Troubleshooting issues

Over time we hope to add more and more sections here. Of course we have had plenty of troubleshooting to do in the lab but it is quite a bit of time to document and present them here. If there is something specific that you don't find here, feel free to ask us if we can help. We hope to eventually have a long list of helpful advice as we mull through problems.

Of course, the isogeochem archives are a great place to find almost anything related to troubleshooting your instrument.

Index:

I]  Mass Spec Problems
II] Peripheral Devices
A) EA

B) TC/EA

C) Heating block for exetainer vials
D) GC-C/TC
E] Dual Inlet

III]  Interfaces

     A) Conflo

     B) Gasbench

     C) GC-C/TC interface

IV]  Other IRMS-related devices


EA: Spikes on m/z 28 and 29

Index: i] Observation

II] What we know

III] Solution/Resolution

I] Observation:

We see ~1mV spikes on the m/z 28 and 29 traces in cups 1 and 2 respectively. These spikes have a period of 45-50 seconds. Instrument setup is a Costech Elemental Analyzer connected to a deltaplus advantage through a conflo 3 interface.

II] What we know:

History: the EA has newly packed reactor tubes and water trap. It was run without problems for about 100 samples after repacking it. It has been in standby mode for about 10 days prior to observing these spikes.

The spikes are not observed at m/z 44, 45 or 46.

The flow rate of the EA is 81.5mL/min and was observed to be stable (+/- 0.1mL/min) over several minutes.

The combustion reactor is at 1000°C, reduction is at 650°C, and the GC is at 55°C.

Adding helium dilution in the Conflo greatly reduces the intensity of the spikes. If we connect a gasbench to the deltaplus we have good backgrounds and no spikes. Consequently, we attribute the spikes to the EA.

If we place the EA in standby mode, the spikes disappear. If we then return the EA to work mode they reappear.

We thought that perhaps they were related to pulsing of the heating elements on the EA furnaces, but varying the temperature of either furnace did not influence the period or amplitude of the spikes.

Here is what they look like:

EAspikes.gif

Here is a closeup of a portion of the above screen shot:

EAspikes_closeup.gif

Here is a TCD trace from the EA with the Gain at high and the attenuation at 1. Nothing obvious here.

EAspikes_TCD_trace.gif

We tried varying the helium carrier gas pressure and here is what we see:

EAspikes_03.gif

Note that at 1.4 bar the period is about 50 seconds, at 1.1 bar it is 100 seconds and at 1.05 bar it is around 140 seconds.

In addition, the spike amplitude is about 1.5mV at 1.4 bar and about 2.5mV at 1.05 bar.

We tried disconnecting the leads to valves 2, 6, 7, and 11 (one at a time) to verify that there is no electronic switching occuring to produce these pulses. There was no effect on the pulses.

We also tried plumbing a similar flow rate of helium from the EA purge line (we don't use the purge normally) in the hopes for one last confirmation that the Conflo is not responsible for this problem. Here are the results:

EAspikes_bypassEA.gif

Definitely looks like the problem is from the EA. We also tried bypassing valve 11 (the standby mode valve) with a union and saw no effect. We had thought that since the TCD trace shows no sign of this pulse, the only valve downstream of it might be the source of the problem.

Now we are going to start bypassing sections of the EA plumbing to try to find the source of the pulses.

II] Resolution/solution:
That last trick we tried was to use a second helium tank to bypass sections of the EA. This is a standard troubleshooting technique to help determine the source of the problem. As we did this, we finally got to a point where we still saw spikes when helium flowed from a second helium tank directly into the water trap, GC column, TCD, and standby valve (valve 11). We had already done a separate test and found that the TCD and standby valve were not the source of the problem (we bypassed them and still saw the pulses). This meant that the problem lied either with the water trap or the GC. When moving to disconnect the bottom of the water trap to bypass it, we saw that the bottom fitting had worked itself loose and was clearly not tight. Simply tightening this fitting removed the spikes.

It is pretty amazing that the conditions set themselves up so well as to generate these spikes with such perfect periodicity! We tried loosening the fitting just to reproduce the spikes but to no avail.

Lessons: The steel tubing that exits the bottom adapter of the water trap builds up torque tension as the bottom adaptor is tightened (it rotates the steel tubing with it). In the future, we need to be more careful to relieve the tension from the steel tubing so that it does not exert an force that would push the bottom cap to rotate open.

Many thanks to Bruno at Costech for helpful discussions on the phone and to all who sent in suggestions from the isogeochem listserv.


TC/EA: Peak shape problems

Current problem with the TC-EA: peak broadening with subsequent analyses.

Context:

A little history of our TC/EA: we have had the instrument for a little less than one year and have performed very few analyses as of yet. It has probably seen at most 300 samples in this last year. The initial interest we have for the TC/EA is for the analysis of phosphates in the form of Ag3PO4. I had run some of these samples at 1450°C about one week ago with no modifications made to the system.

This week I had initially been checking to see if the TC/EA produces any CO2 and then I moved on to taking a look at N2 production in the TC/EA. N2 and CO overlapped significantly in our system but it did seem possible to separate them by changing the GC temperature and carrier gas flow rate. Our carrier gas pressure had originally been set to 30.5psi to obtain a 120mL/min flow. The modifications I made and subsequent results are shown below. On the last run (bottom-right plot below), there was pretty significant peak broadening. I left things alone after that and the next day I tried running a sequence of benzoic acid to check on the peakshape.

separate_N2_and_CO 1.JPG

The results for benzoic acid are shown below. Since seeing the peak broadening I tried baking out the GC column overnight at 250°C (it's max temp rating is 400°C) but that resulted in a 50 second increase in the retention time of CO. I still measure the same flow rate of 120mL/min from the TC/EA with the carrier at 30.5psi as I did before the shift in retention time. If I let the TC/EA sit for a day, then the first peak comes out reasonably narrow and subsequent peak comes out broader again.

BenzoicAcidPeakBroadening 2.JPG

TC/EA: cracking ceramic tubes

Context:

We had used our TC/EA for the analysis of solids for some time without any problems. The same ceramic tube and glassy carbon tube were used many times over. Eventually, we converted our system for the analysis of waters (added liquid injection port and autosampler as well as acquisition of a new glassy carbon tube and bottom feed adapter). Soon after that switch-over the ceramic tube began to show a slight bulge in the center. By the next use it burst and here is what it looked like:

Cracked Ceramic 1.JPG

Naturally, this first ceramic tube burst on a Friday night. I found out about the problem on the following Sunday when the furnace was at room temperature and there was no helium in the conflo.

Since that first blowout we have had a series of additional ceramic tubes cracking (without the bulging). Additionally, a crystallizing deposit forms on the surface of the glassy carbon tube right where the hole had formed in the ceramic tube. When this happens the glassy carbon tube gets jammed inside of the ceramic tube. We have had one glassy carbon tube break while trying to remove it from the ceramic tube. From now on, when this happens we will use a diamond saw to cut the ceramic tube off of the glassy carbon tube which costs about $1000. Here is a picture of our last ceramic tube which cracked after only three days of use. We had a brand new glassy carbon tube in it and now that tube has the same deposit on it.

Cracked Burned Ceramic.JPG
Burned glassy carbon tube.JPG

We believe that the heating element must have been touching the ceramic tube and that that is what left the mark on the ceramic tube and also caused it to crack (perhaps through the rapid heating/cooling that would occur when the element is touching the ceramic).


Dual Inlet troubleshooting

Drift in the reference gas isotope ratio over time. Note: the term "reference gas" used here refers to a point of reference in a sequence: we measure the difference in isotope ratio between the samples or standards and the reference gas. We normalize all results with respect to working standards that have been calibrated against international standards (provided by the IAEA or NIST).

We would eventually like to use the dual inlet bellows as a source of reference gas for continuous flow measurements requiring carbon monoxide or any other toxic or expensive gas. In order to do that we need to verify that the expected shift in the isotope ratio of gas bleeding out of the bellows is either too small to worry about in the time frame of our sequence of analyses (perhaps 6-12 hours), or is easy to characterize and correct. In that light, we measured the isotope ratio of carbon and oxygen in CO2 admitted to both dual inlet bellows with respect to a lab tank of CO2 used as the continuous flow reference gas over almost 48 hours. What we found was that the difference in isotope ratio between the gas originating from the bellows and that originating from a cylinder varied by up to about 0.7 � in a rather erratic way. As we monitor our laboratory environment quite closely we tried to see if the drifting value of the isotope ratio measurement correlated with laboratory temperature. Here is what we found:

Results for δ13C (Note that the reported isotope ratio is that of the gas in the bellows measured against our lab tank CO2. These results have not been normalized to the VSMOW scale.) Results from the left side bellows look the same. Note that the difference in isotope ratio between the two bellows is constant over this time frame (as it should be) and can be viewed at the bottom of this page.

Note: this should be considered "preliminary data". I'm undertaking a more systematic study of this, perhaps to be presented in a poster at the Canadian Continuous Flow Workshop in Victoria, BC this June. Of particular note is that the thermostat in our lab is located some distance from the instrument and CO2 tank so I want to verify just how big the temperature swing is at the instrument and CO2 tank.

Trubleshooting 1.GIF

Results for δ18O

Troubleshooting 2.GIF