Vitamin analysis by hybrid SFC/UHPLC: the complementary strategy

The problem: two chemical universes in one sample

Vitamins are a family of molecules that share only one thing — they are all essential nutrients — but are completely different in their chemistry:

• Fat-soluble (A, D, E, K) — very hydrophobic molecules (log P between 5 and 14), insoluble in water, strongly retained by any RP column and barely retained by HILIC.
• Water-soluble (B-complex, C) — polar or charged molecules (log P between −4 and +0.5) that barely retain on conventional RP but retain well on HILIC or by ionic interaction.

If you try to separate both families with a single LC method, you find that the water-soluble ones almost all elute in the void volume, and the fat-soluble ones need 30+ minutes with an extreme gradient. You have to pick a family or accept a poor compromise.

The hybrid SFC/UHPLC idea

Supercritical Fluid Chromatography (SFC) uses supercritical CO₂ (at ~80 bar and 60 °C) as its main mobile phase, modified with a polar organic solvent (typically methanol). The result is a continuously tunable-polarity mobile phase, ideal for lipophilic compounds that do not dissolve well in aqueous systems.

A hybrid system connects an SFC pump and a UHPLC pump to the same autosampler and the same detector, switching between the two modes with a valve. This lets you:

• Inject a sample and separate it in SFC for the fat-soluble vitamins
• Re-inject the same sample and separate it in UHPLC for the water-soluble ones
• Detect them all with the same mass spectrometer

That is exactly what Edgar Naegele's group at Agilent demonstrated in their application note.

Naegele's experiment (Agilent 5991-9192EN)

Naegele used an Agilent 1260 Infinity II SFC/UHPLC Hybrid System connected to an Agilent 6470A Triple Quadrupole LC/MS. The experimental conditions discussed below are those published in the note.

SFC mode · 7 fat-soluble vitamins

• Column: Agilent ZORBAX SB-C18 3.0 × 100 mm, 1.8 µm (p/n 828975-302)
• Mobile phase: CO₂ + methanol as modifier
• Gradient: 1 → 3 → 15 % B in 4 min
• Flow: 1.5 mL/min · BPR 200 bar at 60 °C · 40 °C column
• Injection: 5 µL

All 7 vitamins elute within ~3 minutes:

What caught our attention was the separation of D2/D3 with Δ tR = 0.038 min — a classic critical pair in vitamin D analysis, and this method resolves them in under 3 minutes. All LOQ values are below 15 ppb, enough to quantify vitamins in fortified foods.

UHPLC mode · 8 water-soluble B-complex vitamins

• Column: Agilent ZORBAX SB-Aq 3.0 × 100 mm, 1.8 µm (p/n 828975-314)
• Mobile phase A: Water + 5 mM ammonium formate + 0.1 % formic acid
• Mobile phase B: Methanol + 0.1 % formic acid
• Gradient: 1 → 70 % B in 7 min
• Flow: 0.7 mL/min · 40 °C · 5 µL injection

The 8 water-soluble vitamins elute in under 5.2 minutes:

Pyridoxamine and thiamine elute almost together (Δ tR = 0.037 min) near the void volume. The two forms of B3 (nicotinic acid and nicotinamide) are well separated (Δ ≈ 0.58 min) — not trivial given their structural similarity. The final pair, folic acid / cyanocobalamin, comes out close together but distinguishable.

Why the column choice matters: ZORBAX SB

Naegele specifically chose the ZORBAX Stable Bond (SB) family — and the choice is not incidental. SB columns have one key feature: they use bulky di-isobutyl groups around the silane bond, which sterically protect the silane from acid attack. The result: they are stable down to pH 1.0, ideal for mobile phases with 0.1 % formic acid (pH ~2.7) without fear of hydrolysis.

The SB-Aq variant adds a specific endcap that prevents "dewetting" (phase collapse) when the gradient starts at 99 % aqueous. That lets you analyze polar compounds compatible with B vitamins without losing retention.

Triple-quadrupole configuration

The detector was an Agilent 6470A QqQ with Agilent Jet Stream. There are two sets of source parameters — one for SFC and one for UHPLC — because the mobile phase reaching the MS is radically different:

All vitamins are detected as [M+H]⁺ except the two forms of Vitamin A (retinol and retinyl palmitate), which ionize as [M+H−H₂O]⁺ at m/z 269. This is a classic water loss in retinoids, a feature that lets you identify them unambiguously.

Practical takeaways

Beyond the elegance of the hybrid system, three takeaways apply even if you do NOT have access to SFC:

1. For fat-soluble vitamins, RP-UHPLC works if you extend the gradient

If you only have UHPLC, you can analyze A, D, E, K with a longer gradient (15–25 min instead of 4) using MeCN/IPA/THF as a strong modifier. You lose speed but gain independence from SFC. The D2/D3 selectivity remains the bottleneck.

2. For water-soluble vitamins, ZORBAX SB-Aq is excellent without HILIC

Traditionally B vitamins were separated by HILIC. Naegele's experiment shows that RP with SB-Aq also works very well with 0.1 % formic acid — you avoid HILIC's incompatibility with aqueous sample solvents and get better tR stability. If reproducibility worries you, RP on SB-Aq is more robust than HILIC.

3. The [M+H−H₂O]⁺ water loss is a general pattern in retinoids

Any compound with an allylic alcohol group (retinol, retinyl esters, calciferols) tends to lose H₂O in the ESI source before fragmenting. Design your MRMs with that loss in mind: the most intense precursor is usually [M+H−H₂O]⁺, not [M+H]⁺.

How to reproduce this in PureAnalyt

In the PureAnalyt simulator we have already added the compounds and the SB-Aq column mentioned in this note:

• Sample: in the selector choose "Water-soluble vitamins → Full B complex (8 vitamins)"
• Column: Agilent ZORBAX SB-Aq 1.8 µm RRHD — 1200 bar · 100% aqueous
• Mobile phase: H₂O + 5 mM ammonium acetate / MeOH + 0.1 % formic acid
• Gradient: 1 → 70 % B in 7 min
• Detector: LC-MS/MS or LC-MS with a 100–700 m/z scan
• The preset "Fat-soluble vitamins (A, D2, D3, E, K1, K2, retinyl palmitate)" is also available to experiment with high-log P compounds on conventional RP.

The simulator does not implement SFC (yet), but you can approximate the UHPLC method at 95 % fidelity.

Limitations and references

The original Agilent note contains far more detail than we could summarize here: the verbatim MRMs (with specific collision energies and fragmentors), the calibration plots for each vitamin, precision data over n = 10 injections, and detailed chemical structures of the 15 compounds. For a serious analysis of the method, download the full note from Agilent's site. The work is by Edgar Naegele and the intellectual property of the method belongs to Agilent Technologies.

Credits and reference

All the technical content you see here is based on:

Naegele, E. (2018). Analysis of Vitamins Using an SFC/UHPLC Hybrid System with a Triple Quadrupole LC/MS for Quantification. Agilent Application Note 5991-9192EN. Agilent Technologies, Inc.

PureAnalyt drew from this note: (i) the ZORBAX Stable Bond family now documented in our column database, (ii) seven vitamins that were missing from our analyte catalogue (pyridoxamine, nicotinamide, retinyl palmitate, ergocalciferol, cholecalciferol, phylloquinone, menaquinone), and (iii) two sample presets that reproduce the two experimental mixtures.

Thanks to Agilent Technologies for publishing application notes with enough methodological detail to replicate and teach the experiment. It is a model of open scientific documentation that benefits the whole analytical community.