Are you sure you need linear steppers or servos, and that they would work? I usually see syringe pumps (even on here) using regular rotary motors and leadscrews. Leadscrews can be very precise—they're what's usually used in micropositioners. If you need super-fast actuation, you could instead use hard drive head-positioning voice coil motors, connected via a connecting rod or similar.

Or you could use a leadscrew for the slow motion (i.e. moving the plunger to continue to apply pressure as the liquid is depleted) and do the fast droplet-ejection pulses using a second actuator, such as a piezo stack between the leadscrew and the plunger, or a hammer that hits a membrane on a second liquid-filled chamber that communicates with the syringe and the nozzle. I think the latter is probably better, because it circumvents syringe plunger stiction, at least for the droplet ejection. You might be able to take up the pressure pulsation caused by syringe stiction using a second more flexible membrane, and somehow (acoustically) route the pulse from the hammer membrane toward the nozzle. Or just get rid of the syringe altogether and use one soft membrane with a leadscrew for slow pushing and a stiffer one with a hammer for ejection.

Also, maybe look at the mechanisms that inkjet printer heads use. Might be hard to replicate without industrial technology, but might inspire something easier to make.

If you want to generate droplets, there are systems available on the market. For example, the pressure driven system, it is able to generate droplets with very good CV.  https://www.precigenome.com/droplet-generation

I'm not quite clear on your application, but if you need a large number of a specific size droplet you could try ultrasonic ejection. An ultrasonic transducer agitates the ferrofluid causing microdroplet ejection into a stream of silicone oil. In may be possible to sort the microdroplets by size using an appropriate magnetic field strength in a manner similar to a mass spectrometer. 

If a magnetic field were used to hold the ferrofluid onto the surface of the ultrasonic transducer, then perhaps the ejected microdroplets could be ejected into free space.

If you have the ability to cast parts out of PDMS I think this becomes a fairly easy project.  What you are doing is actually analogous to droplet formation in a T junction fluidic device.  There are a lot of papers on this sort of work, especially out of the Weitz lab.  This paper might help a bit:  http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3637280/ 

Alternatively if you are just collecting drops and using elsewhere you do not have to do this in fluidics.  You can do a mini emulsion chemistry and create droplets due to shear forces on immiscible fluidics.  The paper you want to start with there is titled Solvent Free Synthesis of Janus Colloidal Particles, Granick is the PI on the paper.  

I would use a peristaltic pump. 

It has the following advantages:

  * No piston and no valve -> less leak.

  * Mechanically easier to to using a stepper motor

  * Ability to control very small volumes