The future of liquid chromatography

[Kostas Petritis]

Admin note: this was split from the thread "Ultra Performance Liquid Chromatography from Waters":
http://www.sepsci.com/chromforum/viewtopic.php?t=972


Nalizer,

The UPLC from Waters is just a LC system that can go to higher LC pressures. The advantage is that with higher pressures you can go down to stationary phase particle size and either increase your peak capacity and/or your analysis time (I can give you some theory on this if you want).

As a result you will have an advantage only if you opearate with smaller particle sizes such as the 1.7 uicro from Waters. This stationary phase has a very narrow particle size distribution so the backpressure is pretty high.

I have heard contradictory versions of one thing and this is if you will be able to operate that system with any type of other than Waters stationary phases (maybe Uwe can comment on this). Now, other companies have started manufacturing sub 2 um particles but the size is not as uniform so you can run them with conventional LC instrumentation. The Agilent 1.8 um particles is an example. Agilent claim increased efficiencies without the drawback of higher backpressures.

Part of our departement have already make the move towards higher pressure even before Waters commercialize their systems (we are currently working at 10000 psi, but we are mainly working in the area of proteomics where high peak capacity is a must.

Now, in your question if industry is moving towards this technology is too early to say but I will give my personal view in the subject and people are wellcome to agree or disagree...

I think that the rules changed about 5 years ago when electrospray mass spectrometry was accepted by the scientific community as the detector of choice for most analytical applications. Several years from now, almost all the analysis will be performed with a mass spectrometer and the rest of the detectors will be there only to supplement for specific applications. As a result what will be the future of the rest of instrumentation (including the separation science as we know it) will depend on what is more compatible with the MS and the needs of biology. The transition of analytical LC to micro and nano LC is an example of this type. The MS is much more compatible with nano flows and biologists wants to analyse as little amount of sample as possible (the future is down to single cell).

The scientific community asks for as much sensitivity as they can get, they want as large dynamic range as they can get and they want to find ways to get rid of the ion-suppression phenomena that occurs during analysis of complex matrices with MS. Dynamic range depends on somehow to your sensitivity and sensitivity is increased the lower your flow rate while due to the higher electrospray efficiency at these flow rates, ion suppression is limited if not eliminated.

Now, while ultra high pressure liquid chromatography was investigated in the first steps of HPLC it was quickly abandoned as highly challenging and believe me there are several challenges associated with it as you need to redesign everything (valves, fittings etc...). Another problem was that for conventional chromatographic columns (i.e. 4.6 um ID and up) are heating up due to water molecule friction when you exceed the 15000-20000 psi. Heating is not a problem for capillary columns.

So why higher pressures? If you are convinced that capillary columns and nano-flow systems will be the (solely) separation front of the MS then there is a bottleneck that has to be resolved:
a) Packing of capillary columns less than 50 um (and especially less than 30 um) is quite challenging even with 3 um -we were able to go down to 15 um with these particles... I assume that the packing of ultra thin capillaries will be easier with smaller and more uniform particle size. Going down to such small ID will serve the purposes of biology (very small injected amounts and high sensitivity) and the purposes of pharmaceutical industry (elimination of ion-suppression) for accurate quantitative analysis and higher throughput.
b) Chromatographic history shows us that there is a continues trend of going in lower and lower particle sizes and I do not think that this trend is going to stop but after a point it would be necessary to have higher pressure operating LC systems... It would be of course more and more difficult to make small particle size porous silica particles but maybe if you can go down to 0.1-0.5 um particles you do not need them to be porous (maybe someone can comment on this).

Finally... I think that another technology that will come in to the play and that is ion-mobility that will be an indermediate gas separation step between the LC and the mass spectrometry. Due to the peak capacity offered (or that will be offered) by ion-mobility you will need very fast, high efficient separations in the front end (LC).

Anyway, this is my point of view. people are welcome to disagree (and I hope that they will... this is a discussion forum after all). I think that 5 years from now we will have a better idea if the above will be at all true (who knows... maybe the "lab on a chip"?).

[tom jupille]

Kostas, I definitely agree with your last statement (it's gonna be an interesting next few years).

On some specific issues, though, we can have a debate :

• - I think you inadvertently got wider and narrower size distributions reversed. For the same nominal particle size, the narrower the particle size distribution, the lower the back pressure (it's the smallest particles that that effectively determine the flow resistance).
  
  - I'm not sure that particle sizes will ever drop much below the 1-2 micron range. For a 1-micron particle diameter, the inter-particle distances are down around 0.25 micron. That's getting into the same range as the pore sizes on conventional packings.
  
  - The other technology to keep in sight is monolith colums, especially in capillary format. Striking the right balance between macro pores and meso pores might allow the equivalent of 1-2 micron particle packed columns with a much lower back pressure. My personal guess is that the real battle will be between capillary monoliths and ultra-high pressure packed columns.

Probably a more contentious topic is the detector debate. I think when you're working on the cutting edge, it's easy to lose sight of the fact that the majority of HPLC separations are run as part of manufacturing/QC rather than R&D. When you have plenty of sample and a reasonably predictable matrix, it's hard to beat the advantages of, say, a PDA compared to MS: no moving parts and a 5x lower initial cost.

I'm sure others will chime in here!

[Yury Zelechonok]

Let’s do the math.
250 mm column with 5 um particles operates usually under 3000 psi leaving about 50% of additional pressure available as a comfort zone in case the column is partially plugged or additional devices attached to the system (capillaries, valves, guard columns, backpressure regulator etc.). From what I learn about UPLC max pressure available in this system is 15,000 psi. To have about 30% comfort zone (not 50%) the max pressure should be about 10,000 psi. With particle size reduced to 1.8 um back pressure will increase at the same column length and flow velocity (5 /1.8 )^2 =7.5 times. Then simple proportion 10000*250/3000/7.5 = 110 mm gives the max length of the column that can be used in this system. So, practical increase in plate count per column (not per meter) is 5*110/1.8/250= 1.2 times. 110 mm is strange number for the column length, the closest one which in common use is 100 mm, and accordingly the gain will be 1.1 times???
For me it can not justify double the price and triple in complexity for the UPLC instrumentation.
For speed or throughput of analysis. Having no pressure problem the monolith columns are advantages. They allow high velocity without lost of efficiency and they don’t require untraditional instrumentation.
Something wrong with my calculations?

[tom jupille]

Yuri, the situation is a bit more complex than that, because run time enters into the evaluation, and the smaller particles have much flatter Van Deemter curves (at least for small molecules!). Instead of thinking of things as slightly better resolution, you can look at it as much shorter run time.

I fired up my copy of the DryLab program and ran one of the standard examples under different scenarios.

The first one is a 150 x 4.6 mm column run at a fairly low flow rate (1 mL/min) in order to get the best resolution. The calculated pressure drop is only about 500 psi, the critical-pair resolution is about 1.3, and the run time is just over 25 minutes (the values are all listed in the "status" box):




The second is the same column, but run at 5 mL/min (about 2700 psi pressure). This drops the run time down to around 5 minutes, but also drops resolution down to just over 1:





The third is a hypothetical uplc column 100 x 4.6 mm packed with 1.8-micron material. Run at 3.5 mL/min, the pressure is just under 10,000 psi, the run time is around 5 minutes, but the resolution is 1.6:




The pressure calculations are optimistic because they assume monodisperse packings (i.e., no fines), but that applies across the board, so the comparison should still be valid.

I agree that the monolithic columns are a strong contender, but I suspect that the big advantages of the monoliths will show up primarily in the longer capillary format, which *does* require "nontraditional" instrumentation (as does any low-volume column).

[Kostas Petritis]

Tom,

I would maybe bet a pizza as well (should we do double or nothing? ...story from another post...) that Agilent manages to have lower backpressures due to wider size distributions. I do not really remember how they were doing it, maybe by introducing an amount of higher particle size particles but the mean was still 1.8. Maybe someone else can clarify this.

I will agree that particle sizes will be difficult to drop in less than 1 micron range and still be porous. These would be quite fragile and won't propably sustain such high pressures. But if you have 0.1 or less micron particles, do you really still need them to be porous (I do not know the answer on this...)?

I am keeping a close look to the monolith technology which I agree it looks promishing. The two problems I see, is that it would be more challenging to make reproducible columns and it would be much more difficult to have well endcapped materials due to the inherited limitations of the procedure (monoliths vs. particles). As a result of the latter, it might be problematic to use them for the analysis of some basic compounds and they might be more sensitive to pH extremes. Finally, if you need to increase your flow rate 10 fold with the monolithic columns in order to match the performances of packed capillaries that conficts with the necessity of MS for lower flow rates...

About the detectors, with the new generation of MS, the industry will start to get rid of the old ones which either will sell for nothing or donate them to academic institutions. My point is that prices will go down and several labs would prefer to buy a cheap or used MS than a PDA. Maybe if any of the manufacturers can share with us what was their sales in PDA's/UV's the last 20 years (year after year) maybe we can see a trend already. Finally for your case of predictable matrix, it will depend on how many analysis you have to make of the same thing. If you tell me 10000 or more I can tell you that throughput will be a concern etc...

Finally, in the case of splittless nano-flow pumps coupled with MS the quantity of wastes (mobile phase) drops near to zero (only what is left in the bottles when you change your application...). Mobile phase wastes is a real headache for the industry and contributes to the indirect price of your instrumentation and all the paperwork you need to do etc...

Thank you for replying to Yury! Your examples were excellent...

PS1: I would add that we need to ultimately go up to 40000 psi to take full advantage of the ultra high pressure LC...
PS2: Tom, could you please make a note at the top that you hijacked the discussion from another post? I do not want people to believe that I started such discussion

[Yury Zelechonok]

To Tom
if throughput is only gain in this new technology than it is not enough to overcome all negatives:

• High price
• Very high pressure (special connector, I assume)
• Relatively low reliability vs. HPLC which by itself is not very reliable technology (to help solve HPLC problems we have this forum)
• Only one supplier of columns
• Poor reputation of the manufacturer (Waters lost during last 10 years their once primary position in HPLC instrumentation to Agilent)
• New technology always has some hidden problems that only time will resolve.

Real throughput is not a number of runs per period of time, but number of run per time when instrument properly works. If down time is significant then throughput is low.
For the money UPLC will cost to a user I will rather have 2 Agilent 1100 or 3 other brand instruments with the same combined throughput and additional flexibility to run more than one analysis at once and high reliability as a proved by market fact.
By the way I am not associated with Agilent by any means.

[tom jupille]

Kostas:

I added the disclaimer to your original post (the reply to the Waters thread) so it will be clear than I'm to blame for hijacking the thread. I already apologized to Nalizer in the initial thread, so I should be in the clear.

Re MS vs UV:
The last numbers I saw were a couple of years ago (circa 2002) from a major manufacturer who claimed that MS represented about 30% of detector sales volume (in dollars). If you factor in a 10:1 price ratio, that drops down to less than 5% of unit sales. You get the same ballpark figures from the number of posts on the Forum (GC + LC is about 7 times "Hyphenated Techniques"), and we (LC Resources) get about the same from the relative enrollments in our LC versus LC-MS courses. I agree that MS is growing, but I'm not convinced that it will overtake UV in unit terms any time soon. In that sense, Yury's comments about throughput, downtime, and reliability are right on the money.

Re sub-micron particles: to a first approximation, the inter-particle distance will be on the order of 1/4 the particle diameter. Once you get below 1 micron, this is in the same ballpark as the pore diameter on conventional HPLC packings. That means that the particles themselves will certainly not be porous.

Re particle size distribution and pressure. Let me see if I can goad someone from Agilent into weighing in here . Worst-case, I'll be out the price of two pizzas.

Yury:

I hear what you're saying, but then I went back and dusted off my old copy of Giddings' 1965 book Dynamics of Chromatography, which was prescient in a lot of resepects. His last chapter basically says that the resolution that can be achieved in a given run time is limited by the attainable pressure. Chromatography (especially HPLC) has essentially spent the past 40 years in advancing the technology to catch up with Giddings' predictions (I'm old enough to remember when successfully packing 10-micron material represented a breakthrough).

At the moment, both the ultra-high pressure systems and the monolith columns represent proprietary formats. Both are new, and both are subject to "teething pains". I'm not sure which (if either!) will win in the end. If I could accurately predict the future, I'd be rich. I'm not rich, so my predictive abilities are obviously questionable.

So what will LC look like in 5 years? Kostas, I hear you saying "MS detectors with very small columns optimized for speed rather than brute force resolution". Yury, I hear you saying "Essentially the same technology as today, but less expensive and more reliable". Or am I misinterpreting?

[Victor]

Tom or anyone else,

I was never very strong on the mechanics of column packings. If you have a column with 1um particles with 100Angstrom pores (0.01um) and the flow channels between the particles are 0.25um, what is the problem here? Why are you suggesting that the particles are non-porous? Do you mean that if the particle size is decreased much below 1um, then having 100Angstrom pores would be mechanically unstable to the pressures under which such a column would have to operate?

Sorry to divert from the main thread of the debate.

[HW Mueller]

One could see more than 5 years ago that MS would have to play a role in biochemistry (and related biofields) akin to its role in organic chemistry. Thus I had gotten instrumental in attempting to place a FTMS into an analytical (mostly) service center at the University at Giessen. It failed, so now it´s partially of "academic" interest to me (also am getting to old for this), but I am sure that especially high resolution MS will be (should be) on the increase, at least for smaller proteins, aptamers (polynucleotides or peptides) etc.
Though I check into analytics regarding proteomics a bit (since I work with antibodies…) I am somewhat surprised that high pressure HPLC should be playing a large role in protein analysis. Now, high pressure and proteins is an interesting combination, already because of life in deep oceans. Also, some analyses don´t have to be concerned with denaturation, dissociation, and whatever. However, there is a need already and will be in the future to analyse proteins near their natural state (for instance in my case, an antibody-Tc complex stability in SEC). It seems then that low pressure high performance LC should be pursued, full blast. That should include monolith-affinity chromatography as well. My guess here is that affinity chromatography with weak ligands of low specificity but with relatively large stationary surfaces would have wider use than present practice (high affinity, high specificity).
Single cell analyses will probably become more important, to do this one has to mimick what cells do, at least partially. Present knowledge would indicate more nano, then. On the other hand, the interplay of cells must remain at least as important as it is now. In short, diversity is a good bet for the future.

A side remark: Some 30 years ago at the U of Colorado a bunch of us students were awestruck when RE Sievers (at the time better known for his work and book on nmr shift reagents) presented, in a seminar, a ~10cm column which contained an in situ polymericed porous organic material, with which he could separate xylenes. I just wonder why it took almost 30 years for the "present" restart? of this technology.

[Uwe Neue]

I highly appreciate that Nalizer brought up this topic, and I’ll be glad to discuss the elements of column design and UPLC. By the way, I learned every little thing that I will be talking about back in 1976, when a colleague gave me a paper by Georges Guiochon on "the Pertinency of Pressure" in HPLC. The most important conclusion of this old triplet of Guiochon papers is the fact that the best ratio of column performance to pressure is achieved, when the column is operated at the minimum of the plate-height vs. velocity curve. Once you accept that this conclusion is correct, column design and column choice become a nearly trivial thought process.
There are two key column characteristics that smaller particles address: column performance and speed of analysis. There is also one major detriment to the use of smaller particles: pressure. Let us just initially keep the column length constant and decrease the particle size. If I do that the column backpressure increases with the square of the particle size at constant flow rate. But since I decrease the particle size, the optimum linear velocity increases also. If I am always operating at the optimum linear velocity (=flow rate), the backpressure at constant column length increases with the third power of the reduction in particle size. If I go from a standard particle size of 5 micron to a particle size of 1.7 micron at the optimal plate count, the pressure increases by a factor of 25 at this fixed column length. That is a lot of pressure for a roughly 3-fold reduction in analysis time. It also increases the plate count by a factor of three, if the details of the instrument design permit you to take advantage of this.

The smarter way to think about a column is to reduce the particle size and the column length using a constant ratio. But now, the speed of the separation for equal performance increases drastically. You can get the absolutely identical plate count from a shorter column at a higher flow rate, which means that the analysis time decreases by a factor of four for a 2-fold change in particle size. The pressure still increases a lot, but not quite as terribly as in the case of constant column length. Here is a concrete example: The maximum plate counts of a 5 cm 1.7 micron column and a 15 cm 5 micron column are the same, but the speed of the analysis increases with the square of the particle size, i.e. roughly 9-fold. In this case, the pressure also increases with the square of the particle size, i.e. 9-fold.

The consequence of all this is that one needs higher pressures, if one wants to get to a higher performance in a shorter time. There is no way around this.

On the issue of particle size distribution: the particle size distribution of the ACQUITY BEH packing is as narrow (or as broad) as that of a larger size packing: +/- 15% around the mean. This is not easy to do at this size. Many manufacturers can’t even get there with the larger (i,e, HPLC) particle sizes.

To Tom: there is still a giant difference between the size of the space between the particles and the pore size. Assuming your estimate of 250 nm is correct, and with the pore size of a standard particle of only 10 nm, the ratio between both is 25. This is still a large factor.

To Yuri: your calculation completely misses the boat on the issue of the combination of performance and speed. The real issue is that now you can get to the maximum column performance in the time frame that is tolerable to the user, instead of having to wait forever. Go read the Guiochon papers!

To Kostas: the real trick with UPLC is not just the column, but the instrument is designed around the column performance to match the capabilities of the column. There are a lot of details in there, besides the higher pressure capability, such as higher detector speed, lower extra-column bandspreading, etc. I also agree with you that one will ultimately need to go to still higher pressures, as Jim Jorgenson has shown already.

[Uwe Neue]

Yuri, you are shooting from the hip without finding out what the story is. Nobody is hindering you or your company to make columns and packings that are compatible with UPLC. Since you think that Waters has such a bad reputation, there should be wonderful opportunity for your company to enter a new market.

[AA]

Misinformation stop gap.

Quote from a previous thread

"For me it can not justify double the price and triple in complexity for the UPLC instrumentation."

There is a premium on the price of the UPLC system of 15-20% over a similar HPLC system . NOT DOUBLE

Operationally, the system is as simple (or as complex) to use as any traditional HPLC system. Otherwise, no one would be able to use it.

[Nalizer]

I guess I opened a can of worms here without the intention. I pass all messages to my boss and he is now reading it daily and even reprints it for his files. I already see one outcome of this discussion - my boss wants to read everything on UPLC. I just do not want to see a heated arguments here (and it might go the "hard" way).

TIA,

Mark

[Yury Zelechonok]

To Tom
In agreement with Giddings' prediction, to get real advantage of 1.8 um particles the pressure limit for UPLC should be in order 6000*5^2/1.8^2=46,000 psi. Exactly what Kostas brought up. Waters stops short on this big time. It is like to buy an expensive sport car with 200 miles optimum speed, but drive only 55 because a speed limit. A nice toy, but mostly useless I think, plus difficult to get in and out.

To Uwe
I thing you “miss the boatâ€

[Kostas Petritis]

Tom,

Re MS vs. UV: It is clear to me that the dynamics will change if they have not already changed in 2004. Of course it is much more difficult to compare in sales per unit due to the cost difference. You have maybe noticed that in HPLC-2004 this year a company was advertising an ESI-ToF instrument for less than 100K$. As for the number of posts... this is maybe the best forum in chromatography but there are several forums more specialised in MS or more biology oriented (ABRF, ASMS, NIH etc.) that you have much more discussions about MS. The next best chromatography forum (yahoo) posts per month what here is posted per day.

HWM,

I think that UPLC will find applications in peptides (bottom-up approach in proteomics) and small molecule analysis. Although analysis of intact proteins is very interesting there are still some challenges associated with their detection and identification with accurate mass measurements (especially for those higher than 40K Da).

To Uwe,

I didn't want to offend your UPLC system. Having worked already in very high pressure LC I know all the challenges associated and take for granted that if you decided to commercialise it you have taken care of these. My point was that if you want to take advantage of very high pressures you have to introduce the necessary stationary phase to go with. In our case, we had to pack 80 cm columns with 3 micron particles to achieve the necessary peak capacity for our proteomic applications.

To Nalizer,

If your boss wants more background in very high pressure LC there are several references out there already. Jorgenson has published some nice papers, our group have published several related to proteomics and peak capacity (search for R.D. Smith and Y. Shen). A good place to start is a recent paper published in LC GC North America (I think Jorgenson is the author). I can find the reference if you are interested...

To Yury,

I do not think that your argument about selectivity is very valid. Saying that you can achieve the same thing by changing the gradient, and/or the organic solvents and/or the column... someone could argue the same for the transition from 10 to 5 micron several years ago... Also, advances in science are almost always made step by step and rarely with jumps. It is up to people to decide if they want to pay the premium of 15-20% higher price (according to AA provided information) for having the amount of advantages discussed earlier.