Liquid Class Optimization

There are good tips here to optimizing liquid classes in-run. In addition to this, here is my gravimetric method:

1.) Reach out to Hamilton to see if they already have a liquid class! If not…

2.) Start with a liquid class similar to the target liquid. Define what you want to do. Dispense jet empty, part volume? What kind of tips? What is the pipetting range?

3.) Perform all testing with the intended labware. Intended tips, reservoirs… To mimic a 96-well dispense plate I will often use 2D matrix racks loaded with 2D matrix tubes. These are easy to weigh and closely mimic most dispense plates.

4.) I’ll use two channels and perform 2-3 transfers at the high volume. I look for…

-Droplets. Do they form immediately when leaving the vessel? If yes, I’ll consider adding an air gap. If they form during transport to the dispense vessel and the liquid is volatile, I’ll consider using anti-droplet control.

-Accuracy. If close, I’ll move to the next step. If no, I’ll start by adjusting aspiration and dispense speed, and then move to other parameters.

5.) Still using two channels, perform 2-3 transfers at the medium and then low volume range. Again my focus is rough adjustment to eliminate unwanted behavior and dial in accuracy.

6.) Begin fine adjustment. Set up a program to dispense the volumes that will make up the calibration curve. E.g. using four 1000 uL channels with 1000 uL tips, dispense 50 uL twice. Loop 6 times, each loop add 190 uL to the transfer volume. 48+ gravimetric measurements will be monotonous. Keep in mind that while you have 8 replicates of each volume, there could be variation between pipetting channels. Once things appear dialed in…

7.) Liquid class validation. Run the fine adjustment but with a minimum of three transfers at each volume. Generate accuracy and precision data. If in a regulated environment, I suggest saving this data in a controlled form.

I find this takes ~2 hours to set up and perform for each tip and each dispense type. Subsequent liquid classes take ~1 hour. Some other suggestions:

-Liquid classes for the single-channel head often but not always translate to the 96-channel head with minimal modification.

-Using 2D matrix racks and tubes saves much time when gravimetrically weighing individual replicates.

-There are faster ways to optimizing liquid classes, but even in my lab where we use a wide variety of solvents I only perform liquid class development ~5 times per year.

-The Artel MVS is a great verification tool. Can be a bit costly. I think it is a must have when you have many systems. As a liquid class development tool though… I work with a lot of organic solvents that are not necessarily dye friendly. Just something to keep in mind if you consider purchasing one solely for liquid class development.

10 Likes