The method of electro-optic detection relies on the Pockels effect (first described in 1906 by Friedrich Pockels). In essence the birefringence of an electro-optic material is modulated by an externally applied electric field. In the case of THz spectroscopy, the incident THz pulse induces a birefringence in an electro-optic medium which is proportional to the electric field of the pulse. This varying birefringence can be measured by observing the change in polarisation state of a probe near-infrared probe pulse. By measuring the degree of polarisation rotation as a function of delay between the terahertz pulse and the near-infrared probe pulse, the terahertz electric field can be mapped.
Components
The electro-optic sampler comprises the following components:
An Electro-optic medium [eg (110) ZnTe]
A broadband quarter-wave plate
A Wollaston prism
Differential detectors [eg balenced photodiodes]
Calibration
Initially the electro-optic sampler is calibrated in the absence of any THz pulse. This involves adjusting a quarter-wave plate to compensate for the equilibrium birefringence of the electro-optic medium. The near-infrared probe pulse will become elliptically polarised as a result of this equilibrium birefringence, and if passed to the Wollaston prism without first going through a compensation plate would give rise to a non-zero detector output. The output voltage of the differential detector is monitored and the quarter-wave plate adjusted until a reading as close to 0V as possible is achieved. This means that subsequent to passing through the quarter-wave plate the probe pulse is circularly polarised i.e. has p and s components of equal magnitude. The following video outlines this calibration procedure:
Output
Once this calibration has been carried out the THz pulse can be measured. For small rotations, the amplitude of the THz-induced birefringence is proportional to the instantaneous THz electric field. By connecting the balanced detector to a data logging system and varying the delay time of the probe pulse with respect to the THz pulse we can sample the THz pulse in the time domain. If required, this time domain representation of the THz pulse can then be Fourier transformed in order to get a frequency domain representation of the THz pulse. However, the time domain detection scheme laid out here means that information about the phase of the signal is not lost; as would be the case for techniques which involve frequency domain spectroscopy.
