Measure 100% of CO2 inside a chamber

[Dave_PT]

Hi hello all.

I need to do something very simple, but at the same time it has become complex.

My task is to measure the CO2 concentration inside a pressurized autoclave up to a maximum of 2 bar.
The concentration should reach 100% (1000000ppm) because the gas will be injected through an industrial cylinder.
The aim is to expose a material to 100% CO2.

The material in test can absorb some CO2, and this difference (nobody knows, but its maybe 2/3%) needs to be measured.

The material can absorb some CO2, and this difference needs to be measured.
After some search, I've found some expensive sensors that do the job, but I need something that gives a rough value with the lowest price possible (of course).

So I found these sensors: MD62 from WINSEN (chinese brand) - ~25€ (aliexpress).



DATASHEET

The sensor, from what I can see through the datasheet, can do the job.
But the datasheet is too simple and without any practical information about the driver circuit, etc ...

Now the questions:
* Do you have any idea (other sensor, technique, etc.) to solve my problem of measure until 100% of CO2?
* Have someone use this sensor? Any feedback about it, please?
* Suggestions?


NOTES:
Inside the autoclave I will have some other sensors, like temperature, humidity, pressure (MS5803-05BA) and to complete, need the CO2 sensor.
The interior of the chamber can reach 90% of humidity and can be heated to 70 ° C.


Thank you all!

[ejeffrey]

I'm not clear what you are actually trying to do.

First off, you won't reach close to 100% CO2 unless you either A) pump the chamber out (vacuum) first, then backfill with CO2, or B) Flow CO2 through it for a while to purge any air present.  If you are trying to measure CO2 absorption by looking at the decrease in CO2 pressure, a continuous flow is going to make that quite difficult.

The sensor you posted is a wheatstone bridge.  So you apply a voltage to the "input" terminals and sense the output voltage with an instrumentation amplifier.  You should be able to find plenty of example circuits for measuring a wheatstone bridge online or in any electronics textbook.

The sensor in question appears to be a thermal conductivity sensor.  So it measures the difference in thermal conductivity of air vs. CO2. But pressure and other cases may also change the reading.

[Ian.M]

It will be very difficult to distinguish between 99% CO2 and 100% CO2 using a CO2 sensor. 

Also that sensor does *NOT* offer CO2 detection with total discrimination against other gasses. e.g. 1% of a hydrocarbon gas can cause a 30mV output change in the opposite direction to that caused by CO2, counteracting the reading one would normally get for a 60% CO2 concentration.  Furthermore the datasheet says prolonged exposure to a high 'gas' concentration will permanently affect the sensor (but doesn't specify whether or not the gas in question is CO2).

IMHO you'd be better off attempting to measure whatever non-CO2 gasses your material outgasses (assuming you know their expected composition), or maybe the residual oxygen concentration as a proxy for the proportion of atmospheric air in the gas mix.

[tpowell1830]

If the material absorbs the gas immediately, this probably won't work. However, if you are raising pressure to 2 BAR in order to infuse part, there is hope. (And as my dad used to say "Hope in one hand and spit in the other and see which one fills up first.")

Step 1. Vacuum chamber down to as near vacuum as you can get.
Step 2. Isolate chamber with valve, turn off vacuum pump.
Step 3. Inject the CO2 into chamber, until you reach a predetermined pressure (2 BAR?).
Step 4. Wait until pressure stops dropping (material has absorbed all it can).
Step 5. Measure final pressure, determine with math how much CO2 is lost into material using volume of gas (derived from delta drop in pressure, cubic volume of chamber (you figure out math part)) as a percentage.
Step 6. Exhaust the pressurized gas from chamber.
Step 7. Remove infused part from chamber, lose all infused CO2 (eventually, depending on permeability and other factors).

Just my 2 cents...

[Dave_PT]

B) Flow CO2 through it for a while to purge any air present.

This is how it is done.
Sorry for misunderstanding.

The pressure sensor are only to give feedback about the pressure inside the chamber.

The sensor you posted is a wheatstone bridge.  So you apply a voltage to the "input" terminals and sense the output voltage with an instrumentation amplifier.  You should be able to find plenty of example circuits for measuring a wheatstone bridge online or in any electronics textbook.

After reading the datasheet I did not realize that it worked like a wheatstone bridge. The image does not help 


@Ian.M
Thank you, you have made very important observations 


@tpowell1830
Thank you. Based on your 7 steps, I will ask some questions about it.
I think that the pressure inside the chamber need to be the same. The industrial cylinder have a mechanical regulator that maintains the pressure in the defined value.

Maybe, what I need is a sensor that measures the quantity of gas that enter in the chamber ...
With this and the chamber volume, I can get a percentage of absorption.


I will come back with more news about this topic.

Thank you 

[ejeffrey]

After reading the datasheet I did not realize that it worked like a wheatstone bridge. The image does not help 

It is the first sentence in the datasheet.

Maybe, what I need is a sensor that measures the quantity of gas that enter in the chamber ...
With this and the chamber volume, I can get a percentage of absorption.

A mass flow controller is that device.  But I doubt you are going to get the kind of accuracy you want.  You are going to have to flow several chamber volumes of CO2 to purge most of the air.  So you will be measuring mass in, mass out, chamber pressure and volume to calculate a missing gas term that is almost certainly going to be too small to see.

If the absorption process is slow enough what tpowell1830 said will work.  Just vacuum/purge the air then fill to target pressure quickly and close the valve to the CO2 tank.  Then wait for the absorption to happen and see the pressure drop.  You need temperature control since pressure is proportional to temperature too.  If the absorption happens quickly, it is going to be hard to do.

I would look at other physical ways to measure this.

The first approach I would try is to measure the weight change.  If it absorbs CO2 it should get heavier.  You should be able to measure that.

Is the process reversible?  It might be easier to measure the CO2 released by e.g., heating the sample.  Can you force as much CO2 in as possible, then put the sample in a vacuum and heat the sample to drive off the CO2 and see how much the pressure rises (assuming no other gases are produced).

[LaserSteve]

For a similar project at work I went with an optical based USB sensor from CO2Meter.com. This eliminated the problem with the old bridge sensor  seeing residual water vapor.  We had two of their probes and I was incredibly happy with them.

Getting from 98% to 100% with a bridge sensor would be difficult, at some point you'd need a RGA or Mass Spec to calibrate it.


Optical far IR sensors using two sensors in a bridge with one looking through a reference tube and the other having an identical tube holding the sample gas get close to 100%, the two tubes have an optical chopper so you can use a lock-in amplifier and something like a globar light source.

I  had a 7 cubic foot oven that had to stay filled with  tightly regulated CO2/N2 mixture for an anerobic bacteria.  I will say this, there is the scientist's expectation and then what you could actually do.   It was not fun holding a balance between an O2 sensor, a pressure sensor for the n2, and the co2 sensor... We ended up consuming voracious amounts of gas every time a student opened the chamber door...

 Not to mention the water in the chamber to keep things humid making life hell, requiring the sensors to be kept warm. In my case a small fan pulled a sample across the sensor chamber..

Ended up with tank Co2 and a Ln2 Dewer to generate enough purge gas.

I'd second the resomendation for a mass flow controller  based means of filling the gas. If you can't pull a vacuum on the chamber you'll easily be purging it 5-6-7 times..


Steve

[Dundarave]

This may or may not be useful:  there's an interesting technique used in medicine for quantifying the ability of the lungs to pass gasses to the bloodstream that involve inhaling a predefined amount of a gas (regular air mixed with a tiny amount of carbon monoxide (CO)) and then measuring the concentration of the exhaled CO to determine how much "stayed in" the patient.  They apparently use CO because of its affinity for red blood cells, and I imagine because it's also easy to detect.

Google "spirometry carbon monoxide" for more info. 

I thought it might be an interesting approach to your problem:  inject a representative gas into the chamber, let it be absorbed by the device under test (DUT) then somehow remove the DUT and use the reduced partial pressure of that gas to determine how much was absorbed by the DUT.  I have no idea of the practicality of this in your case, but I thought I'd mention it.

[CatalinaWOW]

You need to understand the purpose and accuracy requirements before you begin, and those on this forum will also need to know to be most helpful.

Purpose?  Is the purpose to measure how much CO2 is infused into the product, or to measure purity of CO2 in chamber, or to by inference assure that an acceptable level of other species (N2, O2, H2O) are present.  All of these are consistent with what you have said, but each would lead to a different measurement approach.  Those three certainly don't identify all of the possible purposes.  Also required is data update rate.  To illustrate this if the measurement is needed only weekly or monthly, the best approach might be to send a sample off to a lab.

Accuracy?  Your question placed in terms of parts per million implies something in the neighborhood of ppm accuracy (maybe a few tens or possibly even hundreds ppm).   Virtually impossible to achieve with the sensor shown.

[Dave_PT]

Sorry for the delay, but this is a work to one university and some answers take some time.
The need of measure CO2 concentration is because the experiences can be made with different gas concentration.

The first approach I would try is to measure the weight change.  If it absorbs CO2 it should get heavier.  You should be able to measure that.

Right now this is the method that is being done.


@LaserSteve
I saw the CO2Meter.com sensors. In fact I've a very good impression trought the website. But they are not very friendly to "integrate" into a device (the aim is to design a device to record all data during the experiment).
In relation to the NDIR sensors, I have not found one that can read until 100% ... do you have any reference?


@Dundarave
Thank you for the suggestion.
In this case I do not think that it solves the problem ... but it brings some other ideas.

At the end, all of these ideas will be part of the solution 


@CatalinaWOW
Let's make a resume:
So we have a chamber filled with normal air and with a test material dentro. At some point a valve on top is opened to release air and another in the bottom is opened to let the pressurized CO2 enter. The goal is to expose the material at different CO2 concentrations (only mix CO2 with normal air) and see when, with what concentration and how much CO2 the material absorbs.

Relative to the accuracy I think (because no one knows the exact numbers) that 1000 ppm is enough.


Thank you all for the help and suggestions.
Please make more questions, cause maybe I'm not making myself explain.

[CatalinaWOW]

OK.  Assuming you wait long enough for the gases in the cylinder to be well mixed you might do well to measured optical absorbtion.  At peak absorbtion around 4.2 micrometers absorbtion at standard pressure will be near total over only a few dozen cm.  By picking a narrow band near the peak absorbtion you can get an "easily" measured quantity that should be able to directly measure the CO2 density.  Of course this is designing, building and calibrating an instrument, not buying a cheap component.

[LaserSteve]

The Co2meter folks are probably bombarded by the "recreational pharmaceutical" people about their sensors. I could see them limiting development data for that reason.

Based on past experience if you call with a legitimate use, I
Think you would get very good integration support.

Steve

[Dave_PT]

Lets assume that the normal air is (only for simplification) only one gas with a specific "n" (number of moles).

If the chamber has the same volume and the same temperature, from ideal gas law, I can get:
PV=nRT => P=n*(RT/V) => P=n*k

And we can assume that the total pressure inside the chamber is equal to the sum of the gases pressures inside.

So Ptot = Pa + Pb, "a" and "b" are ideal gases.

This is valid for the 2 gases, and if the constant k is equal to both, so I can write:
Pa/na = Pb/nb

I can put all math into a system of equations.

Ptot = Pa + Pb
Pa/na  = Pb/nb

In summary I can control the percentage of the mixture only by looking at the pressures.

By now, let's ignore all the specific calculations and assume that (assuming only) we have 2 gases with the same number moles.
So if I pressure the chamber with 1bar of gas "a" and add a 1 bar of gas "b", I think that is correct to assume that the mixture have 50% of gas "a" and 50% of gas "b" (with a fan inside to mixture the gases).

The next step is decompose the normal air (assuming: 78% nitrogen, 21% oxygen, 1% argon) and solve all the math.
Then the MCU will do the smart calculations and control some electrovalves (gas "a", gas "b" and chamber exhaust) and use the feedback of a good pressure sensor.
With all of this and the logging I think that is possible to achieve the expected results.


Right now, I'm waiting some answers from the university...


Thank you all for the support.

[CatalinaWOW]

Your approach will work if you meet the condition that the exhaust gas is a perfect mixture of the input gases.  It is analogous to fluid dilution in beakers.   The only way I can see to achieve that is to never have an input valve and exhaust valve open at the same time, and to force a delay between opening the exhaust and input long enough to achieve perfect mixing. Which may work for your situation, though you earlier mention continuous flow.  The interval could be reduced with a stirring fan.  You may have to compensate for gas absorbed by fixtures and samples in the chamber (can be detected by anomolous pv=nrt results).  Will definitely need to measure both pressure and temp, s constant temp assumption will not be valid.

[ehughes]

I have done something similar (for Hydrogen) acoustically.  If you can also measure temperature with a decent amount of accuracy and precision then it is pretty straightforward.      Sound speed in C02 is well understood.   Since you have some situation awareness (you know what gases are present) it should be pretty easy to do.        We used a resonator to extract sound speed but a time of flight could work well.     

There really isn't much off the shelf but you could cobble something together.

[Dave_PT]

@CatalinaWOW
Thank you!
The idea is, since the CO2 gas is expensive, open the exhaust valve and the normal air valve (this one connected to an air compressor) and let run for a while to ensure that inside the chamber there is no "CO2". Then close the exhaust valve and start the compression of air inside until reach the calculated pressure. Then close the air valve and open the CO2 until it reached the calculated pressure too and close the CO2 valve.
To maintain the pressure of CO2 (to compensate the material absorption) I will need do some electronic control and create a hysteresis. For example, when the pressure fall below 1.9bar I "stop" the experience and inject more CO2, until reach the 2bar pressure.

The pressure, temperature, humidity and electronic state of each valve will always be part of the continuous logging and more detailed calculations are made after the experiment ends.


@ehughes
Thank you.

For reasons that I can't control, the time slot for testing aren't too big. So I need to develop this (with some level of accuracy) and assembly it in place in a weekend.
But I will read about your suggestion to better know this technique.

[SiliconWizard]

The interior of the chamber can reach 90% of humidity and can be heated to 70 ° C.

The sensor you picked so far would not cut it then. Its operating temperature range is -20°C to +50°C apparently, so I wouldn't count too much on its behavior at 70°C.
As said above, it's a wheatstone bridge basically and the characteristic output at 3V ref. voltage is given on the first graph (mV vs. %CO2 if I get it correctly). Given its nature, I guess it would require calibration as well. Its rated "sensitivity" (around 0.5mV/%CO2) is just given as a ballpark value and should be calibrated.

Getting a reliable measurement of just a couple % difference at around 100% CO2 will be extremely challenging for most sensors as well IMO.

The sound speed idea looks interesting but the "pretty easy to do"... not so sure.

I've thought of some "air quality" sensors (some are rather accurate for CO2 measurement but usually optimized for low CO2 concentration so typically under 5-10%CO2) so... not an option.

[Dave_PT]

@SiliconWizard
Thanks.
After some discussion and study, I think that THIS is the way to have success.