Building a gas chromatograph

[@AdvancedTinkering]

Great Video! I saw a similar design many years ago somewhere on the internet. But your video was grear to watch! It would be interesting to see how far you could push a DIY GC with a longer column (maybe thinner and made from metal), nitrogen as a carrrier gas, other detector designs and maybe even an oven. But your approach to make it smaller and integrate it into a robot is great too!

[@Marbslab]

Thank you. An oven is not hard to build, a heating element, a temperature sensor, a PID controller, and some isolation material. For the robot nose I plan to use a MiCS-6814 gas sensor, which has three fully independent sensing elements on one package.

[@audiaturetalterapars327]

It's interesting how easy it is to build a GC. Thanks for the video.

[@AccidentalScience]

Very interesting build and video. Love the approach of using simple components.

[@GermanMythbuster]

Would love to see a part 2, there is still a lot of performance left on the table, wich would be a interesting journey to see. Love it!

[@Marbslab]

I am working on it, beside of many other things:)

[@simonmoll3843]

Great work! Reducing inner column diameter helps against peak broadening. I experienced poor sensitivity of a MQ135 in a similar setup - i suspect, that the carrier gas flow cooled down the tin oxide layer as i forced it into the wire mesh like you did as well.

[@Marbslab]

Thank you:) For the next version I'll reduce diameter of the column, add an oven, and use a MEMS gas sensor array.

[@peiw.5768]

This is soo cool. I'm studying analytical chemistry this term and learning things about GC. So fascinating to see you build one! I'm curious about your education background and works.

[@Marbslab]

Thank you. I am an engineer and mathematician, chemistry is just a hobby. I am particularly interested in combining chemistry with other fields of science, for example microfluid systems and robotics. Good luck with your study:)

[@videomaker8350]

Amazing project!

[@leonciopina3618]

Excelente trabajo. Muy novedoso. Una herramienta práctica para muchas aplicaciones.

[@christopherleubner6633]

If you want to upgrade it to make an ion mobility spectrometer a very convenient ion source for a diy GC is a Am241 source from a smoke derector for positive ions or one of those gimmicky germ killer lights made for sanitizing cell phones.

[@Marbslab]

Great tip. I used Am241 for my cloud chamber and Geiger counter project:)

[@midi5581]

For miniature version you can use 1/8" PTFE tubing, there is a lot of adapters and connectors for these. Also you can consider making derivatized silica, for example C-18 one using octadecyl-dimethylchlorosilane (procedure is to dry silica first, and then reflux for 24h in toluene your ODS reagent, silica and some base (imidazol or TEA for example), filter, wash and dry).

[@Marbslab]

Very good tips. Thank you.

[@k.o.0]

Excellent! To improve resolution I suggest you use a much smaller ID column within the current tube (that you use as a heated air jacket). These 2 additions will significantly improve resolution. Very creative design!!

[@christopherleubner6633]

Yup thin ptfe tubing or silica tubing is used in most of them.

[@Marbslab]

Thats a great idea! Like a Liebig condenser.

[@VenturaIT]

@@Marbslab modern gc has colum dia of about .1 mm to .53 mm and a length of up to 60meters

[@Fabian-mu3hq]

Amazing work, thanks for the video!

[@Ezio-Auditore94]

Amazing content, so well structured and explained

[@stanislavdenysenko2007]

If you will evolution this project with thermal conductivity detector, it will look very similar to the old Soviet chromatograph Gazochrom 3101 (Газохром 3101) )). There were two columns, two sensitive elements. One of them is thermochemical. A drop of a special substance is applied to the heating coil, which works as a catalyst for the oxidation of combustible gases and increases the response of this element. In particular, the sensitivity to hydrogen was 10 times higher than that of "casual" laboratory chromatographs with a thermal conductivity detector. I do not know how deeply you have dived into chromatography, so, just in case, a few tips that will be trivial if you have dug deep enough. The detector that you made is highly sensitive to fluctuations in the carrier gas flow. Old books on the design of chromatographs advised using two identical columns and two sensitive elements in opposite arms of the measuring bridge. The flow was divided into two by a splitter before the columns. This also helped to reduce fluctuations in the power source. Performing analyses alternating columns helps maintain their identity. Such a scheme allowed to obtain a linear range of the detector of the order of 10^5 on technologies of 60s - 70s. Moisture from the air, accumulating in the column, worsens its operation. A filter-dryer is needed at the compressor inlet. In some cases it is possible to use hydrogen as carrier gas. It reduces analysis by 1,5 .. 2 times with same separation. Probably, this will be somewhat inconvenient for the robot. Modern electrolyzers with platinum catalysts are quite expensive, requiring about 10 W*h per liter of hydrogen and deionized water as source of it. The chromatographs we worked with usually had an internal column diameter of 2...3 mm and a carrier gas flow rate of 20-40 cc/min. One old portable Soviet chromatograph (I don't remember the name) had an interesting layout. The detector was wrapped in thermal insulation (about 10 mm), a column (about 1.5 m) with a diameter of 2 mm was wound around it, again about 10 mm of thermal insulation and an outer casing in the form of a screw-on mug with a volume of about 700 ml. (you can try modern double-walled food containers - that's the thermal insulation) Thus, heating the detector also heated the column. Regarding the electronic nose, about 10-15 years ago I heard about one Russian team, they experimented with straight polycapillary columns about 300 mm long. They allows to get separation 2...3 times better than the column in the video. The analysis time was in units of tens of seconds. The main difficulty is that for an excellent result, the dosing device had to inject the sample into the gas flow in about a millisecond. The thermal conductivity detector is highly sensitive to fluctuations in heater temperature. In my experience, fluctuations of 0.1 degrees are already very clearly visible on the signal. It is better to insulate and not heat than to heat with an inaccurate heater. However, introducing some thermal resistance between the heater, which fluctuates, and the detector body can help reduce detector temperature fluctuations. Experiments are needed. About 25 years ago I talked with an old designer of Czech chromatographs. He told that they did induction heating of the detector by thermal conductivity. At that time (more then 40 years from now), they had to make a casing from special steel over stainless steel. With the frequencies of modern power electronics, you can heat stainless steel "directly" (as it can be said about induction heating)). The second interesting method of making a thermostat I saw in a microcolumn liquid chromatograph - a casing was put on the column, inside which there was a coil of bifilarly wound copper wire with a precisely adjusted resistance. Simultaneously a heater and a temperature sensor.

[@Marbslab]

Thank you very much for all these suggestions for improvement:) After a quick search (using Газохром 3101) I found some papers like this one: mash-xxl.info/page/179247162200251113237025107190238124031173230150/ Exciting stuff.

[@jim2725]

Awesome video! Would it be possible to make a HPLC system as well? Or would the detector be that much more complicated?

[@Marbslab]

Sure it's possible. Detection is usually carried out using UV or fluorescence spectrometer, which is not hard to build.

[@valentinoboi96]

Amazing Video

[@landrec2]

Very cool!

[@ferriswhitehouse1476]

Do you have to use "chromatograph-grade" silica gel? Or would any standard silica gel used for packaging work? Thanks for this awesome video, great work.

[@Marbslab]

Silica gel used for packaging is usually in spherical form. This would not be so good, as many voids would be created when filling the column. Perhaps crush it in a coffee grinder. I have read that people used to use ground cat litter. You would have to experiment with that. The column is indeed easy to replace.

[@ferriswhitehouse1476]

@@Marbslab Gotcha, thanks for clarifying.

[@memejeff]

Very cool idea.

[@GiGaHarrySfotter]

AMAZING!!!

[@markawbolton]

Good work. I have always found the idea of SMD assembly intimifating but seeing your demo makes me reakise it shouldnt be,

[@Marbslab]

Thank you. SMD soldering is actually quite simple using the reflow method. With a little practice, many components can also be soldered by hand. Sooner or later, due to the ever-increasing miniaturization, SMD components will become indispensable.

[@electricalychalanged4911]

Dein sensor muss auf beiden Seiten von Gas umspült werden. Sonst ist der Leitfähigkeitssensor sehr fehleranfällig weildeiin 0-Punnkt schwannkt. Aber spannendes Projekt

[@Marbslab]

Da der Volumenstrom sehr gering ist, wird er das auch einigermaßen. Aber definitiv guter Tip, der in die nächste Version mit einfließen wird. Danke!

[@testboga5991]

Gut und günstig! 😉

[@Marbslab]

Sowieso

[@VenturaIT]

have you been able to convert to a concentration? would this work for ambient ethylene oxide? i need to measure ambient ethylene oxide down to 100 ppt... what's the purpose of the VOC sensor in there?

[@Marbslab]

The concentration can be calculated by the integral of the peak (area below the peak in the gas chromatogram). The VOC sensor serves as a second sensor. It is not as accurate as the thermal conductivity detector, but is easier to handle. Theoretically, the thermal conductivity detector should detect ethylene oxide. However, I would not rely on a simple DIY gas chromatograph for this highly flammable, toxic gas.

[@Yorumcu63]

Great video! Only this video is enough for me to subscribe, and I’ve already subscribed. Have a pcb file (or gerber file) about shield? Could the upcoming videos be about spectrometers? Do you have any plans for that?"

[@Marbslab]

Thanks for the sub:)I am planning videos about microfluid systems, including a colorimeter, a new type of medical device and a mini version of the gas chromatrograph. I will definitely build a spectrometer at a later date, as I am very interested in this.

[@Buongona]

I'm guessing that the first patents on this thing have gone old long time ago... why are they still costing so much if they can be built with such rudimentary things as arduino?

[@Marbslab]

Many scientific devices are very expensive. On the one hand, this may be due to the public funding of scientific projects, from which the manufacturers take a big slice. On the other hand, such analytical devices are high-end. The amateur researcher either falls by the wayside or builds such devices himself.

[@NionXenion-gh7rf]

Nothing to do with arduino. GC as a system is highly sophisticated one. Data logging plays minor part. This on the video is nice semi demonstration but is ultra unprecise. And can't really be gas chromatograph because of too many aspects, not just the air that varries in composition from room corner to room corner...

[@YuriyKrivosheyev]

Great, thx!

[@YO2LYP]

Very educative material. Where do you plan to use this GC? Thanks!

[@Marbslab]

I have some chemistry experiments in mind where I will use it. It's also a test platform for my main goal to build an artificial nose for robotics.

[@JimboJeff-qm7vt]

Galaxy gas be the new universal solvent

[@louistournas120]

What is that gas detector that you placed there?

[@Marbslab]

It's a MQ-2 gas sensor which is sensitive to LPG, i-butane, propane, methane, alcohol, and hydrogen: www.az-delivery.de/en/products/gas-sensor-modul

[@smizmar8]

Huh, you're actually thinking about the artificial nose. I think it might be key to a low power/ low compute method of identifying your "family," well I guess the humans the robot is most often around, in my case assuming some kind of in home assist or pet. I've been thinking about it for some time, glad to see I'm not the only one :)

[@Marbslab]

I think there would be many applications in robotics. Good idea with the pet. I have a cat whose behavior I study every day and it never ceases to amaze me :)

[@benotyourboss]

Verdammt cool! Was hast du studiert/gelernt? 😮

[@Marbslab]

I'm an engineer and mathematician.

[@benotyourboss]

@@Marbslab awesome! And are u german?🇩🇪

[@Marbslab]

@@benotyourboss Yes I am (Ja, bin ich).

[@benotyourboss]

@@Marbslab hahahha ach so. Ja dachte nur weil du mir auf die Deutsch gestellte Frage auf English geantwortet hast. Dachte schon weil einige Produkte so richtig Deutsch sind die du verwendest :)

[@mrobinson9297]

nice.

[@Xsiondu]

Well would ya look at that

[@ВасилийМельников-з8м]

Почему бы не поставить диод параллельно моторчику, и не перевести моторчик в режим неразрывного тока?

[@Marbslab]

The pump is driven by a MOSFET, that has already a protection diode. I am using a MOSFET to control the flow rate by PWM. Please use a translater, so I don't need to translate your comment from Russian to English. Thank you.

[@ВасилийМельников-з8м]

Yes, you're right, of course a MOSFET does contain a diode, but it's not quite what we need here. When the MOSFET is turned on, current begins to flow through the motor. When the MOSFET is turned off, this current is interrupted because the diode inside the MOSFET conducts in the opposite direction. This abrupt interruption of current through an inductive load generates voltage spikes, which you're trying to mitigate by placing a capacitor filter in parallel with the motor. This helps, but it leads to an increase in the current through your switch. If you remove the filter capacitors and instead place a diode with its cathode connected to the positive supply, then when the MOSFET is turned off, the current through the motor's inductance won't be abruptly interrupted but will instead decay smoothly until the MOSFET is turned on again, at which point it will start to rise again. This will improve efficiency and reduce noise levels. You can google for "Continuous Conduction Mode", to clarify what i mean.

[@Marbslab]

@@ВасилийМельников-з8м The interference suppression capacitors are important during the operation of the collector type motor. When operating such motors, interference of a high-frequency nature can occur due to their design, which, depending on their frequency, can propagate more or less strongly via the supply lines or directly. This is common practice. With an electromagnet, a diode would make more sense. If you use a H-bridge to change the direction of rotation a diode would be a terrible idea.

[@MatthewHolevinski]

That all looked extremely simple, except for designing something in kicad :)

[@Marbslab]

The good thing is it's an open source EDA. Once the schematics are drawn it's relativly easy to layout the board. But you can use any other EDA (fritzing is very easy to use). Mayby I publish all files as an open source project:)

[@jujjuj7676]

Absolutely love the video, but can you add voice please to explain what you do, at the very least text to speech with ai is top notch now so you don't have to use your own voice and still sound great. Just google best open source TTS in KZbin.. Would really make us all happy!! 👍

[@Marbslab]

Many people hate TTS in videos, but I am thinking to do a voice over in one of my next videos to see if people like my voice:)

[@AccidentalScience]

​@@MarbslabI for one love real voices, even with a strong accent.

[@Marbslab]

@@AccidentalScience I'll try. Much easier than to handle all the text in the video editor.

[@jujjuj7676]

@@Marbslab im 100% sure ur voice is fine.. Also people hate bad TTS its already different now ..they blow me away they even have breath sounds and laughing down pat now..wonderful stuff.

[@blenderbuch]

Great!

[@measurementsfriend4887]

Hello Marb's Lab. What is the problem with my comment? Why is it not being published? Was it wrong what I wrote?

[@Marbslab]

I just checked it. I don't see your first comment and I didn't delete it. KZbin sometimes behaves weirdly.

[@measurementsfriend4887]

@@Marbslab OK. sorry. Than I try it again.

[@measurementsfriend4887]

Maybe there is a problem with hyperlinks...

[@Marbslab]

@@measurementsfriend4887 Could be. You can email if it's important (but no homework questions😁).

[@sebastianhama5624]

now do liquid chromatography :)

[@Marbslab]

I'm glad you didn't say I should build a mass spectrometer:p

[@odissey2]

​​​@@MarbslabGC is very slow to detect useful substances, as their molecules are heavy and typically need an oven to pass thru the column. The fastest speed for the short column would be approx. 1 minute. For microchannel GC columns you may approach scientists at Sandia Labs On the other hand, building an ion-mobility spectrometer is easier, and its response time for heavy molecules is in 40ms range. I recommend searching for Fourier Transform IMS design by the Sandia Labs as well.

[@odissey2]

​​​​@@Marbslabi recommend an ion-mobility spectrometer. It has much faster response (40ms), which may be more suitable for robot's "nose" . It is also much easier to build, smaller and low-power consumption. The advantage is that it responds to heavy molecules, which are of more interest.

[@Marbslab]

@@odissey2 Puh, that would be a bit of a challenge to build. As far as I know, IMSs have high error rates and a limited range of chemicals they can detect. Anyhow, thanks for the input, I will put it on my list of research topics. Love the ideas I'm getting here:)

[@odissey2]

@@Marbslab the issue with GC - it is slow. Even using microchannel columns it has about 1 min response time. Plus, it needs heating and pumps, which adds size and power consumption. It is true that IMS is very low resolution. However, for molecular mass >200, the home-built GC will also be low resolution as well. Search for FT-IMS, as it does not need fast ADCs.