Sub-Part Per Trillion Headspace Analysis of Odor Producing Haloanisoles, Geosmin, and 2-Methylisoborneol in Water and Wine by GCMS without the need to Contact the Liquid Sample

Application Note: 2016-03

Authors

Daniel CardinThomas X. RobinsonEntech Instruments, Inc.Simi Valley, CA 93065

Abstract A new technique is presented for the analysis of odor producing compounds in wine and water. An adsorbent device performs a vacuum extraction of the headspace providing far superior and nearly exhaustive extraction during analysis. Highly water soluble compounds such as Geosmin, 2-Methylisoborneal (2-MIB), and Trihaloanisoles are effectively extracted between 60-80% to allow detection by GCMS SIM techniques down to 0.01 to 0.1 ng/L when using 50cc of sample. The adsorbent device creates a vacuum tight seal with sample vials as large as 1L, allowing plenty of sample for analysis down to and below sensory thresholds using a single quad GCMS. No contact with the liquid sample is needed, and therefore no after sampling work up or handling of the adsorbent device other than to transfer it to a direct thermal desorption GCMS inlet, or into a tray for automated analysis. The ability to perform sample extractions in parallel allows faster throughput by

decoupling the extraction process from the GCMS cycle times. Salting and heating is not necessary during the extraction, allowing longer adsorbent lifetimes by eliminating the damaging effects of aerosol transferred salts to the absorbent. Data is presented showing calibrations from 0.1 to 20 ng/L for Trichloroanisoles, Tribromoanisole, Geosmin, and 2-MIB, using Geosmin-d3 as the only internal standard. Analysis of spiked and non-spiked drinking water and wine are performed to verify the ability to perform measurements in matrices with varied complexity and affinity for the target analytes. The results show that the ultimate limit of detection was more affected by the ability to chromatographically separate interfering compounds, than in achieving enough signal to reach 0.1 ng/L and below when using a single quadrupole GCMS.

Introduction Many odor producing compounds have been difficult to measure down to olfactory detection limits due to a combination of low vapor pressure and high solubility in the water or beverages that require analysis. Such odor producing compounds include Trihalogenated Anisoles found in wine, and compounds such as Geosmin and 2-Methylisoborneol in drinking water that are produced by bacteria and blue-green algae. Classical methods such as Purge & Trap have been minimally effective due to low recoveries and poor precision leading to unacceptable MDLs. Closed Loop Stripping can reduce detection limits, but is difficult to perform in production laboratories. Techniques such as SPME are challenged by the vast difference in phase ratios between the SPME fiber and the amount of sample required to reach low to sub-ppt levels (40-

50mLs typical), and the rather high affinity of these odor producing compounds for the sample matrix. SPME methods recover so little of these compounds from the sample that isotope dilution is typically required for every target compound in order to make quantitative measurements, and carryover as much as 2-10% makes it impossible to establish method MDLs that are reliable.

The technique presented here approaches the problem differently. An adsorbent inside of a 1/4” tube is introduced within inches of the sample, with no sample path to cause losses or create carryover. Using external seals around the Sorbent Pen adsorbent allows a vacuum approximately 30 times lower than atmospheric pressure to be achieved above an aqueous liquid, permitting chemicals to travel an average of 30 times further between collisions with other gas phase molecules. This creates a fast transfer from the surface of the liquid or solid sample to the adsorbent, greatly reducing the time to reach equilibrium for low volatility or highly matrix soluble compounds. This technique also decouples the extraction from the analysis, allowing samples to be extracted offline into the adsorbent devices to ensure that extractions are at or near equilibrium, with the potential for complete, exhaustive extractions. Extracting to equilibrium under vacuum using a much large phase ratio than possible with SPME creates both higher sensitivity and more reproducible extractions, allowing quantitative analysis without isotope dilution while achieving MDLs 5-50x lower than reported using SPME.

Experimental Geosmin (Supelco) and a mixture of 2-Methylisoborneol, Trichloroanisole isomers, and Tribromoanisole (Veolia Research, France)

were obtained in methanol and diluted to working concentrations in filtered water. Geosmin-d3 was used as the internal standard for all compounds analyzed. Raising target compound recoveries to over 50% eliminates the need to use isotope dilution, since relative run to run error decreases as recoveries increase. Target compounds were spiked into filtered water to create a final volume of 50mL using 125mL bottles, followed by the addition Geosmin-d3 to achieve 10ng/L. Clean Sorbent Pen cartridges (Entech Instruments, Simi Valley CA) containing 60mg of Tenax TA were introduced into the vial headspace and a vacuum was created in the sample vial through the micro seal at the top of each Sorbent Pen. Sample extraction was performed at 25 deg C for periods of 1, 3, 8, and 16 hours to compare rate of extraction for each compound using 20 ng/L concentrations. Analysis was performed by thermally desorbing the Sorbent Pens using a 5800 Sorbent Pen Desorption Unit (Entech) installed on a 7890B/5977 GCMS (Agilent, Palo Alto, CA) to deliver the sample first onto a 5m pre-column (DB1, 0.5um film, 0.25mm ID) with a split vent prior to the primary column (60m, 250um ID, 0.5um DB1) to allow higher flow rates during desorption of the cartridges. The initial GC temperature was 35 deg C with an immediate ramp rate of 20 deg C/min to 280 deg C with a 4 minute hold. Full scan data was collected from 33-350 amu, with approximately 2.5 scans per second. The calibration data from 0.1 to 20 part per trillion was obtained using SIM with an EM Gain of 8. Detection in both unfiltered tap water and wine were performed by both direct analysis, and with a 5ng/L spike of the target compounds to verify invariance to the maxtrix. The red wine was diluted 10:1 with filtered water prior to spiking with 5ng/L to reduce the potential for chromatographic interferences, and because olfactory detection limits are not as low in wine as in drinking water.

Figure 2 - Extractions are performed in parallel to allow much longer extraction times than classical SPME. After extraction, the pens are isolated in a 30 position tray, and then analyzed by direct desorption into a GCMS using a specially designed inlet. After splitless sample injection, a split flow of >50cc/min cleans the pens to less than 0.2% carryover, eliminated the need for further cleanup.

Micro Seal forEvacuation

Silonite coated316 Stainless Steel

60mg Tenax TA

Evacuation (30 seconds)

Sample extracts much faster at 30x below atmospheric pressure, limited by vapor pressure of water at 25 deg C

Desorb Gas Port

Figure 1 - Sorbent Pens containing 60mg of Tenax perform vacuum extraction on 50mL of sample at 25 deg C for 1 to 20 hours, followed by direct, splitless desorption into a GCMS. The tops of the Sorbent Pens were heated to 50 deg C to prevent water condensation.

ISTD/Surr Addition (Optional)

H2/H4

GC EPCInj1 or Inj2

V2 V1 R

ByPass Desorb

V3

Split/Brake Divert

5m, 250um ID, 0.5um DB1

60m, 250um ID, 0.5um DB1

5800 Sorbent Pen Desorption Unit on GCMS

Clean Sorbent Pens

Vacuum Extraction of water and wine using Sorbent Pens. Vacuum remains even after removal of vacuum source

Sampled Sorbent

MS

V4

1 hour

3 hour

8 hour

16 hour

10000

20000

30000

40000

50000

60000

70000

80000

90000

100000

110000

Abundance

Time--> 13.45 13.50 13.55 13.60 13.65 13.70 13.75 13.80 13.85 13.90 13.95 14.00

2-Methylisoborneal

0

10000

20000

30000

40000

50000

60000

70000

80000

90000

100000

110000

120000

130000

140000

150000

Abundance

16.40 16.50 16.60 16.70 16.80 16.90 17.00 17.10 17.20 17.30Time-->

1 hour

3 hour

8 hour

16 hour

Geosmin

15.20 15.40 15.60 15.80 16.00 16.20 16.40 16.60 16.80 17.00 17.20 17.40 17.60 17.80 18.00

0

10000

20000

30000

40000

50000

60000

70000

80000

90000

100000

110000

120000

Time-->

Abundance

1

2

3

2,4,6-Trichloroanisole2,3,5-Trichloroanisole2,3,4-Trichloroanisole

1.2.3.

1 hour

3 hour

8 hour

16 hour

18.80 18.90 19.00 19.10 19.20 19.30 19.40 19.50 19.60

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

Time-->

Abundance

2,4,6-Tribromoanisole1 hour

3 hour

8 hour

16 hour

Figure 3 - Relative responses for target compounds extracted from 20 ng/L standards in 50cc filtered water at 25 deg C, no salt added, 100 rpm agitation, 1/30th atm vacuum, for 1, 3, 8, and 16 hours. A final extraction time of 20 hours was selected for this method with an extraction efficiency of between 60-80%.

13.40 13.50 13.60 13.70 13.80 13.90 14.00 14.10 14.20

0

5000

10000

15000

20000

25000

30000

35000

Time-->

Abundance

2-methylisoborneol

16.30 16.40 16.50 16.60 16.70 16.80 16.90 17.00 17.10 17.20 17.30

0100020003000400050006000700080009000

10000110001200013000140001500016000170001800019000

Time-->

Abundance

GeosminInterferent

15.40 15.60 15.80 16.00 16.20 16.40 16.60 16.80 17.00 17.20 17.40 17.60 17.80 18.00

100

200

300

400

500

600

700

800

900

1000

1100

1200

1300

1400

Time-->

Abundance

2,4,6-Trichloroanisole2,3,5-Trichloroanisole2,3,4-Trichloroanisole

1.2.3.

1

2

3

18.90 18.95 19.00 19.05 19.10 19.15 19.20 19.25 19.30 19.35 19.40 19.45 19.50

100

150

200

250

300

350

400

450

500

Time-->

Abundance

Ion 346.00 (345.70 to 346.70): 16052213.D\data.ms

2,4,6-Tribromoanisole

Figure 4 - SIM of 1ng/L Geosmin/2-MIB, and 0.1ng/L of the 3 Trichloroanisole isomers and Tribromoanisole. Ultimate detection limits for the Haloanisoles were much lower, estimated at 0.01 to 0.02 ng/L due to fewer matrix interferences at the higher, unique mass to charge ratios using unity mass single quadrupole MS detection.

2-Methylisoborneol, 1ng/L, m/z=95

Geosmin, 1ng/L, m/z=212

Trichloroanisole, 0.1ng/L, m/z=212

TribromoAnisole, 0.1ng/L, m/z=246

Table 1 - Calibration curves from 0.1 to 20 ng/L showing good linearity over a 200x calibration range. Geosmin points at 0.3 and 0.1ng/L were omitted due to interferences from contaminants in the filtered water.

Table 2 - Results from Simi Valley, CA tap water and Paso Robles Carbernet Sauvignon wine samples. The wine was diluted 10:1 prior to analysis. All samples had 10 ng/L Geosmin-d3 added as an Internal Standard. The stability of Geosmin in wine is unclear, and the inconsistent recovery of Geosmin-d3 may be the result of losses.

Calibration (ng/L)

0.1 0.3 1 5 20 Average %RSD

2-Methylisoborneol 0.341 0.346 0.351 0.248 0.309 0.319 13.5

Geosmin 0.362 0.414 0.339 0.372 10.3

2,4,6-Trichloroanisole 0.272 0.283 0.240 0.220 0.218 0.246 12.1

2,3,5-Trichloroanisole 0.367 0.386 0.319 0.299 0.301 0.334 11.9

2,3,4-Trichloroanisole 0.251 0.255 0.198 0.201 0.181 0.217 15.5

2,4,6-Tribromoanisole 0.118 0.111 0.095 0.097 0.089 0.102 11.8

Analytical Report (ng/L)

Sample Area Geosmin-d3 2-MB Geosmin 2,4,6-TCA 2,3,5-TCA 2,3,4-TCA 2,4,6-TBA

Tap Water 680178 <0.1 <1 <0.1 <0.1 <0.1 0.22

Tap Water = 5ng/L 697602 3.73 5.64 4.02 3.94 3.88 4.38

Wine (1/10 in Water) 284246 <1 <10 0.11 0.20 0.25 0.17

Wine (1/10 + 5ng/L) 497007 4.39 5.03 5.91 5.71 5.64 6.59

DiscussionFigures 1 and 2 show the concept of how this off-line extraction process works. Samples or standards are extracted simultenously in large quantities, which allows fast GCMS methods to be created to ultimately increase sample throughput well above methods that perform extractions one at a time. A brief evacuation leaves a vacuum of about 0.35 psi absolute in the vial for the remainder of the sampling period, dramatically increasing the rate in which equilibrium between the adsorbent and the matrix is achieved. Figure 3 shows the relative recovery of the target compounds over 1,3, 8, and 16 hours. Even with the enhanced extraction rates under vacuum, the benefit of longer extraction times is clearly seen, especially for the more polar Geosmin/2-MIB, and the lower volatility Trihaloanisoles. Figure 4 shows SIM data for Geosmin and 2-MIB at 1ng/L, and Trichloroanisole isomers and Tribromoanisole at 0.1 ng/L. Interferences from the filtered water were ultimately the limiting factor for Geosmin and 2-MIB, although alternate columns could provide better separation from interfering compounds. In particular, Geosmin nearly coeluted with an interferent that shared the quantitation ion of m/z=112. The higher molecular weight and more unique quantitation ions for TCA and TBA allowed for much lower detection limits, with estimated LODs on the order of 0.01-0.02 ng/L. Table 1 shows the SIM calibration results from 0.1 to 20 ng/L in 50mL of filtered water, using Geosmin-d3 as the internal standard. Chemical noise levels (interferences) for the m/z 112 were too high to allow measurements below 1 ng/L for Geosmin, and 2-MIB also showed significant interferences at 95 m/z, although the 2-MIB in this case showed reasonable linearity down to 0.1 ng/L. In real water samples with unknown interfering compounds, it may be unrealistic to report these compounds below 1 ng/L no matter how much signal is present, as “signal to noise” ultimately determines the practical limits of detection. Although more exact mass TOF analysis can eliminate a lot of these interferences,

1ng/L is below odor thresholds and should be sufficient in most cases. Table 2 shows the results of unfiltered tap water and wine diluted 10:1 with filtered water. Both were run with and without a 5 ng/L spike of the target compounds. The 1.4% alcohol content in the 10:1 diluted wine had an effect on the recovery of the Geosmin-d3 Internal Standard, yet the results showed relatively good recovery of the spiked standard, within an acceptable margin of error. After accounting for the 10:1 dilution, all 4 Trihaloanisoles were visible in the non-spiked wine at 1.1 to 2.5 ng/L, which may have an additive effect in altering the flavor of the wine. The water sample likewise showed an acceptable increase between the unspiked and spiked samples.

ConclusionThe new Sorbent Pen shows great promise in vastly improving the recovery of low volatility, polar compounds from water and wine for more classical analysis without having to perform isotope dilution for each target compound. This is a huge advantage, as it allows the quantitation of compounds for which no isotopically labeled analogs exist. The decoupling of the extraction and analysis process offers a tremendous advantage over other online extraction techniques, as long extraction times are practical while maintaining a high sample throughput. In addition, continued evacuation until all of the liquid is boiled off under the applied vacuum may further increase recovery of high molecular weight compounds. Potentially heavier semi-volatile compounds such as drugs of abuse, endocrine disruptors, carcinogens in foods and beverages, and PAHs, will be possible using extraction times as long as 24-48 hours under vacuum. In general, for all applications where canine detection remains the only viable means of measurement, long vacuum extraction times with Sorbent Pens may offer an alternative.

ReferencesUnited States Geological Survey. 2002. Methods of analysis and quality assurance practices by the U. S. Geological Survey Organic GeochemistryResearch Group – Determination of Geosmin and Methylisoborneol inwater using solid-phase microextraction and gas chromatography/massspectrometry. Open-File Report 02-337.

www.entechinst.com

Entech Instruments, Inc. shall not be liable for errors contained herein or for incidental or consequential damages in connection with the furnishing, performance, or use of this material.

Information, descriptions, and specifications in this publication are subject to change without notice.

© Entech Instruments, Inc., 2016