difference between plasmonic wave plates and usual wave plates plasmonic waveplates are constructed using the generation of surface plasmons. Is there any difference between these plasmonic wave plates and the usual wave plates used in optics, besides the difference in the dimensions? 
Specifically speaking, do these wave plates change the incident wave front in any other way than changing the phase?
 A: It's a little unclear what you are asking, but sometimes one needs to ask an ill formed question of one's colleagues to help formulate a sound question. 
Plasmonic waveplates are different from waveplates relying on a homogeneous linear material's birefringence: the physics is different. The former's birefringence arises from engineered periodic nano-scale inhomogeneities, the latter arises (mostly) from the shape and alignment of constituent molecules (as in quartz or sapphire crystal). 
However, both aim to realize a particular Jones matrix: the matrix of the linear transformation between the $2\times1$ vector defining the polarization pure state input light and that defining the output light for a waveplate is, by definition $\gamma(\theta)=\exp(i\,\theta\,\sigma_z)$, where $\sigma_z=\mathrm{diag}(1,\,-1)$ (the $z$-Pauli matrix) when we write the polarization state of the light as a $2\times1$ vector of superposition weights for the linear polarization eigenstates and $\theta$ is the phase delay for the plate ($\theta = \pi/2$ for a quarter waveplate, for example). 
So the plates are alike insofar that their design goal performance is the same, but, by dint of their different physics, they differ in how well their function is achieved and in their wavelength performance.
A linear homogeneous birefringent material waveplate, such as a layer of quartz, typically has a wavelength performance of the form approximated by $\theta(\lambda) = 2\,\pi\,\ell\,\Delta/\lambda$, where $\Delta$ is the birefringence, $\ell$ the layer thickness and $\lambda$ the wavelength. So a quarter waveplate at 900nm will be roughly a half waveplate at 450nm: the propagation speeds are weak functions of wavelength and so the variation is dominated by the scale of the layer relative to the wavelength. 
One the other hand, the hope for a plasmonic waveplate is that one can engineer the resonances in the nanostructure such that much more sophisticated wavelength variation can be realized: in particular, one can engineer (at least in far infrared) waveplates that are, for example, quarter waveplates over a whole octave of frequencies. Thus we get a waveplate that will be, to a good approximation, a quarter waveplate at all wavelengths. It is hoped that this high wavelength invariance can ultimately be reproduced at optical wavelengths. 
