Analytical Chemistry

High-Performance Liquid Chromatography (HPLC)

Q: How does HPLC separate compounds?

Sample analytes dissolved in the mobile phase partition between the moving liquid and the stationary phase coated on small silica particles in the column. Compounds that interact more strongly with the stationary phase (e.g. more hydrophobic for C18 reverse-phase) travel slower and elute later; weakly interacting ones elute earlier. The number of times an analyte transfers between phases inside the column — the plate count N — determines resolving power. A 150 mm × 4.6 mm column packed with 3 μm particles delivers around 50,000 theoretical plates, enough to resolve compounds whose retention times differ by less than 1%. Detection at column outlet (UV-vis, fluorescence, refractive index, mass spectrometry) gives a chromatogram of peaks vs. time.

Q: What is reverse-phase C18 and why does it dominate?

Reverse-phase C18 columns have silica particles whose surface silanols are derivatized with octadecyl (C18) chains, giving a hydrophobic stationary phase. The mobile phase is polar — water mixed with methanol or acetonitrile — so polar analytes elute first and hydrophobic ones last, the opposite ('reverse') of the original 'normal-phase' polar-silica protocols. C18 dominates because (1) most pharmaceutical drug molecules have molecular weight 200–800 and a balance of polar groups with hydrophobic scaffolds, well-suited to the C18 polarity range; (2) water-acetonitrile mobile phases are inexpensive and UV-transparent down to ~210 nm; (3) reproducibility across vendors is excellent due to standardization since the 1980s. About 85% of all HPLC pharmaceutical methods use reverse-phase C18.

Q: How does HPLC differ from UPLC?

UPLC (Ultra-High-Performance Liquid Chromatography, Waters trademark; UHPLC is the generic term) uses sub-2 μm particles (typically 1.7 μm) requiring back pressures of 1000–1300 bar — versus HPLC's 3–5 μm particles at 200–400 bar. The smaller particles flatten the van Deemter H–u curve, enabling 5–10× faster separations at equal resolution: a 5-minute UHPLC run replaces a 30-minute HPLC run. The instrument must withstand higher pressure and tighter dead volumes, costing 2–3× more than a standard HPLC. For pharmaceutical QC, UHPLC has overtaken HPLC in new method development since around 2010; legacy methods stay on HPLC for regulatory continuity under ICH Q2 method-validation rules.

Q: What is a theoretical plate and what is N for a typical column?

Theoretical plate count N is a measure of column efficiency: N = 16 (t_r/w)^2, where t_r is retention time and w is peak width at base. It comes from a fictitious model treating the column as N stages of equilibrium between mobile and stationary phases — Martin and Synge's 1941 plate theory (Nobel 1952). A standard 4.6×150 mm column at 3 μm gives N ≈ 50,000; a 4.6×250 mm column at 5 μm gives ~25,000; a 2.1×100 mm UHPLC column at 1.7 μm gives ~30,000. Higher N means narrower peaks and more resolved compounds in a given retention window. Plate height H = L/N (typical 3 μm) is the figure of merit being optimized.

Q: Why is HPLC the workhorse of pharmaceutical QC?

ICH guidelines Q2(R2) on method validation, Q3A/Q3B on impurities, and Q6A on specifications all require chromatographic assay of active pharmaceutical ingredients and related impurities at sub-0.1% levels. HPLC delivers the required selectivity, sensitivity, and reproducibility. A typical drug product release assay quantifies the API and 5–15 specified impurities by reverse-phase HPLC-UV in 30 minutes; method validation per ICH involves linearity over 50–150% of nominal, accuracy 98–102%, repeatability RSD <2%, and intermediate precision RSD <3%. Sildenafil purity by USP method, for instance, runs C18 with a 0.05 M ammonium acetate / acetonitrile gradient, UV at 290 nm, with related substances quantified at 0.05% reporting threshold.

Q: What are van Deemter and rate theory in HPLC?

The van Deemter equation H = A + B/u + C u relates plate height H to mobile-phase linear velocity u via three contributions. A (eddy diffusion) — irregular flow paths through packed bed; reduced by smaller, more uniform particles. B (longitudinal diffusion) — analyte spreading by random diffusion in the mobile phase, dominant at low flow. C (mass-transfer resistance) — finite time for analytes to equilibrate between phases, dominant at high flow. The plot of H vs. u is bowl-shaped with a minimum at the optimum velocity u_opt, where the column delivers its highest plate count. Calvin Giddings extended van Deemter into a more general rate theory in the 1960s, treating chromatography as a stochastic transport problem and predicting band-broadening from first principles. Sub-2 μm UHPLC particles flatten C, making u_opt larger and enabling fast runs.

1.7–5 μm silica + 200–400 bar pump separates analytes by mobile-phase polarity — reverse-phase C18 dominates pharma

High-Performance Liquid Chromatography (HPLC) separates dissolved analytes by pumping a liquid mobile phase at 200–400 bar through a column packed with 1.7–5 μm silica particles. Compounds that interact more strongly with the stationary phase travel slower and elute later. Csaba Horváth pioneered the technique in the 1960s and Calvin Giddings developed the rate-theory framework that explains why smaller particles give better resolution. Reverse-phase C18 silica with a water-methanol or water-acetonitrile mobile phase covers about 85% of pharmaceutical applications and is the workhorse of pharmaceutical quality control under ICH guidelines. A 4.6×150 mm column at 3 μm particles delivers ~50,000 theoretical plates, enough to resolve drug products with sub-0.1% impurities.

Pressure200–400 bar (HPLC), 1000–1300 bar (UHPLC)
Particle size1.7–5 μm silica
PioneerCsaba Horváth, 1960s
Standard column4.6×150 mm, 3 μm, ~50,000 plates
Dominant phaseC18 reverse-phase (~85% pharma)
RegulationICH Q2(R2) method validation

Interactive visualization

Press play, or step through manually. The visualization is yours to drive — try it before reading on.

Open visualization fullscreen ↗

Watch the 60-second explainer

A condensed visual walkthrough — narrated, captioned, under a minute.

Why HPLC matters

Pharmaceutical QC backbone. ICH Q2(R2), Q3A/Q3B, and Q6A specifications all rely on HPLC for assay of active pharmaceutical ingredients and related impurities at sub-0.1% reporting thresholds. About 85% of pharmacopeial assays in USP, EP, and JP use reverse-phase C18 HPLC; a typical drug release method runs in 15–30 minutes with UV detection at 210–290 nm.
50,000 theoretical plates. A 4.6×150 mm column at 3 μm particles delivers about 50,000 theoretical plates by van Deemter optimization — enough to baseline-resolve compounds with retention times differing by less than 1%, the criterion for distinguishing drug-related impurities from main API peaks.
Csaba Horváth founded modern HPLC. At Yale in the 1960s Horváth built the first instrument with high-pressure pumps, packed columns, and online detectors — the architecture every modern HPLC still inherits. The renamed Csaba Horváth Award is the field's leading honor.
Calvin Giddings's rate theory. The 1965 monograph "Dynamics of Chromatography" generalized van Deemter into a stochastic transport theory that predicts band-broadening from first principles. Giddings showed why sub-2 μm particles enable UHPLC: smaller particles shorten the mass-transfer length and flatten the van Deemter C term.
Sub-0.1% impurity detection. Sildenafil purity assay (USP) runs C18 with ammonium acetate/acetonitrile gradient, UV 290 nm, reporting threshold 0.05% per ICH Q3B. Drug metabolism and pharmacokinetic studies routinely quantify analytes at low ng/mL using HPLC-MS/MS.
Bioanalytical workhorse. Therapeutic drug monitoring of immunosuppressants (cyclosporine, tacrolimus), antibiotics (vancomycin), and antiepileptics relies on HPLC. Endogenous biomarkers (vitamin D, hormones, neurotransmitters) are quantified by HPLC-MS in clinical labs at sub-pg/mL levels.
Preparative scale. Process-scale prep-HPLC purifies grams to kilograms of pharmaceutical APIs and natural products using 50–600 mm diameter columns. Penicillin, taxol, and many monoclonal-antibody intermediates are purified on prep HPLC at production scale.

Common misconceptions

"Higher pressure means better separation." Pressure is just the cost of pushing liquid through small particles fast — it does not directly improve resolution. The figure of merit is plate count N, set by particle size, packing uniformity, and flow velocity at the van Deemter optimum.
"Smaller particles always give faster results." Sub-2 μm particles enable faster runs only if the instrument can deliver the required 1000+ bar without dead-volume broadening dominating. Standard HPLC instruments cannot exploit 1.7 μm particles — they need a UHPLC system with low-volume detector flow cells and tight tubing.
"Normal-phase is named for being more common." The reverse: normal-phase came first historically (polar silica, nonpolar mobile phase). When Horváth and others inverted the polarities for biological samples in the 1970s the new mode was called "reverse-phase." Today reverse-phase is by far more common, but the name reflects history.
"All C18 columns are equivalent." Vendor C18 phases differ in silica purity, surface coverage, endcapping, and base silica activity — small differences shift retention times by 10–30% and can break a validated method. Pharmaceutical QC methods specify exact column manufacturer/part number for reproducibility.
"Gradient is always better than isocratic." Gradient elution shifts mobile-phase composition during the run to elute strongly retained analytes faster, but introduces re-equilibration time and detector baseline drift. For simple separations with k' < 10, isocratic is faster overall and easier to transfer between labs.
"Detection sensitivity is limited by the column." The detector matters more. UV-vis at 210 nm gives ~ng injected on column, fluorescence ~pg, MS/MS in selected reaction monitoring mode reaches sub-pg. The column controls peak width — a narrower peak helps detection by concentrating analyte in time.

Mechanism

HPLC has six functional components: (1) solvent reservoirs holding mobile-phase A (typically aqueous buffer) and B (organic, methanol or acetonitrile); (2) a high-pressure pump capable of 200–400 bar (HPLC) or 1000–1300 bar (UHPLC) at flow rates of 0.1–2.0 mL/min; (3) an autosampler that injects 1–100 μL of dissolved sample; (4) a thermostatted column (typically 30–40 °C) packed with derivatized silica particles; (5) a detector at the column outlet — UV-vis with diode array (DAD), fluorescence, refractive index, evaporative light scattering, or mass spectrometer; (6) data acquisition software that integrates peaks for quantification. Total system cost is $30k–$80k for HPLC, $80k–$200k for UHPLC, $250k+ for HPLC-MS/MS.

Separation arises from differential partitioning of analytes between mobile and stationary phases. The retention factor k' = (t_r − t₀)/t₀ measures how long an analyte spends in the stationary phase relative to the mobile-phase dead time t₀. For reverse-phase C18, more hydrophobic analytes have higher k'; the relationship k' = k_w × 10^−Sφ ties retention to the volume fraction φ of organic modifier (the linear-solvent-strength model). Doubling the methanol fraction typically halves retention. The selectivity α = k'₂/k'₁ between two analytes determines whether they can be resolved.

The figure of merit is theoretical plate count N = 16(t_r/w)², which derives from Martin and Synge's 1941 plate model (Nobel 1952). Plate height H = L/N is the band-broadening per unit length. Van Deemter (1956) gives H = A + B/u + Cu, where A is eddy diffusion (path length variations through packed bed), B/u is longitudinal diffusion (random walk in mobile phase, dominant at low flow), and Cu is mass-transfer resistance (analyte equilibration time, dominant at high flow). The plot of H vs flow velocity has a minimum at u_opt; sub-2 μm particles lower C and shift u_opt higher, enabling fast UHPLC. Resolution between two peaks is R_s = (1/4) √N × (α − 1)/α × k'₂/(1 + k'₂); R_s > 1.5 gives baseline resolution.

HPLC vs GC vs UPLC vs FPLC

Technique	Mobile phase	Pressure	Particle / Column	Sample type	Typical use
HPLC	Liquid (water + MeOH/ACN)	200–400 bar	3–5 μm packed silica	Non-volatile, polar to mid-hydrophobic	Pharma QC, food/environmental analysis
UHPLC / UPLC	Liquid	1000–1300 bar	1.7 μm packed silica	Same as HPLC, faster	High-throughput pharma, bioanalysis
GC (Gas Chromatography)	Gas (He, H₂, N₂)	1–5 bar	Open-tubular capillary, 0.1–0.5 μm film	Volatile, thermally stable	Petroleum, residual solvents, fragrances
FPLC (Fast Protein LC)	Aqueous buffers	1–10 bar	10–100 μm soft beads (Sepharose, agarose)	Native proteins, peptides	Biopharm protein purification
SFC (Supercritical Fluid)	Supercritical CO₂ + modifier	100–200 bar	3–5 μm silica	Chiral, non-polar	Chiral drug separations
Ion Chromatography	Aqueous, eluted by buffer ion	50–200 bar	Polymer-based ion exchanger	Inorganic ions, organic acids	Water purity, formulations
Capillary Electrophoresis	Aqueous buffer in capillary	(electric field, 10–30 kV)	50–100 μm fused silica capillary	Charged analytes, DNA	Genomics, peptide mapping

C18 reverse-phase vs normal-phase silica vs HILIC vs ion-exchange

Mode	Stationary phase	Mobile phase	Elution order	Best for
Reverse-phase C18	Octadecyl-derivatized silica (hydrophobic)	Water + MeOH or ACN	Polar first, hydrophobic last	Most drugs, peptides, mid-polarity analytes
Reverse-phase C8	Octyl-derivatized silica (less hydrophobic than C18)	Water + organic	Polar first	More polar drugs, hydrophobic peptides
Normal-phase silica	Bare silica (polar)	Hexane + ethyl acetate, etc. (non-polar)	Non-polar first, polar last	Lipids, vitamins, hydrophobic isomers
HILIC (Hydrophilic Interaction)	Polar bonded silica + adsorbed water layer	ACN + small water (5–40%)	Hydrophobic first, polar last	Sugars, polar metabolites, peptides
Ion-exchange (cation)	Sulfonic acid bonded silica	Aqueous buffer with displacing ion	By analyte charge / pK_a	Amino acids, organic bases
Ion-exchange (anion)	Quaternary ammonium bonded	Aqueous buffer with displacing anion	By analyte charge / pK_a	Organic acids, nucleotides
Size-exclusion (SEC)	Porous gel (no interaction)	Aqueous or organic, isocratic	Largest first, smallest last	Proteins, polymers (MW determination)
Chiral CSP	Cellulose / amylose / Pirkle phases	Hexane + isopropanol (NP) or aqueous (RP)	By stereochemistry	Enantiomer separation, pharma QC

Applications

Pharmaceutical QC under ICH guidelines. Sildenafil purity assay (USP <1090>) runs reverse-phase C18 column 4.6×250 mm 5 μm with 0.05 M ammonium acetate / acetonitrile gradient, UV at 290 nm; main peak symmetry <1.5, system-suitability RSD <1.0% (5 replicate injections), reporting threshold 0.05% per ICH Q3B. Validated per ICH Q2(R2): linearity 50–150% nominal (R² > 0.999), accuracy 98–102%, repeatability RSD <2%.
Drug metabolism and PK/PD. HPLC-MS/MS in selected reaction monitoring mode quantifies parent drug and metabolites in plasma at sub-ng/mL levels; supports phase 1 clinical PK studies, drug-drug interaction studies, and FDA bioequivalence filings under 21 CFR 320. Typical runs are 5–10 minutes per sample with 200–500 sample throughput per day.
Biopharmaceutical characterization. Monoclonal antibody peptide mapping uses tryptic digest + reverse-phase C18 + high-resolution MS to verify amino acid sequence, post-translational modifications, and disulfide bonds. ICH Q6B requires this for biosimilar approval. Charge variant analysis uses cation-exchange HPLC; aggregate quantification uses size-exclusion HPLC.
Food and environmental analysis. Pesticide residues in food are quantified by HPLC-MS/MS at low ppb levels under EU regulation 396/2005. Mycotoxins (aflatoxin, ochratoxin) in grains use HPLC-fluorescence at low ppt. Drinking-water contaminant assays under EPA Method 525.2 use HPLC-MS for pharmaceuticals and personal care products in surface water.
Therapeutic drug monitoring. Immunosuppressants (cyclosporine, tacrolimus, sirolimus) in whole blood for transplant patients are quantified by HPLC-MS/MS at clinical labs; results within 4 hours guide dosing. Anticonvulsants, antifungals, and antibiotic levels for critically ill patients drive precision dosing under therapeutic drug-monitoring protocols.

Frequently asked questions

How does HPLC separate compounds?

Sample analytes dissolved in the mobile phase partition between the moving liquid and the stationary phase coated on small silica particles in the column. Compounds that interact more strongly with the stationary phase (e.g. more hydrophobic for C18 reverse-phase) travel slower and elute later; weakly interacting ones elute earlier. The number of times an analyte transfers between phases inside the column — the plate count N — determines resolving power. A 150 mm × 4.6 mm column packed with 3 μm particles delivers around 50,000 theoretical plates, enough to resolve compounds whose retention times differ by less than 1%. Detection at column outlet (UV-vis, fluorescence, refractive index, mass spectrometry) gives a chromatogram of peaks vs. time.

What is reverse-phase C18 and why does it dominate?

Reverse-phase C18 columns have silica particles whose surface silanols are derivatized with octadecyl (C18) chains, giving a hydrophobic stationary phase. The mobile phase is polar — water mixed with methanol or acetonitrile — so polar analytes elute first and hydrophobic ones last, the opposite ('reverse') of the original 'normal-phase' polar-silica protocols. C18 dominates because (1) most pharmaceutical drug molecules have molecular weight 200–800 and a balance of polar groups with hydrophobic scaffolds, well-suited to the C18 polarity range; (2) water-acetonitrile mobile phases are inexpensive and UV-transparent down to ~210 nm; (3) reproducibility across vendors is excellent due to standardization since the 1980s. About 85% of all HPLC pharmaceutical methods use reverse-phase C18.

How does HPLC differ from UPLC?

UPLC (Ultra-High-Performance Liquid Chromatography, Waters trademark; UHPLC is the generic term) uses sub-2 μm particles (typically 1.7 μm) requiring back pressures of 1000–1300 bar — versus HPLC's 3–5 μm particles at 200–400 bar. The smaller particles flatten the van Deemter H–u curve, enabling 5–10× faster separations at equal resolution: a 5-minute UHPLC run replaces a 30-minute HPLC run. The instrument must withstand higher pressure and tighter dead volumes, costing 2–3× more than a standard HPLC. For pharmaceutical QC, UHPLC has overtaken HPLC in new method development since around 2010; legacy methods stay on HPLC for regulatory continuity under ICH Q2 method-validation rules.

What is a theoretical plate and what is N for a typical column?

Theoretical plate count N is a measure of column efficiency: N = 16 (t_r/w)², where t_r is retention time and w is peak width at base. It comes from a fictitious model treating the column as N stages of equilibrium between mobile and stationary phases — Martin and Synge's 1941 plate theory (Nobel 1952). A standard 4.6×150 mm column at 3 μm gives N ≈ 50,000; a 4.6×250 mm column at 5 μm gives ~25,000; a 2.1×100 mm UHPLC column at 1.7 μm gives ~30,000. Higher N means narrower peaks and more resolved compounds in a given retention window. Plate height H = L/N (typical 3 μm) is the figure of merit being optimized.

Why is HPLC the workhorse of pharmaceutical QC?

ICH guidelines Q2(R2) on method validation, Q3A/Q3B on impurities, and Q6A on specifications all require chromatographic assay of active pharmaceutical ingredients and related impurities at sub-0.1% levels. HPLC delivers the required selectivity, sensitivity, and reproducibility. A typical drug product release assay quantifies the API and 5–15 specified impurities by reverse-phase HPLC-UV in 30 minutes; method validation per ICH involves linearity over 50–150% of nominal, accuracy 98–102%, repeatability RSD <2%, and intermediate precision RSD <3%. Sildenafil purity by USP method, for instance, runs C18 with a 0.05 M ammonium acetate / acetonitrile gradient, UV at 290 nm, with related substances quantified at 0.05% reporting threshold.

What are van Deemter and rate theory in HPLC?

The van Deemter equation H = A + B/u + C u relates plate height H to mobile-phase linear velocity u via three contributions. A (eddy diffusion) — irregular flow paths through packed bed; reduced by smaller, more uniform particles. B (longitudinal diffusion) — analyte spreading by random diffusion in the mobile phase, dominant at low flow. C (mass-transfer resistance) — finite time for analytes to equilibrate between phases, dominant at high flow. The plot of H vs. u is bowl-shaped with a minimum at the optimum velocity u_opt, where the column delivers its highest plate count. Calvin Giddings extended van Deemter into a more general rate theory in the 1960s, treating chromatography as a stochastic transport problem and predicting band-broadening from first principles. Sub-2 μm UHPLC particles flatten C, making u_opt larger and enabling fast runs.