GC Troubleshooting: Leaks, Peaks and Retention Time Problems

Diagnose GC problems by comparing evidence before replacing parts or changing the method. Record the symptom, recent events and affected samples; then test a limited set of plausible causes. Retention shifts, low response and abnormal peaks are not unique signatures of a leak or a damaged column. Begin with safety, configuration and non-invasive checks, preserve the original data and define the evidence needed for return to use.

The pathways below are original GuideGxP investigation aids. They do not authorise invasive servicing, universal operating settings or retesting until results pass.

Describe the symptom and preserve context

Compare the suspect chromatogram with a relevant historical run using the same method version, column configuration, acquisition mode and sample type. Determine whether all peaks shift, only some change, peaks disappear, new peaks occur, or baseline noise and shape deteriorate. Record onset and persistence, not just the worst chromatogram.

Collect gas-source changes, maintenance, new consumable lots, sample preparations, software events and instrument messages. Preserve original sequences and processing versions. Separate an actual acquisition change from a different display scale or integration. The experimental UC Davis sensor study found overlapping symptoms from several failure modes; its specialised hardware results are not a pharmaceutical GC repair procedure.

Make the system safe before diagnosis

Assess gas identity, pressure, hot zones, electrical hazards and volatile or hazardous samples before touching the system. Suspected hazardous gas release calls for the site’s emergency procedure and competent response, not continued chromatography. Observe approved isolation, cooling and access instructions. Never search for leaks with a flame, open pressurised connections or bypass protective interlocks.

Use only a leak-check method compatible with the gas and instrument under an approved procedure; a negative result applies to the locations and conditions actually tested. Avoid unapproved liquids that may contaminate the analytical pathway. Work inside heated zones, electrical assemblies or specialised detectors may require authorised service. OSHA’s compressed-gas rule is a US reference; local legal and site requirements determine the applicable safety controls.

Check gas supply and actual control conditions

Confirm the configured carrier-gas identity, pressure/flow mode and relevant settings against the approved method. Compare supply availability and instrument indications before and after the event. Do not equate a source pressure reading with column flow or assume that a cylinder purity certificate proves gas quality at the instrument.

Review regulators, line changes, filters and approved leak-check evidence, especially at recently disturbed connections. Moisture, oxygen or other contaminants can affect performance depending on column and detector. A pressure controller compensating for a disturbance can obscure a simple interpretation. Use suitable reference measurements where justified and safe; define gas and measurement conditions before comparing values.

Separate inlet and preparation contributions

Verify vial identity, preparation calculations, solvent, sample stability and introduction mode. A selective loss may arise from adsorption, incomplete vaporisation or discrimination; a large load may distort peaks. These are hypotheses to test, not conclusions drawn from shape alone. Review inlet consumables and maintenance history without dismantling beyond the authorised operator procedure.

Use planned blanks and suitable controls to locate contamination. A diluent blank, a prepared matrix blank and a control sample answer different questions. A blank after a concentrated sample may reveal carryover but does not by itself locate retained material. If headspace is used, investigate fill, sealing and equilibration before blaming the GC inlet; the dedicated headspace article develops that pathway.

Assess oven, column and connections

Confirm method temperature programme, equilibration and actual configuration. Review whether a column was changed, shortened, reconnected or assigned different dimensions in the software. Thermal-control problems, flow changes and altered column conditions may all move retention times. A common shift suggests a shared influence, but neither its direction nor magnitude proves a unique cause.

Evaluate critical separation and peak shape as well as retention. Suspect adsorption or degradation when the evidence supports it, and consider whether contamination remains upstream. Do not prescribe a universal bake-out or trim the column simply because a peak tails. Such actions can change the system, remove evidence or exceed limits; they require a justified approved procedure and an impact assessment.

Check detector and data without hiding anomalies

Use the detector’s actual technology to select checks: required gas supplies and operating state for flame-based detection, gas compatibility for thermal conductivity, or acquisition mode and appropriate diagnostic evidence for MS. A detector-ready indication does not prove analytical performance. Do not transfer checks or limits from another detector type.

Review signal channel, acquisition start, sampling settings, response range and processing version. Saturation, an incorrect channel or unsuitable acquisition can resemble a chromatographic failure. Retain raw data and justify any processing changes through the established review process. Smoothing, re-integration or peak deletion must not be used to make unexplained behaviour disappear.

Choose discriminating tests and justify return to use

Three-symptom decision tree. Apply the safety assessment first, then follow the branch matching the evidence:

1. Retention shift: confirm method/column identity → check gas change and actual flow/pressure conditions → assess thermal control and recent connections → evaluate remaining column or sample hypotheses.
2. New or ghost peak: compare historical and current blanks → distinguish diluent, preparation, consumables and sequence carryover → investigate the recently changed pathway → use suitable confirmation if identity matters.
3. Low or missing response: confirm sample and acquisition → compare a suitable control → assess introduction/transfer and gas integrity → investigate detector-specific performance with authorised support.

The arrows indicate a reasoning order, not permission to continue when safety or system suitability is compromised.

Simulated case A — retention drift after a gas change. Start with source identity, configured gas and control mode, recent connections and documented operating conditions. If these are correct, assess safe leak evidence and thermal behaviour before replacing the column. A normal cylinder certificate does not close the downstream investigation. The chronological association prioritises checks; it does not prove causation.

Simulated case B — a new peak after maintenance. Start with exactly what was replaced or exposed, then compare appropriate blanks, consumable identity and preparation history. A peak in a blank may implicate the pathway or reagents; its absence does not prove the sample alone caused it. Avoid changing gas and column simultaneously: the next test should distinguish competing explanations. Neither case reports performed experiments or numerical outcomes.

Document hypothesis, planned intervention, expected discriminating observation and decision rule before additional testing. Change one relevant variable at a time where feasible, with justified sequencing. Preserve all results and evaluate contradictory evidence. The FDA OOS guidance supports scientifically justified investigation and rejects testing into compliance; not every instrument anomaly is itself an OOS result.

Return to use requires evidence appropriate to the intervention, satisfactory applicable method controls, review of affected prior data and authorised release. Qualification checks, maintenance records and SST have complementary roles. Record remaining limitations and escalation when the cause is unresolved. A single satisfactory injection does not erase an earlier unexplained failure.

Sources and boundaries

The UC Davis thesis (2024) is primary research on specialised gas-sensor platforms; only its evidence that symptoms can have multiple causes is used, not its intervention settings. FDA OOS guidance is dated May 2022; Chapter 6 provides the applicable EU pharmaceutical QC context. OSHA 1910.101 is US-specific. The diagnostic tree and cases are original editorial proposals, not regulatory troubleshooting procedures.

1. Pimentel Contreras, R. A Diagnostic Case Study for Manufacturing Gas-Phase Chemical Sensors. MSc thesis, University of California, Davis (2024), sections 4.1–4.4.
2. FDA. Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production. Guidance for Industry, May 2022.
3. European Commission — EU GMP Chapter 6, Quality Control (2014).
4. OSHA, 29 CFR 1910.101 — Compressed gases (general requirements). United States.

Related GC checks

• GC qualification and return-to-use evidence
• Headspace preparation and transfer variability
• GC configuration and detector selection