Electrospray Ionization: Source Parameters and How to Tune Them

Electrospray ionization (ESI) is the interface that turns a liquid chromatography effluent into gas-phase ions. Most explanations of it stop at the physics — the Taylor cone, Coulombic fission, the ion evaporation model. That is not what you need when a method that worked last month has lost half its signal, or when the tallest peak in your Q1 scan is a sodium adduct that will not fragment.

This page is a tuning procedure. It covers the six parameters you actually adjust on an ESI source, what each one physically does, how the five major vendors name and unit them differently, and — the part that matters at the bench — which parameter to change when the signal looks a particular way.

Scope note. This page is about the ion source: everything upstream of the first mass filter. What happens after ionization — precursor and product ion selection, collision energy, dwell and cycle time — is the analyser’s job and is covered separately in building and optimizing MRM transitions. If you are still deciding how LC and MS are coupled at all, start with LC-MS explained.

The six parameters that actually matter

Every atmospheric-pressure ESI source, regardless of manufacturer, exposes some version of the same six controls. The vocabulary differs; the physics does not.

1. Spray (capillary) voltage — the potential between the sprayer tip and the counter-electrode. This is what charges the liquid and forms the spray. It sets polarity and, to a point, spray stability.
2. Nebulising gas — a coaxial gas jet at the tip that shears the liquid into a fine aerosol. It does mechanical work on the droplets, not thermal work.
3. Heated drying / desolvation gas — a separate, heated gas stream that evaporates solvent from those droplets so ions can escape into the gas phase.
4. Source temperature(s) — the temperature applied to that drying gas, and on some designs a separate probe or block temperature. This is the main lever against high aqueous content and high flow rates.
5. In-source declustering voltage — a potential applied across the atmospheric-pressure-to-vacuum transition that strips residual solvent clusters off the ion. Push it too far and it starts fragmenting the analyte before the first mass filter sees it.
6. Counter-current / curtain gas — a gas flowing back out of the sampling orifice that keeps neutrals, solvent and particulates out of the ion optics.

Note what is not on this list: the sprayer position. On most sources probe position (depth and lateral offset relative to the orifice) is a physical adjustment rather than a software parameter, but it interacts strongly with flow rate and is frequently the real fix when voltage and gas tuning stop helping. Change it deliberately, one axis at a time, and record where you left it.

The vendor cross-walk: one parameter, five vocabularies

This is the single most common source of confusion when transferring a method between instruments, reading an application note written on another platform, or interpreting a published method section. The same physical parameter carries a different name and often a different unit on every vendor’s software.

SCIEX’s own documentation defines these functionally rather than numerically: GS1 “nebulizes a stream of liquid at the tip of electrospray capillary”; GS2 “transfers heat from the two side heaters… providing the heat to evaporate solvent and ions”; TEM is “the temperature of the heaters applied to GS2”; and CUR “is the gas flowing between the curtain plate and the orifice plate. It acts as a buffer zone and prevents contamination of the ion optics” (SCIEX knowledge base). The absence of recommended numbers in that document is not an omission — it is the point, and the next section explains why.

Three caveats that make this table safe to use

The declustering row is a functional mapping, not an identity. Cone voltage, fragmentor voltage and declustering potential are genuinely the same control: an offset that extracts ions from the atmospheric-pressure region, strips solvent clusters, and at higher settings induces in-source fragmentation. Practitioner literature treats them as interchangeable names for the same knob, with typical operating windows on the order of tens of volts. Thermo and Shimadzu split the function across more than one term, so the row is an approximation on those two platforms — check which parameter your software actually ramps during an autotune before assuming it maps one-to-one.

Some units convert and some do not. Waters expresses desolvation gas in L/h; Agilent expresses drying gas in L/min. Those are the same physical quantity: 1 L/min = 60 L/h, so a Waters method at 600 L/h and an Agilent method at 10 L/min are specifying the same gas flow. SCIEX and Thermo gas settings are instrument units on an arbitrary scale — they are not volumetric flows and cannot be converted to L/min at all. Any application note that appears to give you a direct SCIEX-to-Waters gas conversion is wrong.

Source geometry is a parameter you cannot see in the table. Orthogonal, Z-spray and heated-nebuliser geometries all present the spray to the orifice differently, and identical nominal settings therefore do not produce identical ion populations. Where in-source fragmentation spectra have been compared across instruments, differences in source geometry and residual gas composition are among the reasons the spectra do not reproduce cleanly between platforms.

Why there is no “correct” capillary voltage

The most common bad habit in ESI method development is copying a voltage out of a paper or an application note. Spray voltage is not a property of your analyte; it is a property of your source geometry and flow rate.

SCIEX states this explicitly across its own product line: the OptiFlow Pro source “works efficiently at lower ionization voltages (ISVs), typically in the range of 1500v”, while the Turbo V and IonDrive Turbo V sources are “typically more efficient at 5000v” (SCIEX Community). That is a factor of more than three between two sources from the same manufacturer, for the same chemistry, driven entirely by the sprayer emitter and its distance from the counter-electrode. A published voltage carried across from another instrument carries no information about your instrument.

The same document gives two directional rules that do transfer, because they are about physics rather than hardware:

• Source temperature “will need to be higher with a higher flow rate and/or higher aqueous composition” — water has a much higher heat of vaporisation than acetonitrile or methanol, so a 95% aqueous starting gradient demands far more thermal input than the same method’s organic-rich end.
• Curtain gas “should be set as high as possible without incurring significant signal-to-noise loss” — it is a contamination-control parameter first and a sensitivity parameter second, and the correct approach is to raise it until sensitivity starts to suffer, then back off.

Treat any number in any published method — including the ones in this page’s ranges below — as a starting point for a ramp, never as a setting to adopt.

The tuning procedure, in order

Order matters, because these parameters are not independent. Optimising temperature before you have fixed the flow rate means optimising against a moving target.

Step 1: fix the chromatography and the mobile phase first

Flow rate determines how much solvent the source has to evaporate per unit time, and mobile-phase composition determines how hard that is. Both must be settled before source tuning begins. Additive choice also constrains what ions you can form at all — see the adduct section below. If you are still selecting a column or flow regime, resolve that first via column selection and the HPLC/UHPLC flow-regime comparison.

Step 2: infuse post-column, not neat into the source

Direct infusion of a neat standard in a clean solvent optimises the source for conditions that will never occur again. Tee the infusion into the actual LC flow at the actual mobile-phase composition your analyte elutes in. This costs a few extra minutes and is the difference between parameters that survive contact with a real gradient and parameters that do not.

Step 3: ramp the spray voltage

Ramp across the instrument’s usable range and watch for the plateau, not the peak. A signal maximum sitting on a steep slope is a maximum you will lose the moment anything drifts. Prefer a slightly lower voltage on a flat region over a slightly higher one on a cliff edge. If you see the signal become spiky and unstable at the top of the ramp, you have crossed into corona discharge — back off, and check that the sprayer tip is not damaged or contaminated.

Step 4: set nebulising gas, then check probe position

Nebulising gas controls initial droplet size. Too little and the droplets are too large to desolvate; too much and you can blow the spray past the sampling orifice or destabilise it. Optimise it against a real flow rate, then — only if the plateau is poor — adjust probe position, one axis at a time, and re-ramp.

Step 5: set drying gas and temperature together against the aqueous end of your gradient

These two work as a pair: gas flow supplies the mass to carry heat and sweep vapour away, temperature supplies the energy. Optimise at the most aqueous composition at which any analyte in your method elutes, because that is the hardest case. A method tuned at 80% organic will underperform badly on early-eluting polar compounds.

Watch for thermal degradation on the way up: if signal rises, plateaus and then falls as temperature increases, the fall is usually your analyte decomposing in the source, not poor desolvation. Thermally labile compounds — glucuronides, N-oxides, some esters and carbamates — often want a lower temperature and more gas flow rather than more heat.

Step 6: ramp the declustering voltage last, and ramp it per compound

Cone voltage, fragmentor voltage and DP are the only source parameters that are genuinely compound-specific, which is why they belong at the end and why they must be optimised for each analyte rather than set once for the method. Ramp and watch the intact precursor:

• Rising, then plateauing: you are stripping solvent clusters. This is the region you want.
• Rising, then falling while lower-mass ions appear: you have entered in-source fragmentation. For a targeted method this is loss — you are destroying the precursor before the first mass filter can use it.
• Never rising much at all: the ion may not be the species you think it is. Check for adducts before blaming the voltage.

In-source fragmentation is not always a fault. It is deliberately exploited in some workflows, and on a single-stage instrument a ramped or alternating declustering voltage can be used to generate structural information the instrument cannot otherwise produce. But if you are building a quantitative method, treat every volt past the plateau as lost sensitivity.

Step 7: raise the curtain or sweep gas until sensitivity complains

Per the SCIEX guidance quoted above, set it as high as it will go without a meaningful signal-to-noise penalty. The payoff is not in today’s chromatogram; it is in how long the instrument runs before the ion optics need cleaning, and in how much less your response drifts across a long batch.

Step 8: re-verify under real gradient and real matrix

Run a genuine extracted sample, not a standard in mobile phase. Source parameters optimised on clean solution routinely fail in matrix, and the failure looks like a sensitivity problem when it is actually an ion suppression problem — a different fault with a different fix. That fix is upstream of the source: how much co-eluting matrix reaches the spray is decided during sample preparation, and in particular by how hard the extract was washed — a trade-off worked through in SPE cartridge selection and method optimization. No source parameter recovers signal that matrix is taking.

Symptom to parameter: what to change when the signal looks like this

Adduct control is a mobile-phase decision before it is a voltage decision

ESI does not produce one ion per analyte. It produces a distribution: protonated or deprotonated molecules, ammonium and sodium and potassium adducts, dimers, and in the case of larger molecules a whole charge-state envelope. Which of these dominates is set mainly by what is in the solution, and only secondarily by source voltages.

• [M+H]⁺ and [M−H]⁻ are what you generally want for targeted work: they form reproducibly and fragment predictably.
• [M+NH₄]⁺ is a legitimate and often deliberate choice for compounds that protonate poorly, obtained by adding an ammonium salt to the mobile phase. It is controllable, which is what distinguishes it from sodium.
• [M+Na]⁺ and [M+K]⁺ are almost always a problem. They come from trace contamination you are not controlling, so their abundance drifts between solvent lots and glassware batches — and they are held together tightly enough that fragmentation is inefficient.
• Multiply charged ions are the norm for peptides and proteins, and they are how ESI reaches high masses on analysers with limited m/z range. For tryptic peptides the doubly charged species is usually the productive precursor.

Because adduct formation is source- and solution-dependent, the species you select during infusion must be re-confirmed under your final LC conditions. This is one of the places where source tuning and analyser method-building genuinely overlap; the precursor-selection consequences are worked through in the MRM transition guide.

What source tuning cannot fix

Two failure modes look like poor source performance and are not:

Ion suppression. Co-eluting matrix components compete for charge and for space at the droplet surface. No combination of gas, voltage and temperature recovers a signal lost this way, because the analyte is not being ionised in the first place. The fixes are chromatographic (move the analyte away from the suppression region), preparative (clean the extract up), or compensatory (a co-eluting stable-isotope-labelled internal standard that suffers identical suppression). Diagnose it with a post-column infusion experiment before spending a day on the source.

Analyte polarity mismatch. ESI ionises species that are already ionic or readily ionisable in solution. A genuinely non-polar, non-ionisable analyte will give a poor response no matter how well the source is tuned. That is a source-selection problem, not a source-tuning problem.

When ESI is the wrong source

Atmospheric-pressure chemical ionisation (APCI) ionises in the gas phase via a corona discharge after the sample is fully vaporised, which makes it suit less polar, thermally stable, lower-molecular-weight compounds — and makes it noticeably more tolerant of ion suppression than ESI, because charge competition in the droplet is no longer the mechanism. Atmospheric-pressure photoionisation (APPI) extends further still into non-polar territory. If your compound is non-polar and thermally robust and ESI response is poor, changing source is a faster route than continuing to tune. The comparative case is set out in LC-MS explained, and the analyser-side trade-off between scanning and high-resolution instruments in QTOF vs triple quadrupole.

Record your source conditions as method metadata

Source parameters are the most under-reported part of a published LC-MS method section, and the reason methods so often fail to transfer. Because the vendor vocabulary differs, a bare number is not self-describing: “fragmentor 120” is meaningless to a Waters user unless the platform is stated.

A source record that is actually transferable states, at minimum:

• Instrument make, model and source hardware (source geometry is not implied by the instrument model — many instruments accept several source types).
• Polarity, and the spray voltage with its unit and sign.
• Every gas setting with its unit, and an explicit note where the value is in arbitrary instrument units rather than a volumetric flow.
• All temperatures, each identified by which vendor-specific term it refers to.
• The declustering voltage per compound, under its vendor name, not a single method-wide value.
• The LC flow rate and the mobile-phase composition at which optimisation was performed — without these, none of the above is interpretable.

This is the same reproducibility discipline that applies to any instrument-derived dataset. Where the data will be deposited, the repository’s own metadata schema may already require some of these fields — the Metabolomics Workbench mwTab format, for example, carries structured instrument and ionisation-mode metadata.

Frequently asked questions

What is electrospray ionization in mass spectrometry?

ESI is an atmospheric-pressure ionisation technique that applies a high voltage to a flowing liquid to produce a fine spray of charged droplets. As solvent evaporates from those droplets, the charge density rises until gas-phase ions are released and drawn into the mass spectrometer. Because ionisation happens from solution, ESI suits polar, ionisable and thermally labile molecules and readily produces multiply charged ions, which is what allows large biomolecules to be analysed on instruments with a limited m/z range.

What is the correct capillary voltage for ESI?

There is no instrument-independent answer, and any source that gives you one is misleading you. SCIEX’s own guidance puts its OptiFlow Pro source at around 1500 V and its Turbo V sources at around 5000 V for the same work — a threefold difference between two sources from one vendor. Ramp the voltage on your own instrument at your own flow rate and take the plateau, not the peak.

Are cone voltage, fragmentor voltage and declustering potential the same thing?

Functionally, yes on Waters, Agilent and SCIEX: all three name the offset that extracts ions from the atmospheric-pressure region and strips solvent clusters, and all three induce in-source fragmentation when pushed too high. The numeric values are not transferable between platforms, because source geometries differ. Thermo and Shimadzu distribute the same function across more than one parameter, so treat the mapping there as approximate.

Can I convert gas settings between instruments?

Only where both are volumetric. Waters L/h and Agilent L/min are the same quantity — 1 L/min = 60 L/h — so those convert directly. SCIEX and Thermo gas settings are arbitrary instrument units, not flows, and cannot be converted to or from a volumetric figure at all.

Why is my sodium adduct bigger than my [M+H]⁺?

Trace sodium is ubiquitous — leached from glass, present in solvents and buffers, carried in with the sample. Sodium adducts frequently dominate a Q1 scan. The fix is compositional rather than electrical: add a competing ammonium additive, move to plasticware, use high-purity solvents. Selecting the sodium adduct as your precursor is usually a mistake, because it fragments poorly and its abundance depends on a contaminant you are not controlling.

How do I tell in-source fragmentation from poor ionisation?

Ramp the declustering voltage while watching both the intact precursor and the suspected fragment. In-source fragmentation shows the precursor falling as a lower-mass ion rises, with the crossover moving as you change the voltage. Poor ionisation shows a precursor that is simply weak everywhere on the ramp and does not trade off against anything.

Should I optimise source parameters for each compound in a multi-analyte method?

The declustering voltage, yes — it is compound-specific and most software supports a per-transition value. Gas flows, temperature and spray voltage are shared across the whole method, so they must be a compromise: optimise them at the hardest condition in the run, which is normally the most aqueous composition at which any analyte elutes.

Does ESI source tuning fix ion suppression?

No. Suppression is competition for charge among co-eluting species in the droplet, and no source setting removes the competitor. Change the chromatography, clean the extract up, or compensate with a co-eluting isotopically labelled internal standard. Confirm the diagnosis with a post-column infusion experiment first.

How often should ESI source parameters be re-optimised?

Whenever anything they depend on changes: flow rate, column dimensions, mobile-phase composition or additive, source hardware, or the sprayer tip. Re-optimisation is not a calendar activity; it is a response to a changed input. Routine system-suitability checks tell you whether the existing settings still perform, which is a different and cheaper question.

Related CASRAI resources

• LC-MS explained: how liquid chromatography and mass spectrometry are coupled — the interface, source types and analyser options in overview.
• Multiple reaction monitoring: building and optimizing MRM transitions — what happens after the source, on a triple quadrupole.
• QTOF vs triple quadrupole mass spectrometer — choosing the analyser.
• GC-MS vs LC-MS — when the separation, not the source, is the decision.
• GC and HPLC column selection — the chromatography that must be settled before source tuning.
• Isotope ratio mass spectrometry — a different instrument class with its own reporting conventions.
• Lab equipment — the full instrument and equipment cluster.
• Mass spectrometry proteomics: choosing DDA or DIA, sample prep and run QC — the discovery counterpart, where acquisition is not pre-declared and the QC panel is different.