VERY strange problem with GCMS columns

[BZK]

Hi to all, this is a very strange story.

I'm working with GCMS from several years and on the last two years I observe into the chromatogram four ghost peaks WITHOUT injection, only run : octamethyl cyclo tetra siloxane, decamethyl cyclo penta siloxane, dodecamethyl cyclo hexa siloxane and tetra deca methyl epta siloxane. My instrument is Shimadzu GC-QP2010 and the tested columns with identical results was ZB5MS, ZB5Semivolatiles (Zebron) DB5MS-UI (Agilent), all 60m , 0,25 mm I.D and 0.25 u as film thickness. All columns where conditionated as recommended producer protocol.
My first test to understand the problem was to verify possible contamination coming from GC and I have tried to bypass entirely the injection system by connecting the column with external pressure regulator with no positive result.
The second step was to change helium source (Helium BIP) with three other different gas and no result.
The third step was to replace the in-line triple filter and no result again.
Now, the peaks are quite high (with 1 kV of EM the peak hight could raise 800000).
The first peak is very broad, the second is quite better and the last two are well focalized.
The eluting temperature is around 100°C for the first peak and 200-210 for the last.
I can suppose that the contamination come from the column itself due to heat ramp but seems to me very strange considerating the importance of columns producers.
Any idea ?

Regards,

Robertino Barcarolo, Italy

[Peter Apps]

If you are seeing the contaminants as peaks (as opposed to a steadily rising baseline) it is very unlikely that they come from the column. A much more likely source is the inlet septum, and following that the deactivation of the inlet liner. Although you ran with and external pressure controller (which may itself be a source of quite serious contamination), the siloxanes from the inlet would still have been on the column.

Since you have already invested considerable effort in troubleshooting you might consider eliminating the column phase as a source of siloxanes by testing with a wax type phase.

Peter

[BZK]

If you are seeing the contaminants as peaks (as opposed to a steadily rising baseline) it is very unlikely that they come from the column. A much more likely source is the inlet septum, and following that the deactivation of the inlet liner. Although you ran with and external pressure controller (which may itself be a source of quite serious contamination), the siloxanes from the inlet would still have been on the column.

Since you have already invested considerable effort in troubleshooting you might consider eliminating the column phase as a source of siloxanes by testing with a wax type phase.

Peter

Hi Peter, yes, I have tested a DB-WAX for wine aroma analysis without siloxane peaks. Maybe that the deactivation (new?) used to produce ultra inert column from one hand higly reduce the 207 and 281 M/Z at high temperature, and this is true, but to the other leave not well crosslinked the low boiling siloxanes..maybe as hypotesis...
The strange is that other two very old ZB5 columns (2004) now out of operation doesn't shows on past highly siloxane peaks, only small but the bleeding of 207 was very high.

Another strange issue is : Why Phenomenex and Agilent test the MS column with FID as detector and NOT by fullscan from 33-500 M/Z ? I am sure of that because my columns was delivered with the GC-FID column test included.

Robertino

[Peter Apps]

If you are seeing the contaminants as peaks (as opposed to a steadily rising baseline) it is very unlikely that they come from the column. A much more likely source is the inlet septum, and following that the deactivation of the inlet liner. Although you ran with and external pressure controller (which may itself be a source of quite serious contamination), the siloxanes from the inlet would still have been on the column.

Since you have already invested considerable effort in troubleshooting you might consider eliminating the column phase as a source of siloxanes by testing with a wax type phase.

Peter

Hi Peter, yes, I have tested a DB-WAX for wine aroma analysis without siloxane peaks. Maybe that the deactivation (new?) used to produce ultra inert column from one hand higly reduce the 207 and 281 M/Z at high temperature, and this is true, but to the other leave not well crosslinked the low boiling siloxanes..maybe as hypotesis...
The strange is that other two very old ZB5 columns (2004) now out of operation doesn't shows on past highly siloxane peaks, only small but the bleeding of 207 was very high. If the siloxanes are indeed coming from the column you need to check the condition of the oxygen and water scrubbers on your carrier gas line

Another strange issue is : Why Phenomenex and Agilent test the MS column with FID as detector and NOT by fullscan from 33-500 M/Z ? I am sure of that because my columns was delivered with the GC-FID column test included. Very likely because changing a column on an MS involves venting and pumping down, which takes ages compared to just swapping inlet and detector fittings on an FID system. Also an MS costs a lot more than an FID. It is an interesting point though - siloxanes do not give much of an FID signal, so a column that looks fine for bleed on an FID might be quite poor with an MS

Robertino

[BZK]

[Hi Peter, yes, I have tested a DB-WAX for wine aroma analysis without siloxane peaks. Maybe that the deactivation (new?) used to produce ultra inert column from one hand higly reduce the 207 and 281 M/Z at high temperature, and this is true, but to the other leave not well crosslinked the low boiling siloxanes..maybe as hypotesis...
The strange is that other two very old ZB5 columns (2004) now out of operation doesn't shows on past highly siloxane peaks, only small but the bleeding of 207 was very high. If the siloxanes are indeed coming from the column you need to check the condition of the oxygen and water scrubbers on your carrier gas line Oxigen and water are very low due to high purity helium and in-line triple filter.
The counts of M/Z 32 are about 6000 at 1 kV of EM. My MS range start from 29 (aldehydes) to 400 and then I can see also oxygen.

Another strange issue is : Why Phenomenex and Agilent test the MS column with FID as detector and NOT by fullscan from 33-500 M/Z ? I am sure of that because my columns was delivered with the GC-FID column test included. Very likely because changing a column on an MS involves venting and pumping down, which takes ages compared to just swapping inlet and detector fittings on an FID system. Also an MS costs a lot more than an FID. It is an interesting point though - siloxanes do not give much of an FID signal, so a column that looks fine for bleed on an FID might be quite poor with an MS This is my suspect........

Robertino[/quote][/quote]

[jdezeeuw]

You are seeing here trapped siloxanes that are formed in the inlet
if you keep the column systematically longer at low temperature, you will see these peaks grow linear with the time you keep the column cold.

source is mainly the septum and residue.

make sure you use your septum flush and check it is really working (test flow); Then you may also have septum particles in the liner. This is the biggest issue as they will form the siloxanes as it's pretty hot there. You may check the whole injection port contribution by turning the injection temperature off. The peaks will probably disappear.

height of these peaks will depend on: injection port temperature and Split ratio. If you go splitless the peaks will increase a lot.

One other challenge that we sometimes see. The section of the column that is inside the injection port. Some phases start to degrade which also generate the siloxane. On older columns, this part may have been stripped off during use, so this effect will be minimal.

We, at Restek make also a lot of MS type columns that are all tested with FID. This is because we need to have calibrated bleed measurements and with MS this is very difficult.
Secondly, test costs will increase dramatically and customers will not accept that.

Note that we test every MS-type column for bleed as part of our standard QA procedure. You find several other companies out there, that will write a typical bleed value on the report, but it is only measured by a batch test.(matter of reducing production cost). Even the C'gram is sometimes not from that column, but just a copy of good looking C'gram. It says "representatitive " chromatogram.

jaap de zeeuw, Restek corporation

[BZK]

Hi Jaap, I know well all problem coming from septum. I don't use the liner because i inject directly on 0.53 mm id deactivated precolumn w/o stationary phase, 15 cm long. Anyway to bypass the injector trouble i have connected directly the column (without precolumn) by exiting from an unused injection port to an external manual pressure regulator. No change was observed. The injection side is then excluded in this case. My opinion is that there are problem on crosslinking of stationary phase. I have not tried your column, anyway.

[Peter Apps]

Hi Robertino

To get contaminant peaks (again I stress not a drift in the baseline) the source of the contaminants has to be localised with respect to the the column - in other words it must either be upstream of the column or be only a short section of the column. If the siloxanes arise from the stationary phase, without any extraneous cause such as moisture in the carrier gas, then they must be formed at all points along the column, and then they cannot appear as peaks.

What do you do to remove oxygen and moisture from your carrier gas ?

Peter

[BZK]

Hi Peter, I agree with your consideration about the peak shape and drift. I can be sure that i don't have water and oxygen because my helium is "BIP" type that means He 6.0 into cylinder containing molecolar sieve to achive 7.0 grade and triple moisture, oxygen and hydrocarbon filter by SGT. The tuning of MS shows very low level of 18, 28 and 32 M/Z.
My theory is that during ramp, even at temperature lower than 200°C some into stationary phase broken continously. When the oven decrease the temperature rapidly the less volatile siloxanes are cold trapped while the more volatile are trapped into long tract of column. At high temperature, on effect, I can see the 73, 207, 281 peaks continously.
The peak shape is five minutes long for the first (eluted at 100°C), one minutes for the second (eluted at 125°C) and focalyzed as peak with one minute tailing for the other two.

[jdezeeuw]

I posted that suggestion 30 min ago but it is not listed.. don't know why..

[Peter Apps]

Hi Peter, I agree with your consideration about the peak shape and drift. I can be sure that i don't have water and oxygen because my helium is "BIP" type that means He 6.0 into cylinder containing molecolar sieve to achive 7.0 grade and triple moisture, oxygen and hydrocarbon filter by SGT. The tuning of MS shows very low level of 18, 28 and 32 M/Z.
My theory is that during ramp, even at temperature lower than 200°C some into stationary phase broken continously. When the oven decrease the temperature rapidly the less volatile siloxanes are cold trapped while the more volatile are trapped into long tract of column. At high temperature, on effect, I can see the 73, 207, 281 peaks continously.
The peak shape is five minutes long for the first (eluted at 100°C), one minutes for the second (eluted at 125°C) and focalyzed as peak with one minute tailing for the other two.

Hi Robertino

There is stationary phase evenly distibuted through the whole column length and the temperature is the same everywhere during both the heating and cooling phases of the temperature programme. So any deterioration of the phase must be equal at all points along the column, and there cannot be any cold trapping from one part of the column to another. So I do not see how you can get the siloxanes to elute as peaks.

Peter

[JI2002]

Peter and jaap de zeeuw have given the best advice. When you saw the problem the first time two years ago, what was changed back then? Did the problem start all of a sudden or gradually?

[BZK]

Hi Robertino

There is stationary phase evenly distibuted through the whole column length and the temperature is the same everywhere during both the heating and cooling phases of the temperature programme. So any deterioration of the phase must be equal at all points along the column, and there cannot be any cold trapping from one part of the column to another. So I do not see how you can get the siloxanes to elute as peaks.

Peter

Obvious that the deterioration of stationary phase must be equal into entirely column lenght but when the oven cool some time is need to adsorb thermal energy from column and allow the siloxane condensation. Must be considerate that while air inside oven is rapidly decreased in temperature the inside column temperature can suffer on decreasing by irradiation from oven stainless steel container that remain hot for more time.
During this time the helium flows at 35 cm/s and then the first part of colum is replaced by pure carrier (and the contaminated push on front) while the last remain with contaminated helium for more time. At some time the siloxanes, according to BP, starting to condensate but, at this time, a lot of column is without contamination. This is not exactly a good focalisation but can explaine the peak shape anyway with long tailing.
Remember that the column is 60 m long.

[jdezeeuw]

The only way to get peaks formed in a GC column is when you have a high bleed column, and you cool down with highest speed. You get focusing of bleed products(siloxanes) on ONE SIDE OF THE COLUMN. This is the side where the oven temperature is a few degrees LOWER. When you cool columns, the oven will never cool homogeneously. One side will always be a few degrees warmer then the other side. This where the focusing happens BUT: it happens in every coil.

Ghost peaks derived from this phenomena, look like a a very noisy baseline that starts at 80 C and finishes around 240C. You see many concrete peaks (not 4) and basically the nr of peaks represent the nr of coils you have.

Historically this problem was seen with high bleed phases like 1701 and 624 types.

The fix was to cool with a negative temperature program, which allowed the oven to be at constant temperature when cooling and siloxanes were not focusing on one side.

This does not explain your 4 peaks, I really like to see a chromatogram. If you want, plse sent to: jaap.dezeeuw@restek.com chromatograms tell me a lot more..

[Peter Apps]

Hi Robertino

There is stationary phase evenly distibuted through the whole column length and the temperature is the same everywhere during both the heating and cooling phases of the temperature programme. So any deterioration of the phase must be equal at all points along the column, and there cannot be any cold trapping from one part of the column to another. So I do not see how you can get the siloxanes to elute as peaks.

Peter

Obvious that the deterioration of stationary phase must be equal into entirely column lenght but when the oven cool some time is need to adsorb thermal energy from column and allow the siloxane condensation. Must be considerate that while air inside oven is rapidly decreased in temperature the inside column temperature can suffer on decreasing by irradiation from oven stainless steel container that remain hot for more time.
During this time the helium flows at 35 cm/s and then the first part of colum is replaced by pure carrier (and the contaminated push on front) while the last remain with contaminated helium for more time. At some time the siloxanes, according to BP, starting to condensate but, at this time, a lot of column is without contamination. This is not exactly a good focalisation but can explaine the peak shape anyway with long tailing.
Remember that the column is 60 m long.

I think that I get it. This is a possible mechanism to generate a higher concentration of siloxanes towards the downstream end of the column, the leading edge of the "peak" then corresponds to the slowly rising vapour pressure of the band of siloxanes as the oven temperature increases - NB that there are contaminant siloxanes all the way to the end of the column so this is not a chromatographic process. The back edge of the "peak" is then the trailing edge of the zone of previously accumulated siloxanes, which does move chromatographically. If this is how it works then a slower temperature programme will give a slower rise on the leading edge of the "peak", you might even be able to generate a stepped front edge by using a step gradient in the temperature programme.

Peter