Mirror and beamsplitter phase relationships

As from Fresnel equations, passive optical components, such as mirrors, beamsplitters and lenses, can be described as flat thin layers linearly coupling with the incident light. When light impinges on that surface, both reflection and refraction of the light may occur. A coupling coefficient is the ratio of the reflected, or transmitted, light to the incident light. These ratios are generally complex, describing not only the relative amplitudes but also the phase shifts at the interface.

We define a mirror as the contact plane between two media. We derive the coupling coefficients under normal incidence. To do that, we calculate the phase jump for the transmitted and reflected beam. Finally, we obtain the phase for the transmitted beam as a function pf the position of the mirror. We repeat this for a beamsplitter, which is similar to a mirror except for an angle of incidence.

The phase relationship used is important to conserve energy. A variety exist in literature, the two most common are either a symmetric or anti-symmetric relationship between the two sides of the optic. Many theory papers use the anti-symmetric case which has no phase accumlated on transmission and one of the sides accquiring a 180 degree phase shift. The other symmetric case is where neither reflection accumulates any phase, but the transmission from either side has a 90 degree shift, or a factor of complex \(i\).

Both should be physically consistent, the differences will often be in the detunings you need to use in a model to place an interferometer at the correct operating point. In more complicated scenarios however this phase relationship must be expanded. Below also highlights the relationship between an arbitrary reflecting surface with a refractive index and an angle of incidence difference on each side. The problem is no longer symmetric and must be treated more carefully. By default Finesse will use the below relationships when it sees such cases. This should be taken into account if you are doing analytic comparisons.

Note

You can revert back to Finesse v2 behaviour by changing the model settings. This can be changed using self._settings.phase_config.v2_transmission_phase = False.

Mirror

The coupling of light field amplitudes with a mirror under normal incidence can be described as follows: there are two coherent input fields, \(a_{1}\) impinging on the mirror on the front and \(a_{2}\) on the back surface. Two output fields leave the mirror, \(b_{1}\) and \(b_{2}\).

../../_images/flat_surface.svg

Fig. 5 Schematic for two coherent beams falling on the front and back surface of a mirror, \(a_{1}\) and \(a_{2}\), along with the outcoming beams reflected off either surface, \(b_{1}\) and \(b_{2}\).

The following linear equations can be used to describe the coupling:

\[\begin{pmatrix} b_{1}\\ b_{2} \end{pmatrix} = \begin{pmatrix} M_{11} & M_{12}\\ M_{21} & M_{22} \end{pmatrix} \begin{pmatrix} a_{1} \\ a_{2} \end{pmatrix} \]

Tuning

We define the tuning \(\phi_{0}\) of a surface as the change in the mirror position expressed in radiants (with respect to the reference plane). A tuning of \(\phi_{0}=2\pi\) translates the mirror by one vacuum wavelength (default wavelength set in Finesse): \(x=\lambda_{0}\). The direction of the change is defined to be in the direction of the normal vector on the front surface.

With the mirror position given in meters \(x\), then the corresponding tuning computes as follows:

\[\phi_{0} = k_{0} x = \frac{2\pi}{\lambda_{0}} x = \frac{\omega_{0}}{c} x \]

A certain displacement results in different changes in the optical path for light fields with different frequencies. To take that into account, \(\phi\) can be generalised as follows:

\[\phi = \phi_{0} \frac{\omega}{\omega_{0}} \]
../../_images/flat_surface_tuning.svg

Fig. 6 Tuning of a mirror under normal incidence. The solid line is the reference untuned mirror, the dashed one is tuned mirror.


We define the light-mirror coupling coefficients as:

\[M_{11} = r e^{i\varphi_{11}(\phi)}\\ M_{22} = r e^{i\varphi_{22}(\phi)}\\ M_{12} = M_{21} = t e^{i\varphi_{12}(\phi)} = t e^{i\varphi_{21}(\phi)} \]

where \(r\) is the amplitude reflectance of the mirror, \(t\) the mirror transmittance and \(\varphi_{ij}(\phi)\) the phase jump at the surface for the impinging light, which depends on the surface tuning. We assume \(\varphi_{12}(\phi)\) to be equal to \(\varphi_{21}(\phi)\) not to introduce a preferred direction of propagation. For a loss-less surface we can compute conditions for \(\varphi_{ij}(\phi)\) from energy conservation:

\[|b_1|^{2} + |b_2|^{2} = a_1^{2} + a_2^{2} \]

each term in the equation above is defined as:

\[|b_{1}|^{2} = r^{2} a_{1}^{2} + t^{2} a_{2}^{2} + 2rt\, a_{1} a_{2}\, \cos(\varphi_{12}-\varphi_{11})\\ |b_{2}|^{2} = t^{2} a_{1}^{2} + r^{2} a_{2}^{2} + 2rt\, a_{1} a_{2}\, \cos(\varphi_{12}-\varphi_{22})\]

where we temporarily forget the dependencies. Energy conservation requires:

\[\cos(\varphi_{12}-\varphi_{11}) = -\cos(\varphi_{12}-\varphi_{22}) \]

which in turn requires:

\[\varphi_{12}-\varphi_{11} = (2N+1)\pi - (\varphi_{12}-\varphi_{22}) \]

where \(N\) is an integer. After some simple algebraic steps, we obtain:

\[\varphi_{12} = (2N+1)\frac{\pi}{2} + \frac{(\varphi_{11}+\varphi_{22})}{2} \]

We arbitrarily set \(N=0\) and we will follow this convention throughout this modeling. In general, the conditions for \(\varphi_{ij}(\phi)\) are given by the following equation:

\[\varphi_{12}(\phi) = \frac{\pi}{2} + \frac{(\varphi_{11}(\phi)+\varphi_{22}(\phi))}{2} \]

Phase jumps in the untuned case

At the reference position of the mirror, we arbitrarily set \(\varphi_{11}(0)=\varphi_{22}(0)=0\) and we will follow this convention for the rest of the modeling. The phase gain for a beam transmitted through the mirror at the reference position is:

\[\varphi_{12}(0) = \varphi_{21}(0) = \frac{\pi}{2} \]

Phase jumps in the tuned case

As the mirror is tuned, the phase of a beam reflected off the surface \(\varphi_{ii}(\phi)\) is given as:

\[\varphi_{11}(\phi) = 2n_{1}\phi\\ \varphi_{22}(\phi) = -2n_{2}\phi \]

where \(n_{1}\) and \(n_{2}\) are the indices of refraction of the media on either side of the surface. Substituting \(\varphi_{11}(\phi)\) and \(\varphi_{22}(\phi)\): in the equation for \(\varphi_{ij}(\phi)\), we obtain \(\varphi_{21}(\phi)\):

\[\varphi_{12}(\phi) = \varphi_{21}(\phi) = \frac{\pi}{2} + (n_{1}-n_{2})\phi \]

Beamsplitter

A beamsplitter is similar to a mirror except for the extra parameter \(\alpha\) which indicates the tilt angle relative to the incoming beams.

Reflection

../../_images/flat_surface_tilted_Refl.svg

Fig. 7 Schematic for the beam reflected off a beamsplitter in the reference (solid lines) and tuned case (dashed lines).

Referring to the figure above, we define the following geometrical paths as:

\[a = \frac{x}{\cos(\alpha)}\\[10pt] b = a\cos(2\alpha)\\[10pt] c = \frac{x}{\cos(\beta)}\\[10pt] d = c\cos(2\beta)\\[10pt] \]

where \(\beta\) is the refraction angle given by the Snell’s law:

\[n_{1}\sin(\alpha) = n_{2}\sin(\beta) \]

The phase change for a beam reflected on one side of a beamsplitter is:

\[\left(\frac{\omega}{\omega_{0}}k_{0}\right)\, n_{1}|a + b| = 2 n_{1}\phi \cos{\alpha}\\[10pt] \]

As for a beam reflected on the other side, the phase change is:

\[\left(\frac{\omega}{\omega_{0}}k_{0}\right)\, n_{2}|c + d| = 2 n_{2}\phi \cos{\beta}\\[10pt] \]

As was done for the mirror, we model the beamsplitter-light coupling via linear coefficients \(M_{ij}\):

\[M_{11} = r e^{i\varphi_{11}(\phi)}\\ M_{22} = r e^{i\varphi_{22}(\phi)}\\ M_{12} = M_{21} = t e^{i\varphi_{12}(\phi)} = t e^{i\varphi_{21}(\phi)} \]

The conditions for \(\varphi_{ij}(\phi)\) are given by:

\[\varphi_{12}(\phi) = \frac{\pi}{2} + \frac{(\varphi_{11}(\phi)+\varphi_{22}(\phi))}{2} \]

The phase change for a beam reflected off either side of the beamsplitter is given as:

\[\varphi_{11}(\phi) = 2n_{1}\phi\cos\alpha\\ \varphi_{22}(\phi) = -2n_{2}\phi\cos\beta \]

Substituting \(\varphi_{11}(\phi)\) and \(\varphi_{22}(\phi)\): in the equation for \(\varphi_{ij}(\phi)\), we obtain \(\varphi_{21}(\phi)\):

\[\varphi_{12}(\phi) = \varphi_{21}(\phi) = \frac{\pi}{2} + (n_{1}\cos\alpha-n_{2}\cos\beta)\phi \]