds). Handbook of Vacuum Science and Technology

Подождите немного. Документ загружается.

350 Chapter 3.3: Practical Aspects of Vacuum System Mass Spectrometers

spectrometer. When the grid was degassed above 400°C, the 16 peak substan-

tially diminished, as predicted from studies of atomic oxygen adsorbed on Ag

[22].

Replacement of the grid with a standard molybdenum grid returned the

spectrum to the more conventional series shown in Figure 3. Minimizing this un-

wanted gas desorption can be accomplished by operating at low emission cur-

rents,

minimizing the grid or collector area, and using materials that have a low

propensity for adsorption.

3.3.3.6 Charge Exchange Mechanisms

In addition to electron impact ionization, dissociation, recombination, sputtering

and

BSD,

charge transfer can also affect the actual ionization volume particle mix.

At normal operational pressures for mass spectrometers (p < 1 X

10 ""^

torr), the

mean free path (1) is sufficiently long that ion-molecular impact is not a signifi-

cant process

< 50 cm), but it does occur and can be a significant signal in some

applications. Essentially, the collision of a positive ion with a gas atom or mole-

cule results in an exchange of

energy,

momentum and direction, but also a change

of charge can occur. An energetic ion can "steal" an electron from a neutral and

become an energetic neutral, leaving behind a slow ion. Actually, the cross sec-

tions for charge exchange vary substantially with the difference in mass between

particles. The transfer is more likely to occur in the proximity of resonance or the

smaller the energy exchanged in a collision. Thus, if the incident ions are the

same species as the target atoms or molecules, the exchange has a higher prob-

ability. The amount of energy transferred is equal to the sum of the kinetic and

ionization energies plus the excitation energy of the target molecule (including

vibration and rotation) less the neutralization energy. The charge transfer may

also include other processes such as dissociation, excitation (with subsequent

light emission), and ion formation. Some examples of exchange are [23,24]

He+ + H20-^He + H20+ (12)

H2+ + H2O -^ H2 + H2O+ (13)

H20^

+ H->H30+ (14)

H2 + H+->H3^ (15)

Ar++ Ar->Ar +Ar-' (16)

N2''+ Arf->Ar^+N2 (17)

He+ +Xe->He + Xe-^ (18)

Since hydrogen and water are so prevalent in vacuum systems reactions, (12) to

(15) are often observed. The production of

H^"^

is frequently seen in ion sources

especially with environments of hydrogen-containing species, such as hydrocar-

33.4

Techniques

for

Analysis

351

bons.

Argon and other inert gases have been used for many years in charge neu-

tralization cells for forming energetic neutral beams of atomic oxygen and other

species.

3J.4

TECHNIQUES FOR ANALYSIS

3.3.4.1

Partial Pressure Measurement

The accuracy and overall performance of the mass spectrometer is a function of

each section of the mass spectrometer. As an example, consider a quadrupole-

type instrument that is primarily comprised of an ion source, a set of quadrupole

rods for mass filtering, and an electron multiplier for signal amplification (see

Figure 8).

THE ION SOURCE

For a given ion source geometry, the ionization efficiency is primarily a function

the

number of electrons in the atom or molecule. As previously indicated, the

highest probability for ionization of most gases occurs at approximately 70 eV.

The quantity of ions formed and extracted from the source is a function of the

design of the ion source and is characterized by a source factor (ionization effi-

ciency). Flaim and Ownby [25] have determined an equation that approximates

the ionization efficiency (/).

/ = 4.3 X 10-2 ^ „.2. + 0,4 (19)

where

is the number of

electrons

in an atom of

molecule

i and

rii

is the number

atoms

in molecule

As an example, consider the water vapor molecule, H2O;

it has 2 atoms of

H (1

electron) and

atom of

(8 electrons), thus

/[H2O] = 4.3 X 10-2 [2(1) + 1(8)] -f 0.4 = 0.83

which is the ionization efficiency relative to nitrogen. Figure 9(a) shows the

fit

the equation for several atoms and molecules. There is also a geometrical factor

similar to a gauge constant that reflects the efficiency in which the ions are ex-

tracted from the source and directed to the mass filter.

THE MASS FILTER

When the ions enter the mass

filter,

those ions that match the selected filtering

conditions—i.e., those which have the right mass to charge ratio, entrance vec-

352 Chapter 3.3: Practical Aspects of Vacuum System Mass Spectrometers

Rg.8.

Reflector

Filament

V (ion energy)-<-f

V (focus) -^

• Ion source

+(VQ+V, cos cot)

-(Vo+V,COSCDt)

l\/lass

separator

-3kV -»

|— Electron

multiplier

Actual and schematic sections of a commercial mass spectrometer. (Reprinted with permis-

sion of UTI division of MKS.)

tor, and kinetic energy—have a particular probability of exiting the filter [26]. In

a quadrupole filter, this transmission (T^/e) factor is essentially 1 for the Ught

masses (1-20 amu) and then decreases logarithmically with mass. Figure 9(b)

shows the typical behavior as a function of

mass.

Note there is a band for masses

higher than 20 amu that takes into account variances in molecular degrees of free-

33.4 Techniques for Analysis

353

Fig.

(a)

^HefNe

Xe/^

O Dushman 4

Vexing

a Dushman

A Wag«nor& Johnson

V Sc^ula

0 Anderson

• Vkrion

H Roihe

• UttefbackiGriflim

9 Standwd

•

N.(

number of electrons / molecule '\

(b)

100 150 200 250 300 350

Mass,

amu

(C)

(a) General

fit

of ion source sensitivity versus number of electrons per molecule (after Flaim

and Ownby [25]). (b) Representation of the transmission probability versus mie. Shaded area

represents error

bar.

electron multiplier as a function of mass.

354 Chapter 3.3: Practical Aspects of Vacuum System Mass Spectrometers

dom, ion velocity (time in the filter), rf frequency, and potential uniformity in the

radial and axial dimensions. As shown in the figure, the transmission decreases to

about 0.5 at 75 amu.

THE ION DETECTOR

On exiting the mass filter, the ions are collected by a Faraday cup or amplified by

an electron multiplier. A well-designed Faraday cup should completely collect all

ions and not gain or lose any signal due to secondary emission of electrons. An

electron multiplier (continuous dynode, box and grid, Venetian blind, multichan-

nel plate, etc.) can provide a gain (G) from

X 10-^ to 1 X 10^ depending on the

number of electron collision stages and on the secondary emission coefficient of

the dynode or collision materials. Most often the gain is adjusted to a modest

value, e.g., G~lX10^or less in order to ensure long-term stability. Further, the

gain is very sensitive to ion mass, chemical reactivity, kinetic energy, and total

charge. Some multipliers have been observed to have a very high gain for low

masses and decrease as a function of m~^^^ with increasing mass

[27]

—

see

Fig-

ure

9(c).

In addition, the secondary emission coefficient from impacting ions varies

as the cos^ 6, where 6 is the angle of incidence of the ion. A calculation of the

partial pressure for a given species can be determined from

GiTi

(20)

Considering an actual case where the partial pressure of H2O has been measured

one can compare the calculated value. The primary peaks are m/e =18 (H2O''")

and the fragmentation peak is m/e = 17 (OH^). If the peak height currents are

measured to be

10 ~*

and 2.3 X

"^ A, respectively, and the ion source sen-

sitivity is 1.5 X 10~' A/torr and the electron multiplier gain for N2 is 1 X 10^

then one can calculate a value for the partial pressure for H2O. The G for H2O is

G[H20] = G[N2]

m(N2)

mCHjO)

1/2

(21)

= 1 X loM—- =1.25X105

and for the fragment OH is

''28\"^

G[OH] = 1 X loM— = 1.28 X 105

^17/

Since 7-1 for <20 amu, then r[H20]~l, r[OH]~l

3.3.4 Techniques for Analysis 355

Employing Equation (20), the calculated partial pressure is

/?[H20] =

1X10-^

2.3X10"^

1.25 X 10^(1) 1.28 X 10^(1)

(1.5 X 10--^)0.83

/7[H20] = 1.9 X 10-^4oiT

The actual measurement from the calibrated mass spectrometer was

/7[H20] = 1.5 X 10-^^torr

Although the approximate answer from the calculation is instructive and useful,

there is still significant error. Clearly, the calculation does not contain every cor-

rection necessary to make an accurate assessment of the partial pressure! In gen-

eral, the ion source limitations, the filter and detector errors, and anomaUes make

quantitative measurements quite difficult. Calibrations in the selected gases of in-

terest are absolutely essential to obtain meaningful data.

3.3.4.2 Gas Identification

Two primary techniques enable the user to unambiguously analyze an unknown

gas species or mixture of gas species to clearly quantify the individual fractions

each.

They are the isotope analysis [28] and the fragmentation analysis [29]. In

most general instances, a combination of both methods will be employed. Al-

though isotopic analysis is a useful technique, it cannot supplant the use of frag-

mentation analysis. Fragmentation analysis typically works from a much greater

number of larger-amplitude mass peaks and is thus likely to give a more complete

rendition of the species present. Obviously, to get an accurate species identifica-

tion, a calibration of peak height for parent and fragments (using the instrument

making the analysis) is required. Even without the calibration the spectral stan-

dards are useful as guides for determining species that may be present in the sys-

tem being analyzed. When analyzing for species, one should always attempt iso-

topic analysis first, followed by fragmentation analysis for confirmation.

ISOTOPE ANALYSIS

The main advantage of using the isotope method is that normally no calibration of

the analyzer is required for identification of a species, although it would be re-

quired for quantification of

results.

The isotopic abundances have been accurately

measured for the elements of interest. In using isotopic identification, it is impor-

tant to keep in mind several factors. The analyzer detector system must be capable

of linear output over three to four orders of magnitude for best results because the

356 Chapter 3.3: Practical Aspects of Vacuum System Mass Spectrometers

less abundant isotopes of most of the elements of interest are at the 0.1 to 1.0%

level. Modem amplifiers are linear and stable over five-seven orders of magni-

tude or better, so this factor should not be an issue. It is also important that the an-

alyzer sensitivity not be strongly dependent on the mass number. Most quadru-

pole instruments have a mass sensitivity that decreases at

-2% per amu beyond

12 amu. Most of the isotopic species of interest appear above this range. Species

appearing below the linear range are uniquely identified on other bases. Finally,

due to the low peak height of less abundant isotopic peaks, it is important that

they not interfere with other fragments. Ordinarily this is not a problem because

isotopes of species of interest do not overlap and with few exceptions, the less

abundant isotopes appear at higher mass numbers. One exception is that due to

argon. Natural Ar at 40 amu is the most prevalent isotope and is the decay prod-

uct of an abundant element, thus "^^Ai/^^Ai = 300. The abundances of the iso-

topes of all naturally occurring elements are given in the American Institute of

Physics Handbook, Table 8.6.1 [30]. Most analyses in vacuum systems and on

vacuum processes involve only the lightest 18 elements in the periodic table.

The results of an isotopic analysis at the 1% level should be within a few per-

cent of

the

level predicted. At the

0.01%

level, the results should be within several

tens of percent of the level predicted from natural abundance. If one arrives at val-

ues outside these ranges, then interference is surely present and an isotopic analy-

sis will need to be supplemented with fragmentation analysis.

When looking at pure elemental gases, such as nitrogen and oxygen, one may

find the heavier isotopes to be slightly depleted. This occurs because these pure

gases are obtained by fractionation. The lower-molecular-weight species boil at

lower temperatures. The rarer higher-boiling-point species may be separated and

sold at a greater price of various isotopic tracer experiments. A slight depletion

can also occur if the gases have been introduced by diffusion or molecular flow,

because the flow is dependent on (m)

~^^^.

Any of these effects typically alter the

ratios by a maximum of

~3%.

When looking at an atomic species, the expected ratio of the isotopes can be

obtained directly from the aforementioned table. It is noteworthy that the rarer

isotopes of the elements of primary interest—i.e., hydrogen, carbon, nitrogen,

oxygen, and chlorine with an atomic number (Z) of 18 or

less

—

are

1 or 2 amu

greater than the major isotopes for the element. This means that the largest rare

isotopic peaks for all molecules will appear at 1 or 2 amu greater mass than the

peak of the molecule in question.

calculate the relative abundance of the isotope,

peak height of isotope species ^ ,^^^

Relative abundance (%) = , , . , \ -^---^ — X 100 (22)

peak height of most abundant species

For illustration, consider the spectra of Ne. The spectra for Ne at masses 20, 21,

and 22 amu was observed to have peak heights of

90.92,

0.257, and 8.82 arbitrary

3.3.4 Techniques for Analysis 357

units,

respectively. The relative abundance in percent is, then,

For^ONe-

[^^Ne]

90.92

Fo-^'N. liMx,00 = 5:^X100-0.282

{mie = 21) [20Ne] 90.92

For^^Ne-

[^^Ne]

8.82

imie = 22)

[^''Ne]

90.92

The resulting spectrum of Ne and its isotopes should therefore appear in these

proportions. A second example, CO, will show peaks at masses 28,29,30, and 31

because CO has isotopes of both carbon and oxygen. The relative abundances

should be

For '2C'*0- ["^C] X ['^O] 98.89 X 99.76

^^^ ^ ^- -V-!—\^ X 100 = X 100 = 100

{mIe = 28) ['^C] X ['^O] 98.89 X 99.76

For'-^C'^O + '^C'^O: ['^C] X ['^O]

['-^C]

X ['^O]

imIe = 29)

[

''C] X

[ '^O] [

'^C] X

[

"-0]

1.11X99.76

98.89X0.0374

X 100 + ——-— X 100 = 1.16

98.89 X 99.76 98.89 X 99.76

For'^C'^0+>^C'«0: ['^C] X ['^Q] ["C] X ['«Q]

(m/e = 30) ['2C] X ['^O] ^ ^^ ^ [i2c] x ['^O]

1.11X0.034

98.89X0.204

100

+ ————— X 100 =

0.205

98.89 X 99.76 98.89 X 99.76

For

i-^C'^O-

V^C] X ['^O] 1.11 X

0.204

M. =

31)-

[-^C]X[-^0]^^^ = 98.89X99.76^^^ = ^-^^^^

Clearly, in order to observe the full isotope spectrum contributing to the CO

signal, the instrument must be sensitive enough to detect at least five decades of

signal intensity less than that of the parent peak at mIe = 28 amu. If the partial

pressure of CO is ~1 X 10"^ torr, one requires an instrument with a minimum

detectable partial pressure of ~1 X

10 "'"^

torr to fully see the isotope signal at

rule = 31. In addition to sufficient sensitivity, it is also necessary for the instru-

ment to be linear over the full range to accurately compare the relative peak

heights. Armed with this spectrum, one can now unambiguously identify this gas

if it is the only species in this mass range. Unfortunately, the situation is seldom

this simple unless one is backfilling with CO from a pure source. Molecular ni-

trogen and ethylene are also contributors to mIe = 28, thus requiring the decon-

3S8 Chapter 3.3: Practical Aspects of Vacuum System Mass Spectrometers

volution of the overlapping data. As previously discussed, a leak-tight vacuum

system will not have nitrogen as part of the residual spectrum, but there are in-

stances where a leak may exist or some operation involving N2 may occur. Thus,

in this case, mie = 28 is obviously not pure CO or pure N2. A similar analysis to

that just conducted for CO shows that pure N2 has normalized mass peaks of 100

at mIe = 28,0.75 at mIe = 29, and 1.4 X

10 ~^

at mIe = 30. Assume that the mea-

sured spectrum has 100 units of CO and N2 at mIe = 28, 1.0 unit at mIe = 29,

0.14 units at m/e = 30 and 1.7 X 10"^ units atm/e = 31. The following analysis

can be applied:

Let

= CO fraction and y = N2 fraction;

For m/e = 28: x + y = 100

For m/e = 29: 0J5y +

1.16A:

= 1.0

= 0.61 andy = 0.39

(61%

CO and 39% N2)

To check if

this

is reasonable, the m/e = 30 peak can be examined to see if the ex-

pected relative fraction correctly corresponds to 61% CO. Since for comparable

amounts of CO and N2, the value of the N2 isotope contribution at m/e = 30 is

so small (1.4 X

~^),

the m/e = 30 signal must be predominantly CO. If we di-

vide the observed peak height of 0.14 at m/e = 30 by 61 at m/e = 28, one obtains

0.23 X 10"^, which is in reasonable agreement with the 0.21% for m/e = 30 for

CO.

C2H4

had also been expected as a contribution to m/e = 28, the m/e = 30

peak would have to be used to determine the fraction of CO in the three-way mix-

ture.

Although the fractional contributions in this example give us an idea as to

the relative amounts of each species, this does not mean that the partial pressures

are related in the exact same ratio. The partial pressures would have to be evalu-

ated by calibrating the instrument for these gases. The complexity of calculating

the expected values for isotopic peaks obviously increases as the number of atoms

in a molecule increases. For more detailed analysis of this procedure, please see

the notes of

C. Beavis [28].

FRAGMENTATION ANALYSIS

In general, a clean, dry-pumped vacuum system will have mostly inorganic resid-

ual gases (usually with the exception of CH4). If the system is unbaked, the pri-

mary residual gas will be H2O and perhaps some solvent residues like acetone

and/or ethanol. The inorganic gases are, in most cases, 2- or 3-atom molecules

such as H2, H2O, CO, O2 and CO2 with even mass numbers. Further, ionization

is more probable than is molecular dissociation, so the fragmentation patterns are

dominated by the parent peak. The inert gases are stable, nonreactive atoms. If the

vacuum system is pumped by an untrapped rotary vane and/or a diffusion pump,

the residual gases will be dominated by organic vapors and gases. If the pumps

are well trapped, organics will be present, but dominated by inorganic gases. Or-

3.3.4 Techniques

for

Analysis

359

ganics contain many more atoms than an inorganic molecule and are more likely

to dissociate before ionization, so the parent peak will, very often, not be pre-

dominant. Dissociation usually occurs at the carbon-carbon single bonds, which

means that a CH2 (Am = 14) or a

CH3

(Am = 15) group is the first to break away.

The resulting fragment molecules are usually an odd mass number. For example,

ethanol (C2H6O) is an often-used solvent cleaner in vacuum systems and usually

leaves a residual vapor. Under electron impact, the methyl group in this molecule

is the the most probable dissociation fragment leaving the CH2OH {mie = 31) as

the dominant peak. Subsequent to that, the single bonds associated with hydrogen

are the next most likely to dissociate, so the subsequent fragmentation proceeds

in descending steps of one mass unit.

Consider a stainless steel turbomolecular pumped system that has been moder-

ately baked at 150°C for 24 h. Figure 10 shows the quadrupole spectra for such

a system. The water vapor is substantially reduced and the remaining gases are

primarily H2, CH4, CO, and CO2. Some Ar is also present from an earlier back-

fill. Analysis of this system by fragmentation involves the fragmentation patterns

of all these gases and they are shown in the previous section (see Chapter 3.2,

Fig. 10

10-^

io-«-d

10-^

lO-'M

10-'

Turbomolecular pumped

150° Cbakeout (24 hours)

p~2xl0-'°torr

I I I I I I I I I

I I I I I I I

I I I I I I

TTTT TTT

TTTTTl

0 5 10 15 20 25 30 35 40 45 50

m/e

The mass spectrum of a turbomolecular-pumped system that has been moderately baked

(/7~2X 10-10 torr).