Gas Sensors for Respirometry

Respiring animals consume oxygen and release carbon dioxide in the process of oxidizing metabolic fuels to release energy. However, the exchange of oxygen and CO2 using water as a carrier solvent occurs across moist membranes. Thus, some of that water will evaporate during gas exchange and therefore water vapor also comes into play, also from ambient sources. Thus, the key sensors for respirometry systems are oxygen sensors, carbon dioxide sensors and water vapor sensors. Now do take note, we will only be looking here at aerial measurements of these gases, and we will also only be looking at measurements that yield voltage signals.

Now there are three primary techniques used most often to measure oxygen in the respirometry systems. The first technique is the paramagnetic effect. This was discovered by Michael Faraday working with William Thomas Brand in the mid-1800s. So, oxygen is strongly paramagnetic, meaning that it is attracted to a magnetic field. By contrast, nitrogen is diamagnetic, it is therefore repulsed by a magnetic field.

Thus, the construction of a paramagnetic sensors consists of a set of permanent magnets, these grey blocks, two nitrogen filled spheres positioned between the permanent magnets – and those are quite often made out of little glass spheres. And those glass spheres are mounted on a quartz thread to which a mirror is attached and then there are electromagnetic windings. These are quite often wound around the nitrogen spheres. The sensor is constructed so that the incoming airflow will have minimal effect on the position of these spheres as it is balanced on that quartz thread. The oxygen in the air attracted to the magnets will exert a force on these spheres and therefore generate a torque on the mirror mounted on the quartz thread. In the older paramagnetic sensors, the light source reflects off of the mirror onto a scale that was then read to estimate the oxygen levels. However, due to the nature of magnetic fields this was decidedly nonlinear, and also due to the limitations of the scale itself, it did not yield very fine resolutions.

Modern paramagnetic sensors use the electromagnetic coils around these spheres as weak electric motors. Once oxygen causes the spheres to deviate from a null position, thereby moving the mirror, a current is sent through these spheres. Now based on the deflection of the mirror as reported by a photo sensor, the amount of torque required to return these spheres to their null position and then be converted into an accurate oxygen measure. This can make for a very accurate oxygen measure.

The advantages of a paramagnetic oxygen analyzer are that it can be very accurate, and it also has a very fast response time. However, there are also some disadvantages. First, this is a very delicate system. It is not quite suited for rustic fieldwork applications. Second, it is also orientation and flow rate sensitive. The system has to be calibrated in the orientation that it is placed and going to be used and needs to be recalibrated as soon as its orientation is changed. And also flow rate should be limited to no more than about 200 moles per minute. And then finally it is also a very expensive sensor.

A second technique for oxygen sensing is to use a zirconia cell sensor. The working principle of the sensor is that zirconium oxide, when heated, acts as a solid-state electrolyte through which oxygen can migrate. As the oxygen migrates across it an electric charge is generated across the wall. The sensor is also known as a lambda probe and the signal that is generated by this movement of oxygen obeys the NERNST equation quite accurately and that allows for measuring oxygen quantities. However, it is highly non-linear, but the system can be very sensitive at low oxygen partial pressure.

The most widely known zirconium oxygen cell sensor is the classic S3A made by Joe Weisbach originally from Applied Electrochemistry international (AEI), eventually sold to Ametek. These are the classic analyzer boxes that just take in the electric signals. These are paired with the zirconium cell itself mounted inside its own electric furnace box. Now these sensors are fast and usually quite accurate especially at low oxygen concentrations. However, they are very noisy, and they are actually quite unreliable. They do require very regular calibration. In my own experience, I had to calibrate some of these sensors after every experiment done. And they can also be prone to errors from volatile organic compounds. Due to the heating of the cell those volatile organic compounds can be combusted inside of the cell thereby changing the oxygen content. And these cells are also often at risk of explosion under certain conditions. Most notably either due to excess VOCs or sometimes due to water vapor in the system that then gets pressurized in the heat.

The final technique is the fuel cell oxygen sensor. This is an electrochemical sensor, and this works on the principle that oxygen migrates across an oxygen permeable membrane where it oxidizes a lead anode. Each oxygen molecule released four electrons resulting in a small current. With carefully constructed electronics, this current can be amplified yielding an oxygen measure. However, the sensor is very sensitive to barometric pressure and also to temperature. But with adequate pressure correction, and also with either temperature compensation or temperature control, the fuel cell sensor can yield high resolution measures at a very high accuracy. The advantages of the system are that it is accurate, it is low noise (although it requires skill by the manufacturer), and this low noise is also due to the slow response of the system. And it is also very rugged, making it ideal for rustic fieldwork conditions. The disadvantage of the system is that it has a fairly slow response time, but we’ve already said this can actually be a benefit due to the low noise the slowness generates. But it is also very sensitive, and low drift circuitry is required, and it has a limited lifetime being an electrochemical sensor. The lifetime is run about 18 to 24 months but think about it this way – it is a fairly low-cost replacement to have a new fuel cell and effectively every two years your analyzer will have a brand-new sensor.

So how do we calibrate oxygen sensors? Generally, near atmospheric levels only the span calibration is important. For most cases zero is very close to correct. One exception in this instance is the paramagnetic sensors and those must be zeroed regularly specifically to get that null position. It’s best to do the span correction on dry air. In a room the oxygen percentage is usually around about 20.9 percent whereas outside dry air has an oxygen percentage of 20.94. This is very stable, and it is more accurate than most span gases which can usually deviate by one to five percent.

Now there’s a very important point to keep in mind with oxygen analyzers and oxygen sensors. The biggest source of error is not recognizing that partial pressure is actually measured. Thus, for a given oxygen concentration, when the pressure changes the absolute amount of oxygen molecules in that volume of air will also change. And not recognizing that can cause some serious measurement errors. Thus keep your sensors internal pressure at ambient. Always place the oxygen analyzer last in the analysis chain and never pull air through an oxygen analyzer. That pulling action will reduce the pressure in there making a drift in oxygen, and also never add instruments downstream. The added instruments will increase the resistance in the airflow path thereby increasing pressure and again resulting in oxygen deviations.

Next, CO2 analyzers: The most common CO2 sensors being used are infrared sensors. Typically, CO2 is sensed via infrared absorbance at around about 4.26 microns. This is in the medium infrared range and at that wavelength everything glows. Everything that has a temperature above absolute zero. Imagine taking a photograph with a glass camera. Now we may only be interested in a light coming through the lens forming the image on the sensor, but due to the fact that the camera is made out of glass, light is entering everywhere heating the sensor making for a very dull picture.

In the context of CO2 measurement, consequently the infrared source must be differentiated from the background infrared sources. So how to do this? It would be by modulating the infrared of interest to a known frequency. So, we have an infrared source and an infrared detector, and the source light infrared light is passed through a gas path where it will be absorbed. To modulate the infrared light that can be done mechanically by using a chopper wheel. This is usually a disk with evenly spaced holes in it and that rotates at a known rate creating a known frequency and then the detector only needs to pay attention to the light that is emitted to it by that frequency. The alternative is to eliminate the chopper wheel and directly modulate (electronically modulating) the infrared source and that is what we do with most of the Sable Systems instrumentation.

In most cases substances that absorb light based on the concentration of that substance in a medium follows the Beer-Lambert law where there is essentially a linear relationship between the absorbance of light and the concentration of the substance in question. Not so with CO2. CO2 over a wide range of concentrations is decidedly nonlinear. However over narrower ranges of CO2 concentration the relationships can be linearized. This has important implications for calibrating your CO2 sensor.

So how do we calibrate CO2 sensors? Important two calibration points are required. The most critical for regular calibration is the zero calibration. That can be done by either using nitrogen, zero CO2, or chemically scrubbed air either using Ascarite or soda lime. Soda lime is okay if you use a sufficient quantity of that. And doing a zero calibration where no CO2 is present can also be automated by zeroing your analyzer either by switching in a scrubber column or adding a nitrogen cylinder on a computer control as is done in many of our Promethion products.

For span calibration you should use a span value very close to the measurement range that you want to use. Recall that these sensors can be linearized across very narrow ranges. So do not use a 5% span gas if you are going to use your analyzers to measure CO2 concentrations around 0.25%. For most of the metabolic phenotyping work that’s going to be done. a span gas of around 0.5% is best. And important: always do your zero calibration first and do your span afterwards. And never force a span gas into a system that is being pumped. That will just cause measurement errors. Because the nonlinearity of CO2 gets worse at higher concentrations, minor deviations at the shallow zero end of the curve can translate into significant errors for higher measures. Thus, it is advisable to re-zero the sensor often and to always calibrate zero before span.

Finally, water vapor sensors. But first some background about water vapor as a gas. Mixed in air, water vapor has a partial pressure just like all the other gases that make up air. But above a certain critical partial pressure. water vapor will saturate and then condense into liquid water. This saturated partial pressure is very much temperature dependent. And the temperature at which the measured partial pressure of water vapor will saturate and condense is known as the dew point. That is a key measurement of water vapor.

One of the other common measures is relative humidity. Now relative humidity is actual water vapor pressure divided by saturated water vapor pressure at the temperature of measurement. It is thus very much temperature dependent. The same humidity value – say 50% relative humidity at 20 degrees Celsius – does not indicate the same amount of water vapor in the air than a measurement taken at 35 degrees Celsius. It is thus not a very good quantitative measure to use. Another relevant measure, quite often more useful, is water vapor density. Water vapor density literally is the volume or the mass of liquid water per volume of air. Now all of these units of water vapor in air are interconvertible using fairly complex nonlinear equations. But those are topics for a separate discussion.

So how do we do water vapor sensing? There are two basic techniques. The more accurate technique is the “chilled mirror” water vapor sensor. But the more practical technique is capacitive water vapor sensing. The basic principle of the chilled mirror water vapor sensor is having a channel through which air is flown and mounted inside the channel is a mirror attached to a finely controlled Peltier cooler. The light source is reflected off the mirror to a detector. As the Peltier unit cools the mirror, a critical temperature will eventually be reached where the water vapor in the air will condense. This condensation will thus blur the mirror affecting the light reflected on the photo sensor. The temperature where this occurs is the dew point.

This can be viewed as a graph like this. At the non-critical temperatures, the reflectance of the mirror stays constant until the critical point is used whereby the mirror will fog up greatly reducing reflectance and then staying stable after that. The advantage of a chilled mirror sensor is that it is very accurate. It can actually be used as a primary standard. However, the disadvantage is that a system like this is very expensive. It also has high power requirements to drive that Peltier unit and it needs a very large volume for the water to move through. Consequently, it also has a very slow response and it only measures dew point. If you want to convert to any of the other water vapor units, an additional pressure sensor and also an air temperature sensor is required to make the conversions from dew point to either water vapor pressure or water vapor density.

The capacitive water vapor sensor consists of a bulk polymer that absorbs water vapor and the absorption of this is proportional to free energy for absorption, thus to relative humidity. The absorption of water in this polymer changes the capacitance of the polymer. A change in capacitance is converted to a change in the frequency of an electric oscillator. That frequency can then be converted into a voltage, which can then be measured by our data acquisition systems. If we know the temperature of the capacitive sensor, we can thus calculate water vapor pressure. It is one of those simpler equations where relative humidity can be turned into a more useful measurement unit.

The advantages of the capacitive water vapor sensor are that it is inexpensive, it is light and durable. It’s quite useful for fieldwork applications and it is also quite accurate if it is well calibrated. The disadvantages are that it is not a primary standard and it requires regular careful calibration.

So how do we calibrate these systems? Water vapor sensors can be zeroed with pure nitrogen or with air scrubbed by magnesium chlorate or Drierite. And for the span calibration there are two options. Either use an accurate dew point generator or measure the oxygen dilution effect of water vapor. To span with a dew point generator, it is best to use reverse osmosis deionized water in the dew point generator because any dissolved solids in the water affect the evaporation mechanics of the water. Set the dew point generator to an appropriate dew point temperature, usually somewhere between 5 to 15 degrees Celsius. You do not want to send too much water through the system. Then allow the humid air to flow through the analyzer from the dew point generator and allow sufficient time for your analyzer to equilibrate. This should be at least two hours. Then set your water vapor analyzer to the dew point measurement and then adjust the dew point temperature to a value equal to the setting on your dew point generator.

Calibrating your water vapor sensors measuring the oxygen dilution effect of water can be done by using a very good oxygen analyzer with barometric pressure measurement. In that case we will assume that if accent io2, this is the fractional content of oxygen, is equal to chemically dry fractional content of oxygen in air. In that case water vapor pressure is then the barometric pressure multiplied by the difference in dried oxygen fractional content and humid air oxygen fractional content, and that is divided by the fractional content of oxygen in dry air. Using this technique one can then combine the zero calibration of carbon dioxide and water vapor and you also get a water vapor pressure span value out of this technique. This can all be automated in the Promethion systems.

Now for all of these measurements, oxygen measurement, co2 measurement and water vapor measurement, these are all fractional content measurements, and it is very important to know the flow rate at which these measurements are being made and that will be a topic for the next discussion. Thank you.