Decimal places in linear gradient

[Otavio]

I have a doubt. Recently I programmed a linear gradient on a LC system and the data entry had 2 decimal places (e.g. 8.75-87.75% of acetonitrile). The software allowed me to put two decimal places but when saved, the system showed the linear gradient as 8.8- 87.8%. Is it the software only showing less decimal places or actually injecting 8.8-87.8% and the precision is for only 1 decimal place? It varies from system to system? I know it looks a pretty trivial question but I work with environmental chemistry and I calculate the environmental impact of LC methods. That slight difference can impact slightly the impact calculated by suitable metrics. Thank you.

[tom jupille]

In our lab (admittedly years ago), the PQ specification on both flow and compositional accuracy were +/-1%.
http://www.lcresources.com/tsbible/hs2360.htm
Specifying composition to even one decimal place is pushing things.

[Otavio]

Please calm down. I appreciate your comment, but you don't even know why I save methods like that. Sorry for my lack of clarity. As I said, I work with environmental chemistry and I use alternative and environmentally friendly solvents as mobile phase (ex. vodka, obviously not absolute as MeCN, MeOH or EtOH HPLC grade). To ensure the column would be equilibrated with at least 5% of organic solvent or e.g. to reproduce accurately methods that employ absolute solvents with alternative ones, we need to convert the value regarding the purity of the alternative solvent and sometimes the value gets some decimal places.
Anyway, my question is not about I'm pushing ir not things. It is about the instrument and the software. Basically "the system is injecting with two decimal places or not"? Thank you.

[tom jupille]

"the system is injecting with two decimal places or not"?

The answer is "no".

I would suggest that you run the tests described in the article to confirm the accuracy of your system.

[James_Ball]

It will also depend on the type of pump you are using. A binary system where there are individual pumps for each solvent can be more accurate than a single pump with mixing solenoids prior to the pump.

The system with the mixing solenoids sets the mixture by how long the needle valve is open on each channel, so you have to take into account how fast the solenoid can actuate, how well each solvent flows through each valve and many other variables. This will reduce the accuracy, which means that even one decimal place is probably not what you are actually getting.

Binary pumps control the mixture by the flow rate from each pump. If the flow rate is accurate to 1ul/minute then you could have a high accuracy on the mixing of the solvents, but if the accuracy is 10ul/min then it is less accurate in composition and if the pump accuracy is 100ul/min then it is even worse for accuracy.

As Tom said, most manufacturers only guarantee the accuracy to +/-1%, so if you set it to 80/20% it can be between 79/21% to 81-19% as the final mixture. The software is rounding your entries to one decimal place because it can not control more accurately than that, and then you have to take into account the overall accuracy of the system which is probably giving a final mobile phase mixture accurate to the 1% range.