With over 35 years of refinement, op amps are a staple of analog designers. Have we reached the limit of what can be done with monolithic op amps? For many applications, IC op amps are clearly the best amplifiers available, and some parameters are clearly as good as they can get. Circuit performance is limited by external factors, so no significant improvements can be expected in the future.
DC performance is virtually perfect! Non-chopper amplifiers have dc offsets in the tens of microvolts and drifts below 1 mV/°C. Choppers reduce these drift and offset numbers to a few microvolts of offset and less than 0.1 mV/°C. Bias currents are well below 100 pA over a wide temperature range. Secondary characteristics like common-mode rejection and gain error are also vanishingly small. Further enhancements in these specifications will result in minuscule improvement in the final application. Other specifications may change, but we're hitting the end here.
In the speed category, unity-gain bandwidth has catapulted from 1 MHz in 1970 to 2000 MHz today. With bandwidths this high, feedback is difficult because even a few inches of wire cause enough delay and phase shift for oscillation. Although it's unlikely that we will see faster general-purpose op amps, faster internally fed back amplifiers are possible. Non-speed-related improvements could be combined with high-speed op amps.
So is there room for improvement? Well, op amps aren't perfect in all respects. The best dc performance isn't available on the high-speed devices. Power handling and efficiency can be improved, too. A wider operating voltage range, either low voltage or very high voltage, is open for development. Also, micro-power (or nanopower) operation can still be enhanced. More robust amplifiers for harsh applications are always needed. Manufacturers have yet to figure out how to make an op amp that can't be blown up.
Process developments also fuel op-amp advances. Matched gigahertz complementary transistors have enabled a step function increase in amplifier speed. Improvements in CMOS 1/f noise and stability allow high-performance low-cost CMOS and chopper amplifiers. (While very fine-line CMOS won't be used for standalone op amps, embedded amplifiers are also faster). Better passivation and plastics give encapsulated amplifiers reliability equal to older hermetic packages.
While certain areas still could stand for improved performance in op amps, any changes must be economically driven. Low-cost, high-performance op amps are so good that they leave a small market for new designs that will cost much more than today's op amps. It will take evolutionary change to spur the need for changes in op amp design. When the market does evolve, op amps will also evolve in response.