Improving Non-dispersive Infrared (NDIR) Sensors … › images › documents › CIEJoint2019 ›...

Improving Non-dispersive Infrared (NDIR) Sensors for Greenhouse Gas Monitoring

Howard W. Yoon

1. Motivation

2. Basic concepts, technical approach

3. NIST design for NDIR and comparisons to Picarro CRDS

4. Conclusion and future work

Outline

Keeling curve (from Mauna Loa)

• Measurements using non-dispersive infrared (NDIR) sensor

• Calibrated daily using standard gas bottles

• CO2 concentrations traceable to manometric techniques

http://scrippsco2.ucsd.edu/

400

320

1960 2020

Historical CO2 levels and projection

http://scrippsco2.ucsd.edu/

(2002 CO2 emissions map by Jesse Allen, based on data from the Vulcan Project.)

Ultimate goal of real-time CO2 monitoring using distributed sensor networks

1. In order to reduce CO2 emissions, we need to measure both sources and sinks.

2. Denser sensor grid is needed.

3. Constant calibrations using standard gases are expensive.

4. Better sensors are needed.

Use of commercial NDIR sensors

Use of a commercial NDIR at the heart of the NOAA Tall Tower analysis system

Picarro CRDS system Limited to NIR where lasers exist

Technical Approach

• Gas concentrations measured using Beer-Lambert law • Use of broad-band lamp sources • Dual-band filters (on/off resonance wavelength) • Typically use pyroelectric detectors with modulated source

Licor 820

Beer-Lambert law

𝐴 = 𝜀𝑙𝑐 A absorbance

Molar absorptivity

l path length of the sample

c concentration

Need to adjust path length for desired sensitivity (we use a 60 cm path length)

Calibrations using standard gas bottles (traceable to gravimetric or manometric standards)

Why now?

150 mm focal length ZnSe lens 50 mm diameter

ISP Optics

Temperature-stabilized

Pyroelectric

Detector with

preamp


6 mm diameter

copper

Aperture

100 mm focal length ZnSe lens 25.4 mm diameter

Edmunds Optics


Edmunds Optics


tube (TE cooler)

Laser

Pointer

Foam

Insulation


tube (TE cooler)

Development of the Ambient Radiation Thermometer 8 um to 14 um filter ZnSe lenses Pyroelectric detector

Internal view of the ART

H. W. Yoon, et al. "Improvements in the design of thermal-infrared radiation thermometers and sensors," Opt. Express 27, 14246-14259 (2019)

DN-45 filed May 2019

Repeatability of < 2 mK over 5 days

< 1 mK noise-equivalent temperatures close to ambient temperatures

30 40 50 60 7021.98

21.99

22.00

22.01

22.02

22.03

Tem

per

atu

re,

deg

C

Time (h)

ART

SPRT

Resolution of < 1 mK temperatures at room temperatures

1. Traceable to standard gases 2. Signal of NDIR depends upon

• Internal set temperatures • Detector responsivity • Source stability • Source frequency • Light-pipe stability of transmittance • Humidity

Modification of the ART design for NDIR


Pyroelectric

Detector with

preamp


6 mm diameter

copper

Aperture


Edmunds Optics


Edmunds Optics


tube (TE cooler)

Laser

Pointer

Foam

Insulation


tube (TE cooler)


tube (TE cooler)

Foam

Insulation


tube (TE cooler)


Pyroelectric

Detector with

preamp


6 mm diameter

copper

Aperture


Edmunds Optics


Edmunds Optics


tube (TE cooler)

Laser

Pointer

Foam

Insulation


tube (TE cooler)


tube (TE cooler)

Foam

Insulation


tube (TE cooler)

Schematic of NDIR sensor

Tungsten lamp (5 V 0.06 A) with lifetime of 100,000 hours (11 years of continuous operation) Pulsed with sine-wave signal from a function generator at 3.9 Hz

Copper, straight- tubing light pipe (0.5 inch diameter) 60 cm long

CaF lenses inside a temperature-stabilized tube

One of two filters at 3.9 um (off) or at 4.26 um (on)

ZnSe windowed TE-cooled pyroelectric detector with preamplifier Lockin detection system used

Sealed using sapphire windows at both end of tube

Upper view of setup

Light-pipe effect of tubing

Regular copper straight pipe works well Will also use ultra-polished SS tube and Au-coated Cu pipe

Comparisons to Picarro cavity ring-down spectroscopy (CRDS) instrument

Picarro CRDS 0.05 ppm resolution 0.5 ppm drift over 1 month $30k/unit NDIR sensor 0.15 ppm resolution 1.0 ppm over 3 months (TBD) $2k/unit (TBD)

Off-resonance signal stability is 0.2 ppm of CO2 for hours

Stability of off-resonance signal at 3.6 um

0 20 40 60 80 100

0.09626

0.09627

0.09628

0.09629

0.09630

0.09631

NDIR 7 Hz off resonance 3_6 um.opj

Sig

nal

(V

)

Time (min)

50 uV or 1 ppm

NDIR CRDS comparison 1 (30 hour duration)

5825 5830 5835 5840 5845 5850 5855

360

370

380

390

400

410

28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58

-0.0705

-0.0700

-0.0695

-0.0690

-0.0685

-0.0680

Pic

arro

, C

O2 c

onc.

ppm

Time, hours

Picarro

ART NDIR Picarro comparison with Sapphire windows 60 cm tube heated to 40 deg C and N2 purge 09_01_19.opj

ND

IR

NDIRScaling NDIR vertical axis to have it overlay on Picarro (using Picarro to calibrate the NDIR)

Short-term instability in the Picarro signal due to fluctuations in nitrogen gas purge

Deviation in NDIR due to system drift of ~2 ppm in 12 hours

NDIR CRDS comparison 2 5874 5876 5878 5880 5882 5884 5886 5888

370

380

390

400

410

420

430

76 78 80 82 84 86 88 90 92

-0.0702

-0.0700

-0.0698

-0.0696

-0.0694

-0.0692

-0.0690

-0.0688

-0.0686

-0.0684

-0.0682

-0.0680

-0.0678

-0.0676

pic

arro

picarro

nd

ir

Time, hours

ndir

NDIR sub ppm stability and resolution

5906 5907 5908 5909 5910 5911 5912

372

374

376

378

380

382

384

386

5906 5907 5908 5909 5910 5911 5912

-0.0698

-0.0697

-0.0696

-0.0695

-0.0694

-0.0693

-0.0692P

icar

ro,

CO

2 c

on

c. p

pm

Picarro

ND

IR

Time, hours

NDIR

Constructing a dual-band system

4.26 um 180 nm 3.06 um 160 nm

Lessons learned

Signal of NDIR depends upon 1. Internal set temperatures (stabilize

temperatures of both tube and detector assemblies)

2. Detector responsivity (use interference filters on sapphire substrates and protected pyroelectric detectors)

3. Source stability (use long-life, low-power lamps)

4. Source frequency (use a function generator as a source with sine-wave output)

5. Light-pipe stability of transmittance (use gold-coated Cu pipe or SS pipe)

6. Humidity (condition gas before analysis)

1. NDIR sensors can be improved with better pyroelectric detectors and an actively stabilized detector compartment

2. The improved designs can be competitive with CRDS over short periods (12 hours)

3. The improved design can measure CO2 with 0.2 ppm noise

4. We expect to achieve 1 ppm stability over 3 months with the dual-band detector and a thermally stable and structurally stable setup

5. Other gas species between 1 um to 14 um can be measured with lower uncertainties with the improved design

Conclusion

Improving Non-dispersive Infrared (NDIR) Sensors … › images › documents › CIEJoint2019 ›...

Documents

Transcript of Improving Non-dispersive Infrared (NDIR) Sensors … › images › documents › CIEJoint2019 ›...