The myth that "all C18 columns are the same"

If you have worked in chromatography you know that switching from one C18 column to another C18 can give a completely different separation. The same isocratic method on two nominally equivalent C18 columns can flip your elution order, change Rs, or coelute what used to be well resolved.

Why? Modern C18 columns differ in more than the C18 chain. They differ in:

  • Silica type (Type-A vs Type-B vs hybrid BEH, CSH)
  • Bonding density (how many C18 chains per nm²)
  • Endcapping (how the residual silanols are "capped")
  • Embedded polar groups (in polar-embedded phases)
  • Surface charge (CSH, "charged surface")
  • Pore size (80, 130, 200, 300 Å)

Each of these variables changes selectivity in an unpredictable way. The HSM brought order to this chaos in the 1990s–2000s.

The Hydrophobic Subtraction Model

Proposed by Lloyd Snyder and John Dolan, the HSM describes the relative retention factor of any analyte on any RP column as:

log α = η'·H − σ'·S* + β'·A + α'·B + κ'·C

where:

ParameterColumn propertyPhysical meaning
HHydrophobicityHow strongly the phase interacts with neutral alkyl chains. Standard = 1.0.
S*Resistance to bulky moleculesHow much it rejects "bulky" molecules (steroids, anthracene).
ASilanol acidityHow many active silanols are available to interact with H acceptors.
BBasicity / silanol activityInteraction with bases (protonated amines). The most variable term between phases.
CIonic selectivityIon-ion interaction (salts, residual charges). pH-dependent.

Typical values for a "standard" C18

As a reference, an "average" Type-B C18 has:

  • H = 1.00 (defined this way)
  • S* = 0.00
  • A = 0.00
  • B = 0.00
  • C (pH 2.8) = 0.25
  • C (pH 7.0) = 0.40

The Fs factor — the "how similar are they" metric

The genuinely useful thing about the HSM is comparing two columns. Snyder defined the Fs factor as:

Fs = √[(12.5·ΔH)² + (100·ΔS*)² + (30·ΔA)² + (143·ΔB)² + (83·ΔC)²]

where Δ is the difference between the two columns' parameters. The coefficients (12.5, 100, 30, 143, 83) weight each term by how much it affects the separation.

Practical interpretation

FsInterpretationWhen to use
0–3EquivalentDirect method replacement
3–10SimilarReplacement possible, validate
10–25DifferentMinor selectivity change
25–50Quite differentUseful for fine-tuning
50+Orthogonal2D-LC, purity confirmation

Case 1 · Finding an equivalent replacement

Situation: your validated method uses a column that is no longer sold (discontinued). You need one with Fs < 3 that gives the same separation.

Strategy:

  1. Find the HSM parameters of your original column.
  2. Search the USP database (or the PureAnalyt simulator) for columns with Fs < 3.
  3. Confirm that the dimension and particle size match.
  4. Inject a validation sample and check Rs and tR.

A typical example

If your method ran on a column with parameters roughly H=1.0, A=0.04, B=0.02, C=0.30 and you want to replace it, you might find:

  • A same-vendor hybrid C18 — Fs ≈ 2, equivalent.
  • A standard Type-B C18 — Fs ≈ 4, similar.
  • A different-vendor Type-B C18 — Fs ≈ 8, changes to validate.

Case 2 · Finding an ORTHOGONAL column

Situation: you have a peak you suspect of coelution (is a peak really pure, or are two analytes coeluting?). You need to confirm with a column of opposite selectivity.

Strategy:

  1. Take the HSM parameters of your current column.
  2. Look for one with Fs > 50, ideally with a very different A or B.
  3. Inject the same sample on both columns.
  4. If the peak stays pure in both separations → it is pure.
  5. If the orthogonal run shows 2 peaks → your original peak was a coelution.

Typical orthogonal pairs

  • C18 vs PFP (pentafluorophenyl) — high Fs, a completely different mechanism (π-π on PFP)
  • C18 vs Phenyl-Hexyl — aromatic vs aliphatic selectivity
  • Standard C18 vs polar-embedded RP — Fs ≈ 60+, different behaviour for bases
  • C18 vs CN (cyano) — different retention strength, very different A and B

The pH trick in the C parameter

The C parameter (ionic selectivity) changes with the mobile-phase pH. That is why HSM tables give two values: C (pH 2.8) and C (pH 7.0).

  • At low pH (2.8): silanols are protonated, C is low, and ionic interactions are weak. Good for protonated bases (amines).
  • At neutral pH (7.0): silanols are ionized, C is high, and ionic interactions are strong. Amines are retained more and can tail from active silanols.

For pharmaceutical samples with bases, choose columns with a low C at pH 7 (typically "high-pH stable" or "charged surface" phases).

The practical C18 "families"

To make the choice easier, modern C18 columns group into families:

1. Standard Type-B C18

H ≈ 1.0, low A and B, low C(7). Good for general use and neutral compounds.

2. Charged-surface C18 (CSH)

Designed for pharmaceutical bases: the surface charge repels protonated amines, avoiding silanol tailing. Very low C(2.8).

3. Hybrid C18 (BEH)

Hybrid (ethylene-bridged) silica, stable at pH 1–12. Lets you work at high pH where bases are deprotonated → no tailing.

4. Polar-embedded C18

A polar group (urea or carbamate) embedded near the silica → orthogonal behaviour to a standard C18. Ideal as the orthogonal column in 2D-LC.

5. PFP (pentafluorophenyl)

Not technically a C18, but it competes. Stronger π-π interactions, a completely different retention of aromatics. A natural orthogonal to C18.

How to use the HSM in PureAnalyt

The PureAnalyt simulator includes an HSM database of 60+ columns and an equivalence advisor that computes Fs in real time:

  1. Select your current column in the rack.
  2. The HSM panel appears with its 5 parameters + 2 C values.
  3. In the "Compare with…" dropdown, pick another column.
  4. The simulator shows the Fs factor and an interpretation.
  5. Inject the same sample on both and compare chromatograms.

This saves you weeks of experimental validation when you are looking for a column replacement or want to confirm orthogonality for 2D-LC.

Limitations of the HSM

The model is very useful but not perfect:

  • It is specific to RP phases under aqueous-organic conditions. It does not apply to HILIC, SEC or IEX.
  • It was calibrated with a defined set of test analytes. Very different compounds can show selectivities the model does not predict.
  • The parameters are stable for a "typical" column from the batch. Lot-to-lot variation can shift the values.
  • Temperature matters. The official parameters are reported at 35 °C.

Practical conclusions

  1. Not all C18 columns are the same — selectivity differences are the rule, not the exception.
  2. Use Fs before testing — it saves time and reagents.
  3. Know 3–4 different C18 families — standard Type-B, CSH, BEH/high-pH, polar-embedded. That covers 95 % of cases.
  4. To confirm purity, use an orthogonal column (Fs > 50) — another nominally equivalent C18 is not enough.
  5. Validate after the switch — Fs < 3 is a good guide but not a full guarantee. Confirm with a sample injection.