Chromatography technology and investment research

Technical bottleneck

Physics: why it's hard

The van Deemter equation governs all chromatographic separation:

`

H = A + B/u + C·u

`

where H is plate height (lower = better), u is linear velocity, and:

• A term (eddy diffusion): A ≈ 2λd_p — proportional to particle diameter d_p and packing quality λ. Smaller particles reduce multiple flow path inequalities but require near-perfect packing. Any void or channel in the bed destroys efficiency.
• B term (longitudinal diffusion): B = 2γD_m — analyte diffusion back against the flow direction. Largely independent of particle size.
• C term (mass transfer resistance): C ∝ d_p² — the dominant term. Analyte molecules must diffuse into and out of porous particle pores. Smaller particles reduce diffusion distance dramatically, flattening the van Deemter curve at high velocities. This is why going from 5 μm → 1.7 μm particles enables 5–10× faster separations without resolution loss.

Sub-2 μm particles — the UHPLC revolution: Modern UHPLC uses 1.7–1.9 μm fully porous silica particles (introduced ~2004 with Waters Acquity/UPLC). Minimum plate height H_min approaches ~2d_p — roughly 3.4–3.8 μm for 1.7 μm particles. This means a 50 mm UHPLC column can match the plate count of a 250 mm conventional HPLC column, but at 600–1,500 bar instead of 200–400 bar. The pressure penalty follows Darcy's law:

`

ΔP ∝ η·u·L·(1-ε)² / (ε³·d_p²)

`

Pressure scales as ~1/d_p². Going from 5 μm to 1.7 μm increases pressure ~9× for the same column length and velocity.

The sub-1 μm wall: Below ~1.2 μm particles, pressure exceeds 2,000–3,000 bar, exceeding current pump, valve, and column hardware limits. Frictional heating (power dissipation P ≈ u × ΔP) creates radial temperature gradients — the column center runs hotter than the walls — causing viscosity variations, retention shifts, and peak distortion. This is the fundamental physics ceiling on particle size reduction.

Column packing — the art: Sub-2 μm particles have enormous surface area and strong van der Waals attraction. They aggregate unless suspended in precisely density-matched slurry solvents (~2.2 g/cm³ for silica). Packing is done at 1,000–2,000 bar using automated high-pressure slurry pumps. Even minor inhomogeneity creates trans-column velocity biases (wall effects) that dominate band broadening in narrow-bore columns (2.1 mm ID typical for UHPLC). A single void, channel, or frit clog destroys the column.

Extra-column dispersion: UHPLC peaks are extremely narrow (<1–2 seconds wide, volumes <5 μL). Any band broadening outside the column — connecting tubing (>0.1 mm ID kills performance), autosampler, detector flow cell — can degrade observed plate count by 50% or more. Modern UHPLC systems target <5–10 μL total system dispersion. This is why a great column on a mediocre HPLC system underperforms — the bottleneck has shifted from the column to the instrument.

Economic constraints

• Installed base moat: Once a lab standardizes on a vendor's LC platform, methods, software, and training, switching costs are prohibitive. Waters, Agilent, and Thermo compete on new placements but retain customers for decades.
• Column consumable lock-in: Every column type is application-specific (C18 reversed-phase, HILIC, ion exchange, size exclusion, affinity). A biopharma QC lab may run 10+ different column chemistries. Columns are replaced every 1–4 weeks in high-throughput labs. This is the razor-blade: sell the instrument once, sell columns forever.
• Qualification burden: Regulated labs (pharma QC, clinical diagnostics) must revalidate every method when changing column or instrument vendor. This is the deepest moat — a validated method can lock in a vendor for 5–10 years.

Adoption

Why it matters now

Waters Corporation Q1 2026 (reported May 2026):

• Total revenue: $1.27B (+91% reported, +13% organic legacy/core)
• HPLC global market share ~40%; Infinity III LC platform driving high-single-digit growth
• Absorbing BD Biosciences & Diagnostic Solutions (~$17.5B acquisition) — adds flow cytometry, cell analysis to chromatography core
• Raised FY2026 organic growth guidance to 6.5–8%; mid-teens pharma growth with GLP-1 manufacturing demand
• [SEC] WAT 10-K FY2025

Agilent Technologies Q2 FY2026 (ended April 30, reported May 2026):

• Revenue: $1.83B (+10% reported, +6.3% organic); Non-GAAP EPS $1.49 (+14% YoY)
• LC and GC replacement cycle described by analysts (BofA) as "still in early stages"
• Raised FY2026 guidance: $7.39–7.49B revenue, adj. EPS $6.00–6.10 (7–9% growth)
• [SEC] A 10-K FY2025

Key trends

• GLP-1 manufacturing demand: GLP-1 peptide production requires extensive chromatographic purification (preparative HPLC) — a multi-ton scale demand driver for columns and systems
• Continuous chromatography: Multi-column continuous processes (e.g., simulated moving bed, PCC) replacing batch — higher productivity, lower buffer consumption, enabling continuous bioprocessing
• AI-driven method development: Machine learning predicting optimal column, solvent, and gradient for a given separation — reducing method development from weeks to hours
• 2D-LC: Comprehensive two-dimensional liquid chromatography for complex samples (proteomics, natural products) — combining orthogonal separation mechanisms for peak capacities of 1,000+

Key players

Horizon

• Horizon 1 (0–2yr): LC/GC replacement cycle accelerating (post-destocking); GLP-1 manufacturing pulling preparative chromatography demand; BD Biosciences integration at Waters
• Horizon 2 (3–5yr): Continuous chromatography at manufacturing scale — replacing batch purification in bioprocessing; AI-driven autonomous method development
• Horizon 3: Chip-scale chromatography (microfluidics), clinical diagnostic LC-MS at point of care, sub-1 μm particle columns if hardware catches up

Related Technologies

• Mass Spectrometry — downstream detection; chromatography is the prerequisite separation
• Proteomics — LC-MS/MS is the dominant workflow
• Bioprocessing Consumables — chromatography resins are the single largest consumable category in downstream bioprocessing
• Sequencing — sample prep often involves chromatographic cleanup