Using CRDS to measure stable isotopes of water: an overview of the technology, configurations and applications Kate Dennis, PhD IAEA Isotope Hydrology Symposium Workshop 2: New Developments in Stable Isotopic Measurements by Laser Spectrographs and High Precision Water Isotope Analyses May 13, 2015 Vienna, Austria © 2015 Picarro Inc.

Outline • Stable isotopes and isotope hydrology 

• A few example applications  • (A very brief) Introduction to Cavity Ring-Down Spectroscopy • Picarro CRDS applied to water isotopes (d18O, d2H, d17O and 17O-excess), including the CWS • Challenges associated with measuring water isotopes by laser techniques • More than just water…

© 2015 Picarro Inc.

Game-changing measurement technology Isotope Ratio Mass Spectrometry Laser Spectroscopy

Field deployable

© 2015 Picarro Inc.

Picarro’s CRDS Technology Applied to Water Agricultural Water Use Efficiency

Estuary mixing & Groundwater Hydrology

Glaciology & Climate Change

Atmospheric Science & Water Vapor

Oceanography

© 2015 Picarro Inc.

Freshwater Water Source Mapping

CRDS: time, not absorbance • CRDS utilizes the unique infrared absorption spectrum of gas-phase molecules to quantify the concentration of (and sometimes isotopes of) H2O, CO2, CH4, N2O, CH2O, NH3, etc. • Measure decay rate, rather than absolute absorbance

• Small 3-mirrored cavity ~ 35 cc • Long effective path-length (> 10 km) • Time-based measurement • Laser is switched on and off, and scanned across wavelengths

© 2015 Picarro Inc.

Turning ring-down times into concentrations 1. Select wavelength using l-monitor -6

Detector Signal

5.0x10

-6

4.0x10

Absorbance (1/cm)

2. Measure decay time using CRDS

-6

3.0x10

-6

2.0x10

t I circ (t )  I circ (to ) exp    

Time (ms)

-6

1.0x10

0.0 1549.6

-0.4

1549.8

-0.2

1550.0

0.0 Wavelength (nm) Wavelength (nm)

1550.2

0.2

1550.4

0.4

In this figure, relative to peak

3. Calculate loss (a) 𝛼 = 1 𝑐𝜏

Repeat Gas concentration is proportional to the area under the curve, given constant T and P © 2015 Picarro Inc.

I = light intensity in cavity c = speed of light  = cavity ring-down time a = cavity loss per unit length (ppm/cm)

CRDS: Cavity ring-down spectroscopy

© 2015 Picarro Inc.

6

Picarro CRDS applied to water isotopes L2130-i: Our most popular water isotope analyzer for d18O and dD, in the lab or in the field. L2140-i: High-precision triple oxygen isotope research and dD in waters. With associated peripheries: – High Precision Vaporizer – Autosampler for liquid injection

– Standards Delivery Module – Micro-Combustion Module – Induction Module – ChemCorrect™

– Continuous Water Sampler

© 2015 Picarro Inc.

17O-excess:

a unique tracer for hydrological cycle

Luz and Barkan (2010), GCA, 74, 6276-6286. © 2015 Picarro Inc.

8

17O-excess:

a unique tracer for hydrological cycle

17O-excess

can be used to:

– Reconstruct air mass trajectories – Determine source water regions – Reconstruct past humidity – Identify stratospheric injections of water vapor in the atmosphere

– Constrain evapotranspiration budgets at the leaf scale – Understand cloud convection in the tropics – And more…

Luz and Barkan (2010), GCA, 74, 6276-6286. © 2015 Picarro Inc.

9

L2140-i specifications

Precision [1] Liquids (L2140-i plus High Precision Vaporizer)

Drift (24-hour, for liquids and vapor)

d17O

d18O

dD

17O-excess

0.025 ‰

0.025 ‰

0.1 ‰

0.015 ‰

0.2 ‰

0.2 ‰

0.8 ‰

0.2 ‰

Throughput

Memory (within X % of final value after 4 injections)

up to 160 injections per day [2] 99

Guaranteed precision (1s) at 12,500 ppm (‘normal mode’) Vapor [3]

Guaranteed precision (1s) at 12,500 ppm (‘17O-excess mode’) Measurement range Measurement rate

0.04 ‰ at 300 sec

99

98

0.12 / 0.04 ‰ at 10/100 sec

0.3 / 0.1 ‰ at 10/100 sec

0.04 ‰ at 300 sec

0.1 ‰ at 300 sec

99

0.015 ‰ at 3,600 sec

1,000 to 50,000 ppm > 1 Hz

[1] Determined by calculating the standard deviation of the average of groups of 6 injections over a 8 hour period with no drift correction or calibration. [2] Dependent on the number of replicates, this allows for up to 26 samples per day (or 13 if running replicates for 17O-excess). For ‘normal mode’, high throughput mode of vaporizer gives up to 360 injections per day. [3] Specifications are given for both the ‘normal mode’, i.e., operating as a standard L2130-i, and in the ‘17O mode’. © 2015 Picarro Inc.

Minimal drift leads to long-term performance • Repeated measurements of 2 mL vials of identical water (10 injections per vial), with no calibration • Minimal drift over 60 hours: 17O-excess 1s of all vials = 7.7 per meg

Steig et al. (submitted to AMT)

0

© 2015 Picarro Inc.

10

20 Vial number

30

40

Configuration Examples

SDM for ambient vapor

CWS for continuous, real-time water analysis

IM for matrixbound water Vaporizer and Autosampler for liquid water and vapor MCM for plant and soil waters © 2015 Picarro Inc.

Picarro Continuous Water Sampler (CWS) Basic Design and Operation

Active control of parameters affecting kinetic fractionation across membrane • Easy calibration • Stable

d18O and d2H

Previous use of porous membranes for iH2O measurements: Koehler and Wassenaar, Anal. Chem., 83, 913919, 2011 Munksgaard, et al., Rapid Commun Mass Spectrom., 25, 3706-3712, 2011 Munksgaard, et al., Hydrol Process, 26, 36303634, 2012 Munksgaard, et al., Environ Chem Lett, 10, 301-307, 2012 Herbstritt, et al., Water Resour. Res., 48, W03601, 2012

© 2015 Picarro Inc.

In-lab demonstration: Local tap water variability • Real-time evolution of Santa Clara tap water

Real-time, continuous analysis shows how Picarro’s local water authority switches between locally-sourced groundwater and imported water from the Hetch Hetchy Reservoir in the Sierra Nevada Mountains.

Hetch Hetchy Reservoir* dD = -90 to -95 ‰ d18O = -14 to -13 ‰

The system ran un-attended for 7 days with automated switching to water standards.

*USGS SF Water Budget, 1990

© 2015 Picarro Inc.

Continuous Water Sampler • Real-time continuous analysis of d18O and d2H in liquid water – No discrete sampling required. Pump directly from your water source – in a boat, on a river, from a rain collector – Automated switching from samples to standards for calibration – Reproducibility of 0.4 / 1.0 ‰ for d18O / d2H over 12 hours – Quick and easy field deployment

Spatial mapping of real-time d18O and d2H shows enhanced evaporative enrichment as you move up the channel

© 2015 Picarro Inc.

Field testing in cooperation with the USGS does not imply product endorsement by the U.S. Government.

Challenges and tips for users Measuring water isotopes by CRDS, or any other laser technique, has its challenges: 1. Water is sticky…memory 2. Not all water is created equal…spectral interferences

3. All systems drift with time…calibration 4. Water can evaporate…storing standards 5. Moving parts can fail…syringe lifetime

Useful references include: - Picarro community: http://www.picarro.com/community/community_info - van Geldern and Barth (2012), L&O: Methods, 10, 1024–1036 - LIMS for Lasers - Wasenaar, Coplen and Agaarwal (2014), Environmental Science and Technology © 2015 Picarro Inc.

Challenge 1: What is the memory effect? Isotopic memory: the dependence of the next sample on the isotopic composition of the previous sample – After a certain number of injections, the isotopic composition will stabilize – The magnitude of the memory effect is dependent on the instrument’s design and the isotopic difference between two adjacent samples

• CRDS, OA-ICOS, and some conventional IRMS techniques, are affected by sample-to-sample memory • Two common approaches to dealing with memory: – Ignore first injection(s), average remaining – Mathematically account and correct for memory (e.g., van Geldern and Barth, 2012)

• Picarro’s water isotope analyzers have stable, repeatable and predictable memory effects – Within four injections, the L2130-i is guaranteed to arrive at: • 98% of the final, true value for dD • 99% of the final, true value for d18O

© 2015 Picarro Inc.

Isotopic memory – in real life (d18O) 10

2

d18O, vs. VSMOW (‰)

0

Known value = 0 ‰ Measured value = -0.06 ‰

-10

3

-20

Known value = -18.5 ‰ Measured value = -18.49 ‰

-30 -40 -50

1

Known value = -55.5 ‰ Measured value = -55.49 ‰

-60 *Test conducted on an L2120-i, measured value calculations exclude the first 3 injections

© 2015 Picarro Inc.

• Measure 3 samples, spanning over 50 ‰ • Measured values are within 0.1 ‰ of the known value

Isotopic memory – in real life (d18O) 10

2

-10

3

-20

1.01

-40 -50

• Memory effect is repeatable and predictable

Known value = -18.5 ‰ Measured value = -18.49 ‰

-30

• Measure 3 samples, spanning over 50 ‰ • Measured values are within 0.1 ‰ of the known value

1

Known value = -55.5 ‰ Measured value = -55.49 ‰

-60 *Test conducted on an L2120-i, measured value calculations exclude the first 3 injections

1.00

d18O Memory (%)

d18O, vs. VSMOW (‰)

0

Known value = 0 ‰ Measured value = -0.06 ‰

0.99 0.98 0.97 0.96

0.95 © 2015 Picarro Inc.

Specification > 99 % by 4th injection Apparent noise in the memory effect (e.g., on sample 3 above) occurs when the memory effect is smaller than the measurement precision

17O-excess

memory

Presented at AGU 2014 – continued collaboration between UW and Picarro © 2015 Picarro Inc.

Tips for achieving accurate results • Don’t ignore memory! • If sample throughput is of little concern, run at least 6 injections per vial and ignore the first 3 • If sample throughput is important, consider applying a memory correction based on two or three isotopically distinct standards • Design your run with memory in mind • Calibrate at the beginning of each run, and also consider bracketed normalization • Consider running a QA/QC control standard at the beginning, middle and end of each run

© 2015 Picarro Inc.

Challenge 2: Spectral interferences All optical techniques are impacted by spectral interferences • Signs of Spectral Interference: – Elevated baseline – Additional distinct peak(s) – Change in baseline slope – Change in spectral fit residual

contaminated clean

© 2015 Picarro Inc.

Early evidence for spectroscopic interference during isotopic water measurements Plant and soil water extracts: – West et al. (2010) reported discrepancies between waters measured using Picarro L1102-i and IRMS data • d18O deviations up to 15.4 ‰ (decreased to 11.8 ‰ with activated charcoal treatment) • dD deviations up to 46 ‰ (decreased to 35 ‰ with activated charcoal treatment)

Picarro L1102-i compared to IRMS ‘true’ value

West et al. (2010), RCM, 24, 1948-1954 © 2015 Picarro Inc.

Two-fold approach to spectral interference in water ChemCorrectTM – Software to identify and flag spectroscopic interferences – Tested with methanol and ethanol solutions up to: 5% EtOH, 0.1% MeOH and mixtures

Micro Combustion Module (MCM) – Eliminate organic interferences in-line by combusting alcohols to CO2 and H2O – Connects directly to High Precision Vaporizer (A0211)

Rear of vaporizer with MCM attached below

© 2015 Picarro Inc.

24

Demonstrating the capacity of the MCM on a L2130-i

6 8 5

MCM On, Air MCM Off, Air MCM On, N2 MCM On, Air (2)

6

3

dD (‰)

d18O (‰)

4

MCM On, Air MCM Off, Air MCM On, N2 MCM On, Air (2)

2

d18O = 0.02 [conc] – 0.01

4

2

1

d2H = -1.41 [conc] – 0.06 0

0

-1

-2 0.0

0.2 0.4 0.6 0.8 1.0 Alcohol Mixture Concentration (% w/w)

© 2015 Picarro Inc.

1.2

0.0

0.2 0.4 0.6 0.8 1.0 Alcohol Mixture Concentration (% w/w)

1.2

Comparison to IRMS – plant waters

Presented at AGU 2014 – collaboration between Picarro and UC-Berkeley (Dawson Lab) © 2015 Picarro Inc.

Picarro…more than water • Concentration analyzers: – CO2, H2O and CH4 – Fast CO2, H2O and CH4 – CO2, H2O, CH4 and CO – CO2, H2O, CH4, N2O, NH3 – HF – CH2O – H2O2 – NH3

• Isotope analyzers: – d13C in CO2 – d13C in CH4 – d13C in CO2 and CH4 – d18O, d17O, dD and 17O-excess in H2O

– d15N, d15Na, d15Nb, and d18O in N2O

© 2015 Picarro Inc.

One Analyzer…Many Applications OI 1030W for d13C in DIC & DOC

Costech EA for d13C in bulk materials

SSIM2 for small or high concentration samples

d13C in CO2 and CH4

Combustion module for d13C in bulk materials Closed System kit for recirculation © 2015 Picarro Inc.

AutoMate for d13C in DIC in water and solid carbonates

Innovative ways to measure DIC with a Picarro Semi-continuous and autonomous methodologies: ISO-CADICA: Bass et al. (2012), RCM, doi:10.1002/rcm.6143 • Membrane extraction with acidification • [DIC] and d13CDIC • d13CDIC ± 0.1 ‰ with DIC > 0.3 mM

Isotope dilution method: Huang et al. (2013), L&O Methods, doi: 10.4319/lom.2013/11.572 • 13C spiking method with acidification and membrane extraction • [DIC] only • 4 minutes per sample • [DIC] ± 0.07% (± 0.09% shipboard) • Accuracy better than 0.1% © 2015 Picarro Inc.

DOC in produced brines Isotopic Analysis of Dissolved Organic Carbon in Produced Water Brines by Wet Chemical Oxidation and Cavity Ring-Down Spectroscopy – Randal Thomas1, Christopher Conaway1, Nabil Saad2, and Yousif Kharaka1 1 2

USGS Menlo Park, CA Picarro, Inc., Santa Clara, CA

Picarro G2101-i coupled to OI Aurora at Menlo Park:

© 2015 Picarro Inc.

•

d13C of DOC measurements via persulfate oxidation

•

d13C of DIC measurements by acidification

Dissolved Gas Analysis Membrane contactors

Becker et al. (2012), L&O Methods, doi: 10.4319/lom.2012.10.752

Maher et al. (2013), EST, doi: 10.1021/es4027776

Showerhead equilibrators

r2 = 0.90

Discrete Headspace Warner et al. (2013), Applied Geochemistry, doi: 10.1016/j.apgeochem.2013.04.013 Frankignoulle et al. (2001), Water Research

Reverse-flow marble equilibrator © 2015 Picarro Inc.

Questions? • Feel free to contact us offline: – Kate Dennis ([email protected]) – David Kim-Hak ([email protected])

Picarro isotopic water analyzers: used and recommended by the world’s leading scientists, in industry and academia, and by governmental bodies.

*selected Picarro customers

© 2015 Picarro Inc.

32