Chromatography Forum

This truly is not a post to bash Agilent, but I would like to know what other people's opinions are.

We have two Agilent 6890 GCs, with 5973 MSD and 5975 MSD. I have been fighting the politics at my company to get another system, a Varian ion trap MS. My company wanted me to get some comparison data, so I used a Varian 3800 GC with 2000 MS at a local university. I injected dimethylmethyl phosphonate at concentrations of 5ppb, 10ppb, 100ppb, 500ppb, 1ppm and 10ppm using a DB5 column. The Varian system detected the analyte at all concentration levels in full scan mode, and then when I extracted the TICs for the 79 m/z, my signal to noise values were well over a hundred for all concentrations. I took out the column, installed it into our Agilent 5975 MSD, and then injected the same standards into the system. All temperatures were identical. I could only detect the analyte at 10ppm with the Agilent system, and the S/N of the 79 m/z at 10ppm was less than the Varian system at 5ppb! To confirm that the Agilent system was working properly, I injected the demonstration OFN standard at 1ppb, and did a signal calculation on the 272 m/z and it was over 100. So it meets the specifications.

This isn't the first time I've experienced this. I used to work at an animal toxicology laboratory and frequently test for the toxin cantharidin. Urine concentrations greater than 5-10ppb generally resulted in death. I had no problems with this detection using my Varian MS. However, another animal laboratory that had an Agilent 5973 MSD had to send their samples to us for testing because their instrument was not sensitive enough for the analysis, they could only detect concentrations at 1ppm and above.

I am finally starting to convince my company of the limitations associated with the Agilent MSDs. Just because the demonstration standard works at 1ppb does not mean that all analytes will be detected at this low concentration. We have found for most analytes that you can expect detection only in the higher picogram range, low nanogram.

Does anyone have any comment to make about this?

-Aaron

We had a rep come in to demonstrate his varian ion trap. Using the same conditions and the reps "optimized" conditions, he could not achieve the same levels of detection for a number of volatile compounds we analyze for with our 5973's. Who knows? Maybe there was something we were missing or maybe it was just the quirkiness of that particular instrument. We have lots of 5973's, 6890's, purge and traps, and autosamplers and they all seem to have their own individual behaviors.

Combining a chromatograph and a mass spectrometer tends to produce a beast of very individual characteristics. I suspect much of the angst about brands arises from the original design philosophy of these instruments, and the manufacturers' target markets.

However, for most end users, the actual user interface, consistency, reliability, match with existing toys, and service support will sway the decision, provided all candidate instruments meet their actual and projected needs, and prices are reasonable. If the toy is critical to your business, sensitivity in excess of your needs will probably not be the main decision factor.

Here in NZ, an appliance retailer made his name by the expression,
" It's the putting right that counts ". HP and Agilent toys tend to have reasonable performance and reliability, and very good service support here. Hence they quickly dominated the GC market, so when later looking for a GC-MS, the front part was already familiar.

The HP 5890 was the price-performance GC leader for nearly a decade, and their MS detectors have never been at the bleeding edge in sensitivity. There have always been instruments were more sensitive, some even easier to use, but not many were as reliable. Waters was the same in HPLC with the 6000 solvent delivery system, PE in AAS, etc. Once they capture the market, their support and knowledge bases become larger.

Buying toys is always a trade off, but hyphenated toys tend to be like F1 cars, expensive and tempermental, and good support is critical to consistent success. Some people only want to reach their destination reliably, so different toys exist, and the higher the performance, the more likely there will be individual instrument variations, and a wider range of sensitivities.

It's always nice to read other analyst's views/experiences, but decisions have to be tempered by your own environment.

Bruce Hamilton

I cannot comment regarding the two specific systems you mention however I would recommend all end users to think hard before choosing equipment solely on previous experience and knowledge of the data handling. If you are interested in looking at MDL type levels and/or difficult matricies then having access to both Ion trap and quadrupole technology is beneficial. Varian and Thermo do not both make quadrupoles and Ion traps just to make end users puzzled there are benefits to both technologies

akthmps:

If you inject 1 ppb OFN standard to the Varian system, how does the response compare to that from the Agilent GC/MS?

You didn't mention the injection volume of your tests, if it's 1 ul, 5 picograms is the low calibration point of the target compound. That's the sensitivity we only see in ECD. If you can achieve below 10 pg sensitivity for most of the target compounds in EPA 8270, Varian 3800 with 2000 MS is much better than 6890/5973.

I don't have the Varian GC-MS here to inject the 1ppb OFN into, I had simply used the Varian GC-MS at a local university and don't have access to it. The injection volumes used on both instrument for my comparison was 1uL. I expect I could have detect 1 picogram of OFN with the Varian system if I had injected 1uL of the 1ppb standard. However, I didn't bring the 1ppb standard with me and thus didn't inject it.

On the new Varian 4000 GC-MS, the demonstration upon installation is no longer done with 1 picogram of OFN. The sensitivity is now down to 200 femtograms in full scan! Pretty impressive.