Automated cell culture: keeping a living product alive

Every task on the previous pages moves or reads dead or inert material; cell culture moves something alive, at a temperature and in a medium chosen to make almost anything grow. That single fact reorders the engineering. The robot is the easy part; the design question is how a machine keeps a living culture uncontaminated and on schedule for weeks, including at moments the vessel is opened.

The doubling-time asymmetry

Sterility in culture is not a preference but an arithmetic. Mammalian cells divide in roughly a day; bacterial contaminants in tens of minutes. A single bacterium entering at the start therefore outgrows the entire culture before harvest, while the broth still looks innocent for most of the run — contamination is usually detected late, by then unrecoverable. This asymmetry is why the discipline exists at all, and it dictates the automation strategy. An open manipulation under a laminar-flow hood trades contamination risk for access; a closed system removes the trade by making asepsis structural: welded tube connections, sterile single-use fluid paths, no headspace exchange with room air, and — the point — no human hands, which are the largest spore reservoir in any facility. Closed construction eliminates one class of entry but not all: a contaminant already in the starting material, or inside a welded path, is sealed in with the cells. Closure shifts sterility from operator skill to manufacturing quality of the fluid path, which can be tested before use.

Adherent cells resist because they are a surface process

For suspension cells, a passage is a fluid transfer, and automation inherits the ordinary physics of liquid handling. Adherent cells are harder because their process state is a surface: they grow as a layer anchored to plastic, and passaging means a timed chemical detachment — enzyme or chelator — then neutralisation, uniform resuspension, counting and re-seeding at a defined density. Each step has a window: detach too briefly and cells stay behind; too long and the treatment strips membrane proteins and cuts viability. The machine is dosing chemistry at a layer it cannot see directly, and the clock for the whole operation is set by how covered that surface is. Confluence — the fraction of growth area occupied — is not a cosmetic metric: at high density cells inhibit each other through contact and change state, so passage timing is a biological decision. Vision systems estimate confluence from phase-contrast images, and their errors — uneven patches, morphology changes at overgrowth — become timing errors of the detachment step.

Putting the surface into suspension

The bridge between the two worlds is the microcarrier: a small bead that gives anchored cells a surface while the beads themselves float — converting a surface process into a fluid process that pumps, transfers and samples like any suspension culture, and scaling surface area by scaling volume. What automation cannot engineer away is the environment: cultures live at a set temperature and CO₂ tension that buffers their medium’s pH, and every minute a plate spends on a transport deck is an excursion from that climate plus evaporation from its edge wells. Movement speed and queueing discipline are, again, biological variables — the culture feels the schedule.