RI versus ELSD for Sugar Analysis

[James_Ball]

We have begun to have clients requesting analysis of sugars and polysaccharides plus an AOAC method for Total Dietary Fiber which requires a refractive index detector.

I understand RI can be kinda finicky, so was wondering if anyone has experience with both RI and Evaporative Light Scattering Detectors? I have found sugar applications for both, but would one offer more flexibility over the other, or better ease of use?

These are both new to me so looking for advice before we move forward.

Thanks,

[leadazide]

RI's are a lot cheaper than ELSD's..

With RI you cannot/should not run gradients so you are fairly limited in your chromatography.. but can chose a cheap Iso-pump.. RI's are not as such finiky but very very sensitive to disturbances.. like temperature or air flow will cause the baseline to drift.. heck I would say it runs for the hills! But if you have a stable and controlled enviroment the RI can be a fantastic detector.

ELSD will give you the possibility to run more versitile chromatography but is more expensive.. Also as you are vaporizing the mobile phase you need to consider buffers and additives.. Only volatile additives and I usually also recommed uses going for the more expensive MS grade solvents in general..

So yes.. Both can be used... but differently..

Ease of use go for RI

Flexibility go for ELSD

Ultimate setup = Both detectors and a valve to switch between each detector depending on method you want to run.. Together with either a Quad pump or Bin with solbent selector option you have a very flexible system! Here you might even be able to get the RI for "free" as the discount...

Hope this helps.

[Rndirk]

What kind of samples/matrix are you talking about James?

We analyze so many different kinds of food for sugars each day that HPLC-RI is simply not an option: too much risk of interferences. It's still a century-old GC-FID procedure here, but it runs well.

[James_Ball]

RI's are a lot cheaper than ELSD's..

With RI you cannot/should not run gradients so you are fairly limited in your chromatography.. but can chose a cheap Iso-pump.. RI's are not as such finiky but very very sensitive to disturbances.. like temperature or air flow will cause the baseline to drift.. heck I would say it runs for the hills! But if you have a stable and controlled enviroment the RI can be a fantastic detector.

ELSD will give you the possibility to run more versitile chromatography but is more expensive.. Also as you are vaporizing the mobile phase you need to consider buffers and additives.. Only volatile additives and I usually also recommed uses going for the more expensive MS grade solvents in general..

So yes.. Both can be used... but differently..

Ease of use go for RI

Flexibility go for ELSD

Ultimate setup = Both detectors and a valve to switch between each detector depending on method you want to run.. Together with either a Quad pump or Bin with solbent selector option you have a very flexible system! Here you might even be able to get the RI for "free" as the discount...

Hope this helps.

That alone will probably make it unusable for us. We can easily see a 15F temperature swing in an hour or so when the HVAC acts up, which it does often now. We just got a quote for a complete overhaul of the HVAC, $1.5M, for a building that is only worth about twice that

[James_Ball]

What kind of samples/matrix are you talking about James?

We analyze so many different kinds of food for sugars each day that HPLC-RI is simply not an option: too much risk of interferences. It's still a century-old GC-FID procedure here, but it runs well.

Right now it would be mostly pet food analysis for the nutritional labeling. We have been doing the microbiological analysis for several companies and they are asking about us becoming full service with the chemistries. The sugar analysis is one of the few things we aren't currently equipped for. I tried it in the past on the LCMSMS, but I have too many methods now for the one instrument we have to try it on there again.

I also have a client wanting polysaccharide analysis in Tumeric products along with terpinols, which I believe we would need these for also.

Thanks for all the info, it is still something I am researching and is good to know the pros and cons going either way.

[Andy F]

We were trying to analyse sugars in alcoholic beverages, beer mainly, and found the ELSD was universal, but not selective, so your chromatography had to be nailed down solid or there were too many interferences. As an alternative have you thought about pulsed amperometric with IC separation? This is the way we used to do the sugars and it seemed to work pretty well, though we only went up to maltotriose.

[Rndirk]

Like the poster above me said, you can't run this analysis in routine on universal detectors unless you analyze the same, well-known matrices every time.

I have also been looking into running it on LCMSMS (and posted a topic about it) but it's not a priority for us right now.

[James_Ball]

Like the poster above me said, you can't run this analysis in routine on universal detectors unless you analyze the same, well-known matrices every time.

I have also been looking into running it on LCMSMS (and posted a topic about it) but it's not a priority for us right now.

I was following that thread before, still might try it on LCMSMS and see what I can do there. It would definitely complement either RI or ELSD if the matrix is very complicated.

[Rndirk]

I would consider GC-FID or GC-MS. It's an old, dirty sample prep but the separation and quantification works well. There might be easier procedures out there for GC now but i haven't looked into it - other priorities.

I lost my appetite to developed sugars on LCMSMS after I struggled with matrix effects and a bunch of other weird stuff for artificial sweeteners (steviols, aspartame,..). But i do have to mention that our LCMSMS is getting really old.

[DR]

Cheaper than MSD, more expensive than ELSD and RI - Corona CAD w/ a countercurrent pump if you want quantitative gradient possibilities.

Personally, I HATE RI - always seems to need half a day to equilibrate.

[James_Ball]

Cheaper than MSD, more expensive than ELSD and RI - Corona CAD w/ a countercurrent pump if you want quantitative gradient possibilities.

Personally, I HATE RI - always seems to need half a day to equilibrate.

I am beginning to think that here for us to run RI we would need to put it in a trailer outside with a dedicated A/C to control the temperature.

[DR]

I am beginning to think that here for us to run RI we would need to put it in a trailer outside with a dedicated A/C to control the temperature.

Only if said trailer is put a hundred meters underground.

[leadazide]

The love and joy of qualifying a RI where one of the parameters is drift... The amount of diviations I have written due to temperature fluctuations or drafts..

"Best one" was the customer who wanted RI inside a fume hood..

[James_Ball]

I remember when I was in university back in late 80s we had an HPLC with the ELCD or just Conductivity detector not sure which. That thing would drift with a temperature change of about 0.2C in the room. The operator would enter the room and lock the door, and no one was allowed in until the run was finished. Just opening the door would cause a spike in the signal.

I am beginning to think RI is about the same.

[Jake]

I'm certainly not a big fan of RI, but we spent years making the typical HILIC/aminopropyl/ELSD method work for our sugar feed stocks. We tried nearly every vendors' approach to improved stability, but they all eventually failed after 100-200 samples. Rather than going that route, we ending up just buying the cheapest column we could find and treating them as disposable. We always had to run a full set of standards flanking the samples with every set. Frequently the quality of data for the replicate standards was so poor the entire set needed to rejected and re-run. Really frustrating for all involved. Everyone dreaded having to do the sugar assay. After MANY years of this, we switched to the typical ligand-exchange-RID method. The RA who first set it up was disappointed her coefficient of determination for her seven-point standard curve was only 0.9999. She was able to use the standard curve for many months, and only a single-point calibration was needed. It took us awhile to dial-in the best cation to use for the ion-exchange column due to the high salt content of some of our feed stocks. But once we got that worked out, it was smooth sailing.

A similar situation for us for generating usable, reliable, TAG profiles. Made LC/MS work for awhile, ELSD was completely useless, other than for quantitating a single or few TAGs, and we've settled on NARP/RID. It's dreadful: two 4.6 X 250 mm columns, 2.5 hour run time per sample, must start early in the day so as the method isn't running when the lab gets cold at night, column needs to be rinsed after two to three injections since it's isocratic, 50 mg/mL (!!!) injection concentration, etc. But it always works, and gives good, reliable data.

There's a reason RI is still around. While not the most cutting-edge tool in anyone's lab, sometimes it's still the best way to do something.