A 20 µL injection of mobile phase A, run as a blank on a two-year-old HPLC, produced seven peaks above 0.5 mAU between 3.2 and 11.8 minutes. The method’s reporting threshold was 0.1 mAU, so every one of those peaks would have been integrated as an unknown in a sample run. The column had been installed four days earlier. The ghost peaks were coming from somewhere else, and the blank injection was the only reason we found it before the audit did.
This is the part of lab chromatography ghost peaks, extractables and contamination work that gets skipped in most troubleshooting guides: the source map. Which polymer bleeds which compound, at which temperature, into which solvent. Once you have that map, blank and placebo injections stop being a formality and become a diagnostic instrument with a resolution of one material at a time.
Most labs treat contamination as a hygiene issue. Swab harder, bake longer, buy the “low-bleed” grade. That framing fails because a ghost peak is a specific molecule leaving a specific material under specific conditions. You do not clean your way out of it. You identify the material and either replace it, condition it, or move it downstream of the detector.
The diagnostic frame I use is simple: every ghost peak has a retention time, and every retention time maps to a compound, and every compound maps to a material that contains it as an additive, a monomer, a plasticizer, or a degradation product. The blank injection tells you the peak exists. The material map tells you where to look. The placebo injection tells you whether the sample matrix is masking or amplifying it.
If you are still assembling the basic consumable set for a method, the general reference at Lab Chromatography covers the hardware side. This article is about what those consumables do to your chromatogram after they are installed.
The extractables map: which polymer bleeds what, and at what temperature
Four polymer classes dominate the flow path in a typical HPLC or GC system. Each has a characteristic extractable profile. Learn the profile and you can often name the source from the retention time alone.
Polytetrafluoroethylene (PTFE) and the fluoropolymer family
PTFE is the least extractable of the common flow-path polymers, which is why it is the default for solvent lines and frits. It is not inert. At ambient temperature, PTFE leaches low levels of oligomers and residual processing aids, typically below 0.05 mAU in a 20 µL blank on a 254 nm detector. Above 60 °C, and especially in the presence of amines or high-pH buffers, fluoropolymer extractables climb. I have seen a PTFE-lined transfer line at 80 °C in a 10 mM ammonium bicarbonate mobile phase generate a broad hump at 6.5 minutes that grew by roughly 0.3 mAU per hour of run time.
If your method runs hot and basic, the fluoropolymer is a suspect. A PTFE membrane filter used for mobile phase degassing or sample prep carries the same profile. Filter a blank through it and inject: if new peaks appear, the membrane is the source, not the mobile phase.
Polypropylene and polyethylene
Polypropylene is the workhorse for autosampler vials, pipette tips, and sample cups. It is also the most common single source of ghost peaks in my experience, because it is used in the highest numbers and replaced most casually. The extractable profile is dominated by erucamide (a slip agent), Irganox 1076 and 1010 (antioxidants), and low-molecular-weight oligomers. These elute as a cluster of peaks in the 8 to 20 minute window on a C18 at 1 mL/min with acetonitrile gradient, and they are temperature-sensitive: a polypropylene vial stored at 4 °C for 24 hours releases far less than the same vial held at 40 °C in an autosampler tray for the same period.
Polyethylene, used in caps and some tubing, bleeds similar antioxidants plus residual catalyst residues. The peak pattern overlaps polypropylene enough that you usually cannot distinguish them by retention time alone. You distinguish them by which component you swap.
Silicone and the septum problem
Silicone septa are the classic GC ghost-peak source, and the mechanism is well documented: siloxane oligomers (D3 through D10) elute as a repeating ladder at roughly 100 °C intervals on a nonpolar column. On an HPLC autosampler, the silicone is in the septum of the vial cap, and the extractable profile is different: silicone oil and curing agents that appear as a single broad peak or a small cluster depending on the solvent.
The temperature threshold matters. Silicone septa are rated to about 200 °C for continuous use in GC inlets, but bleed rises sharply above 250 °C and the septum should be conditioned at the method’s inlet temperature for at least 30 minutes before use. A septum that was never conditioned will ghost-peak for the first ten injections and then settle. That pattern (peaks that decay with injection count) is diagnostic of an unconditioned septum, not a contaminated column.
Polyetheretherketone (PEEK) and the tubing question
PEEK is used for fittings, ferrules, and some tubing. It is chemically resistant but not extractable-free: residual monomers and processing residues appear at low levels, and PEEK degrades in concentrated nitric acid and some halogenated solvents. At normal HPLC temperatures (below 50 °C) the contribution is small. At elevated temperature with aggressive mobile phases, it is not.
Temperature is the variable that turns a quiet material into a ghost peak source
The single most useful number in this whole exercise is the temperature at which a material’s extractable release rate roughly doubles. For most polyolefins and fluoropolymers, that doubling happens every 10 to 15 °C above ambient. A vial that contributes 0.02 mAU at 20 °C contributes 0.1 to 0.2 mAU at 40 °C. If your autosampler tray is not temperature-controlled and the room runs warm in summer, your ghost-peak baseline will drift seasonally. I have watched a lab chase a “column problem” for three weeks in July that disappeared in October.
The practical rule: hold sample vials at 4 to 10 °C if the method allows, and if it does not, validate the extractable contribution at the actual tray temperature, not at room temperature. For GC, the inlet temperature is the number that matters. A septum rated for 200 °C used at 280 °C will bleed regardless of conditioning.
Blank and placebo injections: the two-injection diagnostic
Most labs run a blank. Fewer run a placebo, and almost nobody runs them in the right order with the right controls. Here is the sequence that isolates a source in under an hour.
Injection 1: the system blank
Inject mobile phase only, no sample, no vial. Draw from a dedicated glass reservoir through a fresh needle. This tells you what the instrument itself contributes: mobile phase impurities, pump seal extractables, injector rotor residue, detector lamp noise. If the system blank is clean, every subsequent peak is from a consumable or the sample.
Injection 2: the vial blank
Same mobile phase, but drawn from the actual autosampler vial with the actual cap. Now you are testing the vial and septum. Compare peak areas to injection 1. A new peak at 12.4 minutes that was absent in the system blank is a vial or cap extractable. This is the injection that catches the polypropylene slip-agent cluster and the silicone septum contribution.
Injection 3: the filtered blank
Push mobile phase through the same filter type you use for samples, into a glass vial, and inject. If new peaks appear, the filter membrane is the source. This is where a HPLC filter with a documented low-extractable membrane earns its price. A generic cellulose acetate membrane in a 13 mm syringe filter can contribute measurable absorbance at 210 nm; a PTFE or nylon membrane rated for the application usually does not.
Injection 4: the placebo
The placebo is the sample matrix without the analyte. It is the only injection that separates matrix interference from consumable extractables, because it contains everything the sample contains except the thing you are measuring. If the placebo shows the ghost peak at the same area as the sample, the peak is from the matrix or the consumables used to prepare it. If the placebo is clean and the sample is not, the peak is analyte-related or a degradation product. If the placebo shows the peak at half the area, you have a concentration-dependent extractable, which usually means the sample solvent is dissolving the vial.
| Injection | What it isolates | Typical ghost peak signature | Action if positive |
|---|---|---|---|
| System blank | Mobile phase, pump, injector, detector | Broad baseline rise, single peak at void | Replace mobile phase, check pump seals, flush injector |
| Vial blank | Vial body, cap, septum | Cluster at 8 to 20 min (polypropylene), ladder pattern (silicone) | Switch to glass vial with PTFE-lined cap, condition septum |
| Filtered blank | Filter membrane and housing | New peak near 210 nm, membrane-dependent | Change membrane chemistry, pre-rinse filter with 2 mL |
| Placebo | Sample matrix and prep consumables | Peak area scales with matrix concentration | Change prep consumables, check sample solvent compatibility |
Two field cases where the blank injection named the culprit
Case 1: the polypropylene vial that only ghosted above 30 °C
A contract lab running a USP method on a small-molecule API saw a new peak at 14.2 minutes, area 0.8 mAU·s, appearing in about one run in five. The column was new, the mobile phase was fresh, and the system blank was clean. The vial blank was clean at room temperature. The lab had recently moved the autosampler tray next to a heat-generating instrument, and the tray temperature had climbed to 34 °C.
The measurement that identified it: they ran the vial blank at 34 °C and at 22 °C. At 34 °C the peak appeared at 0.8 mAU·s. At 22 °C it was below detection. The peak was a polypropylene antioxidant from the vial body, and the temperature rise pushed its release rate above the reporting threshold. The fix was a glass vial with a PTFE-faced silicone cap, plus relocating the tray. Ghost peak gone in one run.
Case 2: the filter that contaminated the mobile phase
A GC lab running EPA method 8270 saw a repeating siloxane ladder appear in blanks and samples alike, growing with each injection. The inlet septum had been replaced twice. The column was new. The liner was new. The measurement that identified it: they injected a blank that had been filtered through the lab’s standard syringe filter and compared it to an unfiltered blank. The filtered blank showed the ladder at three times the area. The filter housing was silicone-based, and the mobile phase (a dichloromethane extract) was dissolving the housing. Switching to a PTFE-housed filter eliminated the ladder. The relevant guidance is in U.S. EPA SW-846, which specifies the blank and method blank requirements that would have caught this at validation.
The decision checklist for a ghost peak you cannot place
Work top to bottom. Stop at the first positive result.
- Is the peak present in the system blank? If yes, the instrument or mobile phase is the source. Replace mobile phase from a fresh lot, flush the injector with 20 column volumes, and re-run. If it persists, check pump seal age (replace at 6 to 12 months of continuous use) and detector lamp hours.
- Does the peak appear only in vials from one lot? Run a vial blank from three different lots. A lot-dependent peak is a manufacturing extractable. Quarantine the lot and request the certificate of analysis with extractables data.
- Does the peak area change with tray temperature? Run the vial blank at 4 °C and at 35 °C. A temperature-dependent peak is a polymer extractable. Move to glass or lower the tray temperature.
- Does the peak change when you change the filter? Run the filtered blank with two membrane chemistries. A membrane-dependent peak is a filter extractable. Switch to a low-extractable membrane and pre-rinse with 2 mL of mobile phase before use.
- Does the peak appear in the placebo but not the system blank? The matrix or prep consumables are the source. Check pipette tips (polypropylene), sample cups, and any transfer tubing.
- Does the peak area scale with injection volume? If it scales linearly, it is a real compound from a consumable. If it scales non-linearly or appears only at high volume, suspect injector carryover or a rotor seal.
Conditioning and pre-rinse: the numbers that actually suppress extractables
You cannot eliminate extractables, but you can drive them below the reporting threshold before the sample sees them. The numbers I use:
- Septa: condition at the method inlet temperature for 30 minutes with carrier gas flowing before the first injection. For HPLC caps, pre-rinse the septum face with 1 mL of mobile phase.
- Filters: pre-rinse with 2 mL of the sample solvent, discarding the first 1 mL. This removes the surface extractables that would otherwise land in your vial.
- Tubing: flush new PTFE or PEEK tubing with 10 column volumes of mobile phase before connecting to the column.
- Vials: use glass with PTFE-faced silicone caps for anything below 210 nm detection. Polypropylene is acceptable above 240 nm if the tray is held at 10 °C or below.
- Torque: cap vials to 1.5 to 2.0 in·lb (about 0.17 to 0.23 N·m). Over-torquing extrudes septum material into the vial headspace and increases extractable release. Under-torquing lets solvent evaporate and concentrate the extractables that remain.
For consumable selection criteria and extractable data, the technical resources at Restek and the consumables catalog at Agilent publish the kind of lot-level extractable profiles you want in a specification. Thermo Fisher carries comparable data for sample handling consumables. Ask for it before you buy, not after the ghost peak appears.
Where the sample container itself becomes the contamination source
One material deserves its own paragraph because it is the most overlooked: the sample cup or vial used for a non-chromatographic prep step that then feeds the chromatograph. A COD vial used for a digestion or oxidation step, then transferred into an autosampler vial, carries whatever the digestion released. If the transfer is done with a polypropylene pipette tip, you have added a second extractable source. The blank injection on the final vial will show both, and the retention times will tell you which is which if you run the intermediate container as its own blank.
The decision rule
When a ghost peak appears, do not change the column first. Run the four-injection sequence (system blank, vial blank, filtered blank, placebo) and let the peak’s behavior across those four injections name the material. If the peak is temperature-dependent, the source is a polymer and the fix is a material change or a temperature reduction. If it is lot-dependent, the source is a manufacturing extractable and the fix is a supplier conversation with the certificate in hand. If it is membrane-dependent, the fix is a filter chemistry change. The column is almost never the answer, and the blank injection is almost always the question.
References & further reading
Related resources from our lab consumables network:
- Lab Chromatography at labchromatography.com
- Hplc Filter at hplcfilters.com
- Ptfe Membrane Filter at ptfemembranefilter.com
- Cod Vial at codvial.com
External references and standards: