HPLC: Columns, Mobile Phases & Troubleshooting

“High performance” and “high pressure” liquid chromatography are the same technique. HPLC was introduced in the early 1970s as an upgrade to slow, gravity-fed column chromatography, and the original name described the mechanical change: a pump forcing mobile phase through a tightly packed column under real pressure instead of letting it drip through by gravity. As the technique matured, “high performance” became the preferred expansion because pressure was always a means to an end — better resolution, speed, and sensitivity — not the point itself. Searches for “high performance liquid chromatography,” “high pressure liquid chromatography,” “hplc liquid chromatography,” and “liquid chromatography hplc” are all looking for the same instrument and the same technique, and this page covers all of it: how the system works, how to choose a column and mobile phase, and what to do when a chromatogram goes wrong.

What HPLC Actually Does: The Instrument Flow Path

An HPLC system separates the components of a liquid sample by pumping it, dissolved in a solvent (the mobile phase), through a column packed with a stationary phase. Different sample components interact with the stationary phase to different degrees, so they travel through the column at different speeds and exit (elute) at different times. A detector at the column outlet records each component as it elutes, producing a chromatogram: a plot of detector signal against time, with each resolved component appearing as a peak.

The instrument itself is a straight flow path with five functional stages:

Solvent reservoirs and degasser. Mobile-phase solvents are held in reservoirs and degassed — by vacuum, helium sparging, sonication, or an inline vacuum degasser — before they reach the pump, to remove dissolved gas that would otherwise form bubbles downstream.
Pump. Delivers mobile phase at a precise, constant flow rate, either as a fixed composition (isocratic) or a changing blend of two or more solvents over the run (gradient). Conventional HPLC pumps typically run up to a few hundred bar; ultra-high-performance systems (UHPLC/UPLC) run substantially higher to support sub-2-micron column particles.
Injector / autosampler. Introduces a precise, small volume of sample into the flowing mobile phase without interrupting flow.
Column (and guard column). The separation itself happens here, inside a packed bed of stationary-phase particles. A short, replaceable guard column is typically placed just before the analytical column to intercept particulates and strongly retained contaminants that would otherwise foul the analytical column.
Detector. Measures a physical or chemical property of the eluent as each compound passes through, generating the chromatogram. Detector choice (covered below) depends on what the analyte is and whether the method uses gradient elution.

Normal-Phase vs. Reversed-Phase HPLC

The two modes differ in the relative polarity of the stationary and mobile phases, which determines which compounds are retained longest.

Normal-phase HPLC uses a polar stationary phase (e.g., bare silica) with a nonpolar mobile phase (e.g., hexane-based solvents). Polar compounds are retained longest; nonpolar compounds elute first. It is used less often today than reversed-phase, but remains useful for separating structurally similar nonpolar compounds, isomers, and some natural-product classes where reversed-phase gives poor selectivity.
Reversed-phase HPLC (RP-HPLC) uses a nonpolar stationary phase (most commonly a C18-bonded silica) with a polar aqueous/organic mobile phase (typically water mixed with acetonitrile or methanol). Nonpolar compounds are retained longest; polar compounds elute first. Reversed-phase is the default mode for the large majority of pharmaceutical, biomolecular, and general small-molecule HPLC methods because it tolerates aqueous samples, is compatible with a wide range of detectors, and gives reproducible, well-understood retention behavior.

Unless a method has a specific reason to run normal-phase (e.g., an established compendial method, or a separation that reversed-phase genuinely cannot resolve), reversed-phase with a C18 column is the conventional starting point.

Column Chemistry: Choosing Between C18, C8, Phenyl, and HILIC

The bonded phase on the column packing determines retention mechanism and selectivity. This is the single highest-leverage method-development decision after choosing normal- versus reversed-phase.

Column chemistry	Retention mechanism	Best suited for	Practical note
C18 (octadecylsilane)	Hydrophobic interaction; the most retentive common reversed-phase chemistry	The default first choice for reversed-phase methods across a broad range of small-molecule polarities	Widest applicability of any single phase; longer, more strongly retained runs than C8 for the same analyte
C8 (octylsilane)	Hydrophobic interaction, shorter alkyl chain than C18	Analytes that are over-retained or elute with excessive tailing on C18; faster runs where full C18 retention isn’t needed	Useful substitution when a C18 method gives good separation but impractically long run times
Phenyl	Combines hydrophobic retention with π–π interaction with aromatic rings	Aromatic and structurally similar analytes (positional isomers, compounds with similar C18 retention) that co-elute on a standard C18 column	A common first alternative to try when C18 selectivity isn’t enough to resolve a specific pair of peaks
HILIC (hydrophilic interaction)	Polar stationary phase retains polar/hydrophilic analytes from a mobile phase that is mostly organic	Very polar, poorly retained-on-C18 compounds: sugars, small polar metabolites, some peptides and glycans	Mobile-phase polarity logic is inverted from reversed-phase — more organic solvent increases retention, not less

Column particle size and pore size also affect resolution, backpressure, and run time, but chemistry selection is almost always the first decision, made before particle size is optimized.

Mobile-Phase Preparation: Buffers, Degassing, and pH

Mobile-phase quality is a common, under-suspected source of poor chromatography. A few practices matter more than they get credit for:

Water quality. HPLC mobile phase, especially for trace analysis or gradient methods, should use high-purity (Type I / “HPLC-grade”) water, not general laboratory deionized water — see Type I, II and III laboratory water grades and, for the purification systems that produce it, Milli-Q water purification system maintenance.
Buffers and pH. Buffer selection and pH accuracy directly control retention and peak shape for ionizable analytes. Prepare and verify buffer pH with a properly calibrated pH meter — see pH meter calibration and buffer selection — and confirm the chosen buffer’s useful pH range and UV cutoff are compatible with the column chemistry and detector.
Filtration. Mobile phase and samples should be filtered (commonly 0.2 or 0.45 micron) before use to keep particulates out of the pump seals, frits, and column inlet.
Degassing. Dissolved gas in the mobile phase comes out of solution at the reduced pressure downstream of the pump and in the detector flow cell, producing baseline noise, spikes, and inaccurate flow delivery (pump cavitation). Vacuum degassing, sonication, helium sparging, or an inline degasser all address this; it is one of the first things to check when baseline noise appears with no other obvious cause.
Isocratic vs. gradient elution. Isocratic methods hold mobile-phase composition constant for the entire run; they are simpler, more reproducible between labs, and well suited to mixtures with a narrow range of polarities. Gradient methods change the mobile-phase composition over time (typically increasing organic solvent) to elute a wide range of polarities in one run without excessive analysis time; they require a re-equilibration step back to starting conditions between injections and are more sensitive to pump and mixing-hardware precision.

Detector Options: UV/DAD, Fluorescence, RI, and ELSD

Detector	Principle	What it needs	Notes
UV / Diode-Array (DAD)	Absorbance of UV or visible light by the analyte	A chromophore (a UV/vis-absorbing structural feature) in the analyte	The most common HPLC detector; a diode-array version captures a full spectrum per data point, useful for peak identity and purity checks. Compatible with gradient elution provided the mobile phase itself doesn’t absorb strongly at the wavelength used.
Fluorescence	Emission of light at a longer wavelength after excitation at a shorter one	A native fluorophore, or a derivatization step to add one	Highly sensitive and selective for compounds that fluoresce; largely blind to everything that doesn’t, which is itself sometimes an advantage for selectivity in complex matrices.
Refractive Index (RI)	Difference in refractive index between the eluent and pure mobile phase	No chromophore required — effectively universal	Not compatible with gradient elution, since a changing mobile-phase composition shifts the baseline continuously; generally less sensitive than UV or fluorescence.
Evaporative Light Scattering (ELSD)	Nebulizes the column eluent, evaporates the mobile phase, and measures light scattered by the remaining analyte particles	A volatile mobile phase; no chromophore required	Near-universal and compatible with gradient elution, unlike RI, which makes it a common choice for lipids, sugars, and other analytes without a strong UV chromophore. Response is not linear across a wide concentration range.

Peak-Problem Troubleshooting Table

Most day-to-day HPLC problems show up as a specific, recognizable distortion of the chromatogram. This table maps the symptom to the causes worth checking first. Peak tailing is the most frequently reported of these symptoms and has the most overlapping causes, so it gets its own cause-ranked decision tree immediately after this table.

Symptom	Likely causes	What to check / fix
Peak tailing	Column overload; a partially blocked inlet frit or guard column; extra-column dead volume; a column void; secondary interactions with residual silanol groups on the packing; sample diluent stronger than the initial mobile phase	Reduce injection amount or concentration; replace the guard column or inlet frit; inspect tubing and fittings for unnecessary dead volume; check plate count against baseline; consider a column with lower silanol activity; match the sample diluent more closely to the mobile phase — see the cause-ranked decision tree below for the specific test that tells these apart before you start swapping parts
Peak fronting	Column overload at the opposite extreme from tailing; voids or channeling in an aging column bed; injection solvent too strong relative to the mobile phase	Reduce sample load; replace a column showing signs of a void or declining efficiency; verify injection solvent strength against the starting mobile phase
Ghost peaks / unexpected extra peaks	Carryover from a previous injection; contaminated mobile phase, vials, or system tubing; sample degradation between preparation and injection	Run mobile-phase blank injections between samples; check the autosampler needle-wash program; prepare fresh, filtered mobile phase; confirm sample stability under the run conditions
Split or doubled peaks	Partially blocked inlet frit or guard column; an air bubble introduced at injection; poor sample solubility in the injection solvent	Check and, if needed, replace the guard column or inlet frit; confirm the mobile phase and sample are properly degassed; confirm the sample is fully dissolved in a compatible solvent
Baseline drift	Ambient or column-oven temperature fluctuation; a gradient run paired with a detector (e.g., RI) that can’t tolerate composition changes; mobile-phase contamination or slow solvent evaporation from an open reservoir; incomplete column re-equilibration	Use column thermostatting; confirm detector choice suits gradient elution; prepare fresh mobile phase and keep reservoirs capped; extend the re-equilibration time between injections
Pressure rise over a run, or across successive runs	Frit or column-inlet clogging from particulates or precipitated buffer salts; a fouled guard column; precipitation of buffer at high organic-solvent percentage	Replace the guard column first, since it’s designed to take this damage; flush the system with a compatible solvent; check buffer solubility at the mobile-phase composition actually used; filter both samples and mobile phase
Sudden pressure drop	A leak at a fitting or connection; a failing pump seal or check valve; a column frit failure or void allowing mobile phase to bypass the packing	Inspect all fittings and tubing for leaks; check pump seals and check valves; inspect the column for physical damage or a collapsed bed
Retention-time shift / drift between runs	Mobile-phase composition or pH inconsistency between batches; column aging or incomplete equilibration; flow-rate inaccuracy from a wearing pump; uncontrolled column temperature	Prepare mobile phase consistently and verify pH each batch; allow full system re-equilibration before the first injection of a sequence; verify actual delivered flow rate; use a column oven

Diagnosing Peak Tailing: A Cause-Ranked Decision Tree

The troubleshooting table above lists peak tailing’s causes side by side, which is enough to start guessing but not enough to stop guessing. In practice, five causes account for nearly all tailing complaints, and each has one diagnostic test that confirms or rules it out without assuming the answer in advance. Work through them in this order — cheapest and most common first — rather than jumping straight to a column replacement.

1. Column overload. Loading more sample mass onto the column than its linear capacity can hold distorts peak shape as concentration increases, most often as a sloped tailing edge (and, at more severe overload, fronting), along with a retention-time shift toward shorter times as the load rises. Diagnostic test: inject a dilution series of the same sample (e.g., full strength, 1:2, 1:5, 1:10). If the tailing factor improves and retention time increases as concentration drops, overload is confirmed and the column and system are otherwise fine. Fix: reduce injection volume or sample concentration to bring the on-column mass back within the column’s linear range.
2. A partially blocked inlet frit or guard column. Particulates or precipitated buffer salts collecting at the column inlet restrict and channel flow unevenly across the packed bed, producing tailing alongside a telltale second symptom: rising backpressure. Diagnostic test: check the backpressure trend over the last several days or injections, not just the current run — a frit or guard-column blockage shows a gradual upward creep in baseline pressure that overload and silanol activity do not produce. Swapping in a fresh guard column and re-running the same sample is the confirmatory step: if tailing clears immediately with no other change, the frit was the cause. Fix: replace the guard column (routine, and designed to take this damage first) or the analytical column’s inlet frit; filter mobile phase and samples going forward to slow recurrence.
3. Extra-column volume. Unswept volume in tubing, fittings, the injector, or the detector flow cell between the column outlet and the detector adds band broadening on top of whatever the column contributes, and it shows up as tailing that is disproportionately worse on early-eluting, weakly retained peaks and comparatively minor on late-eluting, strongly retained ones — a fixed extra-column volume is a larger fraction of a narrow early peak’s total volume than of a broad late one’s. Diagnostic test: replace the column with a zero-volume union (or the shortest compatible length of tubing) and inject the same sample directly into the detector. Any tailing that survives with the column physically removed is coming from the system, not the packing. Fix: shorten and minimize the internal diameter of tubing between injector and column and between column and detector, check fittings for unswept dead space, and confirm the detector flow-cell volume suits the peak volumes the method produces.
4. A column void. Packed beds settle and can develop a void — a gap, typically at the column inlet — as packing degrades with use, especially under repeated pressure cycling or conditions outside the phase’s stability range. A void adds extra-column-like mixing at the head of the column, producing tailing (and often peak splitting or a leading shoulder) together with a drop in plate count on a system-suitability check compound. Diagnostic test: compare current plate count against the column’s baseline value from when it was new or last qualified — a void typically shows a sharp decline rather than gradual drift. Reversing the column’s flow direction for a single test injection is the classical confirmatory check: if peak shape improves running backward, the void is at what is normally the inlet end. Fix: there is no reliable field repair for a void; replace the column.
5. Secondary interaction with residual silanol groups. On silica-based reversed-phase packings, not every silanol group on the silica surface is bonded to the stationary phase’s alkyl chains; the exposed, acidic silanols left behind can act as weak cation-exchange sites for basic, protonatable analytes, adding a second, slower retention mechanism on top of the intended hydrophobic one. This produces tailing that is specific to basic analytes and largely absent for neutral ones run on the same column under the same conditions — the clearest sign it isn’t a system or physical-column problem. Diagnostic test: compare the tailing factor of a basic test compound against a neutral compound of similar hydrophobicity, on the same column and run. If only the basic compound tails badly, silanol activity is implicated rather than the column’s physical condition. Fix: switch to a high-purity, fully end-capped, low-silanol-activity column chemistry, or add a silanol-masking mobile-phase additive (e.g., a low concentration of triethylamine or an ammonium buffer) appropriate to the detection method and analyte.

The first three causes on this list are the most common and the cheapest to rule out, so most real-world tailing cases resolve within the first two or three tests without ever requiring a new column.

Diagnosing Baseline Drift: Trace Shape, Periodicity, and the Confirmatory Test

The troubleshooting table above lists baseline drift’s causes together, but the four most common ones produce visibly different traces, and each has a specific check that confirms or rules it out. Read the shape and timing of the drift first — smooth and slow versus stepped versus tied to a specific run event — before changing anything on the system.

1. Thermal drift. Ambient-temperature swings (HVAC cycling, sun on the bench, an open door) and column-oven fluctuation change mobile-phase viscosity and density, and change detector response for temperature-sensitive detectors, producing a slow, smooth, cyclical drift that tracks room or column temperature rather than anything in the run itself. Confirmatory test: log ambient or column-oven temperature over the same period as the drifting baseline and overlay the two traces; if the drift’s peaks and troughs line up with the temperature trace, thermal drift is confirmed. Fix: enable or improve column thermostatting, and move the system away from HVAC vents, direct sun, or drafts.
2. Refractive-index (RI) detector drift. An RI detector responds to any difference between the eluent and pure mobile phase, which makes it read the smallest real change in mobile-phase composition as baseline movement — a continuous, monotonic ramp that tracks a gradient’s %B curve, or a slower ramp from solvent evaporating out of an uncapped reservoir and slowly concentrating the mobile phase over a run sequence. Confirmatory test: run a blank gradient (the same %B program, nothing injected) and compare its shape to the drifting baseline — if they match, the detector is faithfully reporting a real composition change, not malfunctioning. Fix: switch to isocratic elution for RI work where possible, or if gradient elution is unavoidable, account for the ramp as baseline in integration; keep reservoirs capped to stop evaporation-driven concentration drift.
3. Air-bubble drift. A bubble trapped in the detector flow cell, or intermittently forming at the pump head from incomplete degassing, does not produce a smooth ramp — it produces sudden step changes or short spikes, often repeating at a fixed short interval tied to the pump’s stroke rather than drifting continuously in one direction. Confirmatory test: watch backpressure alongside the baseline; synchronized spiking in both, or a visible bubble after opening the flow cell, confirms a bubble rather than thermal or chemical drift. Fix: purge and re-prime the pump and flow cell, confirm the degasser is functioning (vacuum, sparge gas, or inline degasser as applicable), and check fittings upstream of the detector for a micro-leak that’s pulling in air.
4. Lamp-ageing drift (UV/DAD only). A deuterium (or equivalent) UV lamp’s output intensity declines gradually over its service life, and as intensity falls the detector’s baseline signal and noise both drift, slowly and monotonically, over days to weeks rather than within a single run. Confirmatory test: check the instrument’s reported lamp energy or intensity value against the manufacturer’s minimum-acceptable threshold; a low or steadily falling reading, with no correlated temperature or gradient event, points to the lamp rather than the mobile phase or environment. Fix: replace the lamp once its intensity reading falls near or below the manufacturer’s threshold; track lamp hours so replacement is scheduled ahead of failure rather than diagnosed after the fact.

Distinguishing these up front avoids the common mistake of re-preparing mobile phase or replacing a column when the actual cause is a lamp nearing the end of its life or an HVAC cycle warming the bench — neither of which a fresh mobile phase or a new column touches.

System Suitability, Qualification, and Data Integrity

For regulated or compendial HPLC work, “the instrument is working” is a formal, documented determination rather than an impression. Two U.S. Pharmacopeia general chapters anchor this: USP General Chapter <621> Chromatography defines system suitability parameters — resolution, tailing factor, theoretical plate count, and injection-to-injection reproducibility — that a system must meet before sample data from a run is considered valid, and USP General Chapter <1058> Analytical Instrument Qualification sets out the design, installation, operational, and performance qualification (DQ/IQ/OQ/PQ) framework used to demonstrate that an instrument is fit for its intended analytical use in the first place. See calibration certificates and metrological traceability for how that qualification chain connects to a specific, documented calibration event.

Calculating the USP <621> System Suitability Parameters

USP <621> names four parameters, but stating them by name is not the same as passing them. Each has a formula, and each formula has a limit the run either clears or does not. The worked numbers below use representative, illustrative values (not a real validated method) so the arithmetic is concrete rather than abstract.

Resolution (Rs) between the critical peak pair

Resolution measures whether two adjacent peaks are cleanly separated. Using peak width at half-height, the common working form is:

Rs = 1.18 × (tR2 − tR1) / (W0.5,1 + W0.5,2)

where tR1 and tR2 are the retention times of the two peaks and W0.5,1 / W0.5,2 are their peak widths at half height. Worked example: peak A elutes at 4.20 min with a half-height width of 0.14 min; peak B elutes at 4.65 min with a half-height width of 0.16 min.

Rs = 1.18 × (4.65 − 4.20) / (0.14 + 0.16) = 1.18 × 0.45 / 0.30 = 1.77

General USP <621> guidance treats Rs ≥ 2.0 for the critical pair (the two peaks hardest to separate on that method) as the default acceptable-separation benchmark, though an individual compendial monograph or validated method can specify a different value. At 1.77, this example run would not meet a 2.0 requirement and the pair is not cleanly resolved — before reporting sample results, the fix is usually a mobile-phase adjustment (organic percent, pH, or ionic strength), a slower gradient through that region, or a column change, re-tested until Rs clears the stated limit.

Tailing factor (T)

Tailing factor is a peak-shape check, using peak width measured at 5% of peak height:

T = W0.05 / (2f)

where W0.05 is the full peak width at 5% height and f is the distance from the peak front to the peak apex, also measured at 5% height. Worked example: W0.05 = 0.42 min, f = 0.18 min.

T = 0.42 / (2 × 0.18) = 0.42 / 0.36 = 1.17

A symmetric peak has T = 1.0; USP <621> commonly cites T ≤ 2.0 as the general acceptable-tailing ceiling, though many validated methods set a tighter working limit (T ≤ 1.5 is a frequently used in-house target). At 1.17 this example passes a 2.0 limit comfortably. A tailing factor that drifts upward over a column’s life — even while still inside the limit — is worth tracking on a trend chart, since it is usually the earliest sign of column fouling or void formation, addressed in the troubleshooting table above.

Theoretical plates (N)

Theoretical plate count is the column-efficiency measure — how sharp the peak is relative to how long it took to elute:

N = 5.54 × (tR / W0.5)²

using retention time and peak width at half height (an older, equivalent form uses N = 16 × (tR/W)² with peak width at baseline). Worked example, same peak B as above: tR = 4.65 min, W0.5 = 0.16 min.

N = 5.54 × (4.65 / 0.16)² = 5.54 × 29.06² = 5.54 × 844.5 ≈ 4,680 theoretical plates

Unlike resolution and tailing factor, USP <621> does not set one fixed, chapter-wide minimum N — the required value is method- and monograph-specific, set during validation for that particular column/analyte pair. N ≥ 2,000 is a commonly cited practical floor in many labs’ own SOPs, but treat that as a rule of thumb, not a compendial number, and confirm the actual figure your method’s validation report or the applicable monograph specifies.

%RSD of replicate injections (system precision)

Precision is checked by injecting the same standard preparation repeatedly (commonly five or six replicate injections) and calculating the relative standard deviation of the peak area (or response factor):

%RSD = (SD / mean) × 100

Worked example, five replicate peak areas: 100200, 101500, 99800, 100700, 100100.

Mean = 100,460. Sample standard deviation (n−1 denominator) ≈ 665.6. %RSD = (665.6 / 100,460) × 100 ≈ 0.66%.

USP <621> ties the acceptable %RSD ceiling to how many replicates were injected: fewer replicates require a tighter limit, more replicates allow a looser one, because more replicates give a more reliable precision estimate. A %RSD ≤ 1.0% for five replicate injections is the most commonly cited default in general assay SOPs, but the exact permitted value for a given replicate count is set out in <621>’s own table and can be adjusted by the individual method’s validation — check the applicable table rather than assuming 1.0% applies at every n. At 0.66% for n = 5, this example passes.

What to Do When a System Suitability Test Fails

A system suitability failure means the run does not get to report sample results — not “report with a caveat,” not “average it out with the next run.” The sequence that avoids compounding the problem:

Stop before injecting samples (or before releasing results, if suitability is checked with samples already in the sequence). Do not proceed on a suitability set that failed any of the four parameters.
Rule out the obvious, cheap causes first: air in the pump head or a slow leak, a degassing problem, an aged or contaminated mobile phase, a partially clogged frit, or a standard preparation that was made up wrong or has degraded. Most real failures resolve at this stage.
Re-inject the suitability standard once the suspected cause is corrected, rather than re-testing the full sample set immediately — confirm the system is back in specification before spending sample volume on it.
If it still fails, treat it as an out-of-specification (OOS) laboratory investigation, not a silent re-run: document what failed, by how much, and what was checked, per your lab’s OOS/atypical-result procedure. Column replacement, mobile-phase re-preparation, or an instrument qualification check (see the DQ/IQ/OQ/PQ framework above) are the next steps if the cheap causes don’t resolve it.
If the method itself is marginal — suitability barely passes on a good day and fails on a bad one — that’s a method robustness problem, not a run-to-run instrument problem, and belongs in a re-validation or robustness study rather than repeated troubleshooting. See ICH Q2(R2) Analytical Procedure Validation for what that re-validation needs to demonstrate.

Where HPLC data is captured and reported through a chromatography data system (CDS) in a regulated environment, the data system itself — not just the physical instrument — is generally in scope for computer system validation and electronic-record requirements under 21 CFR Part 11. See Computer System Validation: GAMP 5, IQ/OQ/PQ, and 21 CFR Part 11 for how that validation is structured.

Frequently Asked Questions

Is it “high performance” or “high pressure” liquid chromatography?

Both names refer to the same technique. The original 1970s name described the mechanical change from gravity-fed column chromatography to a pumped, pressurized system. “High performance” became the more common expansion as the field matured, since the point of the pressure was always better resolution, speed, and sensitivity, not the pressure itself. You will see both in circulation; they are not different methods.

What’s the difference between HPLC and UHPLC/UPLC?

Ultra-high-performance liquid chromatography (UHPLC) uses higher-pressure pump systems and columns packed with sub-2-micron particles to run faster, higher-resolution separations than conventional HPLC. UPLC is a specific vendor’s (Waters’) trademarked name for its UHPLC product line; the underlying technique category is UHPLC. Methods can often be transferred between conventional HPLC and UHPLC with adjustment to flow rate, column dimensions, and gradient timing.

What’s the difference between HPLC and gas chromatography (GC)?

HPLC separates compounds dissolved in a liquid mobile phase and is suited to non-volatile, thermally labile, and larger molecules that cannot be vaporized without decomposing. GC vaporizes the sample and carries it through the column in a gas mobile phase, which limits it to compounds that are volatile and thermally stable at the operating temperature.

Why does the mobile phase need to be degassed?

Dissolved gas comes out of solution as bubbles once the mobile phase reaches the lower pressure downstream of the pump and in the detector flow cell, producing baseline noise, spikes, and inaccurate flow delivery. Vacuum degassing, sonication, helium sparging, or an inline degasser all remove dissolved gas before it becomes a problem.

How do I decide between a C18 and a C8 column?

Start with C18 as the default reversed-phase chemistry; it is the most broadly applicable and most strongly retentive common phase. Switch to C8 if a C18 method over-retains the analyte, produces impractically long run times, or shows excessive tailing that a shorter alkyl chain and correspondingly less hydrophobic retention can resolve.

What’s the fastest way to tell why an HPLC peak is tailing?

Work down the cause-ranked decision tree above in order rather than guessing: run a dilution series first (rules overload in or out), then swap the guard column while watching the backpressure trend (rules out a blocked frit), then remove the column and inject straight into the detector (rules out extra-column volume), then compare current plate count against baseline (flags a column void), then compare a basic versus a neutral test compound’s tailing factor (flags residual silanol activity). Each test is specific enough to confirm one cause and rule out the others, so most cases resolve in the first two or three steps.

What does the shape of an HPLC baseline drift tell me about its cause?

A slow, smooth, cyclical drift that tracks room or column-oven temperature points to thermal drift. A continuous ramp that follows a gradient’s %B curve, or a slow one-directional shift over a run sequence, points to a refractive-index detector responding to a real composition change. Sudden steps or spikes, especially synchronized with the pump stroke or backpressure, point to an air bubble. A slow decline in both signal and noise over days to weeks, unconnected to any single run event, points to an ageing UV/DAD lamp.

Can I fix HPLC baseline drift without replacing the column?

Yes in most cases — baseline drift is rarely a column problem. Thermal drift is fixed with better thermostatting, RI drift is fixed by switching to isocratic elution or capping reservoirs, air-bubble drift is fixed by purging and re-priming the system, and lamp-ageing drift is fixed by replacing the lamp. Reach for a new column only after these four causes have been ruled out.