Because trapped fluid from the previous connection contaminates the next sample.

A conventional quick-disconnect has internal cavities - the space around the poppet, the recess in the sealing face - that retain a small quantity of whatever last passed through. On a process line moving continuously that is irrelevant. On a sample or analyser line it is not: the retained fluid mixes with the new sample and shifts the reading.

At trace concentrations the effect is significant. An analyser measuring parts per million can be pulled well off its true value by carry-over from a much more concentrated previous sample, and the error looks like a genuine measurement.

Instrumentation couplings minimise internal volume and use flat wiping faces that leave nothing behind. Manufacturers publish the dead volume figure, and on analytical work it should be a selection criterion alongside pressure and temperature.

It leaves no recess to hold fluid, and it wipes clean when the halves separate.

On a conventional coupling the valve poppet sits below the sealing face in a recess. When the halves come apart, that recess is exposed and holds a droplet of the line contents - which spills, and which is also the space that will contaminate the next connection.

A flat-face coupling presents a flush surface on each half when disconnected. There is no cavity to retain fluid, the faces can simply be wiped, and when the halves are pushed together the two flat surfaces meet without trapping air between them.

That gives three benefits at once: minimal spillage on disconnection, minimal air inclusion on connection, and negligible carry-over between connections. It is why flat-face designs dominate both instrumentation work and any application where the fluid is hazardous or the environment must stay clean.

Because both sides of the connection need protecting - the process line and the instrument.

When a sample line is disconnected, the line side is usually still live and connected to the process, so it must close or it will discharge. The instrument side contains the analyser's internal volume, which must also close - both to prevent it draining and, more importantly, to stop air being drawn into an instrument whose measurement depends on a sealed sample path.

Double shut-off closes both automatically as the halves separate, so neither is exposed at any point.

Single shut-off couplings are used where only one side needs protection, and are cheaper. On analytical and sampling work double shut-off is normal, and where the fluid is toxic, flammable or under pressure it should be treated as a requirement rather than a preference.

Materials chosen for the process fluid, remembering that the seal is in contact with the sample and can contribute to it.

Body material is usually stainless steel, matching the instrumentation tubing and avoiding the corrosion and metal pickup that would affect a sample. Specialist alloys are used for aggressive or high-purity service.

Seal material is the more consequential choice. It must be compatible with the fluid at the operating temperature, but on analytical work there is a second consideration: some elastomers absorb components of a sample and release them later, or leach plasticisers into it. Either produces measurement error that is very hard to trace back to a coupling.

PTFE and perfluoroelastomer seals are common in analytical service for that reason. State the fluid, its concentration, the temperature and the measurement being made when specifying, not just the pressure and size.

In everything except the basic principle - and they are not interchangeable despite looking similar.

General-purpose couplings are optimised for flow, robustness and cost. They tolerate dirt, they are made in brass and steel as well as stainless, they have internal cavities, and their seals are chosen for durability against common fluids.

Instrumentation couplings are optimised for sample integrity: minimal dead volume, flat wiping faces, stainless or specialist bodies, seals selected for chemical inertness, small bores, and connections that match instrumentation tubing systems.

They are also more expensive per unit, which occasionally leads to a general-purpose coupling being substituted on a sample line to save money. That substitution introduces dead volume and a potentially reactive seal into the measurement path, and the resulting analytical error costs far more to chase than the coupling saved.