A human pipettist adjusts by feel: a little slower for the thick reagent, a pause before lifting so the last drop lets go, a touch to the well wall to shed a stubborn droplet. A robot has none of that intuition. Everything it knows about a liquid has to be written down as numbers. A liquid class is that written-down technique, and building good ones is easier once you hold three ideas in your head at the same time: the transfer mechanism you are working on, the type of liquid you are moving, and the dispense function you are performing.
Start from the transfer mechanism
Not every liquid handler moves liquid the same way, and the mechanism decides how much a liquid class can even control. Air displacement and positive displacement pipetting are the familiar case: a plunger moves a measured volume, and you get command over flow rate, air gaps, delays, blowout, and tip position. This is where liquid class development is richest, and where most of this blog lives.
Other mechanisms expose far less. Acoustic dispensers move nanoliter droplets with focused sound and no tip contact at all, which is wonderfully precise but works over a narrow range of liquids and volumes, so the class is more about calibration than tuning. Peristaltic and bulk multi-dispensers push liquid from a reservoir through tubing, so a class controls little more than dispense speed and height. The lesson is simple: before you tune anything, know which knobs your mechanism actually gives you.
Then classify the liquid
Most reagents fall into one of three families, and each family points you at a different starting configuration.
- Aqueous, or water-like: high water content, low viscosity, well behaved. These take default or near-default settings and are the easiest to get right.
- Solvent: organic and volatile, such as ethanol or DMSO. Speeds can resemble water, but vapor and dripping are the enemy, so you add small trailing air gaps, gentle arm moves, and sometimes a slower aspiration to keep vapor from building in the tip.
- Viscous: glycerol, PEG, concentrated detergents. These demand slow acceleration and deceleration, longer delays, and tip-specific tuning, because a class that dispenses cleanly from a large tip can bubble or flood a small one.
Naming the family first saves hours. If a new reagent behaves like water, you may not need a new class at all. If it behaves like glycerol, you already know to reach for slow speeds and leading air gaps before you touch a single number.
Finally, pick the dispense function
The same liquid needs different parameters depending on how it leaves the tip. There are three functions worth separating.
Wet dispense
The tip touches the liquid already in the well, so cohesion pulls the volume out rather than force. That lets you dispense slowly, which improves accuracy and cuts aerosols and bubbles. It is the default choice when accuracy matters and contamination between wells is not a concern.
Free dispense
The liquid free-falls into the well from above, so the tip never touches the contents. This is what lets you reuse tips for multi-dispensing and keeps a sterile transfer sterile. It needs more speed and often a blowout of a leading air gap to drive the last of the liquid out. Free dispensing a viscous liquid is a losing battle: expect droplets, strings, and short volumes.
Mixing
A mix class aspirates and dispenses repeatedly in one well to homogenize it. Transfer accuracy barely matters here, so a single mix class often serves many liquids. Tune it for vigor relative to tip size instead: enough turnover to mix, not so much that you foam or splash.
When to build a new class, and when not to
The best liquid class is often one you did not write. Vendor defaults usually handle water-like reagents, and if solvent and viscous defaults exist they cover moderately volatile or moderately thick liquids. Reusing an existing class keeps a fleet maintainable and consistent. Build a new one only when the evidence says to: persistent dripping, bubbles, or dispense volumes that will not settle down, or a technique your protocol needs that no existing class offers, such as a tip-touch to clear a droplet or an aspiration that must not disturb a cell pellet.
The quickest development is the development you never have to do, so reach for a new class only when an existing one demonstrably fails your liquid, your volume range, or your technique.
Instrument software shapes how far you can go
How much of this you can express depends on the control software in front of you. Some environments give command-level control over every mechanic per tip type and volume; others expose a friendlier subset grouped by volume range. Neither is strictly better. Deep control helps for hard liquids, but a curated subset is approachable and often just as effective. Plan your development effort around the detail your software actually offers rather than the detail you wish it did.
A workflow that keeps classes trustworthy
- Check what your instrument and software expose, so you tune real parameters rather than imagined ones.
- Decide the liquid type and dispense function you need before you start building the class.
- Look for an existing class that already fits. If one does, stop.
- If not, copy the closest class, rename it for its liquid type and purpose, and change settings incrementally.
- Apply, test on the deck, and repeat. Confirm the class across its full volume range before you rely on it, and record what you validated.