The satisfaction you get from a job well done clearly shows. Their invention „Advanced Relevant Investigation Sampler for Taste & Odor at Tap“ (ARISTOT) has enabled true passive sampling of off-flavors directly from the consumers water faucet. From left, Thomas Thouvenot, David Benanou and Christophe Tondelier from Veolia Environnement in Paris, France.


To demonstrate how simple it is to use the ARISTOT passive sampler, Veolia generated an animated film clip in English language. In it, ”Professor Benanou” and his assistant explain how the sampler works and how it is used.
ARISTOT installed at a water faucet
Snooping out the causes of off-flavors: Analyte concentration is a must. ARISTOT on the test-bench. Water usage patterns had to be established and subsequently simulated in the laboratory.
The contents of two, three, and four Twisters were analyzed in single GC/MS runs using the multi-shot method; the results were compared with standard single Twister desorptions. The Twisters were spiked with 2 ng/L 2,4,6-Trichloroanisole (TCA). The 2,4,6-TCA peak areas were plotted as function of the number of Twisters desorbed. Linear regression of the results showed a correlation very close to 1. “This shows”, says David Benanou, “that the Multi-Shot Method will enable sensitivity enhancements proportional to the number of Twisters used – and with just one GC/MS run”.
The usefulness of ARISTOT was demonstrated by sampling on a water faucet, which was fed water spiked with a mixture of halogenated anisoles at a concentration of 200 pg/L. Sampling was performed over 15 minutes and two Twisters were thermally desorbed for the analysis. The figure shows the resulting Single Ion Mode (SIM) chromatogram. Detection limits are in the low ppQ (pg/L) concentration range.

David Benanou and Christophe Tondelier,
Water Research Center, Analytical Research Department, Veolia Environnement,
Chemin de la digue, BP76,
78603 Maisons-Laffitte,

For more information please contact us by e-mail:

Tap water

Off-flavor compounds collected on tap

Based on stir bar sorptive extraction (SBSE), a solvent-free extraction technique, French scientists have developed and patented an innovative new passive sampler for drinking water. Off-flavor compounds are extracted and concentrated directly at the tap under standard usage conditions and are subsequently determined at ultra-trace levels using thermal desorption GC/MS.

It’s the classic situation: things you have been complaining about for ages refuse to happen when someone has finally come to fix it. Case in point: Your tap water has had an unpleasant flavor for some time. Maybe it could be described as having an “iodine”, “petroleum”, “medicine”, “musty” or “earthy” note, but when a sample was finally taken to analyze the water, nothing
is detected in the laboratory, and there may not even be a detectable bad smell or taste from the sample taken. In such cases, long-term monitoring over several days is the only way to get an answer. Until now, this has not been possible in practice since no suitable technologies have been available.
David Benanou, Christophe Tondelier and Thomas Thouvenot from Veolia Environnement in Paris, one of the largest water suppliers in the world, have come up with a solution. The scientists report their findings in the Journal of Chromatography A (1216 (Tondelier et al. 2009), 2854-2859). A patented passive sampler was developed based on stir bar sorptive extraction (SBSE), with which off-flavor compounds can be extracted and concentrated directly on the consumer’s water tap under standard usage conditions without the use of interfering solvents. The extracted compounds are subsequently determined at ultra-trace levels using thermal desorption GC/MS.

Off-flavors and bad taste in drinking water – whence they come?

In today’s modern world, tap water is generally completely safe to drink and mostly odorless. At the time of Shakespeare, people were used to a heavy background of unpleasant flavors and odors. In those days, water would also often be unsafe to drink. Those who could afford it would drink beer or wine to avoid the threat of being infected with bacteria or parasites. With improved general hygiene and sanitary conditions, however, consumers are no longer used to off-flavors and are unwilling to accept these. In terms of product quality, taste and odor have come to play an important role for drinking water companies. “Consumers assess the organoleptic (sensory) quality of their tap water while drinking it and an unpleasant smell or taste is often mistakenly associated with health hazards”, explains David Benanou. Drinking water companies therefore work hard to identify sources of unpleasant odor or taste compounds in order to take corrective action and/or make recommendations about how to eliminate them.
Off-flavor compounds in drinking water include geosmin, 2-methyl isoborneol and 2,4,6-trichloroanisole, which normally cannot be attributed to a single source. These may well have been introduced by the feed water that is processed into drinking water. Other conceivable sources are algae, waste water, and leaks. Off-flavors are often the product of microbial activity in today’s extensive water distribution networks or directly in small household systems. “In most cases, a concentration level of the solutes in the sub-nanogram per liter range (ng/L = ppt) is sufficient to upset odor and taste receptors”, says Christophe Tondelier. This makes determination of off-flavor compounds difficult: “As a rule, conventional techniques and methods do not offer the required detection limits”, states the water expert. Due to the volatile nature of the analytes, the determination of off-flavor compounds is mainly performed using capillary gas chromatography (GC) combined with mass selective detection (MSD). Simultaneous sensory evaluation is performed using a suitable olfactory detection port (GERSTEL ODP): “GC offers the highest separation power and the MSD and ODP enable highly sensitivity determination of the analytes”, David Benanou points out.

Analyte concentration is necessary in order to determine off-flavor compounds in water

The key to successful determination of off-flavors in water is using the correct sample preparation technique. Analytes must be concentrated prior to GC/ MS analysis in order to provide adequate detection limits and rugged results. Methods that have been widely used are closed loop stripping analysis (CLSA) and Headspace SPME. CLSA involves concentrating analytes on activated charcoal or similar adsorbents after stripping them from the sample, a relatively lengthy process that also requires extensive cleaning of glass-ware. Headspace SPME is performed by placing a fiber coated with adsorbent or sorbent material in the headspace of the sample inside a vial. Volatile Organic Compounds (VOCs) are extracted and trapped in the SPME fibre followed by thermal desorption in a GC inlet and GC/MS determination.
Stir bar sorptive extraction (SBSE) has in recent years increasingly been used for extracting and concentrating trace level VOCs and SVOCs from aqueous matrices. SBSE relies on the GERSTEL Twister, a magnetic stir bar coated with polydimethylsiloxane (PDMS). Organic compounds are extracted into the PDMS phase while the Twister stirs the aqueous sample. The stir bar is then removed, dabbed dry and desorbed in a thermal desorption system (GERSTEL TDU or GERSTEL TDS) prior to GC/MS determination. “Our experience has shown that SBSE is an attractive alternative to conventional stripping methods and SPME”, says David Benanou. “SBSE is based on the same PDMS phase used in SPME fibres, but the sorbent phase volume is significantly bigger leading to much better enrichment and lower limits of detection – even for polar compounds - SBSE helps us determine organic compounds in aqueous matrices dependably and reliably”, the scientist points out.

Developing a passive sampler for tap water monitoring

There was agreement within the Veolia R&D team about which extraction and analysis technologies to use, but developing a passive sampler for tap water proved to be a complicated assignment:
“Uncharted waters had to be navigated while meeting fundamental conditions “, comments Christophe Tondelier. A passive sampler relies on analyte diffusion from the sample into the trap, active pumping of the sample is not used. In addition, the sampler had to be visually appealing and easy to integrate in bathroom and kitchen workflows. The main question was of course how one could insert a passive sampler into the path of the water without influencing or changing the normal flow and water handling conditions at the tap. In the end, Mother Nature provided the model for the patented passive sampler ARISTOT (Advanced Relevant Investigation Sampler for Taste & Odor at Tap). Inspired by the petals of daisy flowers, seven cavities, each 3 mm in diameter and 30 mm long, designed to hold the Twisters were made in a 40 mm long piece of stainless steel. The material was chosen for its inertness to avoid contamination or reactions with the substances in the water. “ARISTOT is like the cylinder of a revolver, except that instead of bullets it contains seven GERSTEL Twisters held in place by a nozzle in the cavity”, says Thomas Thouvenot, flow expert and a member of the ARISTOT team. Mr. Thouvenot adds that seven positions were chosen to ensure best possible flow conditions, while optimizing the interaction surface between water and PDMS and avoiding solute losses.

ARISTOT on the test bench: practice makes perfect

Practical use reveals just how good an innovation really is. Small scale tests were first performed to determine whether the new passive sampler was suitable for routine operation. The scientists developed a pilot unit, which was used to simulate consumers’ habits, which were determined with the help of a questionnaire. To test the sampling efficiency, computer models were used to develop a mixing chamber, that could mix tap water with a defined amount of an off flavor compound standard solution.
Extensive consumer feed-back from questionnaires had revealed that cold water would flow from most kitchen taps about 15 minutes per day at an average rate of 2 liters/min. “Hydrodynamic computer modeling was used to verify that the ARISTOT sampler produced the optimum flow pattern with efficient exchange between water and the PDMS phase”, reports Thomas Thouvenot. The team also made sure that the cavities holding the Twister stir bars remained filled with about 18 mL of water after the water tap had been closed, in order to block contamination from the air from reaching the PDMS phase. The scientists reproduced standard water usage patterns on a pilot unit in the laboratory. Optimal conditions were determined to be a tap water flow rate of 2L/min, cold water, large stir bars – 2 cm long x 1 mm PDMS thickness – and with an enrichment time of up to 120 mins. While the water was dispensed, it was spiked with a standard ethanol solution of off-flavor compounds (200 pg/L). The solution included 2,4,6-trichloroanisole (2,4,6-TCA), 2,4,6-tribromoanisole (2,4,6-TBA), deuterated 2,4,6-trichloroanisole (2,4,6-TCA-d5) as the internal standard, 2,4-dichloro-6-bromoanisole (2,4-DC-6-BA), 2,6-dichloro-4-bromoanisole (2,6-DC-4-BA), 2-chloro- 4,6-dibromoanisole (2-C-4,6-DBA) and 4-chloro-2,6-dibromoanisole (4-C-2,6-DBA). A GC 6890 with MSD 5975 (Agilent Technologies) was used for the analysis in combination with a GERSTEL thermal desorption system (TDS) fitted with an autosampler (GERSTEL TDS -A) and a Cooled Injection System (GERSTEL CIS). 20 mm Twisters with a 1 mm PDMS coating were used for the extractions. The stir bars were subsequently desorbed in splitless mode: 30°C (0.8 min), 60°C/min, 250°C (8 min). The desorbed analytes were cryofocused at –50°C in the CIS. The CIS was then heated up to 300°C at 12°C/min (2 min) and the analytes transferred to the column in splitless mode. The carrier gas was helium with a constant flow through the system of 1.5 mL/min.

GC column: HP5-MS capillary column 30 m long x 0.25 mm ID x 0.25 μ film thickness.
GC oven program: 50°C (2 min) to 200°C at 10°C/min, then to 300°C at 25°C/min (7 min).
Detection was performed in SIM/SCAN mode: the SIM mode was used for quantification and the SCAN mode for confirmation

“To further improve detection limits, we decided to investigate the effects of desorbing multiple Twisters for each GC/MS run”, explains David Benanou. Initially, two Twisters were in the TDS desorption tube instead of one - and the results were encouraging. Sensitivity was doubled with excellent recovery and reproducibility. “When we then moved to desorb all seven twisters and introduce the combined analytes into the GC/MS in just one run, we reached detection limits close to the low ppq range (pg/L). You just can’t get better sensitivity than that.“