development and application of a miniaturised sensor...

Development and Application of a Miniaturised Sensor System for Respiratory Investigations (MAP-RSS)

Project No. AO-99-058 in the frame of the Microgravity Application Promotion Programme (MAP)

RSS-Exec-Sum-March2008 Abstract and Executive Summary to the Final Report to the ESTEC Contract No. 14350/01/NL/SH

Edited by: Stefanos Fasoulas1 and Uwe Hoffmann2

Team members: 1Institute for Aerospace Engineering, Technische Universität Dresden, D-01062 Dresden, Germany

2Physiologisches Institut, Deutsche Sporthochschule, Carl-Diem-Weg 6, D-50933 Cologne, Germany 3ESCUBE GmbH, Nobelstr. 15, D-70569 Stuttgart, Germany

4Medisoft S.A., P.A.E. de Sorinnes – n°1 Grand Routede Ciney, B-5503 Sorinnes, Belgium 5Environmental Physiology Laboratory, Karolinska Institutet, S-10401 Stockholm, Sweden

6Laboratoire de Physique Biomedicale, Universite Libre de Bruxelles, Route de Lennik 808, CP 613/3, B-1070 Bruxelles, Belgium 7PARI GmbH, Aerosol Research Institute, Steinerstr. 15c, D-81369 Munich, Germany

8Institute for Automation Technology, Universität Rostock, D-18051 Rostock, Germany 9Institute of Occupational Health, Universität Rostock, D-18051 Rostock, Germany

Abstract

The MAP-RSS project dealt with the development and application of a new sensor system for human respiratory investigations. Different institutions, including three industrial partners, working on different fields of science and technology, combined their expertise by focusing on two selected applications in the field of spiroergometric exercise testing and lung function diagnostics with subsequent medication. The main goals of this project have been to develop miniaturised oxygen, carbon dioxide and eventually hydrogen or carbon monoxide sensors, to use their capability for simultaneous measurement of total flow rates, to integrate them into a mask for the in-situ measurement of respiratory parameters, and to perform first validation and scientific tests. A small and light-weight device for astronaut fitness surveillance and scientific experiments on-board the International Space Station ISS or other space missions may be targeted for a future development, having also a high potential for different terrestrial applications. This executive summary is intended to give a “stand-alone” overview about the sensor working principles, the performed development efforts for the complete systems, and, based on some selected results, the broad applicability of the system.

1 Introduction and Background

Since 1993 a considerable amount of research and development has been performed at the Stuttgart University and the TU Dresden on the topic of measuring oxygen partial pressures in various space applications using solid electrolyte gas sensors. These sensors have been applied for example to measure the oxygen amount inside plasma wind tunnels, where modern heat shield materials are tested and qualified as thermal protection shields for re-entry space vehicles. In parallel, different space flight experiments were performed to measure the

residual oxygen partial pressure in low earth orbit, which served mainly as precursor experiments for an experiment on-board the International Space Station (“FIPEX on ISS”), which is dedicated to the measure-ment of the natural and induced environment (atomic / molecular oxygen). This experiment has been launched to the ISS together with European module “Columbus” in February 2008. Because commercially available sen-sors did not fulfil the requirements for these space appli-cations, a new sensor development was initiated (Fig. 1). The first results indicate an “as-expected-behaviour” of the experiment and a continuous operation and the corresponding data evaluation is currently ongoing.

Very stimulating for the development of the new sensors was also the high interest which simultaneously arose in industry and science for a variety of terrestrial appli-cations, e.g. for environmental and combustion control, medicine and vacuum applications. The reasons were mainly the miniaturised design and other features of the new sensor elements (e.g. capability to detect different gas species, e.g. O2, H2, CO, CO2, etc.). Modern production techniques and innovative materials were then introduced to allow for high quality, reliable, and minia-turised sensor element design (Fig. 2). Another feature, which resulted from the sensor miniaturisation, is the possibility to simultaneously measure other physical properties, e.g. total flow rates. This characteristic made the new sensor elements very attractive for a variety of medical applications on ground and in space.

Fig. 1: Sensors used for the space experiment FIPEX on ISS.

Fig. 2: Typical sensor element for terrestrial applications.

Executive Summary to the ESTEC Contract No. 14350/01/NL/SH MAP - Respiratory Sensor System

-2-

For example, for many manned space missions, and espe-cially on the ISS, there is a need for a small, light-weight, portable, potentially body-mounted, metabolic gas analy-ser with which periodic fitness/scientific evaluations on the astronauts could be performed. This device would offer high flexibility to the astronauts, enabling them to conduct experiments at every location on-board the ISS while wearing a mask with the sensor system and per-forming respiratory experiments. Thus, the device should be capable of being a stand-alone system providing a real-time data display to permit evaluators to monitor da-ta and archiving the metabolic data for post analysis (Fig. 3). Of course, such metabolic gas analyser systems also have a high potential for terrestrial applications.

Therefore, the new technological solution was proposed to ESA within the Microgravity Application Promotion (MAP) Programme, AO-99-058, by a group of scientists and industry. Already at the beginning of the project, the team members identified the following potential advantages of the new system:

• Short response time allowing “breath-to-breath” or even a single breath analysis.

• Measurement in mainstream, capability of measuring O2, CO2, and flow rate simultaneously.

• Miniaturised and light-weight design. • Possibility of mobile/portable use. • Potential for detecting other gas species. • Potential for “low-cost” fabrication of sensors.

The main objectives of the awarded project have then been to investigate the specific scientific and application benefits of the system (including terrestrial applications), and to confirm, verify, and validate these characteristics on prototype systems, i.e. in some detail:

• To advance the flow-sensing capabilities of the solid electrolyte sensor technology,

• to integrate the sensors into a single, small package, capable of being positioned directly in the inspired/expired flow path of a human subject,

• to optimise the flow-, O2- and CO2-sensor sensitivity, stability and reproducibility,

• to optimise the electronics design, using a smaller amount of components, lower overall power consumption, reduction of housing dimensions, etc.,

• to extend the controller firmware capabilities, including a more complete host-PC interface, a calibration routine, and a flow direction detection,

• to assemble demonstration systems, which shall be used / tested within the ESA-MAP-project, and

• to perform a first assessment for space application (safety, development efforts, data storage etc.).

Taking into account the planned end-use, the project was aimed at demonstrating that the sensor system provides a performance appropriate for metabolic measurements on human subjects, particularly for manned space missions.

In the following a brief summary of the measurement principles, the design of the sensors and some selected interesting results as obtained up to now (March 2008) is given.

2 Measurement Principles

2.1 Oxygen Measurement

The working principle of the sensors is based on solid state electrolysis. For the oxygen sensors, yttria-doped zirconia is used as the electrolyte, which selectively con-ducts oxygen ions at higher temperatures (above about 450°C). Therefore, the sensor has to be heated up to a constant working temperature. Oxygen concentration is then determined by measuring the current that is conduc-ted through the electrolyte at a constant applied voltage (Fig. 4). If a diffusion barrier limits the oxygen flux from the ambient air to the cathode, the resulting electrical current depends linearly on the ambient oxygen concen-tration. However, in order to achieve very short response times, the length of the diffusion barrier should be as small as possible. Thus, in the applied planar design, the printed electrolyte itself is one part of the diffusion bar-rier (Fig. 5) and some additional layers above the electro-lyte assist in adjusting the required porosity. Of course, the sensor signal depends also on the working tempera-ture in a relatively complex manner, e.g. by the tempera-ture dependencies of the adsorption/desorption processes on the electrodes, the ionic conductivity of the electro-lyte, and by several other occurring physical phenomena that are relevant for the sensor signal.

Fig. 3: A complete system for monitoring gas exchange parameters.

A

diffusionbarrier cathode

current anode

O2-

O2

O2

solidelectrolyte

constantvoltage

Fig. 4: Working principle of the amperometric oxygen sensor.

Anode

Electrolyte

Cathode

Substrate

Heater

Planar Design

Size:20 x 3,5 x 0,5 mm

Anode

Electrolyte

Cathode

Substrate

Heater

Planar Design

Size:20 x 3,5 x 0,5 mm

Fig. 5: Main layers of the amperometric oxygen sensor.


-3-

Thus, the heater design itself is optimised to obtain an uniform temperature distribution and to induce only a minimum thermal stress on the sensor. Some of the investigated heater designs are shown in Fig. 6. Of course, in order to minimise heat losses to the surroun-dings and thus to minimise the overall power demand, the sensor must be insulated adequately (see Fig. 7).

The sensor is manufactured by screen printing and sub-sequent sintering of several metallic and ceramic layers on a ceramic substrate. This innovative production tech-nique allows for a mass production and miniaturisation (Fig. 8). The current sensor size is about 20x3.5x0.5mm. It should be mentioned here that the development of the sensors is an iterative and hardly predictable process, because minimal changes in each layer composition and/or sintering procedure may lead to very different results. Therefore, more than 140 different sensor con-figurations have been designed, manufactured and analy-sed. This optimisation process resulted in a remarkable improvement of the sensitivity and stability of the oxygen measurement, leading to an almost ideal characteristic (Figs. 9-11).

Different tests demonstrated the good sensitivity to oxy-gen and the exceptionally fast response to O2 changes (e.g. Fig. 12). Some remaining drift effects are identified to result mainly from inaccuracies in sensor temperature measurement and control, they are not caused by the oxygen measurement itself. However, this influence is relatively low now.

0%

20%

40%

60%

80%

100%

0 2 4 6 8 10 12 14 16 18 20 22Oxygen Partial Pressure by Flow Controller [%]

Rel

ativ

e Se

nsor

Sig

nal

(21%

O2

= 10

0%)

Characteristic 31/10/2000

27/07/2001

24/08/2001

15/05/2002

23/07/2003

Ideal characteristic

0%

20%

40%

60%

80%

100%

0 2 4 6 8 10 12 14 16 18 20 22Oxygen Partial Pressure by Flow Controller [%]

Rel

ativ

e Se

nsor

Sig

nal

(21%

O2

= 10

0%)

Characteristic 31/10/2000

27/07/2001

24/08/2001

15/05/2002

23/07/2003

Ideal characteristic

Fig. 9: Characteristics of the different oxygen sensor generations

illustrating the improvements in the past (0-21% oxygen). The sensor signals are divided by the sensor signals at 21% O2.

0%

20%

40%

60%

80%

100%

120%

0 5 10 15 20 25

Oxygen Partial Pressure by Flow Controller [%]

Rel

ativ

e Se

nsor

Sig

nal

(21%

O2

= 10

0 %

)

Fig. 10: Characteristic of an oxygen sensor (0-21% oxygen, Jan. 2005).

The sensor signal is divided by the sensor signal at 21% O2. The dashed line marks the ideal characteristic, the error bars indicate the errors in the flow controllers used for calibration.

0%

20%

40%

60%

80%

100%

120%

0 20 40 60 80 100

Oxygen Partial Pressure by Flow Controller [%]

Rel

ativ

e Se

nsor

Sig

nal

(90%

O2

= 10

0 %

)

Fig. 11: Characteristic of oxygen sensor (0-100% oxygen, Jan. 2005).

The sensor signal is divided by the sensor signal at 90% O2. The dashed line marks the ideal characteristic, the error bars indicate the errors in the flow controllers used for calibration.

315

320

325

330

335

340

345

350

-30 0 30 60 90 120 150Time [ms]

Sens

or C

urre

nt [µ

A]

15

16

17

18

19

20

21

22

Mas

s Sp

ectr

omet

er O

xyge

n C

once

ntra

tion

[%]O2-Sensor

Mass Spectrometer

Fig. 12: Evaluation of response times in comparison to a mass spectro-meter (test performed at German Sport University, Cologne, with one of the first sensor generations, sensor integrated in mask, all gas exchange effects included).

Fig. 6: Some examples of the heater layout.

Sensor

Thermal insulationCables (6)

Connector

Retaining screwsSensor

Thermal insulationCables (6)

Connector

Retaining screws

Fig. 7: Miniaturised oxygen sensor (overheated at about 800°C) with

insulation and connectors (left) and sensor integrated in the first prototype mask adapter (right).

Fig. 8: Substrate with 54 sensors (broken in three parts for

illustration).


-4-

2.2 Carbon Dioxide Measurement

The development of a respiratory sensor system requires also a fast and reliable measurement of the carbon dioxi-de concentration. The chosen working principle is com-parable to the oxygen sensors, the electrolyte material however is now based on NASICON (Natrium super ionic conductor). Again, a planar sensor design in multi-layer technique is used for a simple reproducible serial production (Fig. 13). Therefore, different pastes have been developed for the different layers. After the printing process, the pastes have to be sintered at high tem-peratures. Again, the numerous parameters of these steps have been determined and optimised in various tests.

However, unlike the oxygen measurement, the voltage between the electrodes of the CO2-sensor depends loga-rithmically on the partial pressure of CO2 (Fig. 14). In order to obtain accurate measurements, the slope of the characteristic has to be as steep as possible and one main emphasis of the sensor development addressed this issue. Thus, the effect of the error induced by temperature variations of the electrodes, which cause minor voltage changes, gets smaller. On the other hand, the temperature variation of the electrodes depends on the flow rate (i.e. the higher the flow rate the higher the influence).

Generally, the sensor works very well for low CO2 concentrations (Fig. 15) and/or low flow rates. Because of the logarithmic characteristic, the sensor meets the requirements for concentrations up to 2-3% CO2. For higher concentrations and especially higher flow rates the characteristics of the sensor has still to be improved or a numerical compensation of this influence has to be performed. In addition, the influence of the flow could be minimised by shielding or placing the CO2-sensor slightly out of the flow.

2.3 Hydrogen Measurement

The working principle of the H2-sensor is the solid elec-trolyte potentiometry, similar to the CO2-sensor. How-ever, a similar electrolyte as for the oxygen sensor is used and only the electrode materials are varied in order to increase the sensitivity of one electrode to H2 (and/or other combustible gases, e.g. CO!). To enlarge the sen-sors’ selectivity, special catalysts, filters or surface addi-tives are also used. Fig. 16 illustrates such a Non-Nernstian probe with a different surface coverage of the two electrodes corresponding to the partial pressures of the reactive gaseous components O2 and CO (exemplary) in the gas phase. Fig. 17 depicts a typical characteristic of the sensor for H2.

Within the MAP-project, different design iterations were necessary in order to increase the measurement range to higher H2-concentrations. Then, also geometrical design adaptations were applied to meet the successful O2 sensor design and thus to minimise cooling flow cross sensitivities.

measurement electrode

alkali carbonate

counter electrode

ionic conductorNASICON

optional reference

carrier substrate

heater

Fig. 13: Planar carbon dioxide sensor design.

-150-140-130-120-110-100

-90-80-70-60-50-40-30-20-10

0

0,1 1 10CO2 percentage

EMV

- Diff

eren

ce in

mV

March 02August 03August 04

Fig. 14: Evolution of CO2 sensor characteristics.

0,0

0,2

0,4

0,6

0,8

1,0

1,2

1,4

860 870 880 890 900 910

Measurement Time [s]

CO

2 Pa

rtia

l Pre

ssur

e [%

]

Fig. 15: Measurement by quickly moving sensor into / out of a free

stream (free stream: 20 % O2, 1 % CO2, 79% N2).

electrode 1 electrode 2electrolyte

ads. Phase:

gas phase:

K1,1 K2,1 K1,2 K2,2

p =90%, p =9%, p =1%N2 O2 CO

O =97%, CO =3%2,ads adsO =1%, CO =99%2,ads ads

US

Fig. 16: Illustration of the working principle of Non-Nernstian sensor

device (here shown for CO).

0

100

200

300

400

500

600

700

0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 2

H2 - concentration in %

sens

or v

olta

ge in

mV

Fig. 17: Characteristic of the CO/H2-sensor to H2 at 21% oxygen


-5-

2.4 Flow Rate Measurement

One basic idea of the solid electrolyte sensor develop-ment is to measure both, total flow rates and gas concen-trations simultaneously using a single sensor element with high time resolution. Therefore, the so-called thin film anemometry was introduced for the total flow rate measurement (Fig. 18). Due to the cooling effect of a gas flow over the hot sensor surface (forced convection), it is necessary to increase the heating power in order to main-tain the temperature on a constant level, which is needed for an accurate gas composition measurement anyway. Thus, the sensor electronics control the operating tempe-rature via a resistor heater. Consequently, by measuring the consumed electrical power for keeping a constant sensor temperature, the original cooling fluid flow rate can be determined. However, although the method is well known, a special adaptation to the new sensor system was incorporated taking into account some basic require-ments such as a high time resolution, the relative high thermal capacity of the sensor, and the high temperature level. Therefore, the theoretical understanding of the relevant physical phenomena, as well as the measurement of signals in addition to the power signal, has been necessary to achieve precise flow rate measurements.

For accurate measurements a “smooth” and stable flow pattern has to be established over a high flow rate range. Therefore, several designs of meshes for flow lamina-risation as well as different positions and orientations of the sensor within the tube were tested, taking also into account the necessity to minimise the volume within the measurement chamber (Fig. 19). However, the most im-portant improvement in flow measurement resulted from the successful implementation of a micro-controller into the sensor control electronics. In combination with va-rious control algorithms the control process showed an almost ideal behaviour and a response time of less than 50 ms was achieved. Fig. 20 shows a typical calibration curve for the flow rate measurement. Figs. 21-22 sum-marize the comparison of the response time and accuracy

relative to available flow rate measurement devices. In Fig. 21 the results of an early experiment with high flow rates are shown, where the flow direction detection was not incorporated yet. Fig. 22 shows the results at relati-vely low flow rates in comparison to a flow turbine.

VFluid

Radiation Free convection Forced convection Heating

Heat conduction Fig. 18: Measurement principle of thin film anemometry.

Sensor hole

Meshes

Honey combs

Sensor hole

Meshes

Honey combs

Fig. 19: A prototype of the flow tube with flow pattern improvement, low breathing resistance, and a dead space of about 25 ml.

0

2

46

8

10

12

14

16

18

20

2 3 4 5 6 7 8 9 10 11 12 13 14Sensor Heating Power [W]

Flow

Rat

e [l/

s]

O2-Sensor:V = 0,471 * (P-P0)

1,608

Error barof reference

signal

CO2-Sensor:V = 0,771 * (P-P0)

1,396

.

.

Fig. 20: Typical calibration curve for the flow rate measurement

(reference: pressure gauge).

0

1

2

3

4

5

6

7

8

9

10

11

12

4,0 4,5 5,0 5,5 6,0 6,5 7,0 7,5 8,0 8,5 9,0 9,5 10,0Measurement time [s]

Vol

ume

flow

rate

[l/s

]

Flow rate derived from sensor signalFlow rate from ultrasonic reference measurement

Fig. 21: Comparison of flow rate derived from the solid electrolyte

sensor heating power and a reference ultrasonic device.

-0,15

-0,10

-0,05

0,00

0,05

0,10

0,15

7380 7381 7382 7383 7384 7385 7386 7387 7388 7389 7390

Measurement Time [s]

Flow

Rat

e [l/

s]

Flow Turbine

RSS

Fig. 22: Comparison of flow rate derived from the solid electrolyte

sensor heating power and a reference flow turbine.

Temperature profile

Temperaturechangedepends onflow direction

Flow direction

Thermoresistor add-onfor measurement of temperature profileon the sensor

Fig. 23: Basic principle for flow direction detection independent of

gas concentration changes.


-6-

Generally, the flow direction cannot be determined by the necessary heating power only. Therefore, two possi-bilities were investigated to obtain this information by other means. The first is based on the rapid gas con-centration changes, i.e. by measuring the concentration gradients in order to distinguish between inspiration and expiration. However, although this method works well for relatively high flow rates, it introduces a small error due to the dead space of the sensor adapter. Therefore, another solution was introduced by measuring the temperature profile on the sensor which depends on the flow direction (Fig. 23). This method works well and is now integrated in the sensor system.

3 Further Hardware, Software and Investigations

In parallel to the sensor development the digital control electronics (see Fig. 24) and the necessary software have been developed. The data flux in the system is illustrated in Fig. 25, a screen shot of the software in Fig. 26. Also, because of the absolutely new concept of measuring res-piratory parameters in-situ, various peripheral hardware, software, and theoretical considerations were developed, improved, and analysed, such as:

• Sensor adapter for the breathing mask (Fig. 19) • Mask adapters considering flow smoothening and

laminarisation aspects • Visualisation software (Fig. 26) • Application specific software • Calibration devices • Simulation devices • Battery driven system (Fig. 27) • Alternative data storage (PDA, USB-Memory-Stick

see Fig. 27) • Identification, definition, and conduction of relevant

tests.

Furthermore, different design studies have been con-ducted in order to allow for an easier and more repro-ducible sensor placement in the mask. An example is shown in Fig. 28.

Different validation tests have been performed in order to analyse the overall sensor system behaviour. Some re-sults were also obtained by testing commercial spiroergo-metric metabolic carts (e.g. respiratory mass spectrome-ter, ultrasonic flow meter) and the existing lab system under the aspects of validity, reproducibility, signal stabi-lity, and handling in comparison to the targeted appli-cations of the sensor system. The results are summarized in Table 1 in comparison to the requirements.

The overall power demand is very important for the plan-ned application of the system. In the current status, the peak power demand of the system at a peak flow rate of approx. 20 l/s, two sensors, and electronics version 1.5, is lower than 30 W (without losses of power supply, i.e. AC/DC-converter). The mean power demand is about 15 W. However, there are several possibilities to reduce this power demand in the future: The positioning of the sensors can be optimised, smaller sensors can be used, and finally integration of the O2- and CO2-sensor on one sensor element could be considered.

Version 0.8 1.0 1.5 Width 105 mm 105 mm 105 mm Depth 175 mm 175 mm 118 mm Height 105 mm 75 mm 39 mm Volume 1.93 l 1.38 l 0.48 l

Power supply incl.

Yes No No

Mass (supply)

1425 g 975 g (600 g)

426 g (600 g)

Fig. 24: Different sensor control electronics developed in the project.

Analogue sensor

signals

Micro-controller

(heat control anddigitalisation)

Sensor(measurement of gas

mixture and flow)

MC-Driver(for data

acquisition andcalibration)

PC

Raw data inRS 232 buffer

Time, Flow,O2 and CO2 datain DDE buffer

UserProgram

(for data processing, storageand visualisation)

Digital sensor signals via serial port

Serial port DDE interface

Current sampling rates:

1000 Hz in MC250 Hz in PC

Analogue sensor

signals

Micro-controller

(heat control anddigitalisation)

Sensor(measurement of gas

mixture and flow)

MC-Driver(for data

acquisition andcalibration)

PC

Raw data inRS 232 buffer

Time, Flow,O2 and CO2 datain DDE buffer

UserProgram

(for data processing, storageand visualisation)

Digital sensor signals via serial port

Serial port DDE interface

Current sampling rates:

1000 Hz in MC250 Hz in PC

Fig. 25: Scheme of the data flux in the system.

Fig. 26: Screen-shot of the software used for visualisation.

Fig. 27: Electronic unit in current status (PRO1.5) with rechargeable

battery pack for an operation time of about 3 hours (left) and communication between the PDA and the RSS-EB (right)

110

110

Fig. 28: Designs for a new sensor connector and a complete sensor adapter.


-7-

Very important cytotoxicity tests concerning the appli-cability of the sensor system were performed already within the frame of the in parallel ongoing MAP-project. These tests have been performed for the CO2 and O2/flow sensors and were conducted according to DIN EN ISO specifications. The test results showed that the sensors do not release substances in cytotoxic concentrations during a 24-hour constant contact period of a 4.5 cm2 surface area with 1 ml physiological fluid.

Other preliminary investigations were related to safety aspects of the high temperature sensors. A safety risk was identified in the case of sensor destruction as soon as hot fragments are small enough to pass through the surroun-ding metal mesh. Here, two scenarios were investigated (Fig. 29): First, during inspiration, i.e. fragments move towards test person passing through the mesh, honey comb and biological filter. Here, the analysis showed that the energy needed to melt the biological filter is higher than the energy content of the largest particle which could pass the laminarisation mesh. Also during expiration, the mesh provides protection from larger particles with higher overall energy content.

4 Complete System Experiments

A huge amount of calibration and qualification experi-ments have been performed to derive isolated sensor characteristics, some of the results have been presented in the previous chapters. However, the key issue of the project has been to simultaneously measure the required properties under real conditions. Thus, in the first phase of the project the focus was laid more on the adjustment of the necessary specifications to the sensor design. Then, in the second phase, more emphasis was laid on the testing of complete systems in different applications. Certainly, this has been a very iterative process in order to meet the very challenging demands for the system, especially concerning the broad applicability. Some of the most interesting results obtained within these test campaigns are presented in the following, namely

• Testing of complete system in a calibration device • Centrifuge tests • Parabolic flight experiments

• Artificial respiration during an animal surgery • System tests with hydrogen sensors • Tests with reduced ambient pressure • Sidestream microchamber application

4.1 Testing of complete system in a calibration device

One of the most recent prototype systems has been tested with the gas exchange simulation system (GESS) desig-ned by the German sports university Cologne (DSH) in the first MAP phase (Fig. 30). Fig. 31 illustrates the high correlation between the RSS flow signal and the signal derived from the position sensor of the GESS. The GESS gives the opportunity to run precisely the same profile in terms of flow and volume. The RSS can be evaluated under “dry” conditions and then within BTPS conditions. As shown in Fig. 32 there are no influences related to hu-midity to the O2 sensor signal and only minor effects to the flow measurement. The CO2 sensor shows however some effects between “dry” and “wet” measurements.

Fig. 30: Extended gas exchange simulator system - GESS.

Fig. 31: Correlation between the RSS flow signal and the signal

derived from the position sensor of the GESS.

Fig. 32: Performance of the O2 and CO2 sensor in dry against wet

conditions.

Fig. 29: The respiratory sensor system – current status. A potential

safety risk exists in case of sensor breaking (A: inspiration, B: expiration). Note: Sensor design (especially heater) is optimised to induce minimum thermal stress. After approx. 5 years of sensor testing, having investigated >2.000 sensors, not a single break-up occurred.

1a/1b: Expiratory / inspiratory calibration syringes 2: Position sensor 3: Connector with 2 one-way valves 4a/4b: Expiratory/inspiratory one-way valves 5: Port to sensor system + temperature sensor 6: Calibration gas reservoir (polyethylene bag) 7: Calibration gas cylinder 8: Heater/humidifier chamber 9: Pumping handle


-8-

4.2 Centrifuge Tests

The centrifuge tests were performed at the DLR Cologne (Fig. 33). The experiment was coordinated by the Ger-man sports university Cologne (DSH). The study was aimed to observe the influence of increased gravity to hu-man capabilities and the learning aptitude to the human motor function. A total sum of 30 persons have been tested, 24 students and 6 pilots. The tests were performed under real conditions and gave good information about the handling of the prototype. The electronics was specially adapted with an additional event marking and analogue outputs. The RSS-prototype was used in continuous operation for about 4 hours and more. The main results of this test campaign have been:

• Stability of the concentrations displayed by the RSS is very good, an example is depicted in Fig. 34.

• The inspiratory and end-expiratory values could be used properly.

• Disturbances in direction detection and offset of switching point did not allow exact volume balance calculations. (Important note: The concentration independent solution for the flow direction detection was not incorporated yet).

• The respiratory rate was taken from concentration changes, which could be done properly.

• All of the RSS equipment was used under 3g!

4.3 Parabolic Flight Experiments

The aim of these experiments was to monitor the effect of gravity changes during parabolic flight to the working principles of the sensor system. The influence of the missing free convection to the heater could be measured and possibly compensated. Thus, the effect of errors

induced by temperature and/or pressure changes should be examined. The main parts of the set-up for the expe-riment were the RSS, syringe, reservoir bag incl. gas cy-linder and two PC laptops. One laptop operated the RSS, the second was used to record further data like g-force, cabin pressure, cabin temperature, housing temperature and temperature in the gas-flow (Fig. 35). For the experi-ment a relatively small rack was used. The reservoir bag was filled with a gas mixture of ~5% CO2 and ~16% O2 in N2. Different ventilation rates were then provided with the syringe. The operator manually moved the syringe during the micro- and hypergravity phases.

In conclusion of this test campaign, the RSS delivered reproducible values over three days. The electronics worked without any difficulties. The influence of gravity changes to the O2 sensor characteristic is very low (Fig. 36) whereas for the CO2 sensor a higher dependency is noted. Nevertheless, if calibrated properly, the system is well applicable also under microgravity conditions.

Fig. 35: Testing of the RSS during a parabolic flight campaign.

Fig. 36: Influence of change in gravity on O2 sensor signal.

Fig. 37: Influence of change in gravity on CO2 sensor signal.

Fig. 33: Centrifuge at DLR Cologne.

Irregular breath measured properly

Misinterpretation in direction detection

0

2

4

6

8

10

12

14

16

18

20

22

500 510 520 530 540 550 560

Time in s

Gas

Con

cent

ratio

n in

%

-5

-4

-3

-2

-1

0

1

2

3

4

5

6

Flow

in l/

s

Fig. 34: Exemplary file of obtained data (blue/top curve oxygen

concentration, black/mid curve flow rate, and red/bottom curve carbon dioxide concentration.


-9-

4.4 Animal Surgery

The RSS was tested also under very special conditions. An animal test with a pig was performed while measuring with the RSS in the artificial respiration tube (see Fig. 38). A comparison system for CO2 measure-ment as well as for flow and total pressure was also integrated in the breathing tube. Additionally, several medical data for blood concentrations of O2 and CO2, heart rate, etc. were recorded.

The tests with the pig represent a good comparison to the physiology of a human subject. However, the size of the pig was about children size (about 25 kg). Therefore, the respired volumes are very small. Nevertheless, the results obtained were extremely informative, revealing results for several breath patterns that have not been tested with the RSS prototypes before. A detailed measurement pro-tocol was designed to obtain as much information as pos-sible about the system capabilities under real conditions. The procedure was defined by the following steps:

1. Variations of inspired oxygen concentration (21%, 40%, 60%, 80%, 100%)

2. Variations in tidal volume (10 ml/kg, 20 ml/kg, 6 ml/kg, 10 ml/kg)

3. Variations in respiratory rate (30 breaths per minute (bpm), 10 bpm, 18 bpm)

4. Variations in artificial respiratory pattern (volume controlled - pressure controlled)

5. Cardiac oscillations 6. Variations of inspired oxygen concentration (100%

to 21% to 100%) 7. Simulation of lung illness by lung lavage and

optimizing of respiratory pressure 8. Second recruitment (similar to 7.) 9. Cardiac oscillation after lavage

Between every change in conditions a respiratory hold was performed to allow for a better coordination of the different measurement devices. Some results of this test are shown in the following. The first step could be seen as remarkable. In the overview the complete data for the O2 and CO2 measurement were observed (see Figs. 39-40). For each step the inspired partial pressure rises immediately, up to a maximum of 750 mmHg (100% at ambient pressure). The expired oxygen level rises slower. Looking at the extreme values, this effect becomes even more clearly. Comparing the minimum level of expired oxygen to the oxygen level in the blood (Fig. 41), a very good analogy was also found. This was postulated by the doctors, but never measured before.

Another interesting effect was observed in step 6, when the respiratory oxygen content was switched from 100% to 21% (see Figs. 42-44). As the inspired oxygen is changed rapidly, the expired oxygen level is higher than the inspired one. This was seen until a break-even point is reached and “normal” conditions appear.

Additionally, a good opportunity to observe the RSS sen-sor response times was given by cardiogenic (or cardiac) oscillations. The heartbeat activates the gas exchange in the lung each beat. This can be observed well in the expired gas flow after holding the breath for a short time. The oscillations can be seen very well in the RSS CO2-signal (Fig. 45). Normally, the oscillations are very clear in the RSS oxygen signal as well. In this case, the inhaled oxygen content was 100%, the corresponding sensor current switched the current measurement gain in the electronics to the next lower level (1/10), so the signal noise was obviously too high.

0

100

200

300

400

500

600

700

800

0 200 400 600 800 1000 1200 1400 1600

Time in s

Part

ial p

ress

ure

in m

mH

g

O2 in mmHgCO2 in mmHg

Fig. 39: RSS data of step 1.

0

100

200

300

400

500

600

700

800

0 200 400 600 800 1000 1200 1400 1600

Time in sPa

rtia

l pre

ssur

e in

mm

Hg

O2 maxO2 min

Fig. 40: Maximum inspired and minimum expired oxygen (step 1).

0

20

40

60

80

100

120

0 100 200 300 400

Breath number

stan

dard

ised

O2

cont

ent

(exp

ired

gas

and

bloo

d)

O2 min norm (RSS)PaO2 (blood value) norm

Fig. 41: Comparison of minimum expired oxygen to oxygen content in the blood (step 1, normalised for better comparison).

Fig. 38: Pig under anaesthesia with RSS and comparison devices.


-10-

0

100

200

300

400

500

600

700

800

0 100 200 300 400 500

Time in s

Part

ial P

ress

ure

in m

mH

g

Fig. 42: Overview over step 6 (100% O2 to 21% to 100%).

0

100

200

300

400

500

600

700

800

25 35 45 55 65 75 85 95 105

Time in s

Part

ial P

ress

ure

in m

mH

g

O2 in mmHgCO2 in mmHg CO2 (comparison)

Fig. 43: Step6: Switch from 100% to 21% O2 inspiration.

0

100

200

300

400

500

600

700

800

0 100 200 300 400 500

Time in s

Part

ial P

ress

ure

in m

mH

g

O2 expO2 insp

Fig. 44: Step6: End-Inspiratory and end-expiratory oxygen levels based

on a breath-to-breath analysis.

0

5

10

15

20

25

30

35

40

45

50

140 145 150 155 160

Time in s

CO

2 pa

rtia

l pre

ssur

e in

Hg

580

590

600

610

620

630

640

650

660

670

680

O2

part

ial p

ress

ure

in H

g

CO2 in mmHgCO2 (comparison)O2 in mmHg20 P Gl i d D h h i (O2 i H )

Fig. 45: Observed cardiac oscillations.

4.5 System tests with hydrogen sensors

For these tests, which have been conducted at the University Libre de Bruxelles (ULB), a modified RSS in-corporating a combination of a H2 and O2 sensor instead of a CO2 and O2 sensor was provided. With the RSS, the respiratory test under exactly the same conditions as it is carried out in a clinical context was performed: This res-piratory test simply consists of a single or multiple breath wash-in test, whereby a tracer concentration of a light, biologically inert gas (usually helium, but in this appli-cation H2) is inhaled. Upon exhalation, this tracer gas is monitored and in the second half of exhalation, a concen-tration gradient is seen with a magnitude that can reflect early changes in the airways of the lung periphery. For instance in lung transplants patients, an abnormally large helium gradient in the second half of expiration has been shown to signal an upcoming chronic rejection origina-ting in the deep lung, and this helium gradient signalled the rejection approximately one year before traditional lung function indices did. The aim of these tests have therefore been to examine whether H2 and helium traces, which should be very similar relative to their inhaled concentration, show a qualitative and quantitative agree-ment that would allow for the calculation of a concen-tration gradient in the second half of expiration.

The tests were performed with either a piston pump or a human subject, using a bag-in-box system, measuring H2 and O2 concentration with the RSS and helium with a mass spectrometer (see Fig. 46). Figs. 47-51 illustrate some of the obtained results.

Fig. 46: Typical test set-up for testing the RSS with hydrogen sensor

(inspiration case from test balloon).

Fig. 47: Comparison of He from mass spectrometer and H2 from RSS

wash-out/wash-in test with syringe at medium test gas concentrations (inhaled wash-in H2 concentration 0.2-0.3%).


-11-


wash-out/wash-in test with syringe at medium test gas concentrations (inhaled wash-in H2 concentration near 0.07%).


wash-out/wash-in test with a human subject (inhaled wash-in H2 concentration near 0.5%).


wash-out/wash-in test with a human subject (inhaled wash-in H2 concentration near 0.04%).

Fig. 51: H2 concentrations normalized to inspired H2 concentration –

numbers next to the traces indicate inspired volumes determinant of H2 dilution.

The main messages from this measurement campaign can be summarized as follows:

• Overall performance of the RSS due to major design changes lead to a much better quantitative and qualitative agreement with mass spectrometry (Figs. 47-50).

• Particularly at the low concentrations, the RSS showed an excellent H2 signal-to-noise ratio which is much better that obtainable by mass spectrometry (Fig. 48 and Fig. 50).

• Human subject experiments were very promising, with acceptable inspiratory plateaus, depending also on the degree of test gas concentration dilution (Figs. 49-50).

• When recalibrating the mass spectrometer curves to make them coincide with the RSS curves during the first long wash-in and wash-out, some quantitative divergences arose between He and H2 traces in other parts of the test. Part of the problem could be the mass spectrometer performance in such low concentration ranges. Hence, an independent way of validating human test results is done by plotting expiratory curves obtained for various concentrations, and normalizing these to inspiratory H2 concentration (Fig. 51). When inspired volumes are similar, these curves should almost coincide. This was not entirely the case, eg, the lowest and 2nd lowest dilution (with inspired volumes 2.1 and 2.0 L) should have almost indistinguishable curves, whereas they differ by 15%. By contrast the expiratory H2 concentration gradients (i.e., the index of clinical interest) were very similar across the different dilutions. Probably the residual dilution problem can be solved with some fine tuning on the correction factors or heater setting.

• Illustrative for the very good performance of the RSS are the concentration oscillations due to the heartbeat which can be identified on the expiratory phase of the RSS H2 trace in Fig. 51.

Certainly, there exist many areas for further improve-ments of the system in order to obtain a valid expiratory concentration gradient. Nevertheless, throughout this project, the H2 sensor has been adapted from its industrial application in a static environment to a probe that can be useful in the context of respiratory experiments at low flows. There has been a major improvement from the first H2 sensor prototype showing poor qualitative agree-ment with an independent measure of a similar gas (helium) to the final prototype which showed a very good agreement. Within the test sequences, it became evident that the H2 sensor shows an extremely good performance at very low H2 concentrations, which is very beneficial for its use in the clinic. Some residual fine-tuning on heater control and cross sensitivity could make this sensor fit for its application in a clinical context.


-12-

4.6 Tests under varying ambient pressure

The RSS was also tested in experiment series in order to investigate its function at lower ambient pressures than normal. For this purpose, a test rig was installed in an altitude chamber at the Environmental Physiology Laboratory, Karolinska Institutet, Stockholm. The test rig comprised the following elements:

• A piston pump generating a tidal volume of 500 ml per stroke and peak flows in both directions of 1029 ml/s. The flow profile was approximately sinusoidal.

• Respiratory filters on both sides of the RSS unit. • The RSS was attached on one side to the piston pump

and on the other to a Douglas bag filled with a test gas mixture.

• In this configuration, the RSS unit was flushed with the following gas mixtures: o air (0,04% CO2, 20,93% O2), o 5% O2 in N2 o 100% O2. o 5 % CO2 + 15% O2 in N2, o 14% CO2 + 86% O2.

Figs. 52 and 53 show the expected calibration curves for the current for the O2 sensor (linear) and the voltage for the CO2 sensor (log-linear plot). These calibrations were made at normal atmospheric pressure, which was 1012 kPa in Stockholm that day (20 m above sea level).

The chamber pressure was then changed from normal to different lower pressures and then back ambient again. The lowest pressure of 285 mmHg corresponds to the pressure inside the Russian space suits for extravehicular activity. The results of the RSS are shown in Fig. 54. The measurements generally reflect how the partial pressures

of the gas mixture in the Douglas bag change with ambient pressure. Thus, the current from the O2 sensor (Is O2) and the CO2 signal converted to % CO2 (CO2 conc.) were reduced with ambient pressure. The curves reflect also the small pressure adjustments that were made at each altitude to compensate for temperature changes resulting from each pressure change.

The flow rate measurement during these tests confirms that the physical principle of flow detection is in general that of a mass flow meter. Fig 55 shows these measure-ments (nominally ±1.029 l/s, here not calibrated separately for the two flow directions). The blue curve is at normal ambient pressure and the magenta curve at 285 mmHg chamber pressure (38% of normal). Flow readings at 285 mmHg are reduced to approximately 1/3 of nominal, which is as predicted.

Finally, another interesting calibration procedure was performed during this test campaign using a metabolic simulator which burns a mixture of propane and CO2 and which is coupled to a breathing pump. Both, CO2 and O2 measurements of the RSS, as shown in Fig. 56, compared very well with mass-spectrometric recording.

Fig. 54: RSS signals for 760, 524, 377, 285, 403, 524, and 760 mmHg

total pressure, respectively (corresponding to altitudes of 0 m, 3000 m, 5500 m, 7500 m, 5000 m, 3000 m and 0 m, resp.).

Fig. 55: RSS flow rate signal with identical volumes at different total

pressures (see text for explanation).

Fig. 56: RSS signals during a test with a metabolic simulator, burning a

mixture of propane and CO2.

Fig. 52: Calibration curve of the O2 sensor in preparation of the

hypobaric test series.

Fig. 53: Calibration curve of the CO2 sensor in preparation of

the hypobaric test series.


-13-

4.7 Sidestream microchamber application

All results presented in this chapter have been achieved by placing the sensors directly into a mask, i.e. by a direct in-situ measurement. During the project it turned out that this simultaneous measurement is very much more ambitious than originally assumed. The key points have been the extreme requirements for a very fast temperature control of the sensors in order to assure a stable partial pressure measurement. However, the sensors showed a very constant behaviour under stable ambient conditions, i.e. at low and constant flow rates. Thus, and in order to have eventually a rapid market entry of a system incorporating the sensors, the team partner Medisoft decided to concentrate on a sidestream microchamber solution. As a side effect, the require-ments for the control electronics are also highly reduced, allowing eventually a very cost optimised solution compared to the existing measurement devices.

Without reproducing the results here, as it is evident that the sensors work in a microchamber as well, it may simply noted that in the meanwhile O2, CO2, and CO / H2 sensors have been integrated in such microchambers showing the expected positive results.

5 Summary and Outlook

Within the MAP-RSS-project the development of mini-aturised sensor systems for respiratory investigations has been certainly advanced. Miniaturised oxygen, carbon di-oxide and hydrogen sensors, capable to simultaneously measure also mass flow rates, have been designed, manu-factured, and integrated into a mask for the in-situ mea-surement of respiratory parameters, and validated in se-veral tests. The sensors seem now applicable for separate measurements of the physical properties in respiratory experiments. Also very important, none of the potential advantages of a fully qualified system had to be withdrawn.

Nevertheless, there are still development and optimisa-tion efforts necessary and possible towards a fully re-liable, reproducible, and accurate, “plug & play”-system. However, the development demonstrated the feasibility of the idea and that the remaining open issues are not a physical constraint but “only” a matter of control, design of sensor heater and layers, and optimisation of the flow through the mask.

Therefore, the optimisation of the system should be pursued in the future in order to further improve the over-all system performance, which in the meanwhile meets almost completely the challenging specifications (Table 1). In parallel, the system development shall be continu-ously accompanied by application tests in order to obtain as much information as possible about the operational and scientific limits and constraints, as well as about the potential benefits in comparison to existing systems.

Although not optimised, the system has already now a very very low mass when compared to other existing devices (Table 2). Also, the preparation time is very short, thus a real mobile and time-effective measurement can be realised. Therefore, it would be highly attractive for a space application, which of course would be one of the most challenging tasks in the future. Additionally, it is assumed that a potential space application would also have a very high impact for the planned market entry and the acceptance of the system on Earth.

6 Acknowledgement

The financial support of the development and the experiments presented in this summary by ESA/ESTEC is gratefully acknowledged.

Volume flow rate Oxygen Carbon dioxide

Requirement Realised / tested up

to now Requirement Realised /

tested up to now

Requirement Realised / tested up to now

Range -20 / +20 l/s -20 / +20 l/s 0…25 % 0…100 % 0…10 % 0…15 %

Response time < 50 ms < 50 ms < 100 ms < 50 ms < 100 ms < 115 ms

Precision ±5% or ±200ml/s,

whichever is greater

± 50 ml/ s for < 1 l/s

± 3% for > 1 l/s < ±0.1 % O2 < ±0.2 % O2 ±0.05 % CO2

±0.1% for < 4%

± 0.3% for > 4%

Resolution < 10 ml/s < 10 ml/s < 0.05 % < 0.05 % < 0.05 % < 0.05 %

Stability within precision within precision < 0.03 %/h < 0.06 %/h < 0.02 %/h < 0.1 %/h Tab. 1: Comparison of designated and realised requirements. The experimental results from which the ranges and accuracies were

derived, were performed in separate experimental set-ups under laboratory conditions.

Item Mass Remark Sensor adapter with two sensors (O2+CO2)

100 g See Fig. 19 (shows only one sensor connected), mass requirement can be further reduced

Sensor harness to electronics

354 g For 2 sensors, not opti-mised, mass can be reduced further

Mask 137 g See Fig. 35 Mask fastening net 42 g See Fig. 35 Biological filter 30 g See Fig. 35 Electronics 426 g Version 1.5, Fig. 24 Power supply to electronics

600 g Standard AC/DC-converter unit for medical applications

RS-232 cable to laptop

122 g Length of 1.5 m

Total 1811 g Tab. 2: Overview of overall system mass. Mass demand can and will

be further reduced in the future.

development and application of a miniaturised sensor...

Documents