UK laboratories monitor nitrates, sulphates, chlorides, fluorides and heavy metal ions at parts-per-billion concentrations. Titration and colorimetric methods cannot reach these levels reliably, so detection depends on more sensitive instrumentation and on sampling designed around where contamination is likely to occur.
The standards come from the Water Supply (Water Quality) Regulations 2016 and the limits retained from the EU Drinking Water Directive. Meeting them requires three things at once: sensitivity below the regulatory threshold, calibration that holds up under inspection, and enough throughput to handle the thousands of samples a large supplier submits each year.
Identify which contaminants a catchment is likely to carry
Detection starts with the sampling plan, not the instrument. A panel that ignores the dominant local risk will return clean results while the actual problem goes unmeasured, so the analyte list should follow the catchment’s exposure.
Match the analyte list to the source
- Agricultural runoff — nitrates, nitrites and pesticide residues, concentrated in arable catchments and strongly seasonal around fertiliser application
- Industrial discharge — sulphates, chlorides and heavy metal ions, usually traceable to a specific consented outfall upstream
- Ageing pipework — lead and copper, which appear at the consumer tap rather than the treatment works and therefore need tap-side sampling to detect at all
- Seasonal events — algal by-products following heatwaves and bloom conditions, which can appear and clear within weeks
Sample at the right point in the network
Where a sample is drawn determines what it can show. Treatment works samples verify that the process is performing; consumer tap samples capture everything the distribution network adds afterwards. Plumbosolvency from old service pipes is invisible in a works sample no matter how sensitive the method. Sampling both points is what makes the difference between confirming treatment performance and confirming water quality as delivered.
Treat private supplies separately
Wells and boreholes carry higher contamination risk and face stricter monitoring than the mains network. Rural properties on private sources have needed additional checks after heatwaves increased algae blooms, and shallow sources respond quickly to rainfall and runoff. Fixed annual schedules routinely miss what risk-based sampling catches, because contamination in these supplies is episodic rather than continuous.
Ion chromatography separates the sample before anything is measured
Separation is what allows one ion to be quantified in a matrix containing dozens of others. Without it, a detector returns a combined signal that cannot be attributed to any single species.
How the separation runs
A sample is injected onto a column packed with ion-exchange resin, which retains ions according to charge and size. Because each species interacts with the resin differently, they elute at different times rather than together. Retention time identifies the ion; peak area gives the concentration. Comparison against a certified standard run under identical conditions converts that area into a reportable figure.
Three refinements that push detection lower
- Suppressed conductivity detection — converts mobile phase ions into a weakly conductive form before the detector, dropping the background so low-level analytes stand clear of baseline noise
- Automated sample preparation — handles dilution, filtration and standard addition inline, removing the manual steps where transcription and pipetting errors enter
- Multi-dimensional detection — pairs conductivity with UV or mass spectrometry to separate species that would otherwise co-elute and be reported as one peak
Technical detail on column chemistry, suppression and detection performance is available from suppliers such as Metrohm.
Establish the real detection limit per matrix
Published detection limits sit within regulatory requirements, but the working figure depends on the sample. A high-chloride matrix raises the effective floor for other anions because the large chloride peak interferes with what elutes near it. Hard water and high dissolved organic content have similar effects.
Run spiked recovery samples on any new source before relying on it for compliance reporting. This establishes both the practical limit of quantification and whether recovery falls within acceptable bounds for that specific water. A limit validated on one source cannot be assumed for another.
Clear the thresholds the 2016 Regulations enforce
The Regulations specify dozens of chemical and indicator parameters, sampled at treatment works and consumer taps, with frequencies scaled to the population served.
Know the limits
- Nitrate — 50 mg/l
- Nitrite — 0.5 mg/l
Measurement uncertainty near these values is the real difficulty. A result at 48 mg/l with a wide uncertainty band tells you very little. Separating a genuine exceedance from analytical scatter is precisely where simpler methods fail, and it is why uncertainty has to be quantified and reported alongside the figure rather than assumed.
Follow the risk-based sampling model
Cryptosporidium outbreaks drove additional monitoring requirements at many treatment plants, and those requirements remain in force under retained EU law. The Drinking Water Inspectorate targets sampling where contamination is most likely rather than distributing effort evenly across the network. Algal blooms and more frequent extreme weather are steadily widening the screening panels labs are expected to run.
Keep instruments running and results defensible
Lost runs become compliance failures, so uptime is part of detection rather than separate from it. A missed regulatory sample cannot be recovered retrospectively.
Controls to build in
- Calibrate against certified reference standards before each analytical run, not weekly
- Work to ISO 17025 validation so results hold up when challenged
- Use external proficiency testing to verify data quality independently of internal checks
- Link instruments directly to lab data platforms so reports draw from source data rather than re-keyed figures
- Run preventive maintenance on a schedule rather than after failure
Costs and training
Columns, reagents and certified standards take a significant share of lifetime instrument cost, often exceeding the purchase price over a full service life. Budgets built on capital cost alone tend to under-provision. Training an analyst to a confident operational standard takes several months, which makes staff turnover a genuine continuity risk in smaller labs.
Automate to raise throughput
Automated sample handling and calibration let a single instrument run unattended through the day and overnight. Remote monitoring flags baseline drift or pressure changes before a batch is lost. Integrated calibration verification cuts manual quality control from hourly to once per analytical run, freeing analysts for method development and investigation work. Select instruments on detection limit and automation compatibility rather than headline specification.
Choose the method and sampling plan your catchment requires
Detecting trace impurities in British drinking water is not solved by sensitivity alone. The three elements have to work together: a sampling plan built around local risk, a separation method validated against the specific matrix, and calibration and maintenance disciplined enough to make every result defensible when an inspector asks.
Where labs come unstuck is usually at the joins. A well-run instrument screening the wrong analytes returns clean data on the wrong question. A broad panel sampled only at the treatment works misses everything the distribution network adds. A validated method carried across to a different source without fresh recovery testing quietly reports against a detection limit that no longer applies.
Practical next steps for any laboratory reviewing its provision:
- Audit the analyte list against the dominant contamination routes in each catchment served
- Confirm sampling points cover both treatment works and consumer taps where plumbosolvency is a risk
- Re-validate detection limits per source with spiked recovery samples, and record the uncertainty attached to each
- Assess throughput headroom against projected sample volumes, allowing for widening screening panels
- Cost instruments over their service life, including columns, reagents and certified standards, rather than on purchase price
Work through those five points and the gaps tend to be obvious. Close them, and problems surface while they are still small enough to fix.
Discussion about this post