HPLC pressure ratings: psi to bar, MPa and what your autosampler needs

To convert HPLC pressures, use 1 bar = 14.5038 psi and 1 MPa = 10 bar = 145.038 psi: divide psi by 14.5 to get bar, multiply bar by 14.5 to get psi. Current HPLC systems are rated for about 345-830 bar (5000-12,000 psi) and UHPLC systems for 1000-1500 bar (about 14,500-22,000 psi). An autosampler needs a rating comfortably above the highest pressure its injection valve sees in your method; as a rule of thumb, keep that pressure at or below 80% of the rating.

Key points

• NIST gives 1 psi = 6894.757 Pa, and 1 bar is exactly 100,000 Pa, so 5000 psi = 344.7 bar and 1000 bar = 14,504 psi.
• Column backpressure rises in proportion to flow, viscosity and column length, and with the inverse square of particle size. Going from 5 µm to 1.7 µm particles multiplies it by 8.7.
• Water/methanol mixtures are far more viscous than either pure solvent, with a maximum near 40-50% methanol; water/acetonitrile peaks much lower. The highest pressure of a gradient is often in the middle.
• Heating the column lowers viscosity and pressure: water is 27% less viscous at 40 °C than at 25 °C.
• Check the pressure at the injection valve, add an allowance for column ageing, and set the pump's pressure limit no higher than the lowest-rated part in the flow path.

How do you convert psi to bar and MPa?

The SI unit is the pascal, and both bar and psi are defined from it. The NIST Guide to the SI lists 1 bar = 1 × 10⁵ Pa (exact) and 1 psi = 6.894 757 × 10³ Pa (NIST). Everything else follows:

bar = psi × 0.0689476      psi = bar × 14.5038
MPa = bar / 10             bar = MPa × 10
MPa = psi × 0.00689476     psi = MPa × 145.038

Worked example: a pump rated 600 bar is 600 × 14.5038 = 8702 psi, and a valve rated 5000 psi is 5000 × 0.0689476 = 344.7 bar, which datasheets round to 345 bar. For mental arithmetic, 1000 psi ≈ 69 bar and 100 bar ≈ 1450 psi.

Conversion table for common HPLC pressure ratings

Examples are vendors' maximum ratings, seen in October 2026; system sources are linked in the system classes section below.

What sets column backpressure?

A packed column is a porous bed, and flow through it follows Darcy's law. The Kozeny-Carman equation (a textbook result) gives the bed's permeability from particle size and packing porosity:

Darcy:          ΔP = η × u × L / K
Kozeny-Carman:  K = dp² × ε³ / (180 × (1 - ε)²)

ΔP pressure drop (Pa), η viscosity (Pa·s), u superficial velocity (m/s) = F / column cross-section,
L column length (m), K permeability (m²), dp particle diameter (m), ε interparticle porosity (about 0.4)

Majors writes the same relationship as ΔP = ηFL / (K⁰ π r² dp²) and notes that "decreasing the particle size by half increases the pressure by a factor of 4" (Majors, LCGC). For everyday estimates, Dolan gives a practical form in lab units (Dolan, LCGC):

P (psi) ≈ 2500 × L × η × F / (dp² × dc²)

L column length (mm), η viscosity (cP = mPa·s), F flow (mL/min),
dp particle size (µm), dc column inner diameter (mm). Divide by 14.5 for bar.

What each parameter does, at otherwise constant conditions:

Worked estimate: 4.6 × 150 mm, 5 µm column at 1 mL/min

With water at 25 °C (0.89 cP):

P = 2500 × 150 × 0.89 × 1.0 / (5² × 4.6²) = 333,750 / 529 = 631 psi ≈ 44 bar

At the methanol/water viscosity maximum (about 1.6 cP) the same column gives 1134 psi, about 78 bar; at the acetonitrile/water maximum (about 1.0 cP), 709 psi or 49 bar. Dolan's own example is consistent: the same column at 2 mL/min with 50:50 methanol/water at 30 °C gives "approximately 2000 psi (140 bar)".

A Kozeny-Carman cross-check with ε = 0.40 and 1.0 cP gives K = 2.47 × 10⁻¹⁴ m², u = 1.0 mm/s and ΔP = 61 bar, against 49 bar from Dolan's equation. Agreement within 25% is as good as such estimates get, because packing density and particle size distribution vary between columns; Dolan notes that "a 0.1-µm difference in a nominally 2-µm particle will make a 10% difference in the calculated pressure".

Worked estimate: 2.1 × 100 mm, 1.7 µm column at 0.4 mL/min

With acetonitrile/water at its viscosity maximum (about 1.0 cP):

P = 2500 × 100 × 1.0 × 0.4 / (1.7² × 2.1²) = 100,000 / 12.74 = 7846 psi ≈ 541 bar

With methanol/water at its maximum (1.6 cP) the estimate is 866 bar, and 1082 bar at 0.5 mL/min, beyond a 1000 bar system. At equal viscosity this column runs at 11 times the pressure of the first example: 1.9 times from flow per cross-section, 0.67 from length and 8.65 from particle size. The system adds more too: with UHPLC, Dolan notes, "narrow-bore tubing and in-line frits can result in 1000 psi or more of pressure in addition to the column".

Why do water/methanol mixtures give the highest pressure?

Mixing water with methanol or acetonitrile gives a mixture more viscous than either pure solvent. Billen et al. measured viscosity across the full composition range of both mixtures at up to 1000 bar and 100 °C (J. Chromatogr. A, 2008). Stoll's summary of those data: for methanol "the maximum viscosity occurs around 50% methanol", for acetonitrile "around 20% acetonitrile", and pure acetonitrile has "about one-third" of the viscosity of water (Stoll, LCGC).

In a methanol gradient from 5% to 95%, viscosity rises from about 0.9 cP to about 1.6 cP near the middle and then falls below 1 cP again, so the maximum pressure comes mid-run, not at injection. And Dolan's rule that "acetonitrile generates approximately 60% of the pressure of methanol" is often the cheapest way to win back margin.

How does temperature change HPLC pressure?

Pressure is proportional to viscosity, and viscosity falls with temperature. As Majors puts it, "with an increase in temperature, column back pressure decreases because of a decrease in η".

Values from the NIST Chemistry WebBook at 0.1 MPa. A method developed at 25 °C without a column oven runs about 13% higher in pressure on a 20 °C morning, and at 40 °C a water-rich mobile phase gives about a quarter less. Heat does not undo the particle-size arithmetic, though: "elevated temperature cannot be used to circumvent the pressures associated with the use of small particles" (Waters).

HPLC vs UHPLC: what are the system classes rated for?

The traditional boundary was about 400 bar. Dolan describes "conventional liquid chromatography (LC) systems (<6000 psi; <400 bar)", and Waters' UPLC primer cites "pressure limitations of conventional HPLC instruments [350-400 bar; 5000-6000 psi]". Today's systems sold as HPLC have moved well above that:

Autosamplers follow the same classes. Agilent's 1260 Multisampler injects "at pressures up to 600 or 800 bar", and Thermo Fisher's Vanquish split samplers are specified at 70 MPa for the C and CT versions, 103 MPa for the FT and 151 MPa for the HT (Thermo Fisher). Columns have ratings too: Agilent's ZORBAX Eclipse Plus C18, 4.6 × 150 mm, 5 µm is rated 400 bar, its 1.8 µm RRHT columns 600 bar and RRHD columns 1200 bar (Agilent).

How to choose an autosampler pressure rating

1. Find the highest pressure in the method

Use a real pressure trace if you have one: a full gradient, at the lowest column temperature you use, ideally on a column with some hours on it. Otherwise estimate it with Dolan's equation at the most viscous point of the gradient and add the system's own backpressure (capillaries, in-line filter, detector cell).

2. Know which parts of the sampler see that pressure

The injection valve sits between pump and column, so it sees practically the full pump pressure. In flow-through needle samplers, the needle seat "must seal to the maximum system pressure level" and the metering device is in the high-pressure path too (LCGC). In fixed-loop samplers with a low-pressure needle, only the valve and loop are at system pressure. More: how an HPLC autosampler works and the 6-port injection valve.

3. Allow for ageing, cold days and injection transients

Columns and frits collect particles, and labs get colder in winter. At injection, a loop filled at atmospheric pressure joins the high-pressure stream and its contents compress: with a bulk modulus for water of about 2.1 GPa at 10 °C (Engineering ToolBox), a 100 µL loop at 300 bar takes up 100 × 300 × 10⁵ / (2.1 × 10⁹) ≈ 1.4 µL, and pressure dips briefly while the pump makes it up. Thermo Fisher describes its SmartInject technology as minimising "column pressure shock" (Vanquish Neo specification). A rule of thumb that covers all of this: keep the expected maximum at or below about 80% of the sampler's rating.

4. Protect the weakest part with the pump limit

HPLC pumps have an upper pressure limit that stops the flow when it is exceeded. Set it no higher than the lowest-rated component in the high-pressure path, whether that is the column, the autosampler valve or a detector flow cell. A blocked frit then stops the pump instead of overloading a valve seal.

Worked example A: a conventional method

4.6 × 150 mm, 5 µm column, 1.5 mL/min, methanol/water gradient, 25 °C:

1. Column at the viscosity maximum (1.6 cP): 2500 × 150 × 1.6 × 1.5 / (25 × 21.16) = 1701 psi, about 117 bar.
2. Add 10 bar for the system (an assumption; measure yours): 127 bar.
3. Allow 30% for an ageing column (also an assumption): 165 bar.
4. At 80% utilisation the sampler needs at least 165 / 0.8 = 207 bar, about 3000 psi.

A 5000 psi (345 bar) valve has ample room. Set the pump limit no higher than 345 bar, the lower of the valve's rating and a typical 400 bar column rating such as the ZORBAX example above.

Worked example B: the same method on 3 µm particles

The column alone now gives 326 bar at the viscosity maximum (2500 × 150 × 1.6 × 1.5 / (9 × 21.16) = 4726 psi). With the same allowances: 336 bar, then 437 bar, so the sampler should be rated about 546 bar. That calls for a 600 bar class system and sampler. To stay within 345 bar you would change the method: acetonitrile instead of methanol and 1.2 mL/min bring the column to 163 bar and the requirement to about 281 bar.

Where the AS-48 fits

The Elutrolab AS-48 has a 6-port, 2-position valve with a 5000 psi (345 bar) design rating*, and its needle line stays at low pressure. That covers many conventional HPLC methods on 3-5 µm columns, such as example A, but it is not intended for high-pressure UHPLC methods or sub-2 µm columns, and example B already needs a 600 bar class sampler. Values marked * are design targets, to be confirmed by testing (specifications). Related: how much you can inject and autosampler qualification tests.