ICP-MS: Challenges with our Agilent 8900

The most basic difficulty has been learning the software. Yes, modern software is very complicated. That’s completely expected. The 8900 is a complicated instrument, used in a lot of different industries, with a lot of different sample introduction methods, and with different data reduction and other requirements. What learning aids did we have? The instrument came with 5 days of on-site training, which was enormously helpful, but it’s hard to remember everything, even with extensive notes. There were some maintenance videos, which were OK as far as they went. There were some PDFs that were also helpful, particularly information about cell gasses and tuning at the end of the hardware maintenance document. We’ve also read a large number of method papers for the 8800 and 8900, trying to understand why one would want to do this as compared to that.

The big disappointment was the built-in software help function. It’s mostly just a listing of what each button does, but almost nothing about why you would want to push a particular button, or what you need to do to accomplish one or another task. Which of the great many settings should we fiddle with to get better results? What are good masses to pick when setting up a method mini-tune procedure? How do you set up standard + blank blocks in the sample list? No help for those in the software help system. We’ve talked to Agilent sales, service, and training people about this, and they’ve told us that a lot of people have complained about the same thing. The help section needs some procedural instructions that go beyond the videos and push-button explanations. The software can do almost anything, but unless you only want to do one thing all the time, be prepared for a very steep learning curve.

The software has been relatively stable, compared to other instrument software I’ve used. We’ve had occasional program crashes, usually after making lots and lots of changes and tests with one method file. One time, a method file seemed OK while working on it, but run results were erratic in different ways with different runs on the same solutions. Re-booting the computer and ICP-MS didn’t help. Building a new method based on an old one solved the problem. I suspect the method file was corrupted somehow, and was sending some misinformation to the instrument. We’ve had no software crashes during runs. Overall, I’m happy with the software except its unhelpful help materials.

Fall 2017

1) When the new ICP-MS was installed, it had the stainless steel backing plate for the nickel skimmer. A brass plate is needed for a platinum skimmer. Unfortunately, the stainless steel backing plate had a yellowish coating on it, I guess from its factory tests. The coating made it look like brass on the front side. Thinking the backing plate was brass, because it looked like brass, I installed our platinum skimmer onto it. The skimmer overheated and developed a green nickel oxide coating, even on the back mounting surface and threads. Wow! I hadn’t realized the skimmers got that hot. I had been told back in my VG PQ2+ days, that the expanding plasma jet was cold.

2) When I realized what the problem was, I looked in our parts box and found a brass back plate. Sadly, it was the wrong one. It was almost the same, but thicker, and with mounting hardware that would not accept our X-lens.

3) This was a nuisance, one mistake mine, sort of, but but the wrong part was a packing mistake before the instrument was shipped to us. The incorrect brass part was exchanged for the correct part without trouble.

4) Although we exchanged the backing plate, the skimmer was our problem. It took some time to clean it. I ultrasonically cleaned it for about 2 hours in DI water, which removed about 90% of the oxide scale, then used very-fine sandpaper to remove most of the rest. Strangely enough, it still works OK.

Summer 2018

We ordered a spare lens assembly, so we could quickly replace the original (left) if we needed to clean it, and especially if we broke an insulator while doing so. We received the wrong part (right). Though a nuisance, it was exchanged without trouble. This makes the second wrong part.

Fall 2018

We kept draining argon tanks even when not running. Our crack technician checked the whole argon system for leaks, and found one in the instrument. The circled metal and plastic thing was loose, leaking argon from around the surface between the metal part and the larger plastic part to its lower-left. This was repaired under warranty.

Fall 2018

One month out of warranty we got a low-speed warning on the small turbo pump, the one used to pump the second quad. We got a replacement, and it was easy to install ourselves.

Winter 2019

Two months out of warranty, close-after replacing the turbo, we started getting hugely-high backgrounds, starting around mass 120 or so. The peak at mass 240 was over 50 million cps, though we were only running the usual 1 ppb tuning solution. Agilent said they’d talk to us only after we set up a rather scary repair authorization PO. Instead, we opened up the vacuum chamber and blew it out with argon (in case there was dust or a metal flake in the detector, or somewhere).

After doing that, the signal intensity was a little lower. We opened up the vacuum chamber again, to look around, an noticed that one of the wires to quadrupole-1 was extremely close to where the chamber lid would be, when installed. We bent the wire down a little, and the problem was gone. Well, how about that? I don’t think it was very nice of Agilent to refuse to talk to us, before asking for a PO with a very high authorization limit. Making money is one thing, but your customers should also like doing business with you.