What compounds should be considered impurities in extraction gasses?

Testing and Analytics
81

Let’s talk about sampling for analyses.

Headspace analysis for water is definitely pretty straight forward, the hardest part is collecting the vapor phase without releasing it to the atmosphere. Drilling a hole in a mason jar then quickly putting in a gas syringe is usually the best practice but that type of sample testing is definitely not ideal for anything other than water.

I sincerely hope no suppliers in the extraction space are producing COA’s via vapor phase headspace sampling. But I don’t know that they aren’t. Another great reason for standardized practices. As @SidViscous mentioned above, such a sampling method would not result in a thorough analysis as it is not representative of the composition of the tank as a whole, especially when looking for heavier hydrocarbons.

However, ensuring single, liquid phase sample introduction is difficult and, as such, may not be a regular practice. To properly pull a single, liquid phase sample we have found it necessary to apply a 1,000 psi helium pad on the constant volume sample cylinder. The current requirement is only 200 psi above the sample pressure but, as soon you open the cylinder to analyze, that pressure starts to fall. The massive pressure pad allows us to avoid any bubble point issues which would produce lower recovery results. Even then, you’re still always limited by the sample collection itself, slow and steady is always best especially with such light hydrocarbons that would rather “flash” into the cylinder. The helium pad definitely helps squeeze all those hydrocarbons into one homogenous single phase sample.

Additionally, any sample drawn is only as good as the sample cylinder being used also. There is an “industry standard” sample cylinder available that, if we so decide, I can link into our ultimate sampling SOP.

Given this, can we safely assume that our standard should include testing only via single, liquid phase samples taken via an SOP that mirrors the above steps?

1 Like
82

We do complete cleans and fresh solvent once a month. We make sure we are at the end of our batch of solvent, then empty our tanks, clean them with water and soap followed by lab grade acetone. They then are ready for fresh distilled solvent to enter. Just got a new system with one large solvent tank so this will make this process a bit easier as its on wheels an one tank vs 4. Im sure there are lots of companies not taking the time to clean their solvent tanks nor using fresh solvent. Im a firm believer that there is carry over from runs when going at scale from time to time and so its needed to have some procedure in place to mitigate any contaminations.

5 Likes
83

You can actually do headspace effectively even looking for heavier compounds as long as your curve is generated using the same/similar matrix. IE if you use gc headspace grade butane and add the impurity standards to that, you can accurately quantify c6+. It will obviously not pick up the mystery oils though as they are nonvolatile. This method should be used to quantify water anyways because it is substantially less volatile than the light alkanes

2 Likes
84

I went to a gas tank filling training class in Denver provided by general air… what you have there is more than likely soot from the inside of the tank. From what was explained to us, the way they clean out gas tanks is by torching the inside of the tanks to eliminate any trace gases and anything else that might have built up inside of the tank. Not too sure how often it’s done, but it definitely comes through on majority of tank fills that we do. I started using the same setup I would for any other run as far as my filters go. I pass the gas through the usual 5 micron screens, followed by a one micron paper, on top of a one micron sintered disc. The collection vessel is always clean after that during tank fills.

4 Likes
85

Learn something new every day!

So would the best practice to be to analyze the liquid phase sample to pick up any potential nonvolatiles then if there is any indication of moisture look to headspace sampling to pin down?

2 Likes
86

Yikes to that cleaning method.

2 Likes
87

Honestly I hate headspace analysis and high split liquid injection is superior IMO but if the method requires GC HS, this is the method that should be used.

2 Likes
88

Perfect! Thank you, again.

1 Like
89

@GasLogix-Adam and I have had a few discussions the limits of detection and quantification of lab equipment. I’m all for standardization and providing good information to users. I’m hoping that my comments on developing a standardized analytical report for the site members and the industry are taken to be constructive.

Normalized to the primary peak or not, the GC-FID test for hydrocarbons identifies all hydrocarbons in the sample and calculates the amount of each in the sample based on the areas / projected areas of the peaks they generate in the chromatogram. Because the test equipment considers every peak it sees, the total %ages of the components will always sum to 100%.

I’m seeing that the group wants the test reports to include all components in the solvent (moisture and sulfur); seen by GC-FID and other testing methods. That’s doable and more accurate than the current standard. Acknowledging and quantifying the presence of unknown molecules though is a problem because it isn’t possible to qualify or quantify a component that is not detected. If the testing equipment shows the presence of an unknown material, the material must be identified, if only to rule out it being harmful. If the lab report ignores the peak, then the analysis is defective.

From a practical perspective though, bringing moisture at 5 ppm, sulfur at 2 ppm, and C6+ at < 1 ppm into the hydrocarbon analysis really has a negligible effect. Instead of the hydrocarbons detected totaling 1,000,000 out of 1,000,000 units: they will now total 1,000,000 units out of 1,000,008 units.

BTW, hydrocarbon tests are done with liquid phase injection.

2 Likes
90

This still doesn’t account for mystery oil and other non hydrocarbon, water or sulfur contaminants though does it?

91

100%! I for one am glad to have you as part of the process. I think it would be unwise to dismiss your experience in the industry.

5 Likes
92

If you know the volume or weight of the liquid injection you can run it against an actual calibration curve, then you do not need to identify an unknown or even see a peak for it.

You would simply calculate the percentage of the sample injected that is X based on the calibration curve, along quantifying any other calibrated peaks (Z)).

100-(X+Z) = percentage of unknowns

4 Likes
93

@Dukejohnson, testing for sulfur compounds is typically done by the manufacturer. Mystery oil(s) are GC-FID detectable, so long as they have carbons in them. We’ve run GC testing out 20 minutes and the last peak we’re

Testing for non-hydrocarbon contamination would initially require a full-spectrum analysis, searching for anything and everything that shouldn’t be in the sample. It would be expensive, but is doable by labs with mass spec analysis capability.

I operate a related company that sells rare gases, neon, krypton and xenon. Xenon is used in light bulbs and one of the larger manufacturers reported failures in their units. A full spectrum analysis showed halocarbon contamination. Investigation into the matter showed that one of the rare gas refining facilities was also purifying CF4 and C2F6 for the semiconductor fabrication industry. As a result of that finding, all rare gases are now analyzed for the presence of fluorine species. Once the problem is identified, the industry changed to ensure the customer got what they needed. It was impossible to fix anything until the problem was identified.

3 Likes
94

What we’re suggesting though is not bringing those tests into the hydrocarbon analysis but performing separate tests for hydrocarbons/sulfurs/moisture etc. and listing the results individually. For instance, if we bring sulfurs into the hydrocarbon analysis, you would never see them due to the massive hydrocarbon peaks.

This would just combine all of the individual analyses into one easy to read stotal analysis to include the composition and potential contaminants. Your point about the total units is exactly right, we would never want to potentially hide very low components in the analysis by trying to combine these.

While the analysis may be “defective” it is still informative. And as we discussed above, if we include this information in addition to the analyses we are currently performing, then are we generating a more full picture for the customer? We also have the option @Chaboes mentioned above in the case that we actually do run into an unknown.

Each analysis is run on its own particular detector and column so there should never be an unknown material. For the compositional analysis, the FID and capillary column will only see hydrocarbons so any unidentified peak will fall between two known peaks and can be referenced to determine the possible suspects. Our external calibration standard has a very wide range of peaks so we don’t observe unknowns even into the hexanes plus retention times. We opted for a natural lot for our hexanes plus therefore we were given a long list of the species present and are able to identify from it. Most calibration standards will only use normals so that is when you can really run into issues of unknown hydrocarbons because you will only have nC6, nC7, nC8, nC9, and nC10. Same thing applies to our sulfur analysis, the instrument only responds to elemental sulfur therefore any unknown peaks can be identified using reference chromatograms; though again the standard in place has a very large variety of sulfur molecules so unknowns do not happen.

At the end of the day, you and I both know that we would never release a product into the market that showed >0% “unknown.” We would resort to further testing to identify and then either remediate or refuse the product. So ultimately all of the COA’s that make it to a customer are going to read 0% unknowns but at least we’re making the distinction between percentage of sulfurs versus percentage of hydrocarbons versus percentage of moisture in the total composition of the gas.

A large portion of the oil and gas testing in the typical lab will be in gas phase. Our particular application doesn’t really fall into the general O&G testing but there are many discussions and posts from other users who mention headspace analysis so yes liquid injection should be a must but doesn’t always happen.

My point about sampling adequacy was more intended to get to a standardized SOP for pulling the sample since, as I’m sure you aware, inadequate sampling methods have a potential to quench results in their subsequent analyses.

2 Likes
95

Time for sulfurs!

The Standard Test Method for Sulfur Compounds in Light Petroleum Liquids by Gas Chromatography and Sulfur Selective Detection (ASTM D5623), which we commonly refer to as a speciated sulfur analysis, necessitates an additional detector equipped in the GC (flame photometric) and can, theoretically, resolve any sulfurs in the gas (the ones @GasGuy-QEG listed above are those of greatest concern).

Generally, if sulfurs are tested at all, it’s done via a total sulfur sulfur analysis (ASTM D6667). For our remediation methods, we need to see exactly which sulfurs and at which concentrations they may be present. We also enjoy having the ability to distinguish sulfurs to determine whether the truck was properly cleaned between loads as we require. (E.g. If we tested a load of high purity propane and found it to have hydrogen sulfide present, we would suspect it to be contamination from a prior load of y-grade product and be able to refuse delivery based on such analysis). However, a total sulfur analysis should be sufficient if remediation is not needed.

A sniff test may be sufficient to verify such an analysis but I think what we’re trying to get at here is the threshold level for what is acceptable for this application. Given that these sulfurs have varying, but all low, odor thresholds (list below) and considering that there is a potential for those odorous sulfurs to be concentrated in our customer’s final product, I would think that threshold should be extremely low, if not 0.000 ppmw. And I would certainly think we can safely assume that sulfur testing is warranted regardless of whether there’s a detectable odor in the gas.

Various sulfur odor thresholds (in ppmv):

Hydrogen sulfide, 0.00047
Dimethyl sulfide, 0.0010
Ethyl mercaptan, 0.0010
Sulfur dichloride, 0.0010
Benzyl sulfide, 0.0021
Methyl mercaptan, 0.0021
Hydrogen sulfide, 0.0047
Diphenyl sulfide, 0.0047
Carbon disulfide, 0.21
Sulfur dioxide, 0.47

Any input from extractors on what an acceptable level, if any, may be for sulfurs?

2 Likes
96

Hey @GasLogix-Adam, thanks for welcoming me to the discussion and the positive words about my history & experience in the specialty gas industry. That said, for Christ’s sake please limit your posts to < 15,000 words ! Do you have a side bet with SD for the longest post in 2021???

The essay you posted on this thread yesterday could be condensed to “an incomplete lab analysis is still informative”.

As for your subsequent post on sulfur, obscure large sulfur species will not be in the refined C3/C4s without one out more of the low order species being present. Companies making odorless aerosol propellants run the desulfurization process to the point where the gas is completely odorless. A) people won’t spray Pam on their cake pans if it stinks; and B) it’s not necessary to test H2S levels in flatulence to parts part billion to know if it the fart stinks or not.

You proposed acceptable total sulfur levels at <0.000 PPM. That would require testing to parts-per-100 trillion level with the pass/fail set at 1,000 PPT / 1 PPB. I’m not a skilled chemist, but ask that one of the skilled chemists on this site please comment on the viability and practicality of such testing?

I’m so looking forward to meeting @GasLogix-Adam in Vegas and thumb-wrestling about levels of qualification and quantification. :laughing:.

97

I counted for you

This post was 605 words including your quotes

This post was 371 words including your quotes

976 words total…. Put your big boy pants on, you got this.

4 Likes
98

Looks like my estimate ran a little on the high side. It sure seemed like more words.

I wonder what the record is… Regarding SD posts, is a picture worth a thousand words?

3 Likes
99

@GasGuy-QEG I’ve said more than once in my diatribes that I don’t have all the answers. My posts have been intended to open the conversation to the community in hopes of determining how to better serve them.

There are more posts in the various gas threads over the last few weeks than words in any of my posts. To me that signals that none of the gas industry is doing a good enough job at producing products that meet the needs/wants of this industry and/or that there has been a lack of communication and education between the supply side and consumer side.

We all get it, you know your shit, but this is a discussion aimed at ultimate betterment, not a wrestling match.

2 Likes
100

I’ll just use words I’ve already written so it doesn’t get overwhelming.

:man_shrugging:

1 Like