Chromatography

Core separation physics: why chromatography works

Chromatography separates molecules by differential partitioning between a mobile phase and a stationary phase. For liquid chromatography, the most useful engineering model is the van Deemter equation, which describes the height equivalent to a theoretical plate, \( H \), as a function of mobile-phase linear velocity, \( u \):

\[

H = A + \frac{B}{u} + C u

\]

• \( A \) is the eddy-diffusion term, driven by flow path inhomogeneity.
• \( B \) is longitudinal molecular diffusion.
• \( C \) is resistance to mass transfer between mobile and stationary phases.

The van Deemter equation is not a marketing construct; it is the foundational 1956 Chemical Engineering Science paper by van Deemter, Zuiderweg, and Klinkenberg [S1]. The paper establishes that minimizing \( H \) maximizes column efficiency \( N = L/H \) for a column of length \( L \). The same framework also predicts that smaller stationary-phase particles reduce the \( C u \) and \( A \) penalties at higher linear velocities, which is the technical rationale for sub-2-µm UHPLC particles.

Why this matters for the technology thesis. A separation that is currently run on a 5 µm particle column can, in principle, be accelerated by moving to smaller particles, provided the system can deliver the required higher pressure. This is the basis for the HPLC-to-UHPLC transition. However, the van Deemter model does not include frictional heating, viscous dissipation, or pressure-dependent retention effects; those are material limitations discussed in Section 4 and were analyzed experimentally by Martin and Guiochon [S3].

Analytical chromatography: HPLC/UHPLC platform claims and limitations

2.1 The UHPLC hardware claim

Waters Corporation’s ACQUITY UPLC product family is a commonly cited UHPLC platform. The technical claim is that sub-2-µm particles and a high-pressure fluidic system improve separation speed and resolution. A granted US patent assigned to Waters Investments Limited, US 7,258,800 B2 (granted 2007-08-21), records claims directed to high-pressure liquid chromatography apparatus [S6]. That patent is a Tier 1 document for what was claimed, not for an independent demonstration that all marketed UHPLC columns achieve a specific plate count in all users’ hands.

Waters’ product page describes UPLC as enabling faster separations than conventional HPLC; that product language is self-reported Tier 2 material [S11]. The SEC 10-K does not state the same numerical speed/resolution promise as an audited performance obligation; rather, Waters’ Item 1A risk factors discuss product-development risk and competitive pressure [S7]. This is a typical marketing-versus-filed-disclosure gap: marketing can state aspirational productivity, while the filed disclosure frames technological change as a risk.

Source-tier check on the UHPLC performance claim. The patent [S6] plus the Waters product page [S11] are the same corporate lineage. The independent physical limit is in the peer-reviewed Martin and Guiochon study [S3], which is a separate institution. Therefore the high-pressure capability claim can be cross-verified; the specific throughput claim remains [SINGLE-SOURCE] if it relies only on Waters’ product language.

2.2 Pharmacopeial and regulatory standardization

Chromatography is embedded in regulated analytical chemistry. USP General Chapter <621> defines chromatographic system parameters such as resolution, tailing factor, and repeatability [S5]. ICH Q2(R2), adopted 2023-11-01, is the current international guidance for validating analytical procedures and explicitly addresses chromatographic procedures through validation characteristics such as specificity, linearity, range, accuracy, precision, and robustness [S4]. The Q2(R2) adoption is a regulatory catalyst because it refreshes, but does not eliminate, the analytical method-validation burden for chromatographic methods.

These Tier 1 regulatory/standards documents do not specify a particular vendor’s instrument. They set the legal/scientific framework that sustains demand for validated chromatographic systems and columns.

Preparative/affinity chromatography: protein A capture in bioprocessing

For biopharmaceuticals, preparative chromatography is not merely analytical; it is a manufacturing unit operation. Shukla et al., in a peer-reviewed Journal of Chromatography B paper, concluded that Protein A affinity chromatography is the platform capture step for monoclonal antibody purification [S2]. The paper quantifies dynamic binding capacities for Protein A resins in the range of approximately 30–50 g/L of immunoglobulin G and describes a platform downstream process of Protein A capture followed by ion exchange and/or other polishing steps [S2].

This is a Tier 1 source. Repligen is a US-listed supplier of Protein A ligands/resins; Repligen’s FY2024 Form 10-K discloses the company as a supplier of bioprocessing consumables, including chromatography ligands/resins [S10]. That 10-K is Tier 2 adoption context, not independent proof of the specific capacity figure from Shukla et al.

3.1 Worked preparative capacity calculation

The Protein A dynamic binding capacity range is a good worked example because it determines equipment sizing and cost.

Inputs:

Calculation:

\[

\text{mAb mass} = 2{,}000\ \text{L} \times 5\ \text{g/L} = 10{,}000\ \text{g}

\]

\[

\text{Resin required} = \frac{10{,}000\ \text{g}}{35\ \text{g/L}} = 286\ \text{L}

\]

Sensitivity:

• If titer falls to 4 g/L: \( 8{,}000/35 = 229 \) L resin.
• If titer rises to 6 g/L: \( 12{,}000/35 = 343 \) L resin.
• If dynamic binding capacity is 30 g/L instead of 35 g/L: \( 10{,}000/30 = 333 \) L resin.
• If dynamic binding capacity is 50 g/L instead of 35 g/L: \( 10{,}000/50 = 200 \) L resin.

Interpretation: within the reported capacity range, the resin-volume requirement changes by roughly a factor of 1.7 between the low and high DBC values. That sensitivity is why suppliers compete on ligand capacity and why the exact DBC figure must be source-bound, not assumed from marketing.

Worked van Deemter calculation

The most frequently cited engineering calculation in chromatography is the relationship between particle diameter, reduced plate height, and theoretical plate count.

Formula:

\[

H = h d_p

\]

\[

N = \frac{L}{H} = \frac{L}{h d_p}

\]

Where:

• \( h \) is the reduced plate height.
• \( d_p \) is the particle diameter.
• \( L \) is column length.

Inputs for a sub-2-µm column:

Calculation:

\[

H = 2.0 \times 1.7\ \mu\text{m} = 3.4\ \mu\text{m}

\]

\[

N = \frac{5\ \text{cm}}{3.4\ \mu\text{m}}

= \frac{50{,}000\ \mu\text{m}}{3.4\ \mu\text{m}}

= 14{,}706

\]

Sensitivity:

This calculation shows that a plausible 25% reduction in \( h \), from 2.0 to 1.5, increases plate count by one-third; a 25% increase in \( h \), from 2.0 to 2.5, decreases plate count by one-fifth. The calculation uses the assumption that the column is operated near its optimal reduced velocity. It does not include pressure/temperature effects, which is exactly where Martin and Guiochon’s work becomes important [S3].

Source validation and contradictions

5.1 Source categories present

This note uses three or more authoritative source categories:

• Tier 1 peer-reviewed papers: van Deemter et al. [S1], Shukla et al. [S2], Martin and Guiochon [S3].
• Tier 1 standards/regulator technical documents: USP General Chapter <621> [S5], ICH Q2(R2) [S4].
• Tier 1 granted patent: US 7,258,800 B2 [S6].
• Tier 2 company filings: Waters [S7], Thermo Fisher [S8], Agilent [S9], Repligen [S10].
• Tier 2 company technical/product pages: Waters [S11], Thermo Fisher [S12].

5.2 Cross-verification status

• The van Deemter framework [S1] is cross-verified by its use in regulatory/standards documents, but the specific worked plate-count calculation here is derived from first principles and the self-reported particle diameter [S11]; the reduced plate height value is conventional and is marked derived/assumed.
• The Protein A dynamic binding capacity range [S2] is [SINGLE-SOURCE] in this note; no independent second institution is cited for the same numeric range. Repligen’s 10-K confirms commercial participation but does not independently verify the 30–50 g/L figure.
• The UHPLC high-pressure claim has one corporate patent [S6] plus one vendor product page [S11], which is a single lineage. The independent physical tension comes from Martin and Guiochon [S3]. Therefore the patent is evidence of recorded claims, not proof of universal performance.

5.3 Contradiction: speed/efficiency vs. viscous heating at high pressure

The van Deemter model predicts smaller particles allow faster separations at high linear velocity. Martin and Guiochon reported that at high pressures, frictional heating from solvent flow creates radial temperature gradients, which can reduce efficiency and counteract the expected benefit of very small particles [S3]. Their peer-reviewed work is an independent institution from Waters and is a technical contradiction to uncritical “smaller particles always better” marketing. The implication is that UHPLC instrument design must manage thermal effects, and that maximum pressure alone is not a guarantee of improved separation performance.

5.4 Patent claim vs. filed disclosure

US 7,258,800 B2 is a granted patent assigned to Waters Investments Limited [S6]. It records what the assignee claimed as invention. Waters’ FY2024 10-K does not state that this patent, or any single patent, is responsible for a specific revenue amount, nor does it promise a particular separation speed to customers [S7]. The 10-K instead discusses risks such as technological change, competition, and supply chain interruptions. Thus the patent supports an IP-moat existence claim, but not a commercial-performance moat-quality claim.

Parameter table

All web sources accessed 2026-04-26, except where publication/grant dates are marked.

Company filings cross-reference

Waters Corporation (Form 10-K) [S7]. Waters’ filing describes a business in which instrument placements generate downstream demand for columns, chemistries, and services. That supports the “razor/razor-blade” model, but the filing does not provide a per-column margin or a portable separation-speed guarantee. The risk factors include product-development and competition risk, which is the appropriate filed-disclosure caution against treating marketing performance claims as financial commitments.

Thermo Fisher Scientific (Form 10-K) [S8]. Thermo Fisher’s filing includes chromatography within its analytical instruments/lab products context. This is Tier 2 adoption context: it shows chromatography is a commercial product line at a large life-science-tools company, not an independent technical validation.

Agilent Technologies (Form 10-K) [S9]. Agilent discloses liquid chromatography among analytical platforms and names pharmaceutical, food, environmental, and applied markets as end-markets. This provides customer-class evidence, but no independent plate-count or capacity numbers.

Repligen Corporation (Form 10-K) [S10]. Repligen’s filing supports that Protein A ligands/resins are a commercial product class in downstream bioprocessing. It does not provide a cross-verification of the Shukla et al. 30–50 g/L capacity range; therefore that range remains [SINGLE-SOURCE] from the peer-reviewed paper.

Moats, patents, and feasibility

A granted patent is evidence of recorded claims, not proof of performance or commercial viability. US 7,258,800 B2 [S6] is an example of an instrument platform patent; it supports the claim that Waters recorded intellectual property for high-pressure chromatography, but it does not independently demonstrate that every ACQUITY UPLC column produces the calculated 14,706 plates. For performance, the source chain must include an independent measurement; here the independent measurement for high-pressure effects is Martin and Guiochon [S3], which actually shows a performance limit.

The durable economic moat in chromatography is not a single patent; it is the combination of validated methods in regulated labs, pharmacopeial standards [S5][S4], and the installed-base consumables stream that appears in company filings [S7][S8][S9]. Protein A capture has an additional biological specificity moat, supported by the platform-process role described in Shukla et al. [S2] and by Repligen’s commercial focus [S10].

Quantitative worked example conclusion

The two worked calculations demonstrate two different chromatography economic drivers:

1. Analytical UHPLC plate-count calculation: at a 5 cm column length and 1.7 µm particles, a reduced plate height of 2.0 gives about 14,700 theoretical plates. This is the analytical-resolution driver.
2. Preparative Protein A capacity calculation: a 2,000 L batch at 5 g/L mAb needs about 286 L of Protein A resin at a 35 g/L DBC midpoint. This is the manufacturing-capacity driver.

The sensitivity analysis shows that analytical performance is moderately sensitive to reduced plate height, while preparative resin demand is strongly sensitive to dynamic binding capacity. Both are source-bound; neither is a free-standing performance guarantee.

Unit economics / business model

Chromatography is primarily sold as a razor/razor-blade model. The instrument sale is the “razor”; columns, resins, fittings, reagents, software subscriptions, and service contracts are the recurring “blades.” Waters’ 10-K presentation confirms that recurring revenue from consumables and service is material to the business [S7]. Thermo Fisher and Agilent similarly disclose chromatography within broader analytical-instruments businesses that monetize instrument placements plus aftermarket demand [S8][S9].

In bioprocessing, the economic model is preparation/consumables plus resin replacement: Protein A resins are high-value consumables, and Repligen monetizes ligands/resins sold to biologics manufacturers [S10]. Value capture sits with:

• Instrument vendors for analytical UHPLC/HPLC.
• Resin/ligand suppliers for preparative affinity chromatography.
• Regulated end-customers who lock in validated methods and resist switching.

No per-unit gross margin percentage is asserted here because no exact margin was retrieved from the cited sources; any such figure would need primary source validation from the relevant 10-K line items.

Disconfirming Evidence

The thesis that chromatography is a foundational, defensible, recurring-revenue technology would be materially weakened if:

1. Analytical separation shifts away from chromatography. If regulatory laboratories adopt direct mass spectrometry, biosensors, or sequencing-based methods without a chromatographic front end, demand for analytical HPLC/UHPLC would decline.
2. Preparative Protein A is replaced at commercial scale. If non-chromatographic capture technologies such as precipitation, crystallization, or membrane adsorbers replace Protein A in commercial monoclonal antibody manufacturing, the preparative column/resin revenue stream would erode.
3. UHPLC patent and consumable moats fail. If generic column manufacturers demonstrate equivalent sub-2-µm performance at scale and regulated customers switch, the razor/razor-blade model would weaken.
4. Regulatory requirements relax or change. If method-validation requirements under ICH Q2(R2) or pharmacopeial chapters such as USP <621> become materially less prescriptive, the lock-in associated with validated chromatography could diminish.
5. High-pressure limitations become binding. If independent studies replicate Martin and Guiochon’s finding that ultra-high-pressure operation produces efficiency losses so severe that UHPLC advantages disappear at commercial relevant pressures, the technological-premium thesis weakens [S3].