Here are a few notable quotes from the article along with my comments in italics.
Quote:Compared to semiconductors, such as silicon and InP (commonly used for today’s optical communications), EO polymers use a different physical effect. They use the Pockels effect, which is a “pure” linear electro-optical effect. The alteration to the electro-optic material is a distortion (hyperpolarizability) to the electron clouds of the molecules without removing them from their orbitals. The response timescale is of the order of 10-13s [100 femtoseconds].
Comment: The Pockels Effect changes the electromagnetic field (E-field) around the polymer (voltage-based). E-fields change at the nearly the speed of light, as it is electromagnetic radiation. As such, its response time is incredibly fast.
Quote:In contrast, silicon works on the basis of injection/extraction of electrons, a relatively slow process (10-10s) [100 picoseconds], resulting in plasma dispersion-induced index changes. Many III-V semiconductor (InP) devices work via the electro-absorption/quantum-confined Stark effect, which is also affected by electron flow.
Comment: The Stark Effect involves "electron flow", i.e., electrical current. All conductors have a property known as inductance. Inductance in electrical systems is analogous to inertia in mechanical systems. This means that it takes time to start moving and to stop moving electrons. Time is our enemy in high-speed communication devices. It degrades the response time of the system. The contrast of response times is on the order of 1000X faster for a polymer modulator exploiting the Pockels Effect vs a Si or InP modulator relying on the Stark Effect. Other factors come into play, so LWLG can't boast a 1000X improvement in response time over Si, but - at least in this regard - it is much better.
Quote:There is another speed limitation for traveling-wave devices where the radio frequency (RF) wave and the optical signal interact as they both travel through the nonlinear material. As the signals travel at various speeds — depending on the RF dielectric constant and optical refractive index, respectively — the RF and optical signals eventually get out of phase, limiting the useful interaction length. In fact, continuing past the point when they are out of phase will actually start to undo the effect.
Higher modulation frequencies get out of phase more quickly. Therefore, for a given electrode length, the phase velocity mismatch can limit the bandwidth. Polymers have better phase velocity matching, resulting in a figure of merit at least double lithium niobate’s 43 GHz-cm. The phase-matching bandwidth limit can be raised by shortening the electrode, but only at the cost of raising the drive voltage.
Comment: The phenomena of various frequencies travelling at different rates through a medium is known as "group delay". It results in a rounding of the edges type of distortion in the eye-diagram. If the edges get too rounded, loss of signal integrity occurs. As quoted above, this is most notable at the higher frequencies. Consequently, the bandwidth of the data signal becomes limited. The fact that polymers exhibit at least double the group delay of LiNbO3 based modulators is remarkable. I know that many are concerned about losing market share to "thin-film lithium niobate" (TFLN) modulators. The contrast of the performance of polymer vs TFLN should help ease your concern in this regard, at least to a certain degree.
The article is full of other interesting observations, but are a bit more technical. The above paragraphs/comments highlight the compelling benefits of LWLG Perkinamine polymers over Si-based ones.