Method and Apparatus for Multiplexing Plural Ion Beams to a Mass Spectrometer

Inventors:

James G. Boyle, Madison, CT Robert A. Valley, Guilford, CT

References Cited

US Patent ] Documents

3,740551 6/1973 Green 250/41.9 ME

3,831026 8/1974 Powers 250/296

4,507,555 3/1985 Chang 250/281

5,179,278 1/1993 Douglas 250/290

5,331158 7/1994 Dowell 250/282

5,420,425 5/1995 Bier 250/292

5,652,427 7/1997 Whitehouse et.al 250/288

5,689,111 11/1997 Dresch et.al 250/287

5,763,878 6/1998 Franzen 250/292

5,811,800 9/1998 Franzen et.al 250/288

Other Publications

Ooms, B. Temperature Control in High Performance Liquid Chromatography, LC-GC (Asia Pacific), vol. 1, No. 1, p. 27-35 (1998).

Lin H. Y., Voyksner R.D., Analysis of Neuropeptides by Perfusion Liquid Chromatography/Electrospray Ion-trap Mass Spectrometry, Rapid Communications in Mass Spectrometry, vol. 8, p. 333-338 (1994).

Chien, B.M., Michael, S.M., Lubman, D.M., Plasma Source Atmospheric Pressure Ionization

Detection of Liquid Injection Using an Ion Trap Storage/Reflectron Time-of-Flight Mass

Spectrometer, Analytical Chemistry, vol. 65, p. 1916-1924 (1993).

Boyle J.G., Whitehouse, CM. and Fenn J.B., An Ion Storage Time-of-fiight Mass Spectrometer for Analysis of Electrospray Ions, Rapid Communications in Mass Spectrometry, vol. 5, p.400-

405 (1991).

Grix, R., Griiner, U., Li, G., Stroh, H., Wollnik, H., An Electron Impact Storage Ion Source for

Time-of-Flight Mass Spectrometers, International Journal of Mass Spectrometry and Ion

Processes, vol. 93, p. 323-330 (1989).

Rights to the Invention

The work leading to this invention was conducted under research sponsored by the United States

National Institutes of Health. The US government shall therefore have the right to practice this

invention.

Background of the Invention

This invention relates to mass .spectrometers and their ability to multiplex between

simultaneously arriving and discrete sample streams without incurring either sample loss or intra-

sample mixing. It concerns itself with the issue of maximizing sample throughput on a mass

spectrometer by creating parallel sample introduction and transmission paths, while at the same time ensuring that no mixing of the individual sample streams occurs. In this manner, chemical

data are uncompromised in terms of cross-stream contamination, while the overall sample throughput is increased substantially.

This invention is applicable to any mass spectrometer which depends upon batch-wise

introduction of samples for performing mass analysis, including but not limited to time-of-flight

mass spectrometers (TOF-MS), fourier transform ion cyclotron resonance mass spectrometers

(FT-ICR-MS), and three dimensional ion trap mass spectrometers (IT-MS). Time-of-flight mass

spectrometers are best suited to exploit this parallel introduction invention because of their

inherent ability to process discrete samples on a millisecond time basis or faster. While FT-ICR-

MS and IT-MS systems require greater periods of time to acquire high quality mass

spectrometric data, these systems could also make use of this invention to improve sample

throughput. Commercial FT-ICR-MS systems are currently capable of generating mass spectra at

a rate of approximately 50 Hz. While several orders of magnitude lower than TOF-MS systems,

this acquisition rate would still permit use of the invention with multiple sample streams, given

that these streams could be sampled frequently enough to reflect any temporally dynamic sample

concentrations present.

This invention is applicable to any mass spectrometer with an external ion source, and is

particularly useful when this ion source produces analytically important ions continuously over

extended periods of time. Examples of external ion sources which can produce ions continuously

include electrospray ionization (ES) and atmospheric pressure chemical ionization (APCI) ₅ both

of which may be coupled to liquid chromatography (LC) in order to first temporally separate different species prior to MS interrogation. When coupled to LC or other chemcial separation

instruments, ES and APCI ion sources generate ions from a temporally dynamic stream of

analyte molecules, ranging in duration from seconds (for very fast separations) to several hours

(for very long separations).

A fundamental principle of time-of-flight mass spectrometry is the extraction of a closely packed

ensemble of ions formed at time zero. These discrete ensembles of isoenergetic and spatially

coherent ions are accelerated from an extraction region and into a field free flight tube for

longitudinal separation based upon their different (constant) velocities and hence mass-to-charge

ratios. Ions created outside the extraction region may be injected into the extraction region, such as from an atmospheric pressure ion source or glow discharge source. Alternately, ions may be

created within the extraction region from neutral molecules, for instance by using a pulsed beam

of photons, electrons or ions. In either case, only those ions that are in the extraction region at the

moment the starting pulse is applied are analytically useful, as only these ions will be imparted

with the proper energy to be detected and properly characterized after field-free flight.

Given this constraint, the direct coupling of a continuously operating ion source to a time-of-

flight mass spectrometer suffers from an inefficient use of the ions created. While one may apply

start pulses to the time-of-flight mass spectrometer at frequencies which match the characteristic

time required to re-fill the extraction region from an external supply of ions, duty cycles may still

be far from unity under certain conditions.

A solution to this mismatch caused by interfacing a continuous ion source and a batch processing

method such as time-of-flight mass spectrometry has been described by Dresch et.al. (1996). In

order to make use of the greatest fraction of ions generated as possible, a multipole ion guide is inserted at the appropriate location between the ion source and the extraction region to store ions

between consecutive start pulses. Owing to the fact that it is a two dimensional device spanning

multiple pumping stages, this device can deliver ions to the extraction region either as a

continuously transmitting ion guide or as a pulsed two dimensional ion trap. In contrast to three

dimensional ion traps described by Lubman (ref) and Douglas (ref), this two dimensional ion trap

can hold a far greater number of ions within its volume before reaching an experimentally

observed critical density. Critical density is characterized in practice by the observation of mass

spectral signals which may be reduced in amplitude, or different due to catastrophic ion

fragmentation, or improperly focussed at the detector due to greater internal energies, or some

combination of the above. For a given flux of ions being delivered from an external ion source,

the higher charge capacity of this two dimensional ion trap allows storage of ions for more time.

This is of the utmost importance to the present invention in affording adequate time for

sequentially introducing multiple independent samples through a single time-of-flight mass

analyzer without loss of information on the chromatographic timescale.

Ionization methods such as electrospray and atmospheric pressure chemical ionization are

utilized regularly to ionize liquid samples containing non- volatile compounds of interest, including but not limited to peptides, proteins, pharmaceutical compounds and metabolites.

The sensitivity, specificity and selectivity of API-MS have made it an essential research tool in

the life sciences and pharmaceutical development, in which the analytical performance of API-

MS systems has most often been categorized in terms of limits of detection, mass resolving

power, mass accuracy, and mass-to-charge range. Previously, little if any regard was paid to

issues relating to automation.

Spurred on over the last several years by pharmaceutical development methods, strictly analytical

performance metrics have been joined by automation metrics. Automation of analytical tests such

as API-MS afford one or more advantages over manual operation, including:

• Reduced labor

• Reduced expertise of labor

• Higher sample throughput

• Better utilization of capital instruments

• Better analytical reproducibility (as measured by the relative standard deviations from

sample to sample) ,

As an example, the automation of API-MS now allows previously untenable sample sizes to be

more rapidly analyzed, thereby supporting technologies such as combinatorial chemistry. which

require very large sample sizes to isolate a compound of interest.

As a result, there have been considerable advances in automating the operation and data

collection of API-MS instruments both at the hardware and especially the software levels. The

latter case is best exemplified by the introduction of Open Access standards for non-expert users.

The former case is best illustrated by the introduction of multiple injector autoinjectors such as

the Gilson 215 instrument (Madison, WI). What has been lacking are the means to accelerate the

throughput

Within the last several years, there has been increasing interest in coupling these continuous

ionization methods to time-of-flight mass spectrometry in order to achieve certain performance

characteristics which would be otherwise unattainable. These include but are not limited to high

mass accuracy, high mass-to-charge detection, quasi-simultaneous detection of the entire mass-

to-charge domain, high pulse rates, high sensitivity, and fewer tuning requirements than scanning

type mass analyzers.

Collectively, these features make time-of-flight mass spectrometers ideally suited as detectors for

temporally changing sample streams. Moreover, the ability to couple liquid separation systems

directly to atmospheric pressure ionization sources such as electrospray ionization and atmospheric pressure chemical ionization allows for on-line processing of these separations

without the need to collect chromatographic or electrophoretic fractions for off-line processing.

In fact, the sampling rate of atmospheric pressure ionization time-of-flight mass spectrometers

with ideal data system architectures can generate complete mass spectra with adequate ion

statistics in far less than 1 second. This speed of acquisition allows faster liquid separation

protocols to be designed and implemented which slower, scanning types of mass spectrometers could not record with adequate chromatographic fidelity.

The desire to introduce multiple samples into a single mass analyzer stems from a combination

of factors. Technically, time-of-flight mass spectrometers are fast enough in "scanning" a useful

mass range that multiple samples can be completely characterized even when these samples are

themselves temporally dynamic (as in the case of a liquid chromatogram). For instance, the vast

majority of liquid samples separated by reversed phase chromatography will exhibit LC peak

widths on the order of several seconds or more. This is ample time for a single TOF-MS to mass ₍

analyze several samples, given its ability to form complete mass spectra in as little as 100

microseconds or less.

This multiplexing capability is inviting for those who wish to (a) achieve higher capacity utilization, (b) lower capital costs, (c) shrink total required laboratory space, (d) centralize data

handling and (e) minimize hardware maintenance.

There are a number of important works which define the state of the art as it relates to this patent

application. These works involve the development of plural ions, parallel mass spectrometers,

and ion storage using two dimensional ion traps.

The use of plural ion beams in either single or parallel mass spectrometer has been demonstrated

by a number of inventors and for a number of distinctly different reasons. Green in U.S. Pat. No.

3,740,551 demonstrated parallel mass separation and detection of different ion beams

simultaneously, principally as a means of performing both high and low resolution mass spectral

scans on magnetic sector type instruments. These ion beams could originate from either a single

chemical sample or from a sample and a reference compound which was used to calibrate the

mass scale of the instrument. In. U.S. Pat. No. 3,831,026 Powers taught the use of a time division

multiplexing apparatus, which sampled alternate ion beams for mass separation and detection in an interleaved fashion. This multiplexing apparatus consisted of either a pair of plates at

controlled voltages or a continuously transmitting hexapole ion optic. By overtly controlling the

portion of time that each ion beam was sampled, relative intensities of the two beams could be

better managed for greatest analytical utility. Chang was among the first to recognize the utility

of plural beams and parallel mass spectrometers in analyzing temporally dynamic samples from

either gas chromatography (GC) or liquid chromatography (LC) in U.S. Pat. No. 4,507,555. Like

the aforementioned inventors, parallelism was sought as a means of extracting different types of

mass spectrometric data from a single sample, especially in circumstances when rapidly eluting

compounds made it difficult or impossible for a slow scanning quadrupole MS to keep pace. One

quadrupole was used to monitor a single target mass-to-charge of interest, as well as to trigger full mass range acquisitions by a second quadrupole should the target ion appear. This improved

detectability over full mass range survey scans by a factor of 100. Using time-of-flight as the

preferred mass separation scheme, Dowel in U.S. Pat. No. 5,331,158 demonstrated the ability to

achieve 100% duty cycle of a flight tube (not an individual chemical sample) by injecting ion

packets from multiple electron impact ion sources in rapid succession to one another.

Several important patents have been issued in the area of two dimensional ion guides and ion

traps, all of which teach important aspects of the science which underpin this patent application.

Douglas in U.S. Pat. No. 5,179,278 taught that two dimensional multipole ion guides were highly

effective devices for trapping and storing off-cycle ions until a three dimensional ion trap mass

spectrometer had completed its analysis of the previous ion bunch. Both pre-selection and

collisional cooling of the stored ions were described as advantageous features. Bier in U.S. Pat.

No. 5,420,425 furthered this argument by demonstrating the relative analytical advantages of two

dimensional ion traps in terms of their storage capacity, circumventing the charge limitations

which less stretched ion traps necessarily suffer due to space charge constraints. Both Whitehouse in U.S. Pat. No. 5,652,427 and Dresch in U.S. Pat. No. 5,689,111 describe the use of

a multistage two dimensional ion guide as an appropriate ion storage device to feed batch-wise

mass spectrometers, including time-of-flight, ion trap and Fourier Transform Ion Cyclotron

Resonance type systems. These patents taught the use of enhanced collisional cooling by close

coupling a multipole ion guide to the free jet expansion of an atmospheric pressure ionization

source. In this way, ions could more effectively be captured while still experiencing viscous

forces in the high pressure region of an atmospheric pressure ion source. After capture, their

cooling and transport to a much lower pressure region would ensure a much more monoenergetic

ion beam which was better suited for injection into energy sensitive MS systems, especially TOF-

MS. Franzen in U.S. Pat. No. 5,763,878 extends the multipole ion trap functionality by both trapping ions within the device and using it as the ion source of an orthogonal TOF-MS. Most

recently, in U.S. Pat. No. 5,811,800, Franzen generates bunches of stored ions from an

atmospheric pressure ion source using RF ¹ coils, this time for the purpose of feeding a three

dimensional ion trap MS system.

The ability to introduce different samples from different separation systems into a single time-of-

flight mass spectrometer was recently introduced by Micromass, Inc. In this design as many as

four different liquid streams are multiplexed, with sample selection occurring at atmospheric

pressure. This concept is commercially advantageous insofar as it makes use of a standard LC-

TOF-MS, requiring no modification of the vacuum system or ion optics to work. However, since

all four liquid streams flow continuously, the selection of any one stream necessarily imposes a

duty cycle limit dictated by the number of streams sampled. For those streams which are "off-

cycle" (i.e. not sampled) any analytical information contained in the off-cycle portions of those

liquid streams is lost and can not be recovered. For a large number of applications currently in

practice involving high concentrations of synthetically derived small organic libraries, analytical

sensitivity is not of paramount concern. Nevertheless, this approach is analytically

disadvantageous in circumstances in which sample amounts or concentrations are especially low.

Proteomics, including both general molecule characterization as well as peptide sequencing, is a

critically important field for which analytical sensitivity is paramount, especially in applications

being reduced to nanoscale dimensions for both separation processes ("lab-on-a-chip") and mass spectrometry (nanoelectrospray).

The present invention arises from the need to mass spectrometrically characterize larger numbers

of distinct samples than is currently possible, but without requiring multiple independent mass

spectrometers. This analytical need is driven in large part by the adoption of combinatorial

chemistry methods by pharmaceutical researchers, who today are the largest and one of the

fastest growing segments of the mass spectrometry market worldwide (Strategic Directions

International, 1996). Due to this shift towards combinatorial chemistry and away from slower,

rational drug design programs, the number of compounds which are being regularly generated

and which require positive identification via mass spectrometric analysis has risen dramatically (Doyle, 1995). This trend is expected to continue for years to come (Hail, 1998).

In the field of functional genomics, the ability to identify and characterize gene products

(proteins) with vanishingly small amounts of material using mass spectrometry will be essential.

Standard separation tools in existence today, including two dimensional electrophoresis, can both

separate and detect proteins in amounts far below the detection limits of any mass spectrometer

,,(ReQ. While more abundant proteins are easily detected, a large portion of all the proteins

contained in mammalian cells exist in copy numbers below the present day capabilities of

dedicated, research grade mass spectrometers. Since many of these low abundance proteins are

likely to have important regulatory functions in cells, their efficient detection using appropriate

staining techniques and their subsequent digestion and analysis using mass spectrometry is vital.

(Herbert, Proteome Research: New Frontiers in Functional Genomics). This need is exacerbated

by the fact that the entire proteome complement of any organism is a function of age, heredity,

wellness, and environmental conditions. Such a dynamic system requires analytical tools which

- can monitor an organism at various stages of its lifetime. This scarcity of sample will limit the future effectiveness of "lossy multiplexing", i.e. the use of multiple sample streams multiplexed

to a single mass spectrometer with duty cycle limits.

Briefly, syntheses of combinatorially created compounds with potential therapeutic value are

carried out using small sets of related starting materials. These sets cover the physical chemical

parameters that are required to optimize the properties associated with a pharmaceutical agent,

such as good oral bioavailability and in vivo stability. The library or array which results from all

possible combinations of these starting materials may be very large in an attempt to cover an

appropriate property space, ranging in size from several hundred to several hundred thousand

distinct compounds. The complete library or some portion of it which meets certain preliminary

screening criteria (the presence or absence of a fluorescence signal, for example) may require

complete chemical characterization, usually by mass spectrometry. Because each of the nominal

library constituents may be a mixture of the intended product, side-products, reactants, and

impurities from various sources, mass spectrometry may be employed in conjunction with a

separation method such as liquid chromatography (LC-MS) to separate in time these various

components. By separating the individual components within a reaction volume, components

elute separately into the ionization source and MS system, generating a mass chromatogram of total ion current versus time. This both simplifies analysis of the data and optimizes the response

of the MS system for each constituent by maximizing the ionization efficiency (i.e. minimizing

charge competition).

While the chemical specificity of an LC-MS system is greater than using an MS system in the

absence of liquid chromatography, there is a time penalty associated with performing an LC

separation, reducing the highest achievable sample throughput. The alternative and faster method

of analyzing individual liquid samples is by flow injection analysis MS (FIA-MS), infusing

liquid samples directly without chromatographic separation.

While the maximum rate at which samples can be sequentially analyzed using either FIA-MS or

an LC-MS varies depending upon the specific protocol being followed, in general FIA-MS

typically requires between tens of seconds and a minute per sample, depending upon the specific

autoinjector hardware being used and the stringency of the inter-sample rinsing. Users in high

throughput settings have demonstrated the ability to analyze as many as 1000 samples per mass

spectrometer per day in this manner. The primary drawback to this approach is the

aforementioned uncertainty in ionization efficiency in the presence of possible impurities. In

instances in which the mass spectrometric response is being used as an indicator of the presence or absence of an expected product, the quality of the mass spectrometric data are vital in judging

the utility of a particular library compound. Typically one looks for an expected molecular ion of

mass M ₁ to verify synthesis confirmation. If this expected mass is obscured or suppressed by the

presence of an impurity with a greater proton affinity of mass M ₂ , then the mass spectrum

generated by flow injection MS may not reveal the presence of the target product. However, if

the liquid solution containing both of these species is first separated by liquid chromatography or

some other appropriate separation which can partition the compounds based upon their physical

or chemical properties, then the resultant mass spectra may likely reveal the presence of each of

these constituents.

In the LC-MS mode, protocols specifically designed for rapid separation of small

molecules.typically require between 5 and 15 minutes, an improvement over traditional 30-60

minutes gradients used before the advent of high throughput screening but still orders of

magnitude slower than other non-mass spectrometric assays. Recently, Banks (1996)

demonstrated more rapid separations of complex mixtures in reversed phase LC-MS using both

normal bore (4.6 mm ID) and microbore (320 μm ID) columns packed with small uniform

spheres of non-porous silica. Separations of 2-3 minutes were typical, demonstrating both high

throughput and very high chromatographic resolution. These faster runs were specifically

designed to exploit the ability of a time-of-flight mass spectrometer to handle very high data

rates. In practice, the compression of chemical separations and the sub-second generation of mass

chromatograms by time-of-flight mass spectrometry is the chemical analog of high speed electronic waveform capture, requiring both the means to generate and record events (ions) at the

high megahertz to gigahertz frequencies. For this reason, high speed separations coupled to MS

have been labelled "burst mode" systems (Banks, 1995). Representative of the current state of the art in high throughput LC-MS, this work clearly shows that radical (order of magnitude or more)

improvements in LC-MS throughput, even with specialized chromatographic methods, are not

easily obtained when operating in a strictly serial fashion. In order to overcome the sample

throughput limitations described here and summarized in Table 1, one of two approaches must be

adopted.

First, additional LC-MS instruments, each operating in a serial fashion, could be brought on-line

to increase throughput in a strictly linear fashion. This requires a proportionate expenditure of

capital and expense funds to purchase and operate multiple machines, as well as requiring

multiple computer systems to run the instruments and acquire and analyze data.

Second, multiple separation systems could be coupled in-turn to a single mass analyzer, allowing

an LC-MS run to proceed with one LC system while a second LC system is re-equilibrated and a

new sample prepared and injected. Such a system has been integrated by the Micromass Division

of Waters Corp. for high throughput applications on quadrupole based LC-MS systems. Such an

approach is a cost effective means of improving specific sample throughput (in terms of samples

per unit time per dollar of realized capital expense), and derives the maximum benefit possible

from the relatively expensive mass spectrometer and data system. However, there are two

significant limitations. First, the net sample throughput operating two LC systems coupled to a

single mass analyzer with a single ion source is far less than two LC-MS systems operating

independently. That is, the time savings per sample is approximately equal to that fraction of the

time that a single LC system spends re-equilibrating and injecting a new sample onto the column (Figure N).

Third, multiple LC systems could be run in tandem and samples from each be sampled by the MS

in turn, using either liquid flow valves or alternating ionization probes to achieve a multiplexing of samples in a single mass analyzer. In the absence of true sample storage, those LC streams

which are not being sent to the mass analyzer at any instant in time are being sent to waste.

Therefore, this time-slicing approach suffers from the fact that by reducing the duty cycle of each

effluent stream, the mass analyzer will be rendered blind to peaks which occur off-cycle. In light

of higher speed and higher plate count methods now coming into wider practice, there would be

an unreasonably high risk of sending to waste complete peaks which would escape mass

spectrometric detection.

The desire to accommodate multiple samples simultaneously in order to achieve higher sample

throughput stems in large measure from the growth of combinatorial chemistry. The Biotage

Corp. of Charlottesville, VA produces a product called Parallex HPLC, intended to allow four

samples to be chromatographically separated simultaneously. In order to interface these four

separate and discrete liquid streams to a mass spectrometer currently, the four streams are routed

through a rotary valve which serially introduces each of the four streams to a mass spectrometer's

ionization source. In order to prevent stream-to-stream mixing, a bolus of make-up solvent (a

"blank") is introduced into the flow in between consecutive analytical samples. For four separate

liquid streams represented by A, B, C, and D, and the make-up solvent represented by S, the

sequence of sample delivery to the mass spectrometer will be

ASBSCSDSASBSCSDSASBSCSDS This necessarily implies that the maximum duty

cycle achievable for any one of the liquid streams is limited to the portion of time it is actively

being sampled, which is one-eight of the total experiment time or 12.5%. For the other 87.5% of

the time, those streams which are "off-cycle" are not accumulated, but rather are discarded as

waste. The time interval required to sample all four liquid streams is on the order of 1 Hz. There are two limitations in coupling such a system to mass spectrometry in order to achieve higher

sample throughput. One difficulty is the immediate loss in sensitivity due to the duty cycle limit.

Moreover, muliplexing the samples in the liquid phase exacerbates this problem due to the need

to introduce inter-sample blanks. The second difficulty is the inability of the multiplexer to select

any given liquid stream at a rate greater than 1 or several Hz. Driven by the need to analyze

samples ever faster, the clear trend in chromatography is towards faster, higher resolution

separations (Ooms). In many cases, separation protocols are now being developed which require

only several minutes even for complex mixtures, with eluants exhibiting peak widths of several

seconds or less. In instances such as this, mass spectrometric sampling of individual

chromatographs at one or several Hz will be inadequate to recreate with any acceptable fidelity

the underlying separation. In practice, it is desirable and in many cases required to sample such chromatographs at a rate far higher than the typical elution time of a peak. Typically, sampling

the chromatograph at a rate 10 or more times faster than the eluant peak width is acceptable to

accurately describe the peak and its fine structure.

The present invention mitigates this time penalty by allowing the simultaneous introduction of

more than one liquid separation to the MS system. Furthermore, because of the ion storage

feature of the invention, no loss of chromatographic fidelity is incurred, even for chromatograms

exhibiting narrow peak widths. This is especially advantageous since high throughput screening

applications favor separation systems which can operate at high linear velocities and/or with high

numbers of theoretical plates, both of which lead to narrow peaks which could otherwise elute

undetected in the absence of ion storage.

One previously described method switches between multiple liquid streams flowing to a single

spray assembly for ionization, consecutively valving to waste all but one of the streams at any

instant in time (Coffey ref). Because of valve mechanics, this sample selection process is limited in the highest frequency it can operate at while preserving analytically important reproducibility,

and moreover creates temporal gaps in the mass chromatograms of the off-cycle streams which

may contain analytically important information. Another previously described method advocates

the use of multiple ionization assemblies each delivering its distinct sample stream in sequence

to a single vacuum orifice. Gating of the individual ionization assemblies may occur by

modulation of a combination of: (1) electric potential to the spray probe; (2) pneumatic gas

pressure and flow to the spray probe; (3) gas pressure, flow and orientation to the countercurrent

bath gas; and/or alignment and positioning of the individual spray probes with respect to the

vacuum orifice.

Making use of the high sampling rate of the time-of-flight electronics and the storage capabilities

of two dimensional multipole ion traps. In this manner, more than one liquid handling system can

continuously infuse its effluent or other the simultaneous introduction of multiple sample streams

to multiple atmospheric pressure ionization spray assemblies.

Brief Summary of the Invention

An object of the present invention is to use a single mass spectrometer to analyze ions from

multiple atmospheric pressure ion sources while satisfying the following two constraints: (1) ion

beams from each of the discrete and separate ion sources are not mixed with one another, thereby

retaining the true chemical profile of each of the analytical samples; and (2) essentially all ions

from each of the ion beams are used for mass spectrometric analysis in turn, regardless of the

number of separate ion beams.

A further object of the invention is to achieve substantially higher sample throughput on a single

mass spectrometer, without mixing the individual analytical samples and without gating various

samples in such a way that duty cycle and hence sensitivity might be compromised.

The means by which this improved sample throughput may be obtained is to employ parallel ion paths and ion storage within the ion optics leading into a single mass spectrometer. Parallelism is exploited by introducing multiple discrete samples through separate and distinct sampling ports, transmitting these ions to separate and distinct ion storage devices, and sequentially gating these separate and distinct ion populations into a single flight tube or other mass analysis device (cyclotron cell, ion trap, etc.) in turn. In this manner, only one set of mass analyzing hardware and electronics are needed to process multiple sample streams, and a user may arbitrarily start or stop experiments on any of the various sampling ports without regard for the experiments being conducted on other unrelated sampling ports. The signals recorded from each of the sample streams are written to different device channels or memory locations, to keep separate and distinct the data associated with each of the aforementioned streams. In this manner, the overall sample throughput which a single mass spectrometer can support will far exceed that of a

mass spectrometer coupled to a dedicated single ion source. Lastly, this multiplexing approach in no way compromises the analytical figures of merit which may be obtained for any given sample when compared to a mass spectrometer coupled to a dedicated single ion source.

This invention has several advantages over existing solutions for obtaining mass spectrometric

data from atmospheric pressure ionization sources coupled to liquid chromatographs. The

existing solutions can be characterized as one of the following: (A) dedicated, (Bl) liquid

multiplexed, or (B2) ion muliplexed at atmospheric pressure. The present invention constitutes a

new and a fourth type of multiplexing, namely (B3) ion multiplexed in vacuo. The properties of these four types of sample introduction systems are shown in Table 1. For mass spectrometers

which mass separate ions in a batch-wise fashion (such as TOF, FT-ICR and ion traps) discrete

samples created in parallel must be submitted serially, lest mixing of multiple unrelated samples

occurs, A timing device is therefore required to multiplex these samples in an orderly and

analytically useful fashion.

The timing of multiple analytical samples originating from separate liquid sample streams,

ionized by an atmospheric pressure ionization process and delivered into a vacuum system for

mass spectrometric analysis may occur in one of three regions. These regions include (a) in the

liquid streams themselves, prior to nebulization and ionization, (b) the atmospheric pressure

region of an ionization source or (c) in vacuum. For all of these multiplexing strategies one may

attain higher throughput than would otherwise be possible using a strictly serial methodology (of

one sample introduced to one ion source coupled to one mass spectrometer). However, unlike the

other strategies, gating in vacuum affords several features which are analytically useful and

unique. The first of these features is the ability to accumulate off-cycle sample (ions) in an ion

storage device, thereby preserving the analytical sensitivity of the system for the compound at

hand. The second of these features is very short switching time. For circumstances in which one

wishes to switch the output of ions from one RF ion guide from "OFF" to "ON" or vice versa, this switch is completed in tens of nanoseconds, a timescale so fast that one may invoke multiple

ion guides to switch multiple times every second without significant loss of duty cycle. This

second feature is critically important for the invention to service multiple sample streams which

may be highly dynamic in nature, such as high speed chromatography exhibiting characteristic

peak widths of a second or less in duration. Exacerbating the sampling demand, one may wish to

mass spectrometrically analyze several such liquid chromatographs simultaneously, each requiring the acquisition of multiple mass spectra every second. If these chromatographs are all

high resolution (i.e. have temporally narrow peaks) and are rapid in nature (multiple peaks

occurring in a short period of time) then it is essential that each of these chromatographs be

frequently sampled by the mass spectrometer to achieve high chromatographic fidelity, preferably

at a rate 5-10 times greater than the typical chromatograph peak width. Unlike other gating

strategies shown in Table 1 which must overcome significant time lags while switching between

sample streams to accommodate the working fluid (air or liquid solvent), invoking an ion gate in

vacuum is essentially instantaneous. This therefore allows one to switch more frequently, which

in turn allows one to monitor a larger number of discrete sample streams with adequate fidelity.

In contrast, switching between liquid samples using a valve must be done at frequencies of approximately 1 Hz or less in order to avoid excessive carry-over from stream to stream. Also in

contrast to the present invention, switching between continuously operating ion sources at

atmospheric pressure will require one to several seconds to accomplish, since these partly

gaseous, partly liquid sprays needs this time interval to stabilize (i.e. begin to deliver analyte ions

to a vacuum orifice) in response to either electrical and/or mechanical shutters.

Compared to dedicated mass spectrometer systems (A) which employ one ion source interfaced

to one mass spectrometer, the subject invention (B3) and other described muliplexing strategies

(Bl , B2) deliver a total sample throughput which is N times greater, where N is the number of

discrete sample streams being sampled for mass spectrometric analysis. But because methods Bl

and B2 offer no means of storing "off-cycle" sample streams until the mass analysis device has

completed its previous analysis, these strategies necessarily lead to loses in duty cycle and hence

analytical sensitivity. For applications requiring high sensitivity, especially those requiring the

detection and characterization of very trace substances such as peptides or metabolites, such

sensitivity losses may be unacceptable. In contrast the present invention risks no loss of off-cycle

information. As an example of multiplexing using strategy Bl, Biotage (Ref) has demonstrated a

commercial instrument which sequentially samples N chromatography streams and delivers the

time-sliced output to a mass spectrometer. The disadvantage of this solution is that any

chromatographic effluent of importance which arrives at the sampling valve "off-cycle" is immediately discarded as waste, thereby degrading the analytical sensitivity of the instrument in

direct proportion to the number of streams sampled, potentially missing important chemical data

altogether. In addition, the speed with which the Biotage system can switch between sample

streams (1-3 Hz) precludes its use for fast chromatographic applications with peak widths of

several seconds or less. Micromass, Inc. has commercialized a multiplexing version of its TOF-

MS product, which uses strategy B2 to switch between different ion sources at atmospheric

pressure. Like the Biotage solution, it too suffers from duty cycle loss, with sensitivity degrading

in direct proportion to the number of streams sampled. Also like the Biotage solution, the

characteristic time to switch between sample streams is limited by the working fluid, in this case

air or nitrogen, to several Hz or less. While ions are continuously generated by several different

spray assemblies, each assembly when selected for MS sampling must be given adequate time for

its spray plume to react to the electrostatics at atmospheric pressure and deliver an adequate

number of analyte ions into vacuum.

In sharp contrast, the present invention may be switched at least as frequently as 1000 Hz, which

is suitably fast to detect many dynamic sample streams with adequate chromatographic fidelity.

This switching capability makes it ideally suited for a growing number of chromatographic

protocols designed for high throughput and high resolution, especially "lab-on-a-chip" based

designs.

I

Brief Description of the Drawings

FIG. X is a tabular comparison of typical sample throughput rates for (1) flow injection analysis

(FIA-MS), (2) LC-MS, (3) fast LC-MS using accelerated separation methods, and (4) parallel

LC-MS using the present invention.

FIG. 1 is a schematic representation of a plural source mass spectrometer.

FIG. 2 is a schematic representation of a preferred embodiment of the invention, in which

multiple atmospheric pressure ionization sources are coupled to a single time-of-flight mass spectrometer. Transmission and storage of ions from each sample stream is accomplished using

multiple two dimensional ion traps which serve to gate the ions into the flight tube in a serial fashion in order to generate unambiguous mass spectra.

FIG. 3 is a timing diagram of the potentials applied to the individual RF multipole ion guide exit

lenses to achieve sequential and non-overlapping injection of their individual ion packets.

FIG. 4 is a schematic representation of an RF hexapole ion guide array for the purposes of

minimizing the aggregate ion beam width admitted into a time-of-flight extraction region.

FIG. 5 shows the cumulative ion storage capacity of a single two dimensional ion trap monitoring

the molecular ion signal observed (Leucine Enkephalin, MW 553.7) versus the total storage

duration.

FIG. 6 is a schematic representation of a worst-case mass spectrometric requirement for a parallel

ion storage time-of-flight mass spectrometer, depicting four simultaneously arriving effluent

peaks of 1 s duration.

FIG. 7 is a listing of relative start times required to achieve simultaneous detection of four

chromatograms with characteristic peak widths of 1 second. A total of 10 integrated mass spectra

per second are obtained for each chromatogram, for a total of 40 mass spectra per second.

FIG. 8 is a comparison of methods to achieve high sample throughput on a single mass

spectrometer for 1 to N discrete sample streams.

PAGE INTENTIONALLY LEFT BLANK

Detailed Description of the Invention

Figure 1 shows an arrangement for conducting mass spectrometric analysis on multiple ion

sources using a preferred embodiment of the invention. In this case a number of samples are

simultaneously injected onto the same number of liquid chromatography columns for separation

of their individual constituents. Each of these sample streams elute and are transferred in line to

its own atmospheric pressure ionization source. These API ion sources are oriented to allow high transfer efficiency of ions between each ionization probe and its respective vacuum orifice.

Likewise, each of these sprayer-orifice pairs is set a suitable distance apart to prevent the

migration of ions from, for example, probe A towards orifice B, which would lead to erroneous

mass spectral data in mass spectrum B by falsely indicating the presence of a compound from

chromatograph A. Each of the API devices converts its respective sample stream into charged

particles which are suitable for transfer into a vacuum system containing a time-of- flight mass spectrometer. Transfer of each ion packet into this common vacuum system is accomplished by

focussing these ion packets through a vacuum orifice and towards an ion optical system

containing at least one two dimensional ion trap for storage and transmission purposes. Because

different ion packets from different samples are prevented from co-mingling within the injector

portion of the instrument, cross contamination of the various samples is therefore avoided.

While a chromatograph is running, ions from each chromatograph are continuously admitted into

the vacuum system, being focussed into their respective two dimensional ion guides. At no point

in time is the influx of charged particles to any two dimensional ion trap turned off, since this

would represent a loss in chemical information. Outflux from the ion traps is allowed serially, the

frequency and duration of which are dictated by different factors. This multiplexing of different

ion packets from different chromatographs into a single TOF mass spectrometer allows one to

simultaneously analyze a number of different samples on a single data acquisition and data

analysis package. This centralized processing allows a single operator to inspect large numbers of

records without relying upon a network to connect multiple instruments.

A depiction of the specific ion optical elements to construct a preferred embodiment is shown in

Figure 2. First, liquid samples are delivered to atmospheric pressure ionization probes from

liquid chromatography or other processes. These samples are converted into separate and distinct

ion clouds by ionization probes, which nebulize and ionize the streams in preparation for their

admission into vacuum. The ions created from these streams are admitted into a common

vacuum manifold through vacuum orifices. A separate and distinct vacuum orifice is dedicated to

each of the liquid sample streams to afford 100% duty cycle and no chemical cross-talk between

the respective streams. As the ions enter Stage 1 of the vacuum system, they are swept forward

by a combination of gas dynamic and electrostatic forces through another vacuum orifice and

into Vacuum Stage 2. As the ions enter Stage 2, they immediately enter a two dimensional

multipole ion guide, which serves to capture and collisionally cool the ions due to the high

pressure at the trap's leading edge. These ions propogate forward due to the high influx of neutral gas molecules at the trap's upstream exit, contained radially all the while by the application of an

appropriate RF potential on the poles of the device. Since the multipole is a multi- vacuum stage

device, after traversing a portion of the ion trap ions are again transmitted though another

vacuum orifice into Stage 3. This differential pumping across the length of the two dimensional

ion trap affords a very large pressure differential across the trap's length. In practice, this allows

one to use the high pressure of the ion trap's upstream section for effective capture and

collimation of ions with a broad translational energy distribution and the low pressure of the ion

trap's downstream section for containment, energy definition, storage and timed injection into

mass spectrometers. Ions which accrue in each of the two dimensional ion traps are held within

the trap and prevented from exiting the low pressure side by the application of a DC potential on

an exit lens. This exit lens may be held "high" to trap ions or "low" to allow ions to exit the trap

as needed. When this exit lens is dropped from its "high" to its "low" state, ions which have

accumulated within the two dimensional ion trap are caused to emit. One or more ion optical

lenses may be used between the exit lens and a mass spectrometer to best focus and transmit the

ion packets forward into a mass spectrometer. When coupled to a time-of-flight mass

spectrometer which employs orthogonal acceleration, it is particularly advantageous to deliver a

packet of ions to the extraction region of the TOF-MS which is monoenergetic, narrow in its

spatial dimension (in the x-y plane) and with little or no velocity component in the axis of the

TOF flight tube. As each of the two dimensional ion traps are pulsed out in turn, an appropriate

time interval is allowed for the ion packets to arrive at the middle of the extraction region,

whereupon a pulse-out lens is then pulsed electrostatically to a suitably high voltage to cause

orthogonal acceleration into a flight tube.

The timing associated with injecting multiple samples into a single flight tube while incurring no

loss in duty cycle for any given sample is strictly defined by the following parameters:

Number of chromatograms N arriving simultaneously;

Time interval t _tra p available for trapping;

Time interval tmg _ht necessary for an ion packet to transit the flight tube; and

Time interval t _em it allowed for an ion packet to be pulsed out of the two dimensional ion

trap;

In practice, one will limit the time interval t _tr ap to prevent overfilling of the ion trap with charged particles, since this has been shown to cause catastrophic fragmentation of the ions and loss of

analytical information. In Figure 5 evidence of this catastrophic fragmentation is evident. The

molecule leucine-enkephalin is used to generate an electrospray ion beam, the ions within which

are comprised primarily of leucine-enkephalin molecules and an attached proton. If a trap is first

emptied, and systematically filled for different periods of time by controlling the ion source's and

ion trap's electrostatic potentials, one may record the relative charge stored by inspecting the signal associated with this molecule. For up to several seconds storage duration, the signal

associated with this ion builds in intensity, until the charge density within the ion trap exceeds the critical density. Beyond this point in time, the ion of interest falls precipitously in amplitude,

signalling a rapid depletion due to space charge repulsion and ion ejection from the ion trap.

Under most analytical conditions, one may trap ions from external atmospheric pressure

ionization sources in two dimensional ion traps without suffering space charge effects and the

aforementioned fragmentation at rates as low as 2000 Hz for traps with internal volumes of

approximately 2 cm ³ (70 mm length and 3 mm inner diameter).

In practice, one will also design the TOF-MS to separate ions over length scales and time frames

which best suit'the analytical figures of merit (mass accuracy, mass resolving power, and

sensitivity). Given standard fabrication processes as well as electronics specifications, this

generally entails a mass separation system which requires tens of microseconds or more to record

an entire mass spectrum. For this reason, the choice of 100 microseconds as a benchmark time

interval for tfπ _g h _t is reasonable for the preferred embodiment.

A depiction of the overall timing for the injection of four separate chromatograms into a single

TOF-MS is shown in Figure 3. It is assumed in the schematic that all ions will be recorded within a 100 microsecond window. This implies that all m/z values are low enough and the flight tube

short enough that no ions will need more than 100 microseconds to arrive at the ion detector. For

most biological applications with commercially viable flight tube lengths and potentials, this

assumption is reasonable. Access to the TOF flight tube is divided equally between the various

chromatograms, although one could preferentially sample certain liquid streams at different

frequencies by altering the pulse-out instruction sequence. Each ion trap and its associated ion

packet is granted access to the flight tube in 100 microsecond blocks. In theory, any number of

sample streams could be accommodated with this method. In practice, for N»4 experimental

conditions would have to be controlled in order to avoid losses due to overfilling. This could be accommodated by injecting fewer charges per unit time, using a larger ion trap volume with

greater charge storage capacity, and/or selectively emptying the two dimensional ion trap while

filling through the use of a low mass, high mass or bandpass filter.

Immediately preceding the time block t _f |j _ght for any sample stream, the ion trap must be opened

for a predetermined period of time t _cm i _t (several microseconds or more) in order to allow an ion

packet to emit towards the TOF-MS. Emission is immediately followed by a time interval tr _a n _s i _t

which allows the ion packet suitable time to enter the TOF-MS extraction region. In practice this

time interval is determined by the ion packet's electrostatic energy and by the physical distance

L _gap from the trap exit to the centerline of the TOF extraction region. For instance, in the case

where Ei _0n = 10 eV and L _gap = 10 cm, tran _s i _t will be approximately 40 microseconds for low molecular weight species under 1000 amu. While ions from the first sample stream are being

separated in the flight tube, the same timing diagram is executed against the second sample

stream, cueing up and delivering an independent and unrelated ion packet as soon as the 100

microsecond flight window expires. For N=4 and the aforementioned assumptions, each of the

four different sample streams may be sampled with zero loss in duty cycle 2,632 times every

second, allowing even rapid time-varying processes to be monitored despite the extreme

multiplexing.

Performance of the orthogonal extraction TOF-MS is strongly effected by the properties of the

incoming ion beam. In order to interface multiple ion beams with multiple points of origination,

two conditions must necessarily be met if the flight tube optics and their voltages are to function

for all N beams. First, the ion packets must be introduced to the extraction region parallel to one

another and varying only in position along the y plane. In this manner all ions will develop the

same electrostatic energies upon acceleration, neglecting field aberrations and other higher order effects. Secondly, the line length L determined by the distance from the centerline of the two

most extreme ion traps should be kept to a minimum. This permits the extraction region to

receive the different ion packets without becoming unduly large or being compromised by

fringing fields which form when pulsed potentials are applied. In this manner, the required

dimension of the extraction region can be held to a reasonable value for typical laboratory

operations, and the different mass spectra resulting from mass separation of each of the ion traps'

ions will be more closely related. In order to minimize the required height of the extraction

region of the TOF-MS (in the y plane) it is advantageous to store ions in two dimensional ion

guides which are closely spaced in the y direction. As shown in Figure 4, a multipole array may

be constructed which takes advantage of shared poles to best compress the required line length L.

For instance, for four hexapole ion traps with individual poles of 1.0 mm diameter and hexapole

diameters of 3.5 mm, one can construct a four ion trap array with a line length L of 9.194 mm.

This value compares favorably to constructing four separate hexapoles with 2mm spacing

between each, which would require over 16 mm of line length and which would further challenge

construction of a compact and efficient extraction region.

To illustrate the utility of the invention, a hypothetical experiment requiring the separation and

detection of four separate liquid streams is shown in Figure 6. As a worst-case scenario, it is

envisioned that one chromatography peak from each of four separate sample streams will arrive

simultaneously, and that each peak will only be 1 second in duration. In order to mass

spectrometrically detect these peaks, and to do so in a manner that faithfully reproduces the time-

varying nature of the samples on a sub-second basis, it is essential that each of these peaks be

repetitively sampled over the course of the 1 second peak elution. As a matter of preferred

practice it is desirable to oversample such LC peaks, acquiring mass spectral data at a rate 5-10 times as fast as the narrowest characteristic peak width. In this example, 10 spectra per second

are desired for each of the four sample streams, requiring the TOF-MS to acquire forty integrated

mass spectra.

The integration of the mass spectra associated with each of the sample streams may be treated

asynchronously with respect to one another, provided each sample stream's raw data are

integrated frequently enough to faithfully reproduce its underlying cliromatogram. Consider the

following example. Four sample streams must be ionized and mass spectrometrically analyzed by

the present invention. However, these sample streams are not stalled at the same time, require

different time intervals to complete their respective separations, and have different characteristic

peak widths. The properties of these four hypothetical chromatograms are shown in Figure 7,

along with relevant pulse and integrated mass spectral rates. This example serves to illustrate that

there may be variation between chromatograms in each of the following:

• Start time

• Duration

• Characteristic peak width, and therefore required MS integration rate

Given these variations, the present invention may be called upon to render differing numbers of

integrated mass spectra every second for each of the sample streams being analyzed. For

instance, in Figure 7, Chromatogram 2 represents a fast, high resolution LC separation, requiring

10 MS spectra per second. Chromatogram 4, in contrast, is a far longer separation with

characteristic peaks that are 10 tikmes as wide. Comparing these two extremes highlights several important facets of the invention. First, each stream, regardless of its characteristic LC time

constants, may be sampled at a fixed and high rate which is determined by the ion capacity of the

two dimensional ion trap, in this case sampled at 2500 pulses per second. Second, varying

number of pulses are added together to comprise an integrated mass spectrum, based entirely

upon the characteristic peak widths expected from the LC chromatogram. In the case of

Chromatogram 2, 250 pulses are added to complete an integrated mass spectrum, yielding the

required 10 spectra per second. For Chromatogram 4, 2500 pulses are added together to yield the

required 1 spectra per second. Both of these integration needs may be serviced simultaneously

with the present invention.

In order to satisfy both this integrated mass spectral rate as well as the pulse frequency rate

described above and shown in Figure 3, it is necessary to add the signals from a number of

consecutive pulses associated with a given sample stream. For example, referring to Figure 3,

sample stream 1 is introduced to the mass spectrometer during Pulse 1, Pulse 5, Pulse 9, and so

forth. Every fourth pulse is added together until the time interval representing the mass spectral rate (in this case 0.1 sec, or 10 spectra per second) has elapsed.

Although the invention has been described in terms of the specific preferred embodiments, it will

be obvious and understood to one of ordinary skill in the art that various modifications and substitutions are contemplated by the invention disclosed herein and that all such modifications

and substitutions are included within the scope of the invention as defined in the appended

claims.