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

Подождите немного. Документ загружается.

300 Chapter 3.2: Mass Analysis and Partial Pressure Measurement

Linearity: Sensitivity may be independent of pressure only in a limited range.

Linearity measures the extent to which the change in output signal is proportional

to the corresponding change in partial pressure. Typically PPAs exhibit sensitiv-

ity independent from pressure within ±15% up to 10~^ torr (~10~^ Pa), above

which the sensitivity decreases with pressure. This pressure is referred to as the

upper operating limit. Above this pressure ion molecule collisions result in in-

creasing ion losses and the sensitivity will gradually approach zero. The linear

range of a quadrupole instrument depends on its construction but also the settings

of ion source and analyzer parameters—ion space charge effects can cause sig-

nificant non-linearity even at pressures orders of magnitude lower than the upper

operating limit. In certain cases the sensitivity may increase with pressure before

the sharp drop at the upper operating limit [95].

Resolving power, is the measure of

the

ability of the instrument to separate ions

according to their M/Q ratio. The ratio of the width of a peak (Am, at 10% of its

height in amu units) and the mass number (m, amu) is the conventional definition

of resolution. (R = mli^m\ R = 200 means that the peak width reaches 1 amu at

mass number 200.) Resolving power of magnetic instruments tends to be constant

along the mass range, while for quadrupoles it often varies proportionally with m

(depending on electrical control settings).

Another way to measure the extent of peak separation is simply to use the peak

width. Am (at 10% of its height), which is especially well suited for quadrupoles.

(Quadrupoles are often used in a way that peak width, Am, is constant over the

entire mass range.) It is to be noted, that as resolution increases, peak width de-

creases—

smaller number for Am corresponds to higher resolution.

Scan rate: The speed at which the information on mass analysis is available

(amu/s). There is a tradeoff between the scan rate and the noise level (random

fluctuation in the output signal unrelated to the partial pressure change). Use of

multipliers enable much faster scan rates at the same signal-noise ratio.

Stability: The change in sensitivity over a certain period of time. Contamina-

tion in the source, the analyzer and the electron multiplier, or shifts in electric

supply voltages are the major causes of degradation of sensitivity (and also re-

solving power) with time and use [95].

The following sections describe the components of a mass spectrometer sys-

tem. These are the inlet system, the ion source, the analyzer, and the detector.

3,2.2

INLET SYSTEMS

The mass analyzer section typically requires vacuum level below 10~^ torr

(~ 10

"-^

Pa). The ion source region may operate at a pressure of

one

or two orders

3.2.2 Inlet Systems 301

of magnitude higher than that using special designs, but still in a large number

of cases some means of sample pressure reduction is necessary. One way to cate-

gorize the different inlet systems is according to the sample pressure range, as

follows.

If the sample pressure is less than 10"^ torr (~10~-^ Pa), such as the residual

gas analysis of UHV systems, or UHV thin-film deposition processes, the ana-

lyzer may be simply immersed into the vacuum chamber or connected to a large-

diameter flange, possibly with an isolating valve. Immersion into the chamber as-

sures that especially with an open-type ion source the atmosphere of the chamber,

instead of an isolated pocket in a tube where filament-related or wall effects may

be more pronounced, is monitored. The possibility to valve the analyzer off can

assure that the analyzer is maintained at vacuum in case of temporary operations

at higher pressures or the cleanliness of the analyzer can be preserved during

short use of noninert gases.

If the pressure is within the range of 10-^-10 torr (--lO-'^-lO-^ Pa) and its

minimax ratio is 1:10 such as the case of certain process-monitoring applica-

tions,

a single, appropriately dimensioned orifice can be used. This is basically a

fixed aperture with an optional isolation valve. The flow through a small orifice

typically used (such as 0.05 mm) will be molecular below 1 torr (—100 Pa) [94],

therefore the flow into the analyzer will be of the same type as the outflow toward

the vacuum pump. (The pumping speed for gases of differing molecular weights

is assumed to follow ideal molecular flow laws.) Fractionation [96,97] of the

sample will not alter the composition unless the sample gas quantity is too small

and lighter components are depleted from the reservoir before the orifice. (In the

molecular flow regime the throughput is inversely proportional to the square root

of the molecular weight.)

If the max. pressure is below 10 torr (~

10-^

Pa), but the min:max ratio is higher

than 1:10, then a parallel arrangement of several orifices of varying size with in-

dividual shutofif valves can cover a pressure range wide enough on the expense of

complexity of the inlet system. It is very common to find a large-diameter high-

vacuum shutoff valve parallel with an orifice to perform residual gas analysis,

leak detection, and process control using the same instrument. Another alterna-

tive is to use variable leak valves that will cover a pressure range several orders of

magnitude wide with a single valve. Even though this is a simpler solution, there

are cases where the benefits of the high stability of a built-in fixed-size orifice,

which provides reproducible pressure ratio, outweigh the benefit of freely adjust-

able pressure in the analyzer. (Some leak valves exhibit large hysteresis and

creep;

their leak rate depends on the direction of adjustment (upward or down-

ward) with a step when changing direction (hysteresis) or the leak rate slowly

changes with time at a given setting (creep).)

If the sample pressure exceeds 10 torr

(—lO-"^

Pa), the simplest solution is again

the application of an adjustable leak valve or a fixed orifice. However, at such

high sample pressures the flow in the leak valve or orifice will be certainly in the

302 Chapter 3.2: Mass Analysis and Partial Pressure Measurement

viscous range while the flow out of the ionizer will be in the molecular range.

Since the viscous flow will carry each component uniformly without altering the

gas composition, while the molecular flow will preferentially carry out the lighter

components, fractionation will occur. This must be accounted for in calculating

the gas composition, or calibration must be performed with a mixture, whose

composition is close to the sample composition, at similar sample and analyzer

pressures. In that case the effect of fractionation will be "buried" in the calibra-

tion factor together with the various ionization and detection probabilities of the

different species.

In place of a simple orifice or a leak valve, a fritted leak or a capillary may be

used to reduce the sample pressure. The end of the capillary may lead directly

into the source, minimizing contamination, but its slow response (usually above

10 seconds, maybe minutes) makes it unpractical in many cases. Fractionation-

related problems must be dealt with as described earlier (calibration or calcula-

tions),

but capillaries do not present clogging problems to the extent associated

with orifices. (An orifice is an extremely "short tube," therefore its diameter must

be much smaller than the diameter of a long capillary of comparable conduc-

tance.) In case of fritted leaks, where many parallel leaks exist, clogging is usu-

ally a slow process, but often observed. Therefore filtering of particulates is highly

reconmiended in any case.

Another way to reduce pressure is to expand a small quantity of the high-

pressure sample (e.g., 1 cc STP) into evacuated larger-volume chambers (a few

liters).

To minimize surface effects and losing trace species several expansion and

pumpout cycles may be repeated. Such a system is referred to as a batch inlet.

The need for repeated expansion and pumpout cycles makes it rather unpractical

in case of environmentally dangerous gases. Correction for fractionation may be

necessary depending on orifice size and expansion chamber pressure [16].

A widely used method to avoid fractionation is the use of a capillary, whereas

the sample flows from the high-pressure side of the capillary toward a "Tee,"

whereas its pressure is reduced by two to three orders of magnitude. The other

side of the "Tee" is connected through an orifice to the ionizer, while the third end

is going through a restriction toward a vacuum pump. This continuous renewal of

the gas sample along with the orifice operating in the molecular flow range makes

it attractive, since through this arrangement much higher sample flow can be at-

tained than if the capillary would lead directly into the ionizer. Response times

are in the order of seconds or less. Several capillaries can be connected to a selec-

tor valve, forming a multicapillary inlet system saving evacuation and pumping

time and enabling fast switching between various samples. An alternative way to

multisampling is the use of zero dead-volume continuous-flow rotary valves (like

column-switching valves used in chromatography) in the standard piping before

the pressure-reducing element. A "Tee" element at the inlet port of the capillary

transfer line can assure continuous venting, while the capillary leads directly into

3.2.3 Ion Generation and Ion Sources 303

the analyzer. This provides the highest venting flow, since the capillary is not

restricting it, but at the expense of slower response time. (For attaining the same

pressure reduction through a single capillary, a smaller conductance capillary

must be used than in the case of the two-step pressure reduction — where a larger

capillary and an orifice are used.) Again, since the capillary operates in the tran-

sient flow regime fractionation will take place and calibration with gas of similar

composition to the sample is necessary. In case of adsorbing or easily condensing

sample, heated tube and heated capillary are recommended.

A membrane inlet is often used when sampling fluids or gases, especially in

medical or biological studies [26-29]. Its dynamic behavior has to be thorougly

tested for correct interpretation of the results. Since substances have different per-

meabilities, a slow response of varying time constants are to be expected for

dif-

ferent components and the depletion of the liquid film on the membrane must be

avoided by agitation. Other problems are the long-term stability and mutual inter-

ference of the permeability of various components. The only remedy is testing the

system over and over again.

Modulated inlet such as pulsed gas sampling [98] can largely eliminate insta-

bility of the background levels resulting from interactions with the sample gas

[19].

Modulated molecular beam coupled with phase sensitive detection improves

the signal-to-noise ratio and also used in the study of chemical reactions.

3.2.3

ION GENERATION AND ION SOURCES

The sample arriving into the ion source faces a fundamental change. Not only its

electrical state is changed, resulting in single- or multiple-charged ions, but often

fragmentation processes take place, i.e., chemical bonds may be broken depend-

ing on their strength. The vast majority of analyzers use the positive ions for sep-

aration because they are formed more easily. The existence of many ions from a

single component of a complex mixture may be confusing during their identifi-

cation; on the other hand, this "cracking pattern" contains useful structural infor-

mation on a molecule. One way to express the efficacy of this process of convert-

ing neutrals into ions is the ratio of

the

ion current and pressure (source sensitivity

[A/torr or

A/Pa],

which is often close to a constant up to a high-pressure limit).

There are several means to generate ions, as follows.

Electron impact is the most widely used method because of its simplicity,

ruggeddness, and high sensitivity. In this source a filament is heated to a temper-

ature high enough for thermionic emission (Figure 1). A grid, held at a potential

higher than the filament, accelerates the electrons (50-150 eV range). The accel-

erated electrons then enter the "ionization chamber," or ion cage, a region where

304

Chapter 3.2: Mass Analysis and Partial Pressure Measurement

Fig.l.

_1L.

3"C

^/ -^ n 1 I 1

I I.

—

L —

1 1 1

H H

111 ill

Electron impact ion source configurations. F = filament, R

L = ion-focusing lens, E = ionizer exit lens.

electron repeller, G = grid.

ion formation takes place, since the electrons have enough energy to form posi-

tive (and to some extent negative) ions from the neutral species present. (The ion-

ization potential of most species falls within the range of 1-25 eV.) There is an

exit aperture (focus plate) held at a lower potential to draw out the ions and pre-

vent electrons from entering the analyzer. Beside these basic features, several

more electrodes may be present depending on design, partly to help to form the

electron beam (such as a repeller plate, to direct the electrons toward the ioniza-

tion region, or Wehnelt electrode to focus the beam), partly to form and focus the

ion beam to prepare it for entry into the analyzer.

Filament materials [99] are selected based on the need to obtain high emission

current at possibly low operating temperature and long filament life.

Tungsten, as the highest melting point metal, has been used extensively, since it

provides high emission current per unit surface area (—0.4 A/cm^) at a tempera-

ture (~2500°K) where average filament life is several thousand hours (6 months

in continuous use is typical). Filament life at constant filament temperature is pro-

portional to the wire diameter used and for a given wire diameter life is a very

strong function of filament temperature (life ~ (fil. temp.)"^^; 10% decrease in

filament temperature (°K) increases life 40 times while 10% increase in filament

temperature shortens life to about 4% of its original value). Unfortunately, tung-

sten is sensitive to the Langmuir cycle, whereas residual water molecules react

3.2.3 Ion Generation and Ion Sources 305

with the hot filament, forming volatile tungsten oxides and hydrogen. The oxides

deposited on cooler surrounding surfaces will eventually be reduced by the hy-

drogen, resulting in a water molecule available for the next cycle and leaving a

metallic deposit behind. This may create unwanted current path on isolators or al-

ter potentials in the source. Another unwanted reaction of tungsten filament is

its reaction with oxygen, partly because of the deposit of tungsten oxide — which

may form insulating layers on metal electrodes. The charge-up of these layers can

result in change in the potentials in the source leading to inefficient source opera-

tion, unsatisfactory peak shapes, etc. Also, the carbon impurities in the tungsten

wire reacts with the oxygen on the surface, while the carbon is continuously

dif-

fusing to the surface. The resulting CO and

CO2

produce an annoyingly high back-

ground level. Molecular hydrogen dissociates at the filament, forming atomic hy-

drogen that is readily adsorbed by most surfaces. As a result, hydrogen signal is

less

—

local "pumping" of hydrogen occurs.

Rhenium is more resistant to water cycle, but its evaporation rate is about 150

times higher than that of tungsten for the same emission current density, there-

fore operation at a lower temperature is recommended. It does not form stable

carbides, therefore the high CO background experienced with W filament can be

lessened. Its resistance to other compounds is also known to be higher, although

halogen compounds will form oxyhalides

[100].

(Tantalum—another high-melting-point material —

useless as a filament

material because of its tendency to absorb large amounts of hydrogen, which at

high temperatures will alter its crystal structure causing extreme brittleness and

short life.)

Thoriated tungsten is a

0.5-1%

Th02 doped tungsten wire. At operating tem-

peratures the thoria will diffuse to the previously carburized surface, providing

free thorium. This monolayer will reduce the work function of tungsten by almost

a factor of

two,

providing high emission current per unit surface area (~4A/cm^)

at lower temperatures, therefore extending life.

Thoria-coated cathodes are made by a different process, wherein the base metal

is coated with a thick layer of Th02, which is activated by operation in vacuum.

The lower work function again enables operation at lower temperatures

[101],

therefore the base metal does not have to be a refractory metal. The choice is

therefore often iridium or rhodium to obtain a much higher level of oxidation re-

sistance. (Accidental exposure to air will not result in sudden burnout.) Even after

repeated exposure to corrosive gases constant emission characteristics were ob-

served. This filament provides advantages when operating at high ionizer pres-

sure because sample-filament reactions are suppressed, but also at extremely low

pressures where the lower filament temperature keeps outgassing of surfaces at a

lower level.

Lanthane

hexaboride-codittd

base metals can (and have to) be operated at lower

temperatures (<1400°K) also, to ensure long life and resistance to active gases.

At higher temperatures, the boron tends to diffuse into the base metal, deteriorat-

306 Chapter 3.2: Mass Analysis and Partial Pressure Measurement

ing its properties and the LaB^ evaporation rate will become high. Rhenium is the

most resistant to this boron diffusion, and so a preferred choice as base metal for

lanthane hexaboride cathodes. As opposed to pure metals, coated cathodes can be

poisoned by aggressive gases, resulting in elevated work functions.

Ion source configurations are as diverse as the special requirements they have

to meet. Still, it is possible to categorize at least the typical arrangements used in

commercial quadrupole instruments.

Axial (radial) sources are usually offered as the standard source. The name ax-

ial refers to either that the sample enters in the axis of the instrument (quadrupole

rods) or that the electron beam is in-axis (and the sample enters through all the

openings in the source). The name radial refers to the radial symmetry of the

arrangement, when ring-type filaments are used. Electrons are generated by a

filament either on axis, or by filament(s) in a plane perpendicular to the axis (ring-

type filament, or several filaments forming a polygon). These are intended for

general gas analysis or work at medium pressure (10~^-10~^ torr, ~10""^-10~^

Pa).

A version called

closed,

gas-tight, or enclosed design is available to improve

trace sensitivity. In this version the ionization chamber is separated from the

filament and the rest of the analyzer by small apertures, enabling it to operate it

at a higher pressure. This will increase sample-background ratio also limiting

sample-filament interactions. The sample is led directly into this chamber through

a small pipe.

Cross-beam sources will lead the sample into the source in a line perpendicular

to the axis of the analyzer. This is especially suited for molecular beams or poorly

coUimated beams of condensable material or aggressive gases. Liquid-nitrogen-

cooled beam stops or built-in beam shields then permit the continuous operation

on vacuum deposition systems, molecular beam epitaxy, etc., without forming

disturbing deposits in the analytical instrument. Again, an "open" and a "closed"

version or versions integrated with ion extraction optics may be available.

Grid,

open, RGA, or UHV source are all names covering an extremely open

design for high pumping speed, low outgassing rate, and low electron-stimulated

desorption. These sources are used at extremely low pressures, residual gas analy-

sis,

and UHV work. Instead of metal sheets, mesh are used for electron repulsion,

cage,

etc. If the instrument is intended as a residual gas analyzer, this may be the

standard ion source.

Figure

shows some typical configurations found in partial pressure analyzers.

(Sources a, d, e are variations of an axial-type source, c is a grid source, and b can

be used as a cross-beam source.) The electrons emitted by the filament (F) are ac-

celerated by the positive voltage applied to the grid (G) and directed toward the

ionization chamber in the center by the negative voltage applied to the repeller

(R) (both compared to the filament potential). Ions are extracted from the ioniza-

tion region by the negative voltage applied to the lens (L) or the exit aperture (E)

(voltage compared to the grid potential).

3.2.3 Ion Generation and Ion Sources 307

Special construction ion sources have been developed for particular applica-

tions,

particular analyzers or trying to deal with the imperfections of standard

sources. The most common ion source used in magnetic

j-^c^c^r

instruments is some

variant of the traditional construction described by Nier

[102].

An improved ver-

sion of the source provides additional focusing to improve transmission through

the analyzer and can be operated at higher accelerating voltages without arcing

[103].

Further improvements replaced the rectangular lens elements with ones

with circular symmetry, constructing the ion beam along the originally long side

of the rectangular element. This resulted in better precision, better transmission

and sharper focusing

[104].

A hemispherical cavity in the focusing (ion exit) elec-

trode acting as a Pierce lens can keep the potentials penetrating into the ion for-

mation chamber nearly constant. Thereby the ions are formed along an equipo-

tential surface, keeping their energy spread low

[105].

A source providing wide

electron energy range (19-70 eV), the incorporation of

spare filament and about

a 10-fold differential pumping factor have been described for space application

[106].

Several computer trajectory calculations were performed, aiming at pro-

ducing an ion beam of higher current density, with mixed success [107,108,109],

or aiming at understanding space charge and mass discrimination effects.

For ion sources operated with quadrupole instruments, trajectory tracing has

also been used to improve ion yield and sensitivity. In an effort to eliminate ions

from electron stimulated desorption

[110],

the electron-accelerating voltage is

modulated and the ion current is measured in a phase-lock loop

[111].

To increase

the high operating limit of the ion source, auxiliary electrodes were installed at

the inlet into the ionization chamber providing short, parallel electron path

[112].

Linear source response has been reported with such a source up to 10~^ torr

(~10-^Pa)[113].

Photo ionization is a "soft" ionization relative to electron impact. Since chem-

ical bonds are generally not broken, the amount of fragmentation or the strong

signal from the matrix can be suppressed. On the other hand, the sensitivities of

these sources are only a portion of the sensitivity of an electron impact source.

Compared to low-energy electron ionization (<15 eV), it is much easier to pro-

duce photons in a narrow energy range than electrons. Also, just above the ion-

ization potential, the efficacy of electron impact ionization is low compared to

the impact with photons. Trace contaminants in air have been measured using the

selective laser ionization — choosing the photon source so that the energy of the

light is above the threshold ionization potential for contaminants but below that

of the main components

[114].

Laser ionization opens up the way of extending

partial pressure measurement in the extreme-high-vacuum range (less than 10"^

torr,

—10""^

Pa) by avoiding filament- and electron-induced desorption related

problems

[115].

Alpha ionization source is a way to avoid hot-cathode-sample reactions. The

alpha radiation from a ^^^Po source provides a "cold" way to produce ions. It

308 Chapter 3.2: Mass Analysis and Partial Pressure Measurement

is extremely stable, being independent from external power supplies. However,

to prevent sublimation of polonium and the resulting increase in noise level of

the multiplier, it is necessary to place the source in a separate HV chamber with

a 2-4 fi Ni window. The sensitivity of the source was only about 10"^ A/torr

(-10-^^ A/Pa)

[116].

Field ionization is another method to avoid thermal effects. However, in its

general form it produces relatively energetic ions (keV range) and if coupled to a

quadrupole analyzer, deceleration is required, to slow down the ions into the

5-50 eV range. Complicated ion optics are necessary to prevent beam losses. A

single pointlike source, where the sample is flowing axially through a volcano-

shaped hole (—20 yLt) of the field emitter, with a grid counterelectrode held at

1-3 keV potential is capable to achieve sensitivity around 10"^ A/torr (~

10"

Pa)

[117].

Field emitter arrays, having a huge number of emitters on a small area

can achieve emission currents around 1 mA. Using this semiconductor-type de-

vice in place of a filament, low-energy ions can be generated since the high volt-

age (~100 V) is between the base and the gate of the semiconductor device and

does not affect the potential of the ionization chamber. Sensitivity approaching

that of a standard hot-cathode electron impact ionizer can be achieved (10~^ A/

torr;

—lO""*

A/Pa), but fluorine outgassing from the semiconductor is a problem

yet to be solved

[118].

Chemical ionization is a soft ionization method that found its way to partial

pressure analysis by the name of gas secondary ion mass spectrometry. In this

method a reagent gas (such as Kr, Xe, or CH4) is ionized in a chamber by electron

impact ionization. The resulting ions are extracted by electric fields and driven to

the charge exchange chamber where, through ion molecule collisions with the

sample, charge transfer takes place. Those constituents of the sample that have

ionization energies below the noble gas ions are ionized without significant frag-

mentation (no electron bombardment). This method, beside being "soft," also

opens the way to distinguish between different ion species of the same MIQ based

on differences in their ionization energies (CO vs. N2 or CH3OH vs. O2).

3.2.4

ION SEPARATION AND ANALYZERS

Analyzers are traditionally divided into categories such as "static" or "dynamic"

depending on whether the force field in the right-hand side of the equation of ion

motion is a function of time or not.

M d'^r

— -ry = E(r, t) + v(r, t) X B(r, t) (1)

Q dt"-

3.2.4 Ion Separation

and

Analyzers

309

(M = mass of the ion, Q = electric charge of the ion, E(r, t) = electric field in-

tensity, v(r, t) = velocity of the ion, B(r, t) = magnetic induction, r = position

vector of

ion,

t = time)

Since the force field used for separation is created by magnetic and/or electric

fields, another conventional way of categorization is referring to "magnetic" or

"electric" instruments. Whereas magnetic instruments usually apply a combina-

tion of magnetic and electric field, instruments that apply electric field only are

mostly of the "dynamic" type such as the electric quadrupole. Dynamic instru-

ments can be further divided into "path stability" or "time stability" instruments

depending on whether the ions of the selected mass-to-charge ratio are separated

from the others based on their path (electric quadrupole, magnetic sector), or

based on the time during which they travel a certain distance (time-of-flight).

3.2.4.1 Electric Quadrupole Analyzers

Quadrupole analyzers composed of four parallel metal rods driven by an RP gen-

erator are the most widely used "path stability" dynamic instruments (Figure 2).

Fig.

^ i,

Source

upply

Uni

SEM

Power

SuppIy

EIect rometer

«r,

c,^,.^^^

Quadrupole

Ion Source Anolyzer

Schematic of

electric quadrupole instrument.

Detector