http://www.diva-portal.org
Postprint
This is the accepted version of a paper published in Journal of Micromechanics and Microengineering.
This paper has been peer-reviewed but does not include the final publisher proof-corrections or journal pagination.
Citation for the original published paper (version of record):
Gatty, H., Stemme, G., Niclas, R. [Year unknown!]
A wafer level liquid cavity integrated amperometric gas sensor with ppb leve nitric oxide gas sensitivity.
Journal of Micromechanics and Microengineering
Access to the published version may require subscription.
N.B. When citing this work, cite the original published paper.
Permanent link to this version:
http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-172968
1 © 2015 IOP Publishing Ltd Printed in the UK
1. Introduction
Amperometric gas sensors are, due to their versatility, applied in wide range of fields from health care to pollution moni- toring [1]. In the health care sector, amperometric gas sensors are finding increased use in gas monitoring instruments [2].
For example, in an asthma measurement instrument that monitors the inflammation in the airways of the lungs, an amperometric gas sensor measures the concentration of nitric oxide (NO) in the exhaled breath [3]. In adults, a NO gas con- centration above 50 ppb typically indicates the presence of an inflammation [4–6]. The asthma measuring instruments cur- rently available in the market today are bench-top instruments that typically are used in healthcare settings. However, if the instrument could be made smaller, asthma patients could have their own personal portable monitor which could be beneficial in the management of the disease. The reason for the large size of the current instruments is that the commercially available parts-per-billion (ppb)-level NO gas sensors have a slow NO gas response time (40 to 60 s) which makes it necessary to incorporate buffering components to store and pump the gas at a low flow rate to the sensor [3]. In addition, commercially
available NO sensors are typically manufactured by serial (sometimes manual) assembly of individual components which leads to relatively costly sensors that are a few cm3 large in size and which, consequently, responds slowly to the gas. Therefore, in order for the instrument to be hand-held and reduce cost there is a need for batch-fabricated, fast sensors, that responds in real time without the need of extra space- consuming buffering components.
Metal oxide gas sensors have a potential to be used in hand- held instruments due to their small size and compatibility with batch fabrication. However, the slow response time, cross sensitivity and high power consumption, are the drawbacks [7, 8] which make them unsuitable for asthma monitoring.
Amperometric microsensors with Nafion™ as a solid electro- lyte, on the other hand, offer higher sensitivity, faster response and lower power consumption than metal oxide gas sensors [9, 10]. However, these solid electrolyte gas sensors are prone to be sensitive to humidity and are therefore not suitable for asthma monitoring. Currently, commercial amperometric gas sensors with liquid electrolyte, which are less sensitive to humidity variations, use multiple layers of Teflon™-based membranes to protect the electrolyte from evaporation
Journal of Micromechanics and Microengineering
A wafer-level liquid cavity integrated
amperometric gas sensor with ppb-level nitric oxide gas sensitivity
Hithesh K Gatty, Göran Stemme and Niclas Roxhed
Micro and Nanosystems, KTH Royal Institute of Technology, Osquldas väg 10, 10044 Stockholm, Sweden
E-mail: roxhed@kth.se
Received 28 April 2015, revised 14 July 2015 Accepted for publication 27 July 2015 Published
Abstract
A miniaturized amperometric nitric oxide (NO) gas sensor based on wafer-level fabrication of electrodes and a liquid electrolyte chamber is reported in this paper. The sensor is able to detect NO gas concentrations of the order of parts per billion (ppb) levels and has a measured sensitivity of 0.04 nA ppb−1 with a response time of approximately 12 s. A sufficiently high selectivity of the sensor to interfering gases such as carbon monoxide (CO) and to ammonia (NH3) makes it potentially relevant for monitoring of asthma. In addition, the sensor was characterized for electrolyte evaporation which indicated a sensor operation lifetime allowing approximately 200 measurements.
Keywords: nitric oxide, amperometric, gas sensor, MEMS, silicon, Nafion (Some figures may appear in colour only in the online journal)
AQ1 H K Gatty et al
Printed in the UK aa0168
JMMIEZ
© 2015 IOP Publishing Ltd 2015
00
J. Micromech. Microeng.
JMM
0960-1317 000000
Papers
00
Journal of Micromechanics and Microengineering IOP
0960-1317/15/000000+10$33.00
UNCORRECTED PROOF J. Micromech. Microeng. 00 (2015) 000000 (10pp)
[1, 11, 12]. The main drawback of this method is an increased response time which inhibits real time gas concentration measurements. Other demonstrated approaches in measuring low concentrations of NO with integrated sensors include potentiometric sensing [13, 14] and indirect measurements of NO2 [15]. Potentiometric sensors have not yet demonstrated relevant detection levels at the short response times needed for real-time measurements. Indirect measurements of NO2 has shown promising results for breath samples, however the applicability for asthma monitoring using this indirect method of NO conversion and measurement of NO2 still remain to be proven in large patient studies.
Recently, we developed a new NO sensor that combined the advantageous effects of Nafion with a liquid electrolyte reservoir to avoid large humidity dependency. This sensor had a fast response and a high sensitivity, detected ppb-levels and was thus potentially suitable for asthma monitoring [16].
However, although the sensor included a small microporous working electrode chip, it had off-chip reference and counter electrodes together with an external liquid electrolyte supply that needed continuous refilling, resulting in a large size and a non-integrated cumbersome manufacturing.
In the present work, a highly integrated, 10 × 10 × 1 mm, ppb-level amperometric NO gas sensor chip with wafer- level fabricated electrodes and liquid electrolyte chamber is described for the first time. The sensor was designed and characterized for its sensitivity to NO gas and for its response time with respect to asthma monitoring application. In addition, the sensor was characterized for selectivity to elec- trochemically sensitive gases such as carbon monoxide (CO) and ammonia (NH3) present in the exhaled breath which, potentially could interfere with electrochemical sensors [17–19]. Typical concentrations of CO and NH3 in exhaled breath are in the range of 0–8 ppm and 0–1 ppm, respec- tively. In order to maximize the sensor operation life time a special sensor sealing mechanism was utilized to minimize
the electrolyte evaporation loss and hence to avoid refilling of the electrolyte.
2. Sensor design
To obtain the required ppb-level sensitivity a high surface area of the working electrode is essential. Here, this is realized through a microporous working electrode with a nanostruc- tured Nafion™ coating. The integration of a liquid electrolyte volume together with the three sensor electrodes provide for constant moistening of the working electrode thus making the sensor less sensitive to changes in ambient humidity [16, 20].
Figure 1 shows a schematic illustration of the sensor design, the amperometric voltage biasing and the current measure- ment arrangement of the three electrodes.
3. Fabrication process of the sensor
This section describes the fabrication process and the assembly of the integrated gas sensor. This includes the wafer level fabrication of the microporous working electrode using a silicon on insulator (SOI) wafer and its integration with a glass wafer containing the counter and the reference electrodes.
Each sensor has an electrolyte chamber, formed by an etched cavity between the SOI wafer and the glass wafer and used for storing the electrolyte. The electrolyte used in the sensor was 5%wt H2SO4 liquid solution together with a Nafion layer cov- ering the microporous grid structure. A combination of Nafion electrolyte and water was reported earlier [20].
3.1. Working electrode and electrolyte chamber fabrication The microporous grid structure of the working electrode was fabricated using an SOI wafer with a 250 μm thick device layer fusion bonded to a 500 μm thick handle wafer having
Figure 1. Schematic cross-sectional (a) and perspective view (b) illustration of the sensor design with the microporous working electrode, assembled to a reference and a counter electrode. The electrolyte is integrated within the cavity formed between the electrodes. Inset illustration shows the triangular grid arrangement of the microporous working electrode. A biasing voltage between the working and reference electrode, Vbias is applied to ensure the oxidation of nitric oxide gas at the working electrode.
a 1 μm thick buried oxide layer as illustrated in figure 2. To begin with, the SOI wafer was oxidized with a 3 μm thick SiO2 layer in an oxidation furnace. The oxidized device layer of the SOI wafer was patterned using photolithography to form a triangular arrangement with a 20 μm beam width and 120 μm distance between the beams, as illustrated in the inset of figure 1(b). The oxide layer was then etched using reactive ion etching (RIE) (Applied Materials Precision 5000 Mark II, USA) for approximately 25 min to expose the device layer.
The device layer was etched through using an ICP Deep RIE etcher (Surface Technology Systems, UK) for 100 min to form the microporous grid structure (figure 2(a)). To protect the side walls of the microporous grid during subsequent processing, the SOI wafer was oxidized with a 3 μm thick SiO2 layer.
The SiO2 layer on the handle wafer was patterned with an 8 mm wide square pattern using photolithography and etched in the same way as above to expose the silicon of the handle wafer. Cavities with a volume of approximately 30 μl were then formed by KOH etching (20%wt KOH solution at 70 °C for 11 h) of the handle wafer down to the buried oxide layer (figure 2(b)). Figure 3 shows a photograph of the cavities formed in the handle wafer. The oxide on the device layer was removed by etching in a 44% HF solution for approxi- mately 15 min, while the oxide layer on the handle wafer was protected by an adhesive tape (Ultra tape 1310, USA). After
rinsing and drying, the adhesive tape was peeled off (figure 2(c)) and the SOI wafer was transferred to an atomic layer deposition (ALD) chamber (Beneq TFS 200, Finland) where a 10 nm thick Al2O3 adhesion layer and a 10 nm thick Platinum layer was deposited on the device layer using the process described in [21]. This resulted in a conformal Pt layer cov- ering all surfaces and micropores of the SOI wafer except the oxide layer on the handle wafer. Finally, the protective SiO2
layer on the handle wafer was removed by etching in 44% HF solution for 15 min (figure 2(d)).
3.2. Counter and reference electrode fabrication
A 300 μm thick Borofloat™ glass wafer was used as a sub- strate to fabricate the counter and the reference electrodes of the sensor in a process flow illustrated in figure 4. To begin with, through glass vias (TGVs), 150 μm in diameter, were drilled through the substrate to enable electrical contacts for the reference and the counter electrodes (figure 4(a)). 500 nm silver was evaporated around and into the via holes using a shadow mask defining the electrode contact areas. The wafer was angled at approximately 45° to the line of deposition to ensure deposition of silver on the side walls of the TGVs (figure 4(b)). Silver counter and reference electrodes were deposited on the opposite side of the wafer using the same deposition process but another shadow mask (figure 4(c)).
A close up view of the counter and the reference electrodes containing the TGVs is shown in figure 5. The liquid elec- trolyte was obtained by diluting 0.5 g of concentrated H2SO4 with 9.5 g of H2O to obtain an approximately 5%wt solu- tion of H2SO4. To prepare the reference electrodes for the amperometric measurements the silver layer surface on the reference electrodes was transformed into silver oxide by dip- ping the wafer in a 5%wt H2SO4 solution with a voltage of 1.0 V applied between a platinum cathode and the intercon- nected reference electrodes for two minutes, as illustrated in figure 4(d). A change in the color from silver to white was observed during the passivation of the reference electrode.
3.3. Sensor assembly
To form an integrated sensor, the glass wafer containing the counter and reference electrodes and the SOI wafer containing working electrodes were assembled by anodic bonding.
Individual sensors were then obtained by dicing the bonded
Figure 2. Illustration of the fabrication flow of the microporous working electrode and the electrolyte chamber using an SOI wafer. (a) DRIE of Si device layer to form the microporous structure. (b) KOH etching of handle wafer to form a cavity for storing the electrolyte.
(c) Etching of the SiO2 layer on the device wafer covering the sidewalls of the grid structure. (d) Deposition of the platinum layer using ALD on the microprous grid structure.
(a)
1 µm 250 µm
3 µm 550 µm
(b) (c) (d)
10 nm ALD Pt
5 mm 20 µm 120 µm
8 mm Si SiO2 Photoresist Platinum
Figure 3. Photograph showing different cavities for storing the electrolyte. Cavities with a volume of approximately 30 μl were formed by KOH etching of the handle wafer down to the buried oxide layer.
wafer stack. Figure 6 shows a schematic representation of a single sensor after the final assembly. Figure 7(a) shows a photograph of the gas sensor chip after dicing. The chip size is 10 mm × 10 mm and approximately 1 mm thick. The SEM image in figure 7(b) shows the cross section of a sensor with the electrolyte chamber integrated in the chip.
In order to achieve the high ppb-level sensitivity, the sur- face area of the working electrode area was dramatically increased by the formation of a highly nanoporous layer of Nafion. This was done by immersing the integrated sensor chip into a beaker containing 5%wt Nafion solution (Sigma Aldrich, USA) which was then transferred to a vacuum desic- cator (Model 550, Kartell, Italy) and pumped to approximately 0.2 bar (abs.). The low pressure in the vacuum chamber des- iccator helps in removing air bubbles that could be trapped within the sensor cavity. The sensor was allowed in the des- iccator for 5 min, after which the vacuum pump was turned off and the pressure was slowly increased to atmospheric
Figure 4. Schematic illustration of the process flow for fabricating the counter and reference electrodes on the glass wafer. The RE contact lines was connected to the positive terminal of the battery in order to passivate the silver. (a) Through glass vias (TGV) of 150 μm diameter drilled in the borosilicate glass wafer. (b) A shadow mask and a 45° inclined silver deposition were used to form the counter and reference electrode contacts. (c) A shadow mask and a 45° inclined silver deposition were used to form the counter and reference electrode areas.
(d) Schematic illustration of the procedure for passivation of silver to form the reference electrode.
(a) (b)
(d) (c)
Counter electrode
External contact
Reference electrode
Glass wafer with contact lines to RE connected to the positive of the battery
Platinum cathode
1v
5 %wt H2SO4 Drill bit
Figure 5. Photographs of fabricated reference and counter electrodes and contacts. (a) Close up of a counter electrode (CE) and a reference electrode (RE). (b) The electrical contacts to the counter (CE) and reference electrodes (RE).
Figure 6. Schematic representation of the integrated amperometric gas sensor after the final assembly and dicing of the bonded wafers.
Counter electrode contact
Working electrode
Si Ag
Reference electrode contact Electrolyte chamber
SiO2
Glass Ag2O Pt
pressure by opening a valve connected to the chamber. The sensor chip was removed from the desiccator and dried using filter paper (Whatman 903® DBS paper, Whatman plc, UK) which, rapidly absorbed excess Nafion solution contained in the electrolyte chamber. The chip was then allowed to dry in air for 30 min after which the TGVs were sealed with a sili- cone sealant (Silicone sealant 7091, Dow corning, USA).
4. Evaluation procedure
To characterize the sensor, gold bond wires were connected to the contacts of the three electrodes using a silver conduc- tive adhesive (Electrolube, UK). In order to fill the electrolyte chamber with electrolyte without trapping air bubbles, the sensor chip was immersed in a beaker containing the electrolyte solu- tion and subjected to a desiccator vacuum treatment to ensure complete and gas bubble free filling of the electrolyte chamber.
The electrolyte-filled chip was then mounted in a custom-made chip holder that holds the sensor chip together with an active carbon fiber filter taken out from a commercial NO sensor (03-2030, Aerocrine AB, Sweden). The role of the filter is to reduce interference from other gases contained in the breath.
4.1. Measurement set-up
Figure 8 shows an illustration of the mechanical sealing module that was used for characterizing the integrated sensor. The module consists of a sensor housing (top part and bottom part) together with a plunger that is used for sealing the sensor. In order to prevent the evaporation of the elec- trolyte, the sealing of the sensor was ensured by a compliant layer consisting of a soft sealing adhesive tape (Double sided acrylic adhesive, Specialist tapes, UK) that was attached to the head of the plunger. For gas concentration measurements, the plunger was manually pulled upwards to allow the gas to reach the sensor.
To test the sensor for different gases and gas concentrations a measurement set-up was arranged as illustrated in figure 9.
In this arrangement a 200 ppb NO in N2 calibration gas (AGA gas AB, Sweden) was mixed with pure N2 gas (99.95% pure,
AGA gas AB, Sweden) and residues of NOx from the N2 gas were removed using a scrubber (Dräger, type 1140, Germany).
To measure the selectivity of the sensor to interfering gases, the NO gas was switched to 45 ppm CO in N2 using a two-way valve. In order to measure the selectivity to NH3 gas, the CO gas bottle was exchanged with a 45 ppm NH3 in N2 (AGA gas AB, Sweden) gas bottle. In order to humidify the gas mixture, a custom-made in-line humidifier consisting of a cylinder con- taining moistened tissue paper (TX609, Texwipe, USA) was used. To keep the humidity constant, small volumes of water was manually injected into the cylinder using a needle and syringe (figure 9) when the measured humidity deviated from the desired operating humidity.
To measure the working electrode current, the sensor was connected to a potentiostat (DY2011, Digi-ivy, USA).
Two mass flow controllers (F201CV, Bronkhorst EL-flow, Netherlands) and two flow sensors (AWM5102, Honeywell, USA) were used to control and measure the flow rate, respec- tively. To measure the operating humidity of the sensor, a humidity sensor (HIH 4000, Honeywell, USA) was placed inside the sensor housing (bottom part) of the mechanical sealing module. A LabVIEW™ program was used to access the data from the humidity sensor and the flow sensors. The humidity around the sensor was held constant to approxi- mately 50% RH for all measurements. The gas flow rate was maintained constant at 550 ml min−1 for all measurements. All the measurements from the sensor were carried out with the working electrode biased at +0.95 V as compared to the refer- ence electrode.
4.2. Measurement methods
The working electrode current from the sensor has a slow drift over a period of time. To compensate for this drift, the back- ground current, i.e. the current at zero NO concentration, was measured directly prior to each gas concentration measure- ment. This was done by measuring the background current in a pure nitrogen flow, then switching to the NO gas concentra- tion to be measured and taking a measurement of the working electrode current at that concentration. After recording the
Figure 7. Photograph of the fabricated sensor. (a) The integrated gas sensor chip obtained after dicing. The footprint of the microporous grid of the sensor is 5 mm × 5 mm. The chip size is 10 mm × 10 mm and approximately 1 mm thick. (b) SEM image showing a cross section of the sensor.
working electrode current, the NO gas flow was stopped and the N2 gas flow was started. A pulsed method of switching back and forth between the NO and N2 gas was used. The output current of the sensor for a given NO gas concentration was then calculated as the difference between the working electrode current and the background current. To determine the sensitivity of the sensor, the output current was measured by sweeping the NO concentration and taking measure- ments at 25 ppb, 50 ppb, 110 ppb and 200 ppb, respectively.
Four such sweeps were carried out leading to four values
of the output current at each concentration which were then averaged. The selectivity of the sensor was estimated by mea- suring the CO sensitivity by adding CO concentrations from 10 ppm to 32 ppm. For comparison, the selectivity was also tested at another bias voltage, Vbias = 0.7 V used earlier [16].
To measure the response time of the sensor, the gases were alternately switched between 200 ppb NO and pure N2. The response time, t90 of the sensor was estimated when the output current was observed to reach an average maximum value of IOmax for 200 ppb concentration.
Figure 9. Schematic illustration of the measurement set-up used for the characterization of the integrated sensor. Data from the flow sensors and the humidity sensor were accessed using a LabVIEW™ program. The humidity was maintained at 50% RH with the help of a humidifier and the flow was maintained at 550 ml min−1. The gases from CO and NO were switched using a two way valve and CO gas bottle was replaced with NH3 for interference measurements.
Figure 8. Schematic illustration of the mechanical sealing module used for the sensor gas measurements. The module is made up of sensor housings (top and bottom part). The top part of the sensor housing contains a plunger and the bottom part of the sensor housing is used for holding the NO sensor chip. Evaporation of the electrolyte is prevented by closing the plunger.
Outlet
Active carbon fiber filter
NO sensor Filter and chip holder Plunger Plunger
movement
Sensor housing (top part)
Sensor housing (bottom part) Inlet
Compliant layer
90 mm
60 mm
Electrode connections
In order to assess the evaporation time of the electrolyte, two types of experiments were performed. In the first experi- ment, the output current and the volume of the electrolyte were measured when the plunger of the mechanical sealing module was kept open. A flow of 550 ml min−1 of N2 was maintained and the humidity was kept constant at 50% RH. Gas with 200 ppb NO concentration was administered at regular intervals to the sensor and the output current was measured. In the second type of experiment, the liquid volume loss was measured by weighing a liquid filled sensor chip using an external preci- sion scale. Both for a sensor with constantly open plunger (at 550 ml min−1 N2 flow and 50% RH humidity) as well as for a sensor with a constantly closed plunger sealing mechanism.
5. Results and discussion
The fabricated wafer contained 44 sensors of which 15 sen- sors were tested for NO gas and found functional. Remaining sensors were tested for reliability issues such as handling, electrolyte evaporation, Nafion coating, and packaging. The sensor was characterized for its sensitivity to NO gas, response
time, and selectivity to CO and NH3 gases. The evaporation of the electrolyte in the sensor was characterized for both open and closed sealing mechanism.
5.1. NO sensitivity
The average value of the output currents for four different concentrations between 0 and 200 ppb is plotted in figure 10.
The output current, IO was found to be linear with the NO gas concentration. Based on a linear fit, a sensitivity of 0.04 nA ppb−1 was calculated, which is in agreement with our earlier fabricated non-integrated sensor [16]. The sensor detects NO gas in the lower limit of 25 ppb and is thus within the limit for detecting asthma. The sensitivity of the sensor can likely be further improved by increasing the pore surface area of the working electrode, i.e. by a denser grid pattern.
5.2. Selectivity to CO and NH3
The selectivity of the sensor to interfering gases in the exhaled breath such as CO and NH3 were determined by calculating
Figure 10. Output current, IO (nA) as a function of NO concentration (ppb). Each error bar represents the measured standard deviation of four measurements from four different concentrations sweeps.
0 20 40 60 80 100 120 140 160 180 200
0 1 2 3 4 5 6 7 8 9
NO concentration (ppb) Output current, Io (nA)
Figure 11. Output current, IO as a function of CO concentration at two different voltage biases. The sensitivity of the sensor to CO gas was found to decrease when Vbias was increased from 0.7 V to 0.95 V.
0 5 10 15 20 25 30 35
0 10 20 30 40 50 60
CO Concentration (ppm) Output current, IO (nA)
Vbias= 0.7 V
Vbias= 0.95 V
the ratio of the respective sensitivities to NO gas sensitivity.
The CO gas sensitivity was measured for two different bias voltages and plotted in figure 11. With Vbias = 0.7 V, the sen- sitivity was calculated using the linear fit and found to be approximately 1.4 nA ppm−1 and with a Vbias = 0.95 V, the sensitivity was calculated to be 0.036 nA ppm−1. Thus, the selectivity of the sensor to CO is approximately 1000 for the higher bias voltage. In exhaled breath, CO gas concentration can be between 500 to 1000 times higher than the NO con- centration, indicating that the sensor has a selectivity to CO within the range to be applicable in asthma monitoring [19].
The reduction of CO sensitivity with increasing bias could be due to the oxidation of the platinum at the working electrode [18, 22, 23]. Since the sensor is exposed to ambient atmo- sphere and real breath samples contain oxygen, processes like this may affect selectivity. For NH3, the output current was found to be below the detection limit and hence was not mea- sureable when the sensor was tested with 45 ppm NH3 gas.
5.3. Response time
The response time of the sensor was determined by estimating the rise time (t90) of the sensor, i.e. the time it takes to reach 90% of the maximum output current, IOmax which was calcu- lated by using the estimated average maximum value of the measured output current. This resulted in a response time of 12 s, as shown in figure 12(a). This is considerably faster than the response times of commercially available ppb-level NO sen- sors. In exhaled breath, the NO concentration profile reaches a stable value after 7 s and a suitable measurement interval of the exhalation phase is suggested to be between 7–10 s [6].
This means that the present design has a response time per- formance which likely is within optimization reach for use in real time asthma detection applications. Figure 12(b) shows a background current drift of approximately 0.8 nA min−1,
which corresponds to 20 ppb min−1 in terms of NO concentra- tion drift. This means that for the 12 s of the measurement time of the response, the drift component of the signal amounts to about 4 ppb. This is an acceptable level of drift for asthma monitoring applications, where the NO concentrations typi- cally are around 50 ppb.
5.4. Operation lifetime
The electrolyte in the electrolyte chamber will evaporate through the microporous working electrode. In order to have the electrolyte liquid last longer a special sealing mecha- nism was incorporated in the mechanical sealing module.
The electrolyte evaporation was assessed with two types of experiments. In the first experiment, the evaporation of the electrolyte and its effect on the output current was measured when the plunger of the mechanical sealing module was kept open. In order to determine the working duration time of the sensor under plunger open condition, the output current, IO
was measured for 200 ppb NO concentration at five different intervals. The output current was measured to be approxi- mately 7 nA for the first four measurements during 80 min, but then dropped to zero after 94 min as shown in figure 13. In a second type of experiment the liquid volume loss over time of a liquid filled sensor chip with an open plunger was inves- tigated. The liquid volume loss was determined by weighing the sensor chip at eight different occasions. The measurements plotted in figure 14 show that the electrolyte chamber has been emptied through evaporation after 100 min. This is in good agreement with the measurements shown in figure 13 where the sensor stopped to work after 94 min, presumably due to a dried out electrolyte chamber. Hence, the sensor seems to be functional even if the electrolyte has partially evaporated.
A reason for this could be the hygroscopic property of the Nafion layer covering the interior of the sensor. This may help
Figure 12. Graph showing the variation of ppb-level NO concentration (a) The output current IO when the sensor is exposed to a step of 200 ppb NO gas. The response time was measured by switching from a zero NO concentration to a 200 ppb NO concentration flow. The rise time, t90 was determined to be approximately 12 s. (b) Working electrode current at four different NO gas concentrations. The drift in the background current was approximately 0.86 nA min−1.
(a) (b)
0 10 20 30 40 50 60 70 80
0 1 2 3 4 5 6 7 8 9
Time (s) Response time, t90 = 12 s Output current, IO (nA)
IOmax
4320 434 436 438 440 442 444 446 448
50 100
200 ppb
110 ppb
50 ppb
25 ppb
150 200 250 300 350 400 450
Time (s)
W.E current (nA)
keeping the electrodes moist even if the electrolyte has par- tially evaporated. It should be noted however, that the sensor signal was only measured at 200 ppb NO and any changes in linearity as a result of evaporation was not studied.
Using the same weighing procedure as above, the electro- lyte volume loss was also measured when the plunger of the mechanical sealing module was kept closed. Figure 15 shows that the sensor lost about 1/3 of the liquid volume in three weeks, indicating after extrapolation that the sensor would still contain liquid after nine weeks. Since the measurements of figures 13 and 14 clearly indicate that the sensor will work as long as there is any electrolyte liquid still left in the sensor it can be functional for NO measurements for nine weeks with this experimental sealing module set-up. The decrease in the volume of the electrolyte in the closed mode is likely caused by small leaks between the sealing plunger and the sensor
chip and can most likely be further reduced by optimizing the sealing material and mechanism.
The measurements illustrated in figures 14 and 15 shows that the liquid loss rate is 0.3 μl min−1 in the open measure- ment mode and 0.5 μl d−1 in the closed mode. Assuming a needed open period of 30 s (allowing both zero-level and sample measurements), an electrolyte liquid loss of approxi- mately 0.15 μl/measurement can be calculated. This means, for example, that the current design would enable approxi- mately 200 measurements. This is a fully sufficient operation lifetime of a NO sensor in asthma monitoring applications.
The sealing of the sensor using the mechanical sealing module was demonstrated to be an effective solution compa- rable to multiple Teflon layers used in the commercial sensors.
Using this method, evaporation of the electrolyte was mini- mized without sacrificing the response time. It is conceivable that the mechanical sealing module can be reduced in size and an electrically controlled actuator mechanism for the opening and closing of the sealing mechanism can be included for the asthma monitoring application.
6. Conclusions
A highly integrated, 10 × 10 × 1 mm, ppb-level detection amperometric NO gas sensor chip with wafer-level fabricated electrodes and liquid electrolyte chamber is described for the first time. Using a mechanical sealing module the elec- trolyte evaporation was minimized resulting in a potential for about 200 measurements. The sensor, which was designed to be applicable for use in real-time measurement of NO gas in handheld asthma detection applications, showed a measured sensitivity of 0.04 nA ppb−1 and a response time of 12 s. A high selectivity to breathing gases such as carbon monoxide (CO) and ammonia (NH3) precludes any interference with the NO concentration measurement. Future work would focus on
Figure 13. Output current of the sensor when the plunger of the mechanical sealing module was maintained in an open position.
Pure N2 gas was continuously supplied except at five intervals where the gas was switched to 200 ppb NO. The output current was stable between 0 to 80 min, which decreased and was not measureable after 94 min presumably due to dried out electrolyte chamber.
0 10 20 30 40 50 60 70 80 90 100 0
1 2 3 4 5 6 7 8 9 10
Time (min) Output current, IO (nA)
Figure 14. Electrolyte sensor cavity volume loss through evaporation when the plunger of the mechanical sealing module was maintained in an open position while gas was flowing. The electrolyte in the sensor cavity was completely evaporated after 100 min.
0 20 40 60 80 100
0 5 10 15 20 25 30
Electrolyte volume (µl)
Time (min)
Figure 15. Electrolyte sensor cavity volume loss through evaporation when the plunger of the mechanical sealing module was maintained in a closed position. The volume of the electrolyte was measured once a week. The electrolyte is still present in the electrolyte chamber after three weeks of measurement indicating a much longer life time of the sensor. A higher liquid volume than the nominal value of the electrolyte volume was measured for this experiment.
0 0.5 1 1.5 2 2.5 3
0 5 10 15 20 25 30 35 40
Number of weeks
Electrolyte volume (µl)
investigations using breath samples from healthy and asth- matic patients.
Acknowledgment
This work was supported by the Swedish innovation agency VINNOVA, the European Research Council (ERC) through the Advanced Grant No: 267528 and Aerocrine AB, pro- ducer of handheld diagnostic tools for breath monitoring. The authors would like to thank Kjell Norén and Mikael Bergqvist for the help in the preparation of the mechanical sealing module.
References
[1] Stetter J R and Li J 2008 Amperometric gas sensors a review Chem. Rev. 108 352–66
[2] Brooks C R, Brogan S-B M, van Dalen C J, Lampshire P K, Crane J and Douwes J 2011 Measurement of exhaled nitric oxide in a general population sample: a comparison of the Medisoft HypAir FENO and Aerocrine NIOX analyzers J.
Asthma 48 324–8
[3] Hemmingsson T, Linnarsson D and Gambert R 2004 Novel hand-held device for exhaled nitric oxide-analysis in research and clinical applications J. Clin. Monit. Comput.
18 379–87
[4] Dweik R A et al 2012 An official ATS clinical practice guideline: interpretation of exhaled nitric oxide levels (FENO) for clinical applications Am. J. Respir. Crit. Care Med. 184
[5] Spergel J M, Fogg M I and Bokszczanin-Knosala A 2005 Correlation of exhaled nitric oxide, spirometry and asthma symptoms J. Asthma 42 879–83
[6] A T S Documents 2005 ATS/ERS recommendations for standardized procedures for the online and offline measurement of exhaled lower respiratory nitric oxide and nasal nitric oxide, 2005 Am. J. Respir. Crit. Care Med.
171 912
[7] Eranna G, Joshi B, Runthala D and Gupta R 2004 Oxide materials for development of integrated gas sensors—a comprehensive review Crit. Rev. Solid State Mater. Sci.
29 111–88
[8] Fine G F, Cavanagh L M, Afonja A and Binions R 2010 Metal oxide semi-conductor gas sensors in environmental monitoring Sensors 10 5469–502
[9] Buttner W J, Maclay G J and Stetter J 1990 An integrated amperometric microsensor Sensors Actuators B: Chem.
1 303–7
[10] Otagawa T et al 1990 Planar microelectrochemical carbon monoxide sensors Sensors Actuators B: Chem. 1 319–25 [11] Cao Z, Buttner W J and Stetter J R 1992 The properties and
applications of amperometric gas sensors Electroanalysis 4 253–66
[12] I. s. technology 2008 Electrochemical sensors Available: www.
intlsensor.com/pdf/electrochemical.pdf
[13] Mondal S P, Dutta P K, Hunter G, Ward B, Laskowski D and Dweik R 2011 Development of high sensitivity potentiometric NOx sensor and its application to breath analysis Sensors Actuators B: Chem. 158 292–8 [14] Chang C, Maduraiveeran G, Xu J, Hunter G and Dutta P
2014 Design, fabrication, and testing of MEMS-based miniaturized potentiometric nitric oxide sensors Sensors Actuators B: Chem. 204 183–9
[15] Magori E et al 2011 Fractional exhaled nitric oxide measurement with a handheld device J. Breath Res.
5 027104
[16] Gatty H K, Leijonmarck S, Antelius M, Stemme G and Roxhed N 2015 An amperometric nitric oxide sensor with fast response and ppb-level concentration detection relevant to asthma monitoring Sensors Actuators B: Chem.
209 639–44
[17] Kharitonov S A and Barnes P J 2002 Biomarkers of some pulmonary diseases in exhaled breath Biomarkers 7 1–32 [18] Jensen G C, Zheng Z and Meyerhoff M E 2013 Amperometric
nitric oxide sensors with enhanced selectivity over carbon monoxide via platinum oxide formation under alkaline conditions Anal. Chem. 85 10057–61
[19] Horvath I et al 2003 Comparison of exhaled and nasal nitric oxide and exhaled carbon monoxide levels in bronchiectatic patients with and without primary ciliary dyskinesia Thorax 58 68–72
[20] Van der Wal P, De Rooij N and Koudelka-Hep M 1996 Extremely stable Nafion based carbon monoxide sensor Sensors Actuators B: Chem. 35 119–23
[21] Pardon G, Gatty H K, Stemme G, van der Wijngaart W and Roxhed N 2013 Pt–Al2O3 dual layer atomic layer deposition coating in high aspect ratio nanopores Nanotechnology 24 015602
[22] Ho K-C, Liao J-Y and Yang C-C 2005 A kinetic study for electrooxidation of NO gas at a Pt/membrane electrode- application to amperometric NO sensor Sensors Actuators B: Chem. 108 820–7
[23] Blurton K and Sedlak J 1974 The electro-oxidation of carbon monoxide on platinum J. Electrochem. Soc. 121 1315–7 AQ2
AQ3
AQ4
QUERIES
Page 1 AQ1
Please be aware that the colour figures in this article will only appear in colour in the online version. If you require colour in the printed journal and have not previously arranged it, please contact the Production Editor now.
Page 10 AQ2
We have been provided funding information for this article as below. Please confirm whether this information is correct. (1) European Research Council: 267528; (2) VINNOVA.
AQ3
Please check the details for any journal references that do not have a link as they may contain some incorrect information.
AQ4
Please provide page/article number for reference [4].
