Optimum methods for the analysis of 1,4-dioxane and the potential for false positives

1
Optimum methods for the analysis of 1,4-dioxane
and the potential for false positives

• Charles Carter, Ph.D.
• Vice President of Quality and Technical Services

September 16, 2014
2
Topics

• Background on 1,4-dioxane as an environmental
  contaminant
• Basic chemistry of 1,4-dioxane
• A story
• Optimum methods for the analysis of 1,4-dioxane
• Field sampling issues that can create false
  positives for 1,4-dioxane
• Another story
• Sample and laboratory conditions that can create
  false positives for 1,4-dioxane
• Achieving standard reporting limits with reduced
  volume samples
• An update

3
Background

• Regulatory levels in between 1 ppb and 3 ppb
• EPA clu-in site indicated detection in 73 of 702
  sources sampled in California
• Above action limit in 28 of these sources
• 67 sites in New Hampshire with concentrations
  from 2 ppb to 11,000 ppb
• Probable human carcinogen
• Used as a solvent stabilizer, as a solvent in
  various chemical processes, and is present as a
  by product in a variety of other materials.

4
Chemistry

• Formula and structure
• C4H8O2
• Boiling point 101.1 C
• Infinitely water miscible
• Miscible with most organic solvents
• Boiling point in the range of standard volatiles,
  e.g toluene boils at 110 C
• Low Henrys Law constant due to miscibility with
  water. Fairly volatile in pure form, not very
  volatile from water.

5
Story 1

• Site contaminated with trichloroethene
• Years of monitoring
• Abrupt appearance of 1,4-dioxane at
  concentrations up to 800 ppb, field blanks and
  laboratory blanks clean
• Data made no sense, but now 1,4-dioxane was the
  major risk driver
• Please review the data
• Perfect
• Please check for contamination
• None
• Please reanalyze the samples and have a separate
  analysis done in a different lab
• Same results
• Now what?

6
Optimum method

• Some criticism from EPA on method choice
• 8260 or 8270, SIM or full scan?
• Poor purge efficiency, but poor extraction
  efficiency as well
• Poor purge efficiency accounted for in
  calculation of response factor
• Is purge efficiency consistent and largely
  independent of matrix?
• Lots of opinions! Lets get some data.

7
Results

• Pulled all matrix spike results for a 10 month
  period

Summary data points Avg. Rec. Avg. RPD
8260 full scan 24565 93.9 19.3
8260 SIM unheated purge 326 98.9 11.3
8260 SIM heated purge 933 103.2 6.2
8270 - 3510 762 39.7 9.1
8270 - 3520 2278 52.4 14.8
8
Test any materials in contact with the sample

• SPLP extracts of well casing, well fittings,
  gloves, cleaning solutions, and other materials
  (in Edison)
• Analysis by 8260 SIM with heated purge (in
  Irvine)
• Numerous materials leached 1,4-dioxane including
  nylon fittings and the FLUTe well liner
• Only those materials that had been either leak
  checked or washed prior to use showed
  contamination
• The leak checks were done with Dawn dishwashing
  detergent, and the field equipment
  decontamination was done with an old bottle
  labeled Alconox.

9
1,4-dioxane in surfactants

• Dawn 7000 ppb
• Alconox 710 ppb
• Ethoxylated surfactants a known issue at FDA
• Ajax Dish Lemon Liquid 1200 ppb
• Clairol Herbal Essences Body Envy Volumizing
  Shampoo 24,000 ppb
• Era 2X Ultra Laundry Detergent 14,000 ppb
• Tide Laundry Detergent 55,000 ppb
• Healthy Times Baby's Herbal Garden Honeysuckle
  Baby Bath 5300 ppb

10
Field blanks

• If this issue is widespread, one would expect to
  find an increased frequency of 1,4-dioxane in
  field blanks.
• Pulled all client sample results where the client
  sample name included FB.
• Results 1204 field blanks, no hits for
  1,4-dioxane above the reporting limit

11
Story 2

• Switched analysis location from one network lab
  to another
• Charlie we have never seen 1,4-dioxane at this
  site and the location never used 1,4-dioxane, but
  now we have positive results for 1,4-dioxane.
  Could you find out if these are false positives
  or laboratory contamination?
• Retention time is perfect
• Spectrum is perfect
• Blank is clean

12
But there are some oddities
13
8 runs later
14
Reanalysis

• Different instrument
• Soil mode with purge in VOA vial instead of water
  purge with standard sparge chamber
• All non-detects
• Decided it was a one time anomaly
• Happened again in 30 days
• Time for a root cause investigation!

15
The known variables

• Desorb temperature
• Old lab 210 C
• New lab 250 C
• Sparge times and flows
• Identical
• Sparge configuration
• Old lab water mode
• New lab both, but 1,4-dioxane only appears with
  water mode
• Unusual detail sulfuric acid is used as the
  field preservative as opposed to hydrochloric
  acid

16
The theory

• If H2SO4 somehow got into the purge gas stream
  and accumulated on the trap, heating could cause
  a dehydration reaction creating gas phase SO3
  during the desorption step
• Methanol is used as the solvent for our internal
  standards and surrogates, so some methanol is
  always present
• Plausible reaction mechanism

17
Three step mechanism

• Methanol reacts with sulfur trioxide forming
  formaldehyde and sulfur dioxide
• CH3OH SO3 ? CH20 SO2 H2O
• Methanol and formaldehyde have a dehydration
  reaction with sulfuric acid to form acetaldehyde
• CH3OH CH2O SO3 ? C2H4O H2SO4
• Acetaldehyde does a cyclic dimerization in the
  acidic gas phase

18
(No Transcript)
19
Is there any supporting evidence?
20
How would sulfuric acid get on the trap?

• Bubble ejection
• Bursting bubbles create aerosols
• Could carry raw sample and preservative into
  purge and trap system
• Water purge versus soil purge
• Smaller bubbles make smaller ejected droplets
• Same flow through each
• Water sparge vessel had 1 cm ID
• VOA vial gt 2 cm ID
• Water sparge vertical velocity more than 4 times
  greater than soil vertical velocity
• Droplets created in VOA vial are more likely to
  settle out. Larger droplets with a lower
  vertical velocity

21
We can test this!

• 5 KCl solution
• Sparge under different conditions into an
  impinger
• Measure chloride in impinger
• Calculate the amount of sample in the purge gas

22
Results

• Coming soon!!

23
Topic 3

• Achieving Standard Reporting Limits with Reduced
  Sample Sizes
• What determines the sensitivity of our methods
• Instrument modifications to increase sensitivity
• Lower Volume Initiative LVI
• SW 846 3510C 3520C with 250 and 125 ml sample
  volume
• SW 846 3511 with 40 ml sample volume
• QA White Paper
• State Certifications
• Unanticipated benefit extraction kinetics

24
Method sensitivity

• Key items are the volume of sample which the lab
  starts with, final volume of the extract and the
  volume of the extract injected onto the GC
  column will determine the Reporting Limits the
  laboratory will provide
• Typical injection volumes are 1 2 microliters
• Injection port modifications 4 to 8 microliters
• What to do with all this additional sensitivity?
• Lower reporting limits
• Increase final extract volume
• Reduce sample size

25
LVI Standard Bottle Sizes
40 ml
26
Advantages

• Efficiency Water samples can be collected at
  more efficient rates especially from low
  production groundwater wells
• Easier Field Logistics Smaller bottles, less
  bulk to transport to sampling site
• Maximized Shipping Capacity More samples will
  fit into a shipping container
• Reduced Shipping Costs Minimized sample weight
  due to reduced volume
• Reduced Breakage Smaller containers are much
  more durable

27
Approvals

• White paper
• Verified acceptability with states
• Demonstration packages

28

• Method 3520
• Heavier than water solvent boils in flask
• Vapor condenses in condenser and drips through
  sample
• Solvent extracts analytes as is falls through the
  sample
• Condensed solvent accumulates in bottom of
  extractor body and flows back into the
  distillation flask

29
Kinetics of extraction

• Change in water concentration is mass removed in
  solvent divided by the volume of sample
• The volume of solvent is the distillation rate
  times the elapsed time, so Vs Q x ?t
• The solvent concentration is the sample
  concentration times a partition coefficient, so
  Cs Cw x k
• Substituting gives
• ?Cw -(k xCw avg x Q x ?t) / V
• dC -kCQ/V dt
• Rearranging and integrating gives the rate
  expression
• C/C0 exp(-kQt/V)

?Cw -(Vs x Cs_avg)/Vw
30
Extracted as a function of time at different
values of k
31
extracted over time as a function of sample size

• C/C0 exp(-kQt/V)

32
The simple version

• Solvent to water ratio
• At a constant distillation rate, the solvent to
  water ration will increase by a factor of 4 if
  the sample size decreases by a factor of 4.
• When the sample size decreases by a factor of 4,
  the distillation time can decrease by a factor of
  4 and we will keep the solvent to water ratio
  constant.

33
Results

• 81 analytes
• 6 replicates at 1000 ml for 18 hours
• 6 replicates at 250 ml for 4 hours

34
Implementation considerations and advantages

• Precedent for reduced time in 3520 accepted by
  EPA and incorporated into the CLP SOW
• Arguably less labor with 3520, definitely better
  recoveries
• With 36 hours of extraction time, 24 hour
  turnaround was not possible
• Many laboratories maintained both procedures
• With 8 hours of extraction time 24 hour
  turnaround is possible

35
Conclusions

• 1,4-dioxane chemistry and optimum methods
• Potential field contamination that can result in
  false positives for 1,4-dioxane
• Specific laboratory conditions that can create
  false positives for 1,4-dioxane
• Reducing sample sizes and unexpected benefits