Note: Descriptions are shown in the official language in which they were submitted.
~0~5Z53 ~ :
1. Field Of The Invention
This invention relates generally to the field of com-
pound identification and, more particularly, to a novel system
; having the capability to rapidly, automatically, and spècifi-
~ 5 cally identify and ~uantitate any one of a set of pre-selected¦ compounds from unknown and impure samples. The present in-
vention is described with respect to an embodiment incorporat-
ing a mass spectrometer.
2. Prior art
The use of mass spectrometry for the identification
of compounds and determination of their molecular structure
is well known in the art. In a mass spectrometer, a sample
gas is partially ionized by electron impact or other means in
an ion source. For each compound in the sample, a set of
fragment ions are typically formed, each one having a par-
ticular mass to charge ratio, m/e, where m is the mass of the
ion in atomic mass units and e is the charge of the ion deter-
! mined by the number of electrons removed therefrom by the
ionization. The mass to charge ratio, m/e, is usually re-
~¦ 20 ferred to as "mass".
The ions are separated by electric, magnetic or com-
! bined fields (in a mass analyzer) into defferent species
¦ according to their respective masses. In the usual arrange-
ment of the mass analyzer, ions of one mass at a time are
' 25 transmitted to a suitable detector, typically an electron
multiplier, for measurement and/or recording. Usually, the
mass analyzer controls are manipulated so that the m/e values
are repeatedly and continuously swept over a selected mass
range. A plot or tabulation of ion current or intensity vs
, 30 m/e is referred to as a mass spectrum and is the basic data
.. . .
.; ~
'
1045Z53
output from a mass spectrometer. It should be understood
that references to mass peaks made herein may apply to the
amplitude of said ion intensity, the integral of ion intensity
with respect to m/e, or any other quantitative measure of the
S presence of ions. If the mass separative power or resolution
of the mass analyzer is such that integral values of m/e can
be separated but fractional values cannot, the technique is
referred to as low resolution mass spectrometry.
Interpretation of the mass spectrum has two related
but somewhat different objectives. Identification is the
process of determining which (if any) compounds in a pre-
determined list or library are present in the sample, by
means of a comparison of the sample spectra with previously
recorded spectra of known pure compounds. In structure
determination, part or all of the molecular structure of an
unknown compound are deduced from the mass spectrum. This
invention relates to the identification function of mass
spectrometry (and the quantitation of the compounds identified).
The mass spectrum analysis for identification purposes
may be performed manually or with the assistance of electronic
analysis means. For a manual analysis, a skilled mass spec-
trometrist is normally required to study the data for features
which suggest the possible identity of the compound sampled.
Tables, computations, and application of rules of the forma~
tion of mass spectra are typically used. The final identifi-
cation is usually made by a comparison of the selected sample
spectrum with a published or measured spectrum for the com-
pound identified.
Mass spectrum identification by electronic analysis
~0 - means is typically accomplished by encoding or contracting
--3--
.~ .
- 7
S253
the mass spectrum according to any one of a number of rules.
Examples of some of the rules used include: (i) selection of
"n" most intense peaks; (ii) selection of one or two of the
most intense peaks in each mass range of 14 amu; and (iii)
binary encoding (indicating presence or absence only~ of all
peaks. The encoded spectrum is then compared with each of
a number of similarly-encoded library spectra. Based on some
criterion of similarity, the compound whose library spectrum
most closely resembles that of the sample is identified as
the compound which was sampled. Often more than one possible
identification is provided, with the final identification -
being left to the operator based on comparisons of complete
or encoded spectra.
Use of mass spectrometry for identificat~on presently
requires purification of the sample by physical or chemical
means. One or more abundant ions from an impurity could
conceivably misdirect the identification of the compound
sought. The purification may be accomplished before intro-
~` duction of the sample into the apparatus, or a separative
device may be attached to the mass spectrometer. Externalmeans of sample purification include extraction (using con-
trol of pH and suitable solvents to preferentially dissolve
the compound(s) of interest), distillation, recrystallization
and thin layer chromatography. The separative device most
commonly used with a mass spectrometer is a directly inter-
facing gas chromatograph (GC), providing the well known gas
chromatograph/mass (GC~MS) instrument. Recently, a liquid
chromatograph has been used in conjunction with a mass
spectrometer for separation of the sample.
In a gas chromatograph, a sample comprised of one or
--4--
. . . .
: , '
1~45253
more compounds is injected by a syringe or valve into a
heated chamber (the flash evaporator or injector~ or directly
into a chromatographic column. The sample is vaporized (if
it is not already a gas) and transported through the column
by a suitable inert carrier gas. The column is a glass or
metal tube, usually packed with a powdered support material.
The tube or the support material therein is coated with an
organic liquid, called the stationary phase or liquid phase.
The liquid pahse has the property of absorbing and desorbing
each of the constituent compounds in the sample at different
rates, thereby causing a different rate of slowing of each
compound as it passes through the column. As a result, the
~ifferent constituents comprising the sample pass through the
column at different rates and emerge therefrom at different
times. Under fixed operating conditions (column type and
temperature, flow rate, etc.) each compound has a character-
istic, reproducible retention time, or delay from injection
to elution at the column outlet. In this manner, a mixture
is separated into its constituent compounds. Each compound
then flows into the mass spectrometer for identification.
An interfacing device or separator is usually re-
quired between the outlet of the gas chromatograph and the
inlet to the mass spectrometer, because the pressure at the
column outlet is typically one atmosphere, while the mass
spectrometer must operate in a vacuum of the order of lQ-8
atmospheres. The separator transmits a reasonable fraction
of the compounds of interest while excluding most of the
carrier gas. Usually, a suitable non-selective detector is
used to indicate GC peaks, that is, the elution of each of
the constituent compounds. At least one mass spectrum is
--5--
104~ 53
taken as each GC peak is detected.
The principal disadvantages of the instruments known
in the prior art are as follows:
(a) Where the mass spectrometer is not equipped with
a separative device such as a gas or liquid chroma-
tograph, lengthy, complex and tedious sample puri- .
fication methods must be employed.
(b) Where a separative device is used, the device it-
self imposes additional limitations. For example,
only those constituent compounds of a sample which
can be successfully separated by the separative
device can be analyzed. Secondly, the operation
of the separative device typically requires at
least several minutes; thereby, it introduces a
time delay. Moreover, the separative device in-
creases the complexity of the instrumentation and
the skill required to operate it.
(c) Manual interpretation of mass spectra requires
a great deal of time and skill.
(d) Computing means for aiding the interpretation
of mass spectrum used in the prior art, suffer
from one or more of the following disadvantages:
(i) a significant amount of operator interven-
tion is necessary and, therefore, the interpreta-
tions are not fully automatic. In addition, a
relatively high degree of operator skill is re-
quired. (ii) The computing means are typically
used to make comparisons between a measured
spectrum and library spectra, or characteristic
peaks thereof. Such comparisons are made on the
-6-
. - .-
~.~)452S3
basis of an assumption that the spectrum measured
is that of a pure sample. Thus, in order for the
computing means to be effective, it requires puri-
fication of the sample (typically by means of a gas
S chromatograph). (iii) In making comparisons, the
quantitative criteria of similarity typicl]ly applied
by systems of the prior art do not reflect d~rectly
the probability that the identification is correct.
(iv) The complexity of the comparison method re-
quires the use of a high capability computer and/or
temporary storage so that analysis can be completed
after the data measurement is completed. Thus, the
analysis cannot ordinarily be done in "real time",
that is, as the measurement is being made. (v~ Prior -~
art computing means for identification of a compound
from the mass specturm typically fail to make use
of some of the information available therein, in-
cluding the absence or weakness of characteristic
peaks and the differing significance of peaks as
a function of their mass and intensity. (vi) The
relatively long time required for the comparison
process usually precludes an analysis of the plural-
ity of spectra derived from mass measurement of the
gas chromatograph effluent.
The present invention overcomes substantially all of
the above-described limitations and shortcomings of the prior
art instruments and methods. This invention enables the
identification of any one of a number of a preselected "target"
compounds in unknown mixtures of compounds with little or no
sample purification. Thus, for one thing, it enables the
_7_
. . .
~J452S3
elimination of a gas chromatograph or, at the least, a sub-
stantial reduction in its complexity. As a result, this in-
vention provides a system which is less complex and less ex-
pensive than the corresponding systems of the prior art. In
addition to the elimination or simplification of the gas
chromatograph, the present invention further reduces the cost
and complexity of compound identification by eliminating bulk
data storage means. This is the result of its incorporation
of means capable of real time analysis.
The present invention also enables greater specificity
in the identification of compounds than that possible using
the systems and methods of the prior art. This is due to the
novel analysis means incorporated into the invention, means
which (i) apply probabilistic techniques to the comparisons
màde; (ii) carry out exhaustive analysis of each spectrum
measured; (iii) make use of negative information, e.g., the
absence or weakness of characteristic peaks; and (iv) make
use of calibration data ~learned" from the invention itself.
In addition, the analysis means of this invention can pro-
vide a confidence index, consistent for all target compounds,which quantitatively indicates the probability that the
identification is correct.
The present invention enables automatic identification
of compounds. This has the advantage of reducing to a mini-
mum the skill and attention required of the operator. Inaddition, the invention enables rapid operation. For example,
in applications where direct mass analysis gives satisfactory
results, i.e., where chromatographic separation is not neces-
sary, analysis of a sample can require as little as one second
and is typically completed in thirty (30~ seconds, including
104S'~3
data printout. Moreover, a greater variety of samples can be analyzed in
situations where the chromatograph is eliminated than has h0retofore been
possible. Even in situations where some chromatographic separation is
necessary, less analysis time is required than in the prior art because the
degree of separation necessary to achieve satisfactory results is substan-
tially reduced by the invention.
One further advantage of the present invention lies in its making
possible the quantitation of identified compounds even when the target com-
pound mass spectrum is largely obscured by other compounds in the mixture.
While some instruments disclosed by the prior art overcome some
of the disadvantages described above, there has heretofore been no system
which combines in one structure all of the features and advantages found
in the present invention.
According to a first aspect of the present invention, there is
provided in a mass spectrometric system comprised of (i) means for measuring
the mass spectra of sample compounds; (ii) means for introducing a sample
of a pure compound or a mixture of compounds into said measuring means;
(iii) means for controlling the operation of said measuring means so as to
measure the intensities of one or more mass peaks of said sample; and (iv)
means for data input and output electrically coupled to said control means
and to means for analyzing said mass peaks, said analysis means comprising:
(a) first means for storing the mass spectrum of at least one target com-
pound or a contracted mass spectrum thereof; (b) second means for storing
at least one spectral matching criterion; ~c) third means for storing a
predetermined sensitivity factor with respect to each mass peak of said
target compound's spectrum, said sensitivity factor relating the measured
intensity of each mass peak to a quantity of said target compound; (d)
means for determining a measure of quantity of said target compound in said
sample based upon the intensities of said measured mass peaks and said
corresponding sensitivity factors, the mass peak at which the least quantity
of said target compound is determined being stored in said second storage
means as a reference peak, said quantity determination means being electric-
Qi~ ~ g
. .. . : :
~, . ~ '' ''''- ~
. . .
1~45253
ally coupled to said second and third storage means and to said measuring
means; ~e) means for predicting the intensity of a mass peak of said sample
at at least one other mass position of said target compound's mass spectrum,
said prediction being based upon the intensity of said reference peak and
said mass spectral data for said target compound, said prediction means
being electrically coupled to said first and second storage means; and ~f)
means for comparing the intensity of each measured mass peak to the inten-
sity predicted therefor with respect to a predetermined tolerance, said
comparison means being electrically coupled to said prediction means and to
said measuring means, whereby, a mass peak is considered con~aminated if
its measured intensity exceeds said predicted intensity by more than said
tolerance and said mass peak is considered uncontaminated if its measured
intensity agrees with said predicted intensity within said toleranceJ
and said target compound is identified as being present in said sample or
as not being present therein as a function of the number of uncontaminated
mass peaks measured at mass positions of its spectrum and their uniqueness.
~.j
According to a second aspect of the invention, there is provided
in a mass spectrometric system comprised of (i) means for measuring the
.,
, mass spectra of sample compounds; (ii) means for introducing a sample of a
pure compound or a mixture of compounds into said measuring means; and (iil)
~ means for controlling the operation of said measuring means so as to measure
~ the intensities of one or more mass peaks of said sample; and (iv) means
` for the data input and output electrically coupled to said control means
. and to means for analyzing said mass peaks, said analysis means comprising:
~a) first means for storing the mass spectrum of at least one target com-
pound or a contracted mass spectrum thereof; (b) second means for storing at
' least one spectral matching criterion; ~c) third means for storing a pre-
. determined sensitivity factor with respect to each mass peak of said target
compound's spectrum, said sensitivity factor relating the measured intensity
~ 30 of each mass peak to a quantity of said target compound; ~d) means for deter-
; mining a measure of quantity of said target compound in said sample based
upon the intensities of said measured mass peaks and said corresponding
h ~ -9a-
1~45Z53
sensitivity factors, the mass peak at which the least quantity of said
target compound is determined being stored in said second storage means as a
reference peak,said quantity determination means being electrically coupled
to said second and third storage means and to said measuring means; (e)
means for predictingthe intensity of a mass peak of said sample at at least
one other mass position of said target compound's mass spectrum, said pre-
diction being based upon the intensity of said reference peak and said mass
spectral data for said target compound, said prediction means being
electrically coupled to said first and second storage means; ~f) first means
for comparing the intensity of each measured mass peak to the intensity
predicted therefor with respect to a predetermined tolerance, said compari-
son means being electrically coupled to said prediction means and to said
measuring means, a mass peak being considered contaminated if its measured
intensity exceeds said predicted intensity by more than said tolerance and
uncontaminated if its measured intensity agrees with said predicted intensity
within said tolerance; ~g) means for determining a probabilistic measure of
the likelihood of the presence of said target compound based upon said un-
contaminated mass peaks measured, said means for determining a probabilistic
measure being electrically coupled to said first comparison means; and (h)
second means for comparing said probabilistic measure to said spectral
matching criteria; said second comparison means being electrically coupled
to said means for determining a probabilistic measure and to said second
storage means; whereby, said target compound is identified as being present
in said sample or as not being present therein in accordance with said
spectral matching criterion.
In a preferred embodiment of this invention, the mass spectrometer
comprises an electron-impact ion source, quadrupole mass analyzer, electron
multiplier detector and an ion pumped vacuum system. The sample inlet
device is comprised of a flash evaporator and a membrane-type molecular
separator. The control and analysis means may be implemented by a hard-
A wired, digital logic and control system or a programmable digital computer.
In either case, appropriate input and
-9b-
. ~
' ' , . ~ :
iO45;~
output devices are required.
The invented system is arranged and operated so as
to (i) measure and analyze the mass spectral data necessary
to determine the presence or absence of one or more of a
limited number of target compounds; (ii) measure the quan-
tities of the compounds found; and (iii) display the results
of the analysis rapidly and automatically. The system will
operate properly with pure samples, mixtures, or partly
separated mixtures. In addition, the invented system gen-
erates a confidence index which indicates the probability
that the identification, if made, is correct.
The invented system handles mixtures or pure sub-
stances in the form of gases; liquids, or solids in solution
in microgram quantities. A dual inlet arrangement allows
gases to be introduced directly, or liquids to be injected
into a flash vaporizer. In the case of liquid samples, the
sample vapor is transported by a dry nitrogen stream. Trace
amounts of material, even as low as 10 nRnograms or less, can
be identified in the sample.
Gas samples or vaporized liquid samples flow to an
enrichment device, which preferentially extracts a large
fraction of the organic compound contained in the sample
stream from the permanent air gases. The sample-to-carrier
enrichment ratio is very high, permitting routine sampling
from atmospheric pressure environments. The enriched sample
passes directly to an evacuated ionization region of the
mass spectrometer where it is examined to acquire the spectral
data necessary for the analysis, and then exhausted from the
spectrometer. Operation of the mass spectrometer is con-
trolled by the electronic control means, while analysis of
.~ .
--10--
.~ ~
..... . . . .
:, ., . :
10~5ZS3
the spectra taken is by the electronic analysis means, as
more fully described below. Moreover, the control means is
responsive to information generated by the analysis means,
thereby providing closed loop control of the mass spectro-
meter as a function of the results of the analysis performedon the data theretofore acquired.
The electronic analysis means provide the unique
capability of searching for and identifying the presence of
specific target compounds in a sample by analyzing the mass
- 10 spectral data output by the spectrometer in real time, despite
the presence of a great many confusing mass peaks attributable
~ to residual background and sample impurities.
; Identification reliability is improved by taking into
` account the occurrence probabilities of the masses found to
be present, the accuracy of the fit of the relative intensity
pattern as compared with a stored reference pattern for the
'! pure compound, and the compound concentration in the sample.
`^ In addition to using positive information, the matching pro-
cedure also makes use of negative information in the spectrum,
basing the identification on the fact that particular peaks
are absent, as well as that others are present, in the mass
spectrum of the unknown sample.
The characteristic mass spectral "fingerprint" stored
for each target compound is a contraGted spectrum. Just as
an experienced mass spectrometrist looks for characteristic
t
$ peaks to determine if a particular compound is present, the
present invention examines a predetermined set of masses
most characteristic of the target compound, and compares the
measured mass and their relative intensities with the stored
reference spectrum. In the comparison, peaks which have
-11-
'i .,
` ~045253
excessive relative abundance are identified as contaminated
and eliminated from the analysis.
Limiting the search to the most characteristic masses
saves time which would otherwise be wasted measuring peaks
which give no or less information as to the presehce or ab-
sence of the target compound. To further reduce search time,
~ .
only enough masses to assure reliable identification are used.
A confidence index is generated by the analysis means
as a measure of the likelihood that the target compound is
present Even in non-random situations, as in discrimination
among closely related compounds, the confidence index proves
to be a useful indication of the confidence of mass spectral
identification, which previously could be obtained only
through the careful study of the mass spectral data by a
trained mass spectroscopist. Presence of the target compound
;
is indicated by a relatively high confidence index, depending
î on the quantity found and the degree of contamination. Other
.,, . :~
compounds, even when of similar molecular structure and
present in high concentration, give relatively low values of
the confidence index by this analysis method. Typically,
reliable identification can be obtained over a concentration
ratio of one hundred to one.
The quantity estimate is computed from the intensities
~.~ of the uncontaminated peaks identified by the analysis means.
In this manner, quantitation of a specific compound is possible
even in a mixture which contains other compounds having many
of the same masses. The response of the mass spectrometer is
nearly linear over several decades. Thus, after a target com-
pound is identified, its quantity is determined in units of
weight (e.g., micrograms) by use of a stored scale factor.
.~ ,
-12- ~
1~)45Z53
The scale factor is derived from the intensity measured for
a known quantity of a pure sample of the target compound
during an earlier calibration.
When the confidence index value indicates that the
compound is problably absent, the quantity figure sets an
independent upper bound on the target compound concentration.
This is often very useful, as an extremely low quantity con-
firms the negative finding.
The present invention operates in at least four modes;
i.e., start-up, calibration, identification, and a data system
mode, In the start up mode, the invented system is placed in
a condition for operation; e.g., operating temperatures and
pressures and various parameters are established. In the
calibration mode, the system automatically measures and re-
cords the reference mass spectral data for a given compoundwhen an authentic sample is introduced. In addition, the
background peak intensities of the system are measured in the
absence of any sample material.
The identification mode is used to assay samples for
. !
the presence of one or more of the target compounds in either ~ `
of two sub-modes. In a first such sub-mode, a confirmation
mode, the system repeatedly attempts to match the spectrum
- of the injected sample with the characteristic spectrum of
` one pre-selected target compound. The sample is either con~
tinuously or sequentially injected into the sample inlet de-
vice and spectral masses measured repeatedly by the mass
spectrometer. The analysis means generates a confidence
index by application of the matching criteria and rules. If
` the confidence index is above a preestablished threshold for
any target compound, a display is activated to indicate its
-13-
~. .
. .- ~
:
: 1~145253
presence. In addition, further information, such as the
quantity of the target compound found, the magnitude of the
confidence index, and/or the mass peak intensity may also be
output by the system by conventional input/output devices.
In the second identification sub-modes, a search mode, a
probabilistic matching of the spectrum of the injected sample -
is attempted with each of a subset of the total number of
; target compounds whose spectral "fingerprints n are stored in
the system.
In the data system mode, the invented system measures
and displays the entire mass spectrum of an authentic sample,
and ranks the mass peaks as to their significance for identi-
Z fication purposes by means of built-in probability tables.
The spectral data so acquired may then be stored in the sys-
tem, thereby adding that compound to the set of target com-
pounds used for identifying unknown samples.
The present invention finds utility in a wide variety
of applications, including (i) the forensic sciences, where
.
it enables fast and accurate analysis of abused drugs in
street sample mixtures; (ii) pharmacology, where its high
sensitivity enables the measurement of drugs and metabolites
in body fluids for clinical therapeutic study as well as
urine screening; (iii) clinical toxicology, where its fast
response time enables rapid identification, in a hospital
` 25 environment, of drugs taken by comotose overdose patients,
thereby facilitating their emergency treatment; (iv) indus-
trial toxicology, where its automated capability enables 24-
hour monitoring for multiple toxic compounds in a manufactur-
ing plant. In addition, the invented system can also be
utilized advantageously for the detection of air and water
`
1045253
pollution, pesticides and explosives, even at relatively low
concentrations.
Novel features, and advantages of the present in-
vention will become apparent upon making reference to the
following detailed description and the accompanying drawings.
The description and the drawings will further disclose the
characteristics of this invention, both as to its structure
and its mode of operation. Although a preferred embodiment
of the inventon is described hereinbelow, and shown in the
accompanying drawing, it is expressly understood that the
descriptions and drawings thereof are for the purpose of
illustration only and do not limit the scope of this inven-
tion.
In the accompanying drawings, a preferred embodiment
~ 15 of the present invention is illustrated:
- FIGURE 1 is a general block diagram showing the five basic components of the invented system.
., .
FIGURE 2a is a cross-sectional view of a flash
i evaporator used in the sample inlet device of this invention.
FIGURE 2b is a cross-sectional view of a separator
.~, .
used in the sample inlet device of this invention.
.~,
FlGURE 3 is a more detailed functional block diagram
of the invented system.
~ FIGURE 4 is a schematic representation of the mass
-~ 25 scan converter portion of the invented system.
FIGURE 5 is a schematic representation of the elec-
trometer amplifier portion of the invented system.
i FIGURE 6 is a schematic representation of the peak
stretcher portion of the invented system.
FIGURE 7 is a functional block diagram of the control
~ . . .
: : . . - . .
)4SZS3
means and analysis means of the invented system.
FIGVRE 8 shows the relationship between a complete
mass spectrum of a particular compound and its contracted
mass spectrum.
FIGURE 9 shows the voltage waveform at the output of
the mass scan converter.
,
FIGURE 10 shows the voltage waveform at the output
of the electrometer amplifier.
, FIGURE 11 shows the voltage waveform at the output
of the peak stretcher.
FIGURE 12 shows a time profile of the sample partial
pressure after injection into the sample inlet device.
The present invention is comprised of the
five basic components shown in FIGURE 1, namely (i) a sample
inlet device 10; (ii) a mass spectrometer 12; (iii) means
14 for controlling the operation of the mass spectrometer 12;
(iv) means 16 for analyzing the mass spectra obtained from
;~ the samples; and (v) interfacing means 17 between the mass
spectrometer 12 and the control and analysis means 14 and 16.
.l 20 The sample inlet device 10 is described with refer-
ende to FIGURES 2a and 2b. It is comprised of a flash
f evaporator 18 and a separator 20. The flash evaporator 20 is
required for samples which are in either a liquid or solid
'J) state. It is comprised of a heated metal tube 22 having a
glass insert 23, an injection port 24, an outlet port 26, and
a carrier gas port 28. The injection port 24 is sealed with
a septum 30, preferably made of a silicone rubber. The
septum 30 is secured to the tube 22 by a septum nut 33.
Liquid or dissolved or suspended solid samples are injected
through the septum 30 with a syringe. The carrier gas port
-16- -
,', - '
~-
1~a~5253
28 is coupled by conventional means to a source of carrier
gas under a slight pressure. A suitable carrier gas is
nitrogen. A conventional heater (not shown) maintains the
temperature of the flash evaporator 18 at a level suitable
for evaporation of the sample.
In operation, volatile components of the sample are
evaporated and swept by the gas carrier through the tube 22
~j and through a glass wool filter 32 disposed within the tube,
,i wherein particulate matter in the sample is trapped. The
outlet port 26 of the flash evaporator 18 is coupled to the
separator 20.
. ~
If the sample is gaseous, the flash evaporator 18 is
unnecessary. The flash evaporator 18 may be replaced with an
inlet tube and the sample flowed directly into the separator
20. The separator 20 is comprised of a housing 36 having an
inlet port 38, an outlet port 40, two thin polymer membranes
42a and 42b stretched across the interior of the housing 36 -
between the inlet port 38 and the outlet port 40; i.e., across
the path of flow of the sample. An exhaust port 44 located
in the housing 36 on the inlet side of the first membrane 4Za
enables the non-transmitted carrier gas and vaporized solvents
- to be vented to the atmosphere. A small suction pump 51 may
be coupled to the exhaust port 44, and used to establish the
~ flow of gaseous samples. Suction pump 51 is unnecessary for
`~x 25 liquid or solid samples because the carrier gas provides the
flow medium. A vacuum port 46 i8 provided in the housing be-
tween the membranes 42a and 42b. By means of a mechanical
pump 47, coupled to the vacuum port 46 through valve 49, a
pressure of about 10-2 torr or less is maintained in the in-
terior space between the two membranes. The pressure in the
,
`; ~ -17-
~,
.~
... . :: : .
': :: .. - ,. . . .
.: .: . ,
1045Z53
mass spectrometer 12 is typically maintainea at about 10-5
torr or less by an ion pump 48 (as shown in FIGVRE 3). Thus,
a pressure differential is maintained across each membrane.
In starting up the system, mechanical pump 47 is used to
evacuate the mass spectrometer 12 through valve 50 (as shown
in FIGURE 3). The membrane temperatures are typically held
between 150C and 220C, by a second conventional heater, the
:
specific temperatures being a function of the target compound
being sought.
Through preferential permeability, the higher mole-
cular weight sample vapors are transmitted through the mem-
branes 42a and 42b much more efficiently than the carrier gas
and solvents. Most of the latter are vented to the atmosphere
, or pumped from the region between the membranes, while an
appreciable fraction of the sample vapor passes out of the
`~ outlet port 40 and enters the mass spectrometer 12. Sample
enrichment by a factor of 106 is possible and a large number
of compounds are concentrated by a factor of 105 or more.
Pressures, temperatures, and the carrier-gas or sample
flow rate are parameters which are monitored and controlled
by conventional means.
A number of ~ther spparators, known in the art, are
suitable for use in the present invention in lieu of the
separator 20 described above. Such other separators include
effusion separators, jet orifice separators, and a single
membrane separator like that disclosed in the United States -
Patent No. 3,751,880, granted to Michael Holm.
Solid samples are generally prepared by dissolving
them in a suitable solvent. Aqueous or low-molecular weight
solvents such as methanol, ethanol, acetone or ether are
s
-18-
.
..
~045Z53
preferable. In complex or dilute mixtures, it is advantageous
to use a suitable extraction procedure. Silylation, methyla-
tion or other standard derivatization methods may improve
sensitivity or specificity.
Liquid samples may be introduced by continuous or
discrete injection. If the concentration in the original
matrix is too low, or if there is serious contamination of
the sample, standard laboratory methods of extraction and
concentration are used. For example, body fluids such as
blood or urine are prepared for drug analysis by solvent ex-
traction at an appropriate pH, or non-ionic resin column
isolation.
1 The invented system, while not requiring purification
`,~ fo the sample by means such as a gas chromatograph, is never-
theless adapted to operate with a gas chromatograph; thereby,
users of the present invention have a choice between a direct
analysis mode and a gas chromatograph/mass spectrometer (GC/MS
` analysis mode. In the latter mode any suitable adapter known
~. :
in the art enables the continuous real time analysis of the
effluent from the gas chromatograph. The time related separa-
~; tion of the sample into its constituent compounds improves
~i the specificity of identification when mixtures are involved,
and the retention time data generated by the gas chromato-
graph is useful as an independent check on the identifications
made.
MASS SPECTROMETER
A detailed description of the mass spectrometer 12
is now made with reference to the system block diagram shown
in FIGURE 3. As is well known in the art, a mass spectrometer
is basically comprised of an ion source, a mass analyzer, and
. .
1045253
a detector. In a preferred embodiment of this inventon, the
ion source is an electron impact ion source 52, wherein the
sample gas is partially ionized by a beam of electrons. The
ions so formed are electrostatically removed from the source
S and formed into a beam which is projected through the mass
analyzer to impinge on the detector.
A preferred mass analyzer is a quadrupole mass analyzer
54 comprised of two pairs of metal rods 56 which, when ex-
cited by the proper combination of radio frequency (rf) and
dc voltages, produce electric fields which cause the tra-
jectories of all ions, except those in a narrow range of mass
to charge (m/e) ratio (referred to herein as the "mass posi-
tion") to be unstable. Thus, only ions at a selected mass
position are allowed to reach the detector at a given in-
~`
^~ 15 stant. All other ions are deflected into the mass analyzer -~
rods 56 or walls and, thereby, are undetected at that instant.
A continuous channel electron multiplier 58 is uti-
lized as a detector of the mass spectrometer 12. The ion
current which reaches this detector 58 and impinges upon its
surface is amplified through the phenomenon of secondary
electron multiplication. Typically, the output of the detec-
tor is an electron current equal to as much as 106 times the
detected ion current.
The mass spectrometer 12 is provided with adjustable
power supplies 60 for energizing the ion source filament and
electrodes, and the electron multiplier detector 58. The
adjustable power supplies 60 may be adjusted by the control
, means 14 or manually. The overall sensitivity of the invented
system is established by adjusting the power supply voltage
to electron multiplier detector 58. This adjustment is done
.;
-20-
;., .
,. ~ .
-
1(~4`~:
by looking at the intensity of a known mass peak of a known
sample, typically the carrier gas. Moreover, changes in the
gain of detector 58 may be compensated for by adjusting power
- supplies 60. Mass spectrometer 12 is also provided with an
rf/dc power supply 62 which produces rf and dc voltages to
excite the quadrupole mass analyzer 54. The rf and dc voltages
are determined by a mass control voltage which is an anlog of
the mass position, (m/e); i.e., the mass position is related
to the mass control voltage by a substantially constant scale
factor. Thus, any mass position may be selected by providing
~ . .
the corresponding mass control voltage to the rf/dc power
.~ . , .
supply 62.
A number of suitable mass spectrometers are available
. . .
-~ in the trade. Moreover, it should be understood that the -
present invention contemplates any type of low or high resolu-
tion mass spectrometer. For example, chemical ionization may
be used instead of the electron impact ionizer 52; likewise,
-~ a magnetic sector analyzer instead of the quadrupole mass
analyzer 54, and a discrete dynode electron multiplier or an
ion collector instead of the continuous channel electron
multiplier 58. In addition, any vacuum system design capable
` of maintaining the necessary vacuum and having adequate pump-
ing speed may be used.
;`
`'~ INTERFACING MEANS
.~, .
The interfacing means 17 are comprised of (il a mass
scan converter 64; ~ a gain-controllable electrometer
~,- amplifier 66; and (iii) a peak stretcher 68, interconnected
as shown in FIGURE 3.
}
(a) MASS SCAN CONYERTER
~` 30 The mass scan converter 64, shown schematically in
:, .
-21-
, .
`: :
~:: . .
....
., . , . - ,
1~45ZS3
FIGURE 4, is disposed between the output of the control
means 14 and the input control line of the rf/dc power supply
¦ 62. As indicated above these rf and dc voltages control the
¦ scanning operation of the mass spectrometer 12 by means of
~,' 5 its quadrupole mass analyzer 54. The control means 14 pro-
vides to the mass scan converter 64 a mass control sig~al,
preferably in digital form, which is an analog of a particular
mass position at which a mass peak is to be sought. The con-
; trol means also provides a start signal to the mass scan con-
verter 64. In response to the start signal, the mass scan
converter 64 generates, as an output to the r/dc power supply
62, the mass control voltage in the form of a ramp having an ,,
initial voltage approximately equal to that indicated by the
mass control signal.
A typical mass control voltage generated by the mass
scan converter 64 is 25 millivolts per atomic mass unit tre- ,
,,i ferred to herein as the "mas control scale factor"). Thus,
if the mass position to be scanned is 100 amu, the mass con- ~ , ,
trol voltage will be 2.5 volts. Because of the inherent un-
, 20 certainty in establishing the precise mass position by an
analog mass control voltage, a scan is required beginning at
a mass position just below the selected mass position and end-
ing just above it. This is accomplished by the ramp voltage.
Such a scan ensures that the presence of a mass peak at a
3 25 selected mass position will nctgo undetected because small,
but cumulative, errors in the system cause a mass measurement
to be made at a position slightly shifted from the selected
~ mass position. Typically, the scan width is about 3/4 amu
`~ centered on the selected mass position. Ths initial mass con-
trol voltage in the foregoing example would be about 10
-22-
. .
1~45Z53
millivolts below 2.5 volts.
If should also be understood that the mass control
transfer function in an actual system is not perfectly linear
over the entire mass range. Moreover, there may be a slight
5 offset at zero volts and, in addition, this offset and the
mass control scale factor may vary slightly with time and
from unit to unit. In the preferred embodiment of the present `
invention the control means 14 generates scale factor and
offset correction signals, preferably in digital form, on the
basis of calibration data obtained by measuring the actual
scan control voltage required to measure mass peaks at several
precisely known mass positions. The scale factor and offset
correction signals are stored in the mass scan converter 64
which correspondingly adjusts the initial mass control voltage.
.. . .
Remaining errors due to non-linearity of the mass control
i transfer function are accommodated by the mass scan converter
and peak stretcher designs described below.
,.
, With reference to FIGURE 4, the mass scan converter
64 is now described in greater detail. First, second and
third digital to analog (D/A) converters 70, 72, and 74 store
~.
and convert the mass control, scale factor correction and
offset correction signals respectively, received from the con-
,3 trol means 14. D/A converter 70 produces an output, vl, which
is proportional to the value of the mass control signal, D/A
converter 72 outputs an analog reference voltage, Vref, to
D/A converter 70. Tnis reference voltage determines the
ratio between voltage vl and the value of the mass control
signal, which ratio is an analog of the mass control scale
factor. Thus, control means 14 can compensate for the varia-
tions of the mass control scale factor by providing the
'.,
-23-
5253
: appropriate value of the scale .factor correction signal to
D/A converter 72. D/A converter 74 produces an output voltage,
v2, which is proportional to the offset correction signal.
I Voltages vl and v2 are electrically coupled to a conventional
3 5 operational amplifier 76 through resistors Rl and R2 respective-
ly. A feedback resistor R8 is electrically coupled between
the input and output of operational amplifier 76, thereby
making it an adder of its input voltages, as is well known
in the electronics art.
A second conventional operational amplifier 78 pro- -
duces an output voltage, v3, which is coupled to operational
:l amplifier 76 through resistor R3. A feedback capacitor Cl ~-
is coupled between the input and output of operational ampli-
. fier 78, thereby making it an integrator of a constant dc
voltage, VR. Thus, voltage V3 is a ramp voltage, equal to ~ :
VR t , where Ri is the input resistance of operational amp-
~ lifl~er 78 and t is time. The value of Ri, and therefore the
;; slope of the ramp, or rate of the scan is a function of the
I states of conventional binary switches SWl SW2, and SW3 which,
: 20 when closed, short out input resistors R5, R6, and R7 re-
, spectively. The states of switches SWl, SW2, and SW3 are
:~ controlled by the states of corresponding flip-flops in a
:~l storage register 79, wherein a digital "scan rate" signal,
.? received from control means 14, is stored. Thus, Ri can be
:~ 25 as low as R4, and as high as R4 + R5 + R6 + R7.
A conventional binary switch SW4 is connected across : -
-. feedback capacitor Cl. The state of switch SW4 is determined
~ by the start signal received from control means 14. When
'~ switch SW4 is closed, capacitor Cl is shorted out and V3
~ 30 equals zero. When switch SW4 is opened, the ramp voltage V3
. : ~
-24-
'
. ~ . .
10~5~53
appears as the integration of voltage vR begins.
The output voltage, vm, of operational amplifier 76
is the mass control voltage which is electrically coupled to
the rf/dc power supply 62. Since operational amplifier 76
.
functions as an adder,
'vl V2 V3
m = Rl+ R2 +R3 R8
Thus, voltages vl and v2 determine the dc level or pedestal
of vml while V3 provides the ramp portion thereof.
Binary switches SWl, SW2, SW3, and SW4 may be relays
or transistor switches. Many suitable electronic switches
are available in the trade.
(b~ ELECTROMETER AMPLIFIER
The gain-controllable electrometer amplifier 66,
shown schematically in FIGURE 5, is disposed between the
detector 58 of the mass spectrometer 12 and the peak stretcher
68. It converts the output current from the detector 58, re-
presenting an analog of the intensity of the ions at the mass
position selected by control means 14, into a voltage analog
thereof (referred to herein as the "ion intensity signal").
The electrometer amplifier 66 is adapted to having its voltage-
~, to-current sensitivity or gain digitally set to one of a
number of values, as required to amplify the ion current peaks
j to appropriate levels for measurement. In this preferred
embodiment, any one of six gain settings for the electrometer
`~ 25 68 may be selected by means of a "gain select" signal from
~ the control means 14. However, manual control is also con-
`~ templated by this invention.
~` With reference to FIGURE 5, the gain-controllable
` electrometer amplifier 66 is now described. It is comprised
of a conventional operational amplifier 80 having resistors
'1
-25-
.,
.~, ~. . . . . .
1045Z53
Rg, Rlo~ Rll~ R12~ R13~ and R14 electrically coupled between
its input and output. Corresponding conventional, binary
switches SW5, SW6, SW7, SW8, and Swg are coupled across re-
stors Rg ~ Rlo~ Rll, R12~ and R13 respectively. When any
switch is in its closed state, the corresponding resistor isshorted out. The states of switches SW5, SW6, SW7 SW8, and
; SWg are controlled by the states of corresponding fip-flops
: in a storage register 82, wherein a digital gain select
signal, received from the control means 14, is stored. Thus,
the feedback resistance across operational amplifier 80 can
be any one of six values, depending upon the states of the
switches SW5 - SWg. The voltage output of operational ampli-
fier 80, vii, the ion intensity signal, equals the ion cur-
rent ii multiplied by the magnitude of the feedback resistance.
(c) PEAK STRETCHER
~:
The peak stretcher 68, shown schematically in FIGURE
6, is disposed between the electrometer amplifier 66 and the
analysis means 16. The peak stretcher processes the ion in-
.~
; tensity signal which is output by the electrometer amplifier
66. The peak stretcher 68 has two principal operating modes,
; ~ descriptively referred to as "hold~' and "blank". In the hold
-,l mode, hhe ion intensity signai is filtered and amplified, and
~ .~
the maximum amplitude of the signal; i.e., the peak intensity
~ during a mass scan, is held and presented to analysis means
¦ 25 16 by peak stretcher 68. In the blank mode, the input to the
peak stretcher 68 is shorted, thereby preventing the ion in-
tensity signal from reaching the analysis means 16. In addi- ~ -
tion, the previously stored signal is removed in preparation
for the next scan.
In the preferred embodiment described herein, the
:
1045253 ~ -
peak stretcher 68 has four values of selectible voltage gain
! and four values of selectible filter time constant. Selection
of the appropriate gain and filter time constant is by the
control means 14, although manual selection is also contem-
plated by this invention.
With reference to FIGURE 6, the peak stretcher 68 is
now described. The ion intensity signal, vii, is input to a
conventional operational amplifier 84 through input series
resistors R15, R16, and R17. Coupled between the input and
output of operational amplifier 84 is a voltage resistor Rlg.
A capacitor C2 can be connected to a resistive divider Rlg,
R20, and R21, via analog switches SW13, SW14, or SW15. Con-
ventional binary switch SWlo is connected to one side of
` resistor R15. When SW10 is closed, vii is shunted to gro
through resistor R15. This is the state of switch SW10 when
peak stretcher 68 is in its blank mode. Another switch, SWll,
. ~ .
~ is coupled across input resistor R17. In its closed state,
. .
~, switch SWll shorts out resistor R17, thereby affecting the
gain of operational amplifier 84.
When switches SW13, SW14 and SW15 are open and SW12
~; is closed, resistor R23 merely shunts the input of amplifier
84 - there is no filtering and the filter capacitor C2 is dis-
charged. With switch SW12 open and SW15 closed, the full
filtering time constant R18C2 is effective. Closing only
SW14 or SW13 causes a feedback voltage derived from the R21,
R20, Rlg divider to be applied to C2, effectively multiplying
C2 by the attenuation of the divider factor, and thereby re-
ducing the effective filter time constant of amplifier stage
84 without changing its gain.
The voltage at the output of operational amplifier 84
:
-27-
`~
~1)4SZS3
is electrically coupled to the input of a second opera-
tional amplifier 86 through series input resistors R24 and R25.
; Coupled between the input and output of operational amplifier
86, through filed effect (FET) transistor Tl, is a feedback
resistor R26. The gate of transistor Tl iS electrically
coupled to the output of operational amplifier 86 through low-
leakage diode Dl or switch SW17, when closed. A capacitor
C3 is connected between the output of operational amplifier
86 and circuit ground; thus, it charges to the output voltage
thereof. The output of peak stretcher 68 is taken at the
source of transistor Tl. When switch SW17 is closed, the
output of the peak stretcher 68 is the filtered and amplified
ion intensity signal vii. When SW17 is open, capacitor C3
cannot discharge through low leakage diode Dl or the gate of
lS FET Tl; therefore, it will retain the most positive voltage
applied to the gate of Tl. Through the feedback action of
resistor R26, the corresponding most positive value of the
vii will be retained at the output of peak stretcher 68.
Conventional binary switch SW16 is coupled across
input resistor R25. In its closed state, switch SW16 shorts
out resistor R25, thereby affecting the gain of operational
amplifier 86.
The states of switches SW10 and SW17, the mode switches,
are controlled by the state of corresponding flip-flops in a
storage register 88, wherein a digital "mode select" signal,
received from control means 14, is stored. The states of
switches SWll and SW16, the gain control switches, are con-
trolled by the states of corresponding flip-flops in a
register 90, wherein a digital "gain select" signal from
control means 14 is stored. Four values of gain are possible
,
-28- -
1~5Z53
based upon the four combinations of the states of sw~tches
SWll and SW16. The states of switches SW12 - S~15, the time
constant (and filtering) select switches, are controlled by
the states of corresponding flip-flops in a storage register
92, wherein a digital "time constant select" signal from con-
trol means 14 is sotred.
CONTROL MEANS
.~
The control means 14, described with reference to
FIGURE 7 controls (i) the mass scanning of the sample in the
;i
mass spectrometer 12; and (ii) certain operating parameters
of the system. The control means 14 is comprised of a memory
means 92, a sequence timer 94, and a parameter determination
means 96, all of which can be implemented by a programmable
. digital computer utilizing known techniques of computer pro-
lS gramming or by a hard-wired, logic and control system utiliz-
ing known techniques of logic design and available integrated
circuit logic components.
Each target compound selected for identification by
the invented system has a unique mass spectrum. Each such
mass spectrum is comprised of mass peaks located at various
mass positions. A particular subset of all the masses can be
predetermined as being especially characteristic of each tar-
get compound. This set of masses is called the "contracted
-~, mass spectrum". The relationship between the complete mass
spectrum of a hypothetical target compound and its contracted
`~ mass spectrum is illustrated in FIGURE 8. In the background
is shown a full mass spectrum of a particular compound. In
the foreground is the selected subset of mass peaks comprising
the contracted mass spectrum of the compound. The criteria
for selecting the contracted mass spectrum is discussed
"
., .
. . .
~` ` -29-
.,
.,
1()~5ZS3
hereinbelow.
The masses for the contracted mass spectrum,of a
given target compound spectrum, are selected and ranked ac-
cording to how characteristic they are of that particular com-
pound. This "characteristicness" is a function of (i~ howunusual, on a probabilistic basis, the vary existence of a
I peak at that mass position is, and (ii) the magnitude of the
¦ intensity of the peak (referred to as "abundance"). The con-
;l tracted mass spectrum for each target compound, i.e., the mass
¦ 10 position of each peak thereof, is stored in the control mem-
¦ ory means 92. In any event, for each target compound being
sought in a sample, the parameter determination means 96
initiates a mass scan sweep about each mass position for a
predetermined scan period and width. The mass position re- ~;~
lating to the most characteristic mass peak for the target
`~ compound sought is the first one selected by the parameter
determination means 96. After the sample is analyzed for
` tbe first and most characteristic peak, the mass position
; relating to the next most characteristic peak is selected
and a scan about it initiated. The parameter determination
means 96 continues operating in this manner until all of the
; mass peaks in the pre-stored contracted spectrum of the
particular traget compound have been scanned, unless the
sequence is modified by the analysis means 16, as more fully
described below.
` The waveform of the mass control uoltage output of
the mass scan converter 64 is shown in FIGURE 9. The wave-
form relates to scans about mass control voltages Vl and V2,
corresponding to two mass positions at about 40Vl amu and
40V2 amu respectively (based upon a mass control scale factor
..,
-30-
. .
.
., : .
.. , . , ..
. . . .
l,
lO~SZ53
~, of 25 millivolts per atomic mass unit). The start time for
the first scan is tl; the scan ends at t2. The scan period
is a variable determined by the parameter determination means
96. At about time t2, a mass step occurs to a voltage cor-
responding to a mass position just below the second selected
~! mass position, 40V2 amu. The second scan starts at t4 and
ends at t5.
i,
The inter-scan interval (t4 - t2) is comprised of
two delays. The first delay, (t3 - t2~, allows the rf/dc
power supply 62 to respond to the step in the mass control
voltage. The second delay, ~t4 - t3), allows the electrometer
amplifier 66 to recover from the spurious ion intensity sig-
¦ nals which occur between t3 and t2 as a result of the mass
`~l sweep of the mass spectormeter 12. (These spurious signals
are shown in FIGURE 10, which shows the waveform at the out-
¦ put of the electrometer amplifier 66 in the same time frame
l as that of the mass scan converter output.) Even if the same
;:1
mass position is bieng rescanned, an interscan interval is
still required although the first delay would be relatively
'.3 20 small. The first delay is a function of the magnitude of the
mass step. The second delay is a function of (i) the previous
time constant setting of the filter in the peak stretcher 68,
which filter must be discharged before the next scan; tii~ the
re~overy time of the el~ectrometer 66 after responding to the
spurious ion intensity signals; and (iii) any changes in the
settings of the gain of electrometer amplifier 66 and/or the
., .
gain and filter time constant of peak stretcher 68, required
for the next measurement. The delays are determined by the
parameter determination means 96 and clocked out by the
~; 30 sequence timer 72.
i~
~ .
-31-
.,
:
1045Z53
With reference to FIGURE 10, the maximum values of
the electrometer output voltage, Vel and ~e2, occur during
the first and second mass scans respectively; i.e., during
scan periods (t2 - tl) and (t4 - t3), representing the ion
intensities at the selected mass positions. However, spuri-
ous signals also occur at the massstep times t2 and t5. These
spurious signals are suppressed by placing the peak stretcher
68 in its blank mode for periods covering the times t2 and t5.
This is accomplished by the parameter determination means 96
issuing a mode select signal to the peak stretcher 68 at
these times.
The corresponding waveform at the output of the peak
~; stretcher 68 is shown in FIGURE 11. At scan start times tl
and t4, the parameter determination means 96 places the peak
stretcher in its hold mode. In this mode, the maximum values
of the electrometer amplifier output during each scan, after
~.
filtering and amplificaiton, are held at the peak stretcher
output for digitization and storage by the analysis means 16.
The peak stretcher voltages Vpl and VP2 shown in FIGURE 11 are
proportional to the intensities of the two ion peaks at the
two selected mass positions respectively. After the peak
values are stored in the memory portion of the analysis means
~.
16, the peak stretcher 68 is reset, to prepare for the next
....
scan. This is accomplished by the parameter determination
means 96 issuing an appropriate mode select signal to the
peak stretcher 68 at times t6 and t7 to place it in its blank
~! mode.
; The sensitivity of the invented system is adjustable
by selection of the gain of the electrometer amplifier 66
and/or the gain of the peak stretcher`68. The sensitivity
. ;, .
~., .
-32-
,.. -, . . . - . , .
.,. . ~, .. .. . . .
104S253
must be adjusted so that the peak intensity measured is "on
scale", i.e., neither clipped by saturation nor so low as to
be obscured by noise. The scan rate and the filter time
constant of the peak stretcher 68 are interrelated functions
of the sensitivity setting of the system. When the sensitivity
of the system is reduced, there is a need for greater filtering
by the peak stretcher 68 in order to enhance the signal-to-
noise ratio. Accordingly, for each sensitivity setting, the
parameter determination means 96 provides a time constant
select signal to the peak stretcher 68, so as to select the
appropriate filtering required for that sensitivity setting.
The amount of filtering, in turn, determines the scan period
inasmuch as the frequency response of the filter must be
;~ compatible with the reequency content of the ion intensity
.,
signal. Thus, as the filtering is increased to enhance the
signal-to-noise ratio, the scan rate is correspondingly de-
creased by the parameter determination means 74 so that the
filter can pass the ion intensity signal. The interrelation-
ships between spectrometer sensitivity, i.e, electrometer
and peak stretcher gain, filter time constant and scan rate,
may be derived analytically and/or empirically. These rela-
tionships are stored in the control memory means 92 and
utilized by the parameter determination means 96 to generate
the appropriate parameter select signals. These signals are
provided, at the appropriate times, to the mass scan con-
verter 64, the electrometer amplifier 66 and the peak stretcher
` 68.
The parameter determination means 96 also receives
the temperatures of flash evaporator 18 and of the separator
20. These temperatures are converted to voltages by
. .
- -33-
10452S3
appropriate température transducers and converted to digital
form for presentation to the parameter determination means
96. The parameter determination means 96 typically controls
the on-of~ cycles of the heaters associated with flash
evaporator 18 and separator 20, as required to maintain the
desired temperatures.
CONTRACTED SPECTRUM
Before describing the analysis means 16 and its
operation upon the ion intensity signals output by the peak
stretcher 68, the criteria used to contract the mass spectrum
of each target compound is described.
It is well knwon that the existence of mass peaks in
mass spectra are much less common at certain mass positions,
especially those of higher atomic mass units. Thus, if a
target compound has large peaks at mass positions tm/e) 43
and 243, most mass spectrometrists, for identification, would
look first in the unknown spectrum for the presence of the -
. peak at mass position 243, knowing it to be a much more selec-tive criterion if the sample could be any one of a relatively
random selection of compounds.
In order to define this "unigueness" as a more quanti-
tative "V" value, it is necessary first to select a particular
universe of all compounds of interest and to obtain their
known mass spectra from available reference sources such as,
for example, the 17,124 compounds found in the "~ight Peak
Index of Mass Spectra" published by the Mass Spectrometry Data
Centre, AWRE, Aldermaston, Berkshire, England, 1970, The
target compounds, of course, are a subset of the compounds in
the selected universe. The U values will only be applicable
to spectra measured under conditions comparable to those
,
~l -34-
.
. -- .
., .
10~5'~53
employed to measure the spectra of the compounds in the
i universe of interest.
Abundance is defined as the ratio of the intensity
of a mass peak, at a given mass position, to a base intensity,
the latter being the maximum normalized intensity in the
entire spectrum. The U value is based on the probability
that, at a particular mass position, a spectrum taken at ran-
dom would have a mass peak with an abundance greater than 50%.
More specifically, U is defined, for each mass position, to be
the logarithm to the base 2 of the number of randomly selected
¦ mass spectra (taken from the universe of interest) which
; would have to be examined to find one having a peak at thatj mass of greater than 50% abundance
! Uj = log2 Nj; (1~
Where Uj is the U value at the jth mass position and Nj is
the number of randomly selected spectra which would have to
be examined as described above. It, therefore, follows that:
¦ P~i) = N ; (2)
~! where p(j~ is the probability~that, at the jth mass position,
a mass peak of 50% abundance exists; i.e., that any given
spectrum in the universe of interest will satisfy the fore-
going condition. Thus, since:
Nj = 2U; , (3~
p(j)= ~ (4~ `
For each mass position, the value of Nj can be deter-
mined from examination of the library of spe~ral data for the
universe of interest, and the Uj values derived from Nj. For
`~ simplification, the Uj values are rounded to the nearest
integer value.
It has been observed that the uniqueness of a
-35-
.: - ,: - . ,, . . ................. :
. " ` , ~ . ~,
1045'~53
particular mass peak is a function of its abundance level.
For example, in the Aldermaston universe of compounds, the
data shows that approximately one in 32 spectra (U=5) have
a mass peak of greater than 50% abundance at mass position 45
amu. However, at that mass position, nearly half the spectra
have a peak of greater than 1~ abundance. Thus, at any mass
position, the number of randomly selected mass spectra which
would have to be examined to find one having a mass peak of
a particular abundance generally decreases as the abundance
level decreases. Thus, the probability function can be ex-
pressed as - `
P(j~ = 1 ;
(Uj - Aj~
where Aj is an abundance term reflecting the effect of the
abundance level of the peak on the probability of occurrence.
At abundance levels between 50% and 100~, Aj = o. The values
of Aj can be determined from examination of the library of
spectral data for the universe of interest. A large sampling
of spectral data shows that, for most masses, the following
Table I gives a satisfactory approximation of the abundance
factor A necessary to adjust the probability of occurrence as
a function of the mass peak abundance. The abundance factor
A is substantially independent of the mass position of the
peak.
Table 1
Abundance Range A
50 - 100%
-19 - 50~ 1
7.1- 19% 2
2.7- 7.1% 3
1.0- 2.7% 4
0.38-1.0% S
-36-
. . ~
~14S2S3
The known spectral data for each target compound is examîned
and the value of Uj - Aj (referred to as Vj for convenience)
is determined for each mass peak thereof. For example, if the
target compound is known to have a mass peak of 30~ abundance
at mass position 45, U45 = 5 tin the Aldermaston universe of
compounds). From the above table A4s = 1. Thus, V45 = 4.
The subset of all peaks in the mass spectrum of the target
compound which have the highest values of V are the most unique
and, therefore, are selected to comprise the contracted mass
spectrum of that compound. Ordinarily, the subset of peaks
is chosen in decreasing order of Vj. However, any available
supplementary information, such as the fact that certain peaks
may relate to key structural features of the molecule, may
be used in the selection process. In addition, only those
peaks which can be reliably measured when a threshold quantity
of the target compound is introduced should be selected. The
nu~ber of peaks selected is a function of the confidence level
desired for an identification.
ANALYSIS MEANS
Analysis means 16 is comprised of analysis memory
means 98, spectral matching means 100, and an analog to digital
(A/D~ converter 102, all of which can be implemented by a
programmable digital computer utilizing known techniques of
~` computer programming or by a hard-wired, logic and control
system utilizing known techniques of logic design and availa~le
integrated circuit logic components.
The primary function of analysis means 16 is to
analyze the measured spectral data of a sample for the presence
of one or more of the target compounds. The contracted mass
spectrum for each target compound, i.e., the mass position
-37-
. , ' ,
~o45z53
and relative intensity of each peak thereof, is stored in the
analysis memory means 98. The measured spectral data of a
sample is o~tained at each mass position of each target com-
pound sought as described hereinabove. The magnitude of the
ion intensity signal output by peak stretcher 68 to analysis
means 16 is, of course, the value of the mass peak at thet
mass position. The ion intensity signal is first converted
to digital form by A/D converter 102 and then fed to spectral
matching means 100 and stored in analysis memory means 98.
The first mass peak in the sequence is measured at an
arbitrary and predetermined system sensitivity (determined by
the gain of the electrometer amplifier 66 and peak stretcher
68). The scan rate and filter time constant are set by para-
meter determination means 96 to be appropriate for that sensi-
~! 15 tivity. If, after the first sca~, the ion intensity of the
first peak is too large or small for accurate measurement,
~- the spectral matching means 100 outputs a rescan signal to
the parameter determination means 96 indicating that the scan
should be repeated at a higher or lower sensitivity as the
case may be. The parameter determination means 96 responds by
adjusting the gain of the electrometer amplifier 66 and/or
that of peak stretcher 68 upward or downward as appropriate
and the same mass position is rescanned. When the intensity
of the first mass has been established, the maximum quantity
of the target compound present is estimated from the known
response sensitivity of the system to quantities of that
compound. Based on the estimated quantity of the target com-
pound which appears to be in the sample, and the relative in-
tensity pattern of the target compound's contracted mass
spectrum stored in analysis memory means 98, the intensity
.
-38-
.
lO~SZS3
of the next mass to be sought in the contracted mass spectrum
of target compound is predicted. The predicted intensity of
the next mass peak in the sequence is output from spectral
matching means 100 to parameter determination means 96. The
latter, in response thereto, adjusts the system sensitivity
for the next mass peak and the mass position thereof is
scanned.
The value of the second mass peak measured is com-
pared with the predicted value. If it exceeds the predicted
value by more than a predetermined tolerance, the mass peak
is identified as contaminated, that is, consisting of ions
produced by at least one compound other than the target com-
pound, as ~ell as (possibly) the target compound. If, on the
other hand, the measured intensity of the second mass peak is
too small to measure accurately, the spectral matching means
100 outputs a rescan signal indicating that the scan should
be repeated at a higher sensitivity. The sensitivity is in- -
creased appropriately and the mass position is rescanned.
If the accurately measured intensity of the second mass peak
is smaller than the predicted value by more than a predeter-
mined tolerance, the estimated maximum amount of the target
compound present is reduced based upon the known sensitivity
response of the system to quantities of the target compound.
In such an event the intensity of the previous (first) mass
peak is predicted on the basis of the known relative intensity
pattern of the target compound's contracted mass spectrum.
The predicted intensity of the previous mass peak is then com-
pared to the previously measured value thereof. If the pre-
vious mass peak exceeds the predicted intensity by more than
a predetermined tolerance, then the previous peak is identified
-39-
,:
lV45253
as "contaminated".
The above-described sequence is continued until the
mass peaks at each of the mass positions in the contracted
mass positions in the contracted mass spectrum have been
measured. During and/or after the measurements described,
a confidence index relating to the likelihood that the target
` compound is present, and an estimate of the quantity of the
target compound which may be present, are determined from the
mass peaks found to be uncontaminated. This is described more
fully below.
For greater accuracy, the intensities of all mass
peaks measured (including calibration intensities~ are cor-
rected by subtracting from the measured intensities the pre-
^~ viously measured background intensities at all mass positions
to be scanned for identification purpose. The background in-
tensities are measured under conditions which are identical
to those under which the sample identificatian is to be made,
. . .
except that the sample is absent. Such background data is
stored in analysis memory means 98.
In the event that a mass peak intensity at a particular
` mass position is predicted by spectral matching means 100 to
^ be below the background intensity at that mass position,
spectral matching means 100 issues a "skip to next mass"
signal to parameter determination means 96. The latter re-
sponds by initiating a mass step at the appropriate time.
The reason for skipping a measurement in view of such a pre-
dication is as follows. If the prediction is correct or too
high the intensity can't be measured since it is in the
"noise" of the system. If the prediction is too low, i.e.,
if the intensity of the peak were measured and it were higher
-40-
. .
1()4S2S3
than predicted (and measurable), it would nevertheless be
identified as "contaminated" (since it would be greater than
the predicted value) and would, therefore, contribute nothing
to the confidence of the identification.
CONFIDEN('E INDEX
;l
The probability that the mass spectral data examined
for a sample is due to the compound sought is determined in
the form of a confidence index, X. The calculation of K in- -
volves a number of assumptions and approximations. Within
these limitations, ~he probability that the mass spectral
data measured could arise from a compound selected at random
from the universe of compounds of interest is 1/2K. Thus, K -~
is the logarithm to the base 2 of the probability of a random
occurrence of the observed data.
In analyzing the unknown sample spectrum fro the pre-
sence of a target compound, Kj values are determined sequen-
tially by spectral matching means 100 for each of the selected
peaks sought. It is assumed that the various statistical
contributions to the probability are independent, so that the
overall probability is the product of the probabilities for
each contribution. Therefore, K is a linear combination of the
corresponding logarithms for uncontaminated peaks. Thus,
K = Kj = (Vj + Wj - Dj), where Vj is the logarithm
of the probability that a mass peak of a particular abundance
will occur, at the jth mass position, as discussed above; Wj
is a window tolerance factor, reflecting the degree to which
the matching of measured peaks to known peaks of the target
compound occurs; and ~j is a dilution factor based upon the
amount of the target compound found relative to the quantity
of sample introduced.
-41-
.
10~5ZS3
-One requirement for the indentification of the target
compound in the unknown mixture is that the relative abundances
of the peaks in the unknown spectrum must be consistent with
those of the corresponding peaks in the contracted spectrum of
the target compound. The probability that the required
abundance has occurred instead by chance will depend on the
expected degree of matching of these abundances. This, in
turn, is a function of the tolerance to which matching is
sought; i.e., the "window tolerance".
The range of abundance values is a function of the
dynamic range of the system. Whatever the range of abundance
values, it may be arbitrarily divided into a number of windows
as a function of the tolerance of matching. Thus, for example,
if the abundance range of the system is from 0.5% to 100%,
there are 8 windows reflecting a + 30% tolerance. These win-
dows are (1) 0.5% - 1.0%; (2) 1.0% - 2.0~; (3) 2.0% - 4.0%;
(4) 4.0% - 8.0%; (5) 8.0% - 16%; (6) 16% - 32%; ~7~ 32% - 64%;
and (8) 64% - 100%. It should be noted that the window of
highest abundance is truncated since no peak can have an
abundance greater than 100%. It is seen that any abundance
value within one of these windows is within + 30% of the
center value thereof. A similar breakdown of the abundance
range can be done with respect to other tolerances, such as,
for example, + 10%, + 20% and + 50%.
It is assumed that it is equally probable that the
abundance of a measured mass peak will fall within one of the
arbitrarily defined tolerance windows. (A more sophisticated
approach might consider that the probability of a measured
abundance falling within a particular tolerance window is a
variable, and a determination made of that probability
.
-42-
.
10~5253
function by a study of the library of spectral data com-
prising the universe of interest.) In any event, the prob-
ability that an observed peak will fall within the correct or
predicted abundance window by chance is 1 in n, where n is
the number of windows in the abundance range (a function of
the tolerance as described above). Thus the window tolerance,
W, is defined as the logarithm to the base 2 of n. The values
of W as a function of tolerance is shown in the following
Table 2:
Table 2
Tolerance n W
+ 10% 32 5
+ 20% 16 4
+ 30% 8 3
15 + 50% 4 2
If, at the jth mass position, the measured mass peak
; abundance (intensity) falls within the tolerance window of
the predicted mass peak abundance, the "closeness" of the
. match is a function of the tolerance. The smaller the
tolerance, the higher the probability that sample is the tar-
get compound (since the relative intensity patterns appear to
match). The W value reflects this probability function by
being additive to the value of Vj (i.e., Uj - Aj).
The magnitudes chosen for Uj and Aj are based on the
spectra of pure compounds. However, if the target compound
represents only part of the sample, its actual abundance tand
thus the uniqueness) of its peaks wi~ll be reduced. In such
a case, the uniqueness contrubution to K of each peak should
be less than Vj. Thus, if a mass peak measured for an un-
` 30 known sample is due, in part, to a second compound, the
, ~
-43-
`:`
lU~5253
abundance factor must be reduced ~y a dilution factor D, the
latter being subtracted from the Vj values for each peak of
the target compounds contracted spectrum. ~hus, the dilution
factor D is based on the quantity of the target compound
found Q relative to the quantity of the sample, Qs, introduced
into the system, typically a standard quantity. The dilution
factor D is defined as follows:
D = - log Q ; (7)
If Q = Q5~ the sample is pure, and D = 0. For mixtures, Q is
less than Qs; The definition of D is based on the observation
that is there are 2D different compounds present in the sample
in equal amounts, the observed quantity of any one compound
would be reduced by the factor 2D and there would be about 2D
times as many peaks of any given intensity than if the sample
were pure.
With reference to the genreation of the confidence
index K, consider the first mass peak. For that peak, K1
equals Vl - Dl; (Wl equals zero since the first peak must
serve as the reference for predicting the abundance of the
next peak). The value of Yj at each mass position in the
spectra of the target compounds is stored in analysis memory
means 98. Thus, if a first mass peak is measured, spectral
matching means 100 obtains the value of Vl from analysis
memory means 98. The value of Dl is then determined from
the maximum estimated quantity of the target compound Q and
the known quantity of the sample Q5, in accordance with equa-
tion ~7), and subtracted from the value of Vl.
Next, the intensity of the second mass peak is measured.
If its intensity is within the window tolerance of its pre-
dicted value (based upon the intensity of the first peak), X2
1045Z53
equals V2 - Dl ~ W. Dl and D2 should be approximately equal
Since the estimate of the quantity of the target compound
from the intensity of the second peak should be about the
same as that estimated from the intensity of the first peak.
The value of V2 is obtained from analysis memory means 98
and the value of K2 determined by subtracting therefrom the
values of Dl and W, the latter in accordance with Table 2.
At this point the overall confidence index, Kl, equals Kl+ K2.
If the intensity of the second peak is greater than
the value predicted by more than the window tolerance, the
second mass peak is deemed to be contaminated and K2 is set
equal to zero. If, on the other hand, the intensity of the
second peak is below the predicted intensity value by more
than the window tolerance, the intensity of the second peak
becomes the new reference peak, and a value of the intensity
of the first peak is then predicted agian on the basis of
the intensity of the seoond peak and the stored spectral data
of the target compound sought (which data reflects the relative
intensities of the various peaks thereof). The measured
value of the intensity of the first peak is next checked to
determine whether it exceeds the predicted value by more than
the window tolerance. If it does, the first peak is deemed
contaminated and Kl is set to zero. X2 then becomes equal to
V2 ~ Dl,W2 being set to zero since the second peak is now
the reference peak. The measured intensities for all sub-
sequent peaks are treated in the same manner by spectral
matching means 100. Thus, if Ij, the intensity measured at
the jth mass position is below the predicated intensity value,
the intensities of all previous uncomtaminated peaks are re-
checked against the new reference intensity Ij.
-45-
.
1~)45ZS3
Any previous peaks which are now above the allowed window
tolerance are termed contaminated, and their K values are re-
duced to zero. The x values of the other peaks must be re-
calculated to reflect the new value of Dj, and then K is re-
determined.
If measured intensity Ij is equal to or less than the
background intensity at the jth mass position, Bj, Kj is set
to zero. However, if Bj if lower than the intensity pre-
dicted by the reference peak, the intensity of the jth peak
is assumed to be Bj and the intensities of all previous peaks
rechecked based upon the intensity value of Bj as the refer-
ence peak. The final confidence index R equals the summation
of the individual Xj values. Spectral matching means 100
does the summation of Kj values after the last measurement is
made at the last mass position in the contracted mass spec-
trum of the target compound.
The final value of confidence index K is output from
spectral matching means 100 to a suitable display which is
part of conventional input/output means 104. This confidence
index informs the operator that, on average, 2K compounds in
the universe of interest would have to be selected at random
and examine in order to find data which would match the target
compound's contracted spectrum to the same degree as does the
unknown sample.
QUANTITATION
After contaminated peaks are eleminated, spectral
matching means 100 estimates the quantity, Q of the target
compound identified. One method for doing this is to average
the values of the estimated quantity, Qj determined at each
uncontaminated peak. Thus,
-46-
1~)45Z53
Q = m im=lQj
Alternatively, the quantity Q could be determined
from the intensity of the peak finally used as the reference
peak. The quantity based upon the reference peak intensity
must be considered a maximum value, since this peak could be
contaminated. However, the probability of such contamination
will be small if the confidence index K is relatively high.
MODES OF OPERATION
The preferred embodiment of this invention is capable
of operating in four basic modes; (i) start-up; (ii) calibra-
tion; (iii) identification; (iv) data system. Mode selection
is done via conventional switches on input/output means 104.
In the start-up mode, proper operating temperatures
of the sample inlet device 10 are established; the separator
20 and the mass spectrometer 12 are evacuated; the system
sensitivity is established by the parameter determination
means 96; and the mass scan converter scale factor and offset
corrections are made so as to enable accurate mass peak
measurements. Other necessary preliminary functions are also
-20 carried out by control means 14 and/or by the operator.
In the calibration mode, the intensities of the con-
tracted reference spectrum for each target compound are
measured and stored in the analysis memory means 98. A known
sample quantity of each target compound is first injected
into the sample inlet device 10. The calibration is then
performed as follows:
The first mass in the identification set is repeatedly
measured, with appropriate adjustments in system sensitivity
to keep the peak on scale. When the intensity of the first
mass peak exceeds the stored background intensity at that mass
11~)4S253
position, by a specified factor, the intensity of the peak is
monitored until it reaches a maximum and begins to decrease.
; At this point the remaining masses in the set are measured in
sequence and stored, after subtracting the corresponding stored
background intensities.
The relationship between the intensities of the mass
peaks measured and the known quantity of the calibration sample
is used by spectral matching means 100 to determine the
systems response sensitivity at each mass position in the
contracted mass spectrum of the target compound. Thus, if
the intensity of the ith mass is Ij, and the calibration quan-
tity is Qc' the sensitivity of the jth mass to the target
compound is Sj = Ij/Qc~ The values of Sj are stored in
analysis memory means 98 and used in the identification mode
to determine estimated quantities of the target compound
present, as described above.
The measurement of background intensities, as des-
cribed above, can be considered to be done in the calibration
mode. However, the system must, of course, be purged of any
residual sample material.
The identification mode is used to assay samples for
; the presence of one or more of the target compounds. It is
subdivided into a confirmation mode and a search mode.
In the confirmation mode, the invented system is used
` 25 to confirm the presence of a single target compound. Thus,
the mass analysis to determine K and Q described above is run
repeatedly for a period of time determined by the operator
~through input/output means 104), during which time the
sample is introduced. The highest values of K and Q are re-
tained as the final confidence index and quantity estimates.
-48-
- . .
1~)45253
This value of K is compared with a predetermined threshold
value KT; if K exceeds KT, the identification is positive;
that is, the target compound is indicated to be present in
the sample. The amount of the target compound estimated to
be contained in the sample is the final Q value. The results
of the analysis are then displayed on a display position of
input/output means 104, preferably on a set of illuminated
indicators labeled with the names of the target compounds.
If one of the target compounds is identified, the correspond-
ing indicator is illuminated. The result may also be printedout, if desired, on a conventional teletype. In addition,
the confidence index, mass peak intensities and quantity found
may be displayed and/or printed out.
In the search mode, each target compound in the entire ~-
set of target compounds, or a subset thereof, is searched for
sequentially in the sample. For each compound sought, the
spectral analysis of the sample is carried out as described
above, with two exceptions; first, the spectral analysis is
terminated by spectral matching means 100 with respect to
any target compound sought if (a) the confidence index K,
based upon the peaks analyzed to that point, exceeds a thres-
hold index KT; or (b) the quantity estimate of the target
compound at any time is less than a threshold quantity QT. In
the former case, the identification is deemed to be sufficiently
positive. In the latter case, the identification is deemed to
be negative because the target compound is not present in a
sufficient quantity. When the spectral analysis is terminated
under either of the foregoing conditions, spectral matching
means 100 issues a "skip to next compound" signal to the para-
meter determination means 96, causing the latter to proceed
-49-
.. ~. ,
104S253
to the next target compound; i.e., to analyze the sample at
the mass positions of the next target compound's contracted
spectrum. If an identification of a target compound is
positive, the quantity estimate thereof is the maximum esti-
mated at any peak satisfying the condition, K5 KT.
By proper selection of the mass peaks used in the
identification, it is usually possible that only one, or at
most a few, peaks need be examined to establish the absence
of a target compound. As a result, the present invention can
analyze a sample and search a plurality of target compounds
for an identification match in a relatively short tmme. For
example, 16 target compounds can be searched in much less
time than that normally required to measure once the entire
mass spectrum (e.g.400 amu) of the sample.
As in the confirmation mode, illuminated indicators
and an optional printout may be used to convey the results
of the analysis to the operator. When no positive identifica-
tion occurs, the operator may then proceed to the next sample.
~ In the identification mode, the input/output means
`I 20 104 permits the operator to select the target compound or
compounds to be confirmed or sought, in the sample.
In the data system mcde, the present invention can
be used to generate new or special compound identification
programs, as well as for conventional mass spectrometry or
mass fragmentography applications. In this mode, one can
rapidly scan all or part of a mass spectrum in any sequence,
print or plot results, and analyze the data to derive a con-
tract mass spectrum.
First, the control memory means 92 is loaded with all
or a portion of the known mass positions found in the spectrum
-
' : ; , . ~ .
1045ZS3
of a particular compouna. A known quantity of a pure sample
of the compound is then injected into the sample inlet device
10 and the mass peak intensities are measured at each of the
mass positions stored. Measurements can be made repeatedly
or at the operator's request. The measured mass intensities
are corrected by subtraction of the background intensities
present in the system.
For each mass peak measured, spectral matching means
100 looks up the uniqueness factor Uj, stored in analysis
memory means 98. In addition, based upon the relative inten-
sity of each mass peak measured, the abundance factor AJ is
also determined for each peak from Table I, which is also
stored in analysis memory means 98. Spectral matching means
100 reduces the value of each Uj factor by the corresponding
abundance factor Aj to yield a-Vj value; as indicated above,
Vj = Uj - Aj. The values of Vj, Ij and Sj at each mass peak
measured are displayed through the input/output means 104.
The operator may then select the most unique mass peaks ti.e.
those having the highest values of Vj) to make up the con-
tracted mass spectrum of the compound measured. The spectraldata of the contracted mass spectrum of that compound may
then be stored in control memory means 92 and analysis mem-
ory means 98 for subseguent use in identifying that compound
in an unknown sample. This invention also contemplates the
automatic selection of the most unique mass peaks based upon
one or more pre-stored selection criterion, and the automatic
storageof such selected mass peaks in memory means 92 and 98.
Thus, in the data system mode, any compound of interest may
be analyzed and the necessary spectral data obtained to en-
able that compound to be made a target compound in the invented
-51-
1~45ZS3
system. In this manner, the operator can develop his own set
of target compound spectra.
Operation of the invented system in the calibration,
identification and data system modes requires introduction
of the sample into the sample inlet device 10. Samples are
introduced in such a way that the partial pressure of the
sample vapor in the mass spectrometer is maintained at a
level sufficient for detection above the spectrometer back-
ground for the 1 to 10 seconds required to perform an analysis.
The temperatures of the flash evaporator 18 and separator 20,
the carrier gas flow rate and the vacuum system pumping speeds
are chosen so that the partial pressure of the sample vapor
in the mass spectrometer ion source has the general time be-
havior shown in Figure 12. Gaseous samples may be valved into
the carrier gas stream so as to produce a similar pressure
` pattern.
A number of variations on the structure and operation
of the present invention are possible without departing from
its essential scope and spirit. Some of the possible varia-
; 20 tions are described as follows:
.~ .
(1) The analysis of data for identification purposes
could be done "off-line" rather than in real time.
That is, the data could be measured and stored, and
` analyzed later.
(2) An alternate comparison or matching method could
be used, such as, for example, the "Biemann - MIT"
system which utilizes the two largest peaks in each
14 mass unit interval (described by Hertz, Hites, and
Biemann in Analytical Chemistry Vol. 43, page 681~1971).
(3) The entire spectrum, rather than a partial
`
-52-
,. -
~45253
spectrum, could be measured. For each eompound of
¦ interest the appropriate measurements could be ex-
tracted and compared with the reference data.
(4) Identification data generation, as described with
respect to the data system mode above, could be made
entirely automatic. That is, an unknown sample (pure
or mixture) could be introduced and, in lieu of the
; operator, the system could be designed to select the
most appropriate (highest Vj) peaks for the con-
tracted mass spectrum. The speetral data associated
with sueh peaks would then be stored for use in the
subsequent assay of unknown samples.
The inventive principles embodied in the present in-
vention are also applieable to other physical measurement
methods which produce a "spectrum" or a pattern of eharacter-
istic relative intensities as a function of some parameter.
j Examples of sueh physieal measurements inelude emission a~dabsorbtion optical spectrometry (ultraviolet, visible, in-~
frared), nuclear magnetie resonance and X-ray spectrometry.
Although this invention has been disclosed and des-
eribed with reference to a particular embodiment, the
prineiples involved are susceptible of other applications
whieh will be apparent to persons skilled in the art. This
invention, therefore, is not intended to be limited to the
particular embodiment herein disclosed.
.,
.
~ -53-
