Image courtesy of Bruker Daltonik GmbH
Interfaces for LC–MS
Filip Lemière, Dept of Chemistry, University of Antwerp, Belgium.
Introduction
When searching for a suitable techniquefor the analysis of mixtures, often
containing unknowns and/or analytes inlow concentration, the combination ofliquid chromatography (LC, separation)with mass spectrometry (MS, sensitivityand structural information) appears to bean obvious choice. However, LC–MS is an“odd couple” (1). First, there is the natureof the analytes. LC is preferred over gaschromatography (GC) because of thehigher polarity and lower volatility of thesamples. One of the prerequisites for massspectrometric analysis is the formation ofvolatilized ions. Second, and a harderproblem to solve, is the necessary
elimination of the mobile phase. A flow-rate of 1 mL/min water is converted in1244 mL/min vapour at atmosphericpressure, which is far too much to behandled by the standard MS vacuum
systems. Third, salts and other additives ofthe mobile phase are often involatile (e.g.,phosphates, NaCl etc.).
Several coupling methods have beendeveloped to solve these problems.
Methods such as particle beam (PB) andmoving belt/wire (2) rely on removal of thesolvent prior to entering the MS. Couplingwith continuous flow fast atom
bombardment (cf-FAB) (3, 4) or directliquid introduction (DLI) (5) reduces theflow entering the MS using some kind ofsplitting device. A serious drawback of thelast approach is the reduction in sensitivitycaused by the split factor. The entire flow-rate, about 1 mL/min, used with a classic4.6 mm i.d. column is only tolerated bytechniques such as thermospray (TSP) andatmospheric pressure chemical ionization(APCI). Electrospray (ES) has a workingrange from nL/min to 0.2 mL/min, whichcan be extended to mL/min flows. Themost attractive difference with other
interfacing techniques is that the ES signalis dependent on the concentration ratherthan the amount injected. Thisphenomenon makes ES virtually
independent of flow-rate allowing the useof splitters to reduce the amount of eluentintroduced into the MS source and
opening the possibility of miniaturizing theLC technique to capillary (300 µm i.d., 4 µL/min flow-rate) and nano (75 µm i.d.,200–300 nL/min flow-rate) dimensions.The same features also allow coupling ofcapillary electrophoresis, which has a flow-rate in the nL/min range, with MS. A stranger in this series with most of itsapplications in ‘inorganic’ analysis is
inductively coupled plasma (ICP)–MS. Forthis reason a special section will bedevoted to this technique (See “The
Inductively Coupled Plasma Ion Source forLC–MS”). Most LC–MS interfaces andapplications are focused on the analysis ofintact molecules, often organic materials.However, biomolecule samples areanalysed using this technique (6).
Apart from being an inlet system for theMS, an LC–MS interface is also the couplingof a detector (MS) to a chromatograph.The choice of LC–MS interface strongly
influences the characteristics of the MS asa detector for LC. Therefore, we shouldkeep in mind what characteristics are idealfor an LC detector (Table 1).
The First LC–MS Interfaces
The first experiments to couple LC to MSdate back to the late 1960s. Thoughfascinating at the time of their
development, the earliest LC–MS interfacesare now almost obsolete. The introductionof techniques that allow delivery ofthermolabile biomolecules into the MSshow an exponential increase in the
number of publications employing LC–MS.However, in order to give a somewhat
wider overview older LC–MS interfaces willbriefly be described.
Direct Liquid Introduction
The first attempts to introduce a liquid intoan MS using the classic electron impactionization (EI)/chemical ionization (CI)source were based on the simple principlethat by minimizing the amount of liquid,the vacuum system would remove the
solvent leaving the analyte in the gas phasefor ionization. By using larger pumpsystems and differential pumping,
through two differentially pumped vacuumlocks. A heater in the ion source evaporatesthe sample from the belt allowing MSanalysis (Figure 2). Most moving-beltanalyses deal with volatile analytes usingCI/EI; however, less volatile molecules suchas nucleosides and nucleotides are analysedusing this system (9).
Thermospray
The TSP interface was developed by
M. Vestal and co–workers (10–13). A majoradvantage of TSP over other LC–MSinterfaces is its ability to handle the highflow-rates delivered by LC (up to 2 mL/min).As the name thermospray implies, heatingthe liquid flow leaving an LC systemcreates a spray of superheated mistcontaining small liquid droplets. Severaltechniques are developed to heat and
vaporize the effluent (12, 13); however, themost successful method involves directingthe liquid flow through an electrically
heated capillary (11), which can be directlyintroduced into the MS ion source. Thedroplets are further vaporized as theycollide against the walls of the heated ionsource. This ion source is equipped with amechanical pump line opposite to thespray in order to evacuate the excesssolvent vapour (Figure 3). The rapidheating and protective effects of thesolvent allow the analysis of non-volatilesamples without pyrolysis. The analyte ionsare sampled into the MS through asampling cone, if necessary aided by anapplied electric field (repeller oraccelerating electrode).
Ionization of the analytes in TSP occurs by means of several processeswherein two classes of ionization type can be distinguished: one without anexternal ionization source, so–called ‘real thermospray’ and one with anexternal ionization.
The real thermospray uses a volatile
Figure 1: Scheme of the DLI interface. 1 ϭconnection to LC column, 2 ϭdiaphragm 5 µmopening to MS, 3 ϭneedle valve, 4 ϭcooling region, 5 ϭto UV detector or waste.
Figure 2: Schematic showing the principal components of a moving-belt interface.
Figure 3: Thermospray interface. (a) configuration for ‘real-TSP-ionization’ (filament off) orexternal ionization (filament on). (b) configuration with discharge electrode for external ionization and repeller electrode. (Adapted from reference 14.)
Figure 4: Schematic showing the principal components of a particle beam or MAGIC interface.
Figure 5: Overview of a differentially pumped API source coupled to a mass spectrometer.
using a differentially pumped momentumseparator. The PB interface allows flow-rates from 0.1–0.5 mL/min. Mostanalytes that are amenable to
PB LC–MS can be analysed using GC–MSas well.
Atmospheric Pressure Ionization
The earliest LC–MS techniques (DLI, TSP, moving belt, PB), although
commercialized, were often difficult touse, had limited sensitivity and were notrobust; however, they were very useful forspecific applications. The overwhelmingincrease in LC–MS applications is mainlythe result of the sensitivity and
ruggedness of atmospheric pressure
ionization (API) LC–MS techniques. API isa general name for all ionization
techniques in which the ions are formedat atmospheric pressure. Though verypopular today, ionization processes atatmospheric pressure (flames, dischargesetc.) have been studied using massspectrometers for many years (18–20).In modern LC–MS applications we findtwo major techniques: ES and APCI.Electrospray can be subdivided into
techniques such as pneumatic-assisted ES,ES, multiple sprayer ES etc., that differmainly in the formation of a spray fromthe LC flow. However, all ES variants relyon the same mechanism(s) to form ionsfrom the droplets at atmospheric pressure.The ions formed at atmospheric pressureare transported from the source to thevacuum of the analyser through one ormore differentially pumped stages
separated by skimmers (Figure 5). The ionsare focused and guided through the
skimmer openings into the MS by applyingappropriate electric fields. Various sourcedesigns, ion optics configurations,
pumping systems and other experimentalparameters (21) are used, but the basicfeatures can be found in all instruments.Where ES has its optimal performance atlow flow-rates (nL/min range) APCI
operates happily using mL/min flow-rates.ES and APCI perform differently underdifferent chromatographic modes.
The advantages of API were summarizedby Voyksner (22) in four points:
1. “API approaches can handle volumesof liquid typically used in LC”
2. “API is suitable for the analysis of non-volatile, polar and thermally unstablecompounds typically analysed by LC”3. “API-MS systems are sensitive, offeringcomparable or better detection limitsthan achieved by GC–MS”
4. “API systems are very rugged and relatively easy to use.”
Figure 9: ES spectrum of the -chain of bovine haemoglobin.
biomolecules such as peptides, proteins,oligonucleotides etc., with molecular
weights of tens and hundred of thousandsamu. Unlike most ionization techniquesthat yield monocharged ions, ES of thesebiopolymers yields multiply charged ions.Because MS measures a mass-over-chargeration, these large masses can be
measured using standard MS equipmentwith a limited mass range (quadrupole4000 amu, TOF 20000 amu) (Figure 9).Although these multiple charge spectralook somewhat odd at first, molecularweight information can be obtained usingsome simple mathematics.
Unless special additives are used, mostmolecules including peptides and proteinsare charged by (de)protonation. Thisimplies that successive peaks (M 1/Z 1and M 2/Z 2) in the envelope of multiply chargedions differ by 1 charge unit and the massof 1 proton. Therefore, we know that forthese peaks the following holds:
M 1Z 1 ϭ M ϩ 1.0079*Z 1[1]M 2Z 2 ϭ M ϩ 1.0079*Z 2
[2]
where M is the mass of the unchargedmolecule. For the electric charges we knowthat
Z 2 ϭ Z1 ϩ 1
[3]
This allows us to calculate the chargestate of the ion at M 1/Z 1
Z 1 ϭ (M2 Ϫ 1.0079)/(M 2 Ϫ M 1)
[4]
Once we know the charge status themass of the multiply charged ion can becalculated and together with the chargestate (equals number of protons) themolecular weight of the analyte can beestablished.
Another highly uncommon characteristicof ES is its ‘softness’; that is, very labilestructures can be carried as ions into thegas phase without disrupting their
structures. ES can be used to study proteinfolding status, non-covalent bonding, DNAduplexes etc. (38). For the same reason ESspectra contain little or no structuralinformation because of the absence offragmentation. Molecular weightinformation is obtained in the first
instance. If more structure information isneeded, for example, sequence
information of peptides, fragmentationmust be induced. This is most convenientlydone by applying tandem MS. Startingfrom doubly charged peptides product ion
spectra can be obtained from which theamino acid sequence can be deduced(Figure 10).
Atmospheric Pressure ChemicalIonization
Using APCI the liquid flow from the LC issprayed and rapidly evaporated by a coaxialnitrogen stream and heating the nebulizerto high temperature (350–500 °C).
Although these temperatures may degradethe analytes, the high flow-rates and
coaxial N2-flow prevent breakdown of themolecules. Ions already present in solutioncan be carried into the gas phase,
however, additional ionization is achievedusing a corona discharge (3–6 kV) in thespray. This discharge can ionize not onlythe analyte molecules, but also the solventmolecules. These solvent ions can reactwith the analytes in the gas phase in thesame way samples are ionized in a CIsource by the reagent gas. In positive ionmode protonated molecules and adductsare formed; in negative ion mode ions areformed by deprotonation, combinationswith anions or electron-capture. Theionization efficiency is better comparedwith CI because it occurs at high pressure(atmospheric) so the collision frequency ishigh compared with the process in astandard CI source, and ionization
efficiency is higher. The different ionizationmodes can be exploited to further enhancethe sensitivity of the technique. One of themost sensitive GC–MS techniques iselectron capture negative ion CI. Theanalytes are derivatized with an electron-capturing group, generally a
pentafluorobenzyl derivative. Becauseelectrons are produced in the coronadischarge used in APCI it was anticipatedthat these derivatives would give goodsensitivity using APCI as well. For a series of biomolecules and drugs it isshown that one can obtain detection limits in the attomole (femtogram) range (41).
Unlike in ES, the solvent-evaporation andion-formation processes are separated inAPCI. This allows the use of solvents thatare unfavourable for ion formation. Theselow-polarity solvents are commonly used innormal-phase chromatography with lowpolarity samples that can generally beevaporated for APCI ionization. Anothermajor difference between APCI and ES canbe found in the LC flow-rates that areused. APCI is a technique with optimalperformance at high flow-rates (1 mL/minand higher). Lower flow-rates can be used;however, when flow-rates are too low thestability of the corona discharge may
Figure 10: ES product ion spectrum of a tryptic digest peptide Glu-Fibrino peptide B showingthe sequence ions.
Figure 11: Basic elements of a cf-FAB probe.
Future Developments
A very clear goal since the beginnings ofLC–MS and still an important trend in
newly developed instruments is robustness.Both separation science and mass
spectrometry are very specialized researchdomains and often scientists are focusedon only one of them. Thus, when applyingthe hyphenated LC–MS techniques theother half “just has to work”. The
chromatographer wants MS to work as areliable detector that can be hooked up toa column (no matter what flow-rate orkind of separation), whilst the massspectrometrist needs a system forintroducing these liquid samples
(sometimes mixtures) containing polar,thermolabile, involatile biomolecules,pharmaceuticals, environmentalcontaminants, pesticides etc.
As sample availability and sensitivity arealways an issue in analytical (bio)chemistry,miniaturization will be a continuing trendin LC–MS. Miniaturization of the
separation techniques and consequentdevelopment of the appropriate interfaceswill proceed, including chip-basedtechnology for both separation andinterfacing to MS. Techniques that aretoday still considered as off-line
techniques, for example, MALDI, 2D gelelectrophoresis etc., will be modified andnew techniques developed to couple withexisting MS and LC–MS systems (49, 50).References
(1)P .J. Arpino, Trends Anal. Chem., 1, 154–158,
(1982).
(2)R.P .W. Scott et al., J. Chromatogr., 99,
395–405, (1974).
(3)Y . Ito et al.,J. Chromatogr., 346, 161–166, (1985).(4)R.M. Caprioli, T. Fan and J.S. Cotrell,
Anal.Chem. , 58(14), 2949–2953, (1986).(5)G.V . Tal’roze et al., J. Phys. Chem., 42,
1658–1664, (1968).
(6)C. Siethoff et al., J. Mass Spectrom., 34,
412–426, (1999).
(7)L.R. Snyder and J.J Kirkland, Introduction to
Modern Liquid Chromatography, (Wiley, NewYork, USA, 1974).
(8)E.L. Esmans et al.,Biomed.Mass Spectrom.,
12(5), 241–245, (1985).
(9)D.E. Games, Biomed. Mass Spectrom., 8,
454–462, (1981).
(10)M.L. Vestal, Science , 226, 275–281, (1984).(11)M.L. Vestal and G.J Fergusson, Anal. Chem.,
557, 2373–2378, (1985).
(12)C.R. Blakley, M.J. McAdams and M.L. Vestal,
J. Chromatogr., 158, 261–276, (1978).(13)C.R. Blakley, J.J. Carmody and M.L. Vestal,
Anal. Chem., 52, 1636–1641, (1980).
(14)P . Arpino, Mass Spec. Rev., 9, 631–669, (1990).(15)F . Lemière et al., J. Chromatogr., 647(2),
211–218 (1993).
(16)K. Vanhoutte et al., J. Mass Spectrometry, 30,
1453–1461, (1995).
(17)R.C. Willoughby and R.F. Browner, Anal. Chem.,
56, 2625–2631, (1984).
(18)P .F . Knewstubb and S.B. Sugden, Proc. Roy. Sec.
A , 255, 520, (1966).
(19)M.M. Shahin, J. Chem. Phys., 45, 2600, (1966).
(20)E.C. Horning et al., J. Chromatogr. Sci., 12,
725, (1974).
(21)A.P . Bruins, TRAC, 13(1), (1994).
(22)R.D. Voyksner, in “Electrospray Ionization Mass
Spectrometry: Fundamentals, Instrumentation,and Applications”, Ed. R.B. Cole, (Wiley-Interscience, New York, USA, 1997).
(23)K. Tang and A. Gomez, J. Aerosol Sci., 25(6),
1237–1249, (1994).
(24)S. Zafran, J.C. Beynon and P.W. Kidd, Prototype
One–Millipound Colloid Thruster System, AIAA11th Electric Propulsion Conf., New Orleans,Louisiana, USA, Paper No. 75–392, (1975).(25)C.J. McNeal, R.D. Mcfarlane and E.L. Thurston,
Anal. Chem., 51, 2036, (1979).
(26)N.J. Felici, Recent Developments and Future
Trends in Electrostatic Generation., DirectCurrent, 4, 3–12, (1959).
(27)J.J. Thomson, Rays of Positive Electricity and
their Application to Chemical Analysis, (Longmans Green, London, UK, 1913).(28)J. Zeleny, Phys. Rev., 10, 1–6, (1917).
(29)A.P . Bruins, T.R. Covey and J.D. Henion, Anal.
Chem. , 59, 2642–2646, (1987).
(30)G.A. Clegg and M. Dole, Biopolymers , 10,
2240–2249, (1971).
(31)B.A. Thomson and J.V. Iribarne, J. Chem. Phys.,
71, 4451–4463, (1979).
(32)J.V . Iribarne, P.J. Dziedzic and B.A. Thomson,
Int. J. Mass Spectrom. Ion Phys., 50, 331–347,(1983).
(33)J.V . Iribarne and B.A. Thomson, J. Chem. Phys.,
64, 2287, (1976).
(34)M. Dole, L.L. Mack and R.L. Hines, J. Chem.
Phys. , 49(5), (1968).
(35)M. Yamashita and J.B. Fenn, J. Phys. Chem., 88,
4451–4459, (1984).
(36)M. Yamashita and J.B. Fenn, J. Phys. Chem., 88,
4671–4675, (1984).
(37)M.L. Aleskandrov et al., Dokl. Akad. Nauk.
SSSR , 277, 379–383, (1984).
(38)‘Mass Spectrometry of Biological Materials’, Ed.
B.S. Larsen and C.N. McEwen, (Marcel DekkerInc., New York, USA, 1998.)
(39)K. Vanhoutte et al., Anal.Chem. , 69,
3161–3168, (1997).
(40)C.G. Edmonds et al., J. Chromatogr., 474,
21–37, (1989).
(41)G. Singh et al., Anal.Chem. , 72, 3007–3013,
(2000).
(42)S. Perez and D. Barcelo, Chromatographia ,
53(9–10), 475–480, (2001).
(43)B.A. Ingelse et al., J. Chromatogr. A, 918(1),
67–78, (2001).
(44)M. Barber et al., J. Chem. Soc. Chem.
Commun. , 325, (1981).
(45)Y. Ito et al., J.Chromatogr. , 346, 161, (1985).(46)S.M. Wolf et al., J. Am. Soc. Mass Spectrom., 3,
757–761, (1992).
(47)L.J. Deterding et al., Anal.Chem. , 64(21),
2636–2641, (1992).
(48)M.J.F . Suter and R.M. Caprioli, J. Am. Soc. Mass
Spectrom. , 3, 198–206, (1992).
(49)J. Preisler et al.,Anal. Chem., 72, 4785–4795,
(2000).
(50)B. Zhang et al., Anal. Chem., 72, 1015, (2000).
Dr Filip Lemière works in the NucleosideResearch and Mass Spectrometry Unit,
Department of Chemistry at the University ofAntwerp, Belgium. His research interestsinclude the miniaturization of LC–MSprocedures particularly for the analysis ofDNA adducts.