gas sensor systems for indoor air quality monitoring 1... · gas sensors for indoor air quality...

LAB FOR MEASUREMENT TECHNOLOGYProf. Dr. rer. nat. A. Schütze

COST Action TD1105 – EuNetAirWG Meeting: New Sensing Technologies and Methods for Air-Pollution Monitoring

Gas Sensor Systems for Indoor Air Quality Monitoring

European Environment Agency, Oct. 3 – 4, 2013, Copenhagen

Andreas Schütze Saarland University, Lab for Measurement Technology

SlideGas Sensors for Indoor Air Quality Monitoring – EuNetAir-Meeting – Copenhagen, Oct. 3-4, 2013 2

> Outline

� Introduction: indoor applications and air quality

� Gas measurement systems – more than just sensors� The three “S”� Gas measurement systems� Signal processing and evaluation� Calibration and field testing

� Indoor Air Quality monitoring� Target gases and concentrations� Potential sensor solutions

� Novel developments� Novel sensors: versatile GasFETs� System self monitoring� Multifunctional multisensor systems

� Conclusions

10/7/2013


> Introduction: Indoor Air Quality

Why worry about indoor air?� Safety

� Gas leak detection (combustible gases, e.g. CH4)

� Fire detection (various gases)

� Hazardous gas detection (e.g. CO)

� Malodor detection (kitchen & bathroom ventilation)

� HVAC systems

� Reduced air circulation for greatly reduced energy consumption

�CO2 monitoring for fresh air

� Increased levels of VOCs lead to sick building syndrome

�Selective (formaldehyde, benzene etc.) and sensitive (ppb level) detection

� Systems have to be adapted to the specific room use scenario

10/7/2013


> Introduction: Indoor Air Quality

Sensor requirements� Low cost

� Networked systems (in major buildings, but also private homes)

� Long lifetime: >10 years without maintenance for private homes

Which sensors are used today?� Safety

� Gas leak detection: pellistors (ind.), human nose (in Japan: MOS)

� Fire detection: various sensors, mostly optical; gas sensor systems under development (EC, MOS, GasFET)

� Hazardous gas detection: EC, MOS

� Malodor detection: MOS

� HVAC systems

� CO2 monitoring: NDIR (in major rooms/buildings), EC, GasFET

� VOCs: MOS (total VOC), GasFET (emerging)

10/7/2013


> Gas measurement systems – more than sensors

The three “S”� Sensitivity

� Broad spectrum from ppb (for malodors, ozone, hazardous VOCs) up to 1000 ppm (gas leak, CO2)

� Selectivity

� False alarms are primary concern for fire detection (ratio 10:1)

� VOC detection: hazardous (formaldehyde) vs. neutral (alcohol vapor, cleaning agents) vs. wanted (odorants)

� Stability

� Industrial applications: maintenance interval < 6 months

� Public buildings: annual or bi-annual tests (if that)

� Private homes: 10 years lifetime w/o regular maintenance?

10/7/2013



indoor

atm

osphere

sensor

electronics

signal processing

and evaluation

chemical and/or IR gas sensors

r.h.

T

display

data interface

A generic modular gas sensor system

10/7/2013


Gas measurement systems – more than sensorsTemperature Cycled Operation (TCO) – hardware

• Heater temperature controlHeater resistor RH(T) controlled for exact temperature control of (micro-)hotplates

• Sensor resistance read-outGas sensitive layer: RS(gas)

– Closed-loop control mode� constant voltage drop across RS

– Temperature compensation mode� large dynamic range of 26 bits

• Software controlled

• 16 bit PWM outputs used to apply signals to RH and RS

� now commercialized“OdorChecker” by 3S GmbH

Hardware platformPuMaH for exact temperature control and large dynamic range data acquisition -Pulse-width-modulated Measuring and Heating Unit

Th. Conrad et al., IEEE Sensors Conference 2005

10/7/2013


0

10

20

30

Benzene

0 5 10 15 200

10

20

30

40

50

Diethyl Ether

Iso-Pentane

Methyl Alcohol

0 5 10 15 20

Methyl Tert-Butyl Ether

Con

duct

ance

[a.u

.]

Time in Temperature Cycle [sec]

70% r.h.30% r.h.

Signal

evaluation:

1. Normalization of the response curves ⇒ reduces sensor drift2. Generation of secondary features, i.e. levels, slopes etc.3. Suitable patterns are extracted for further evaluation

A. S

chüt

ze, A

. Gra

mm

, T. R

üh

lIE

EE

Sen

sors

Jou

rnal

, Vo

l. 4

, No

. 6, 2

004

0 5 10 15 20

Propylene Oxide

10/7/2013

Gas measurement systems – more than sensorsTemperature Cycled Operation (TCO) – software

SlideGas Sensors for Indoor Air Quality Monitoring – EuNetAir-Meeting – Copenhagen, Oct. 3-4, 2013

Evaluation of sensor data based on temperature cycling (example)� Virtual multisensor

Characteristic features of the curve shapes (i.e. slope at the end of the high temperature phase and curvature during the low temperature phase) are evaluated, to discriminate between different gases in several steps.

Note: the decision tree reflects the chemical composition of the solvents starting with the alkane pentane and the aromatic benzene (both pure CH-compounds), then the alcohol (R-COH) and finally the three ether compounds (R1-O-R2). This indicates that an expansion might be possible to classify many different molecules.

Decision 3

Synthetic air, Iso-Pentane, Benzene, Methyl Alcohol, Methyl Tert-Butyl Ether,

Diethyl Ether, Prop. Oxide

Decision 4

Decision 1

Decision 2

Methyl 4-Butyl Eth.

clean air & interferents

Propylene Oxide

Diethyl Ether

Benzene

Iso-Pentane

Methyl Alcohol

T-Cycle 2

T-Cycle 1

Sou

rce:

A. S

chü

tze,

A. G

ram

m, T

. R

üh

lIE

EE

Sen

sors

Jou

rnal

, Vo

l. 4

, No

. 6, 2

004

910/7/2013

Gas measurement systems – more than sensorsTemperature Cycled Operation (TCO) – software

SlideGas Sensors for Indoor Air Quality Monitoring – EuNetAir-Meeting – Copenhagen, Oct. 3-4, 2013 1010/7/2013

Target appl.

E. Leo

nhardt, T

hesis p

roject, U

dS

-LMT

, 2007

.

Many possibilities for optimization:

• Sensor selection

• Operating mode

• Data acquisition

• Signal preprocessing

• Feature extraction

• Separation

• Classification

…and always testing under real application conditions (field testing)!

GMA

Sensor

Messhardware

Datennormierung

Merkmalsextraktion

Unterscheidung/Trennung

Klassifizierung

T-Zyklus

Messdaten

Statische

Messung

dynamische

Messung

Signalverarbeitung

Entscheidungs-

prozess

sensors

electronics

normalization

feature extract.

discrimination/separation

classification

gas test system

raw data

T-cycle

Signal processing

dynamic

meas.

Classification

Static

meas.

Gas measurement systems – more than sensorsTemperature Cycled Operation – system design


Calibration: novel gas mixing system for VOC testing down to sub ppb-level

1110/7/2013



Novel gas mixing system: results of reference measurement (zero air)

1210/7/2013


compound CAS no c [µg/m 3] c [ppb]benzene 71-43-2 0.17 0.053toluene 108-88-3 0.06 0.016chlorobenzene 108-90-7 0.26 0.056camphene 79-92-5 0.29 0.052benzaldehyde 100-52-7 0.2 0.046phenol 108-95-2 0.3 0.06benzonitrite 100-47-0 0.61 0.144octanal 124-13-0 0.1 0.019benzyl alcohol 100-51-6 0.19 0.043acetophenone 98-86-2 0.62 0.126naphthalene 91-20-3 0.24 0.046bicyclol[2.2.1]-heptane,2-chloro-2,3,3-trimethyl

465-30-5 16.2 2.6

TVOC 24.3


Novel gas mixing system: results of first sensor tests

1310/7/2013


10000 20000 30000 40000 500003,0x10-3

4,0x10-3

5,0x10-3

6,0x10-3

7,0x10-3

8,0x10-3

9,0x10-3 cond. UST 1000 conc. formaldehyde

time [s]

cond

ucta

nce

[A/V

]

upper range value of ADC

1E-4

1E-3

0,01

0,1

1

10

con

cent

ratio

n [p

pm]

40000 50000

4,1x10-3

4,2x10-3

cond. UST 1000 conc. formaldehyde

time [s]

con

duct

ance

[A/V

]

Sensor reaction to 1 ppb formaldehyde

Relevance?Legal limits in France: Formaldehyde25 ppb in 2015;Benzene 0.6 ppb in 2016


> Indoor Air Quality monitoring

MNT-ERA.net project VOC-IDS� Volatile Organic Compound Indoor Discrimination Sensor

� Scenario specific detection of hazardous VOC

� Integration of sensor system into KNX building automation networks

1410/7/2013


VOC-IDS: ppb-level detection of VOCs against background gases?

Similar requirements: Fire detection in coal mines� CO/C2H4 mixtures at 10 ppm/100 ppb respectively

� Background: up to 1% methane + interfering gases: r.h., CO, H2, NO2 etc.

different temperature levels for the target gases

ramps between these levels allow max. reproducibility

feature extraction

slopes and mean values from these sections

0 10 20 30 40 50 60150

200

250

300

350

400

450

185°C

225°C

185°C

387°C

tem

pera

ture

set

poi

nt [°

C]

time [s]

326°C

C2H4

C2H4 and CO

CO

long term stability

P. R

eim

ann

, Ph

D th

esis

, Ud

S-L

MT

, 201

1

1510/7/2013

> Indoor Air Quality monitoring


> Application examples

gas atmosphere

raw-datafeature extraction

1st LDA

2nd LDA

30% 50% 70%

0.5%

-20 -10 0 10 20

-8

-6

-4

-2

0

2

4

6

1

2

3

70%r. H.

50%r. H.

30% r. H.

UST 1330;: 500ml/min

air flow

disc

rimin

ant f

unct

ion

2

discriminant function 1

1%0.1%

� discrimination along function 1� similar for CH4

P. R

eim

ann

, A. S

chüt

ze:

Sen

sor

Rev

iew

Vol

. 32

Iss:

1,2

012

1610/7/2013


> Application examples

� alarm decision

� discrimination along function 1 & 2

� additional step(s) for interfering gases

3rd LDA

NF

0.1% 0.5% 1%

F NF-100 0 100 200 300 400

-5

-4

-3

-2

-1

0

1

2

3

4

5

disc

rimin

ant f

unct

ion

2

discriminant function 1

fireCH4+CO+C2H4 (1 % + 4 ppm +

ramp 0,06..0,16 ppm)

non-fireCH4 (1 %)

non-fireCH4+CO

(1 % + 4 ppm)

UST 1330air flow: 500 ml/min

increasing concentration

Field test: correlation with existing sensors shown, stability needs to be improved

P. R

eim

ann

, A. S

chüt

ze:

Sen

sor

Rev

iew

Vol

. 32

Iss:

1,2

012

1710/7/2013


> Novel developments

Novel sensor materials and transducer principles� Nanotechnology for improved gas sensitive layers

� Novel transducer principles

� Gas ionization (can be electronically controlled!)

� GasFETs commercially available

Source: Micronas

1810/7/2013



SiC Gas-sensitive Field Effect Sensors (Linköping U, SenSiC)� high temperature operation

� allows temperature cycled operation as for MOS

� (nano-)porous platinum and iridium gate materials

1910/7/2013



SiC GasFETs (Linköping U, SenSiC): high sensitivity

2010/7/2013

Sensor reaction to 2 ppb benzene



System integration: temperature and gate bias variation for SiC-GasFETs

C. Bur et al.: Influence of a Changing Gate Bias on the Sensing Properties of SiC Field Effect Gas Sensors, IMCS 2012

0 1 2 3 4 50

40

80

120

160

200

240 5% O2 in N 2

40 s ramp

curr

ent i

DS (

µA)

gate bias U G (V)

200°C

0 1 2 3 4 50

40

80

120

160

200

240

curr

ent i

DS (

µA)

gate bias U G (V)

400 ppm CO40 s ramp

200°C

0 1 2 3 4 50

50

100

150

200

250

300

150°C

curr

ent i

DS [µ

A]

gate bias U G [V]

5% O2 in N 2

40 s ramp

0 1 2 3 4 5

10

20

30

40

50

60

70

80

90

curr

ent I

DS [µ

A]

gate bias U G [V]

5% O2 in N2

120 s ramp

200°C

0 20 40 60 80 100 120 140

0

1

2

3

4

5

6 Ugate

gate

bia

s U

gate

(V

)

time (s)

2110/7/2013



A. Schütze et al.: Improving MOS Virtual Multisensor Systems by Combining Temperature Cycled Operation with Impedance Spectroscopy, ISOEN 2011

• Electrical Impedance Spectroscopy yields similar improvement in selectivity as Temperature Cycled Operation (but time scale is completely different)

TCO dataEIS data

shifts caused by 100 ppb ethene in 1% methane

2210/7/2013



Sensor self-monitoring with combination of TCO and EIS

A. S

chüt

ze e

t al.:

Imp

rovi

ng M

OS

Virt

ual

Mu

ltise

nso

r S

yste

ms

by

Co

mb

inin

g T

emp

erat

ure

Cyc

led

Op

erat

ion

with

Imp

edan

ce

Sp

ectr

osc

op

y, IS

OE

N 2

011

classification classification

featureextraction

featureextraction

discrimination discrimination

output

diagnostic

T-cycle EIS

deci

sion

pro c

ess

yes

no

difference deci

sion

pro

ces s

hardware for EIS-poor mobility-high cost-long acquisition time

hardware for TCO operation available at reasonable cost

2310/7/2013



A. S

chüt

ze e

t al.:

Imp

rovi

ng M

OS

Virt

ual

Mu

ltise

nso

r S

yste

ms

by

Co

mb

inin

g T

emp

erat

ure

Cyc

led

Op

erat

ion

with

Imp

edan

ce

Sp

ectr

osc

op

y, IS

OE

N 2

011

sensor stimulation data acquisition

Low cost EIS hardware realization: general approach� Implementation using an FPGA (field programmable gate array)� MLS signal refined with variable amplification using dedicated circuit� Data acquisition using high speed ADC (sample rate 200 MHz)

2410/7/2013



MOS-IR measurement system:�Gas filled chamber (9 cm length)

�MOS gas sensor (MICS 5131, e2v)

�Transmission: Thermopile (HIS A21 F4.26, Heimann Sensors)

�Ambient condition monitoring (p, r.h./T)

�Electronics controlled by a microcontroller

�Configuration settings set by a GUI (LabVIEW)

�Data evaluation offline using Matlab

K. Kühn et al.: Investigations on a MOX Gas Sensor as an Infrared Source for an IR-based Gas Sensing System, IMCS 2012

2510/7/2013



MOS signal:

Raw data DC resistance → Shape of the response curve from temperature cycle

0 1 2 3 4200

300

400

500

600

700

Sig

nal i

nten

sity

(a.

u.)

Time (s)

synth. air 40,000 ppm CO2

0.48 0.49 0.50 0.51 0.52

0

1M

2M

3M

4M

DF

T m

agn

itude

|Y

(f)|2 (

a.u

.)

Time (s)

synth. air 2,000 ppm 5,000 ppm 40,000 ppm

0 10k 20k 30k 40k

2M

3M

4M CO2

DF

T m

agn

itud

e |Y

(f=

f A)|

2 (a

.u.)

Concentration (ppm)

Transmission signal:Raw signal (thermopile) → DFT analysis of raw data (90 ON/OFF cycles) → |Y(f=fA)|2(c)

0 1 20.00

0.05

0.10

0.15

0.90

0.95

1.00

1.05 125 ppm CO 120 ppm C

2H

4

No

rma

lized

ele

ctric

al c

ondu

cta

nce

(a.u

.)

Time in temperature cycle (s)

0 500 1,000 1,500 2,000

300

400

500

600

700

800

900

gasair

CO (25, 75, 125 ppm) C2H4 (20, 70, 120 ppm)

Ma

x. c

yc.

val.

elec

tric

al c

ondu

ctan

ce (

a.u.

)

Time (s)

air gas air gas

K. K

üh

n e

t al.:

IMC

S 2

012

2610/7/2013



Transmission signal, fA = 6 Hz square wave mod. of the MOS gas sensor (MICS 5131, SGX)

600k

800k

1M

1M

Transmission signal CO2 (Thermopile detector)

DF

T m

agni

tude

|Y(f A

)|² (a

.u.)

0 5,000 10,000 15,000 20,000 25,000 30,0000

100200300400

75 % r.h.RT

CO2 C

2H

4 NH

3 NO NO

2 CO

Ca

rrie

r g

as

[pp

m]

Time (s)

1.2M

05k10k15k20k25k

CO

2 [ppm

]

K. K

üh

n e

t al.:

Inve

stig

atio

ns

on

a M

OX

Gas

Sen

sor

as a

n

Infr

ared

Sou

rce

for

an IR

-bas

ed G

as S

ensi

ng

Sys

tem

,

IMC

S 2

012

2710/7/2013


> Conclusion

� Indoor applications are of increasing interest, especially for improving energy efficiency and health

� Gas measurement systems are more than just sensors

� Multifunctional, intelligent multisensor systems can address emerging applications

� Application specific development still required

� Field testing is an absolute must for any chemical sensor system

� Field tests of VOC-IDS sensor systems are starting now

� Chemical sensor systems can become ubiquitous in modern building environments and a key to Indoor Air Quality

2810/7/2013

LAB FOR MEASUREMENT TECHNOLOGY Prof. Dr. rer. nat. A. Schütze

Thank you for your attention.

gas sensor systems for indoor air quality monitoring 1... · gas sensors for indoor air quality...

Documents