Gradient vs isocratic: when to use each?

The two options in one sentence

In reversed-phase (RP) chromatography you have two elution modes:

• Isocratic — the mobile-phase composition (% organic) does not change during the run. For example: 60 % MeCN for the full 15 minutes.
• Gradient — the composition increases progressively. For example: start at 5 % MeCN, reach 95 % MeCN in 20 minutes.

Snyder's 2× rule

Lloyd Snyder proposed a simple rule to choose between the two modes:

If the isocratic chromatogram has a k' range > 10 (i.e. the last peak has a k' more than 10 times that of the first), use a gradient. If the range is below 10, use isocratic.

In day-to-day lab practice this translates into the 2× rule: if the last peak elutes before twice the retention time of the first peak, isocratic is enough.

Why the rule works

Under isocratic conditions, peaks broaden in proportion to their tR. Very late peaks become wide and blend into the noise. If tR(last) > 2× tR(first), the method is already too long and the final peaks lose sensitivity — a gradient fixes both problems.

When to USE isocratic

Typical cases:

• QC analysis of a pure product: 1–3 analytes of similar polarity, a method you run every day. Isocratic is simpler, more stable and lets you re-inject without re-equilibration.
• Quantifying a single active + known impurities: if all impurities are similar to the active (not very polar nor very lipophilic), isocratic separates them well.
• Fixed compendial methods: many pharmacopoeias (USP, EP, JP) prescribe isocratic for identity and purity testing because of its simplicity.
• Detectors that tolerate composition changes poorly: the RID (refractive index) is essentially incompatible with a gradient — the baseline becomes unusable.

Concrete advantages of isocratic

When to USE a gradient

Typical cases:

• Multi-residue screening: 30+ pesticides, 100+ PPCPs, metabolites. A very wide polarity range → a gradient is mandatory.
• Complex-sample analysis: plant extracts, biological fluids, natural products. You take what you get and you need to elute everything.
• Developing a new method: even if you will end up isocratic, start with a gradient (5 → 95 % B) to "see" which analytes you have and where they elute.
• Very late peaks under isocratic: if your last peak comes out at 45 minutes isocratically, a gradient brings it to 12–15.

Concrete advantages of a gradient

Disadvantages of a gradient

Before jumping to a gradient "just in case", consider what you lose:

• Re-equilibration time: typically 5–10 column volumes (~2–5 extra minutes per run).
• Lower reproducibility: any deviation in gradient formation (pump mixing, system delay) shifts the tR.
• Does not work with RID: a composition change → a refractive-index change → an unusable baseline.
• More expensive in consumables: you use ~1 L of organic per day instead of ~0.3 L.
• Harder to transfer: two LC systems with different delay volumes produce different profiles from the same nominal program.

The complete decision tree

Follow this mental order when designing a new method:

1. Do you know the analytes and their log P? If they all sit within a log P range of ±1, try isocratic first.
2. What is the detector? If it is RID or conductivity, go isocratic (a gradient is impractical).
3. Is the method repetitive (QC) or exploratory? QC → isocratic if it works; exploratory → gradient.
4. Run a diagnostic gradient of 5 → 95 % B in 20 min. Look at where the analytes elute:
   • All between 30 and 60 % B → isocratic at 40–45 % will work
   • Spread over a wide range → gradient
   • All at the start → a weaker phase (HILIC, normal phase)
   • All at the end → less retention (C8 or a higher composition)

Gradient slope: the 2–10 % B/min rule

Once you decide you need a gradient, the slope affects resolution and duration. A practical rule:

• For better resolution: 2–3 % B per minute (slower)
• For high resolution and screening: 4–7 % B per minute (typical)
• For fast screening: 8–15 % B per minute (sacrifices Rs)

The slope changes selectivity, not just time: two analytes can swap their elution order depending on the slope. If you are going to optimize selectivity, try two different slopes.

Snyder's Φ (phi) parameter

Underneath the model lies the LSS (Linear Solvent Strength) equation:

log k' = log k'_w − S · φ

where φ is the organic fraction (0 to 1), k'_w is the theoretical retention factor in pure water, and S is the "strength" of the organic (analyte-dependent, typically 4–10 for ACN in RP).

In a gradient, φ increases linearly with time, and the effective retention results from integrating this equation. That is why gradient peaks have more uniform widths — the elution "strength" rises just as the analyte starts to elute.

How to decide it in PureAnalyt

In the PureAnalyt simulator you can test both modes without spending reagent:

• Load a sample (e.g. organic acids, pesticides, vitamins).
• Configure column + detector + % organic.
• Inject first in isocratic with a mid % B. Check whether the method meets the 2× rule.
• If not, switch to gradient and try different slopes. Compare total time, Rs between peaks and peak width.
• The simulator shows k' in the table — compute your own rule: k'(last) / k'(first) > 10 → you need a gradient.

Common mistakes

1. Jumping to a gradient without trying isocratic: you add complexity with no gain. Most QC methods are isocratic for a reason.
2. A very steep gradient "to finish fast": compressed peaks, poor Rs — you save 3 minutes but lose quality.
3. Not re-equilibrating between runs: 5–10 volumes minimum. Skip this and the first analyte's tR of the day is irreproducible.
4. Forgetting the delay volume: your program says "5 → 95 % in 15 min" but the analyte sees something slightly different for a few minutes (mixer dead volume).