betaFeedback

WorksPhysics and Astronomy

Misalignment and mode mismatch error signals for higher-order Hermite-Gauss modes from two sensing schemes

Abstract

The locking of lasers to optical cavities is ubiquitously required in the field of precision interferometry such as Advanced LIGO to yield optimal sensitivity. Using higher-order Hermite-Gauss (HG) modes for the main interferometer beam has been a topic of recent study, due to their potential for reducing thermal noise of the test masses. It has been shown however that higher-order HG modes are more susceptible to coupling losses into optical cavities: the misalignment and mode mismatch induced power losses scale as 2n+12n+1 and n2+n+1n^{2}+n+1 respectively with nn being the mode index. In this paper we calculate analytically for the first time the alignment and mode mismatch sensing signals for arbitrary higher-order HG modes with both the traditional sensing schemes (using Gouy phase telescopes and quadrant photodetectors) and the more recently proposed radio-frequency jitter-based sensing schemes (using only single element photodiodes). We show that the sensing signals and also the signal-to-shot noise ratios for higher-order HG modes are larger than for the fundamental mode. In particular, the alignment and mode mismatch sensing signals in the traditional sensing schemes scale approximately as n\sqrt{n} and nn respectively, whereas in the jitter-based sensing schemes they scale exactly as 2n+12n+1 and n2+n+1n^{2}+n+1, respectively, which exactly matches the decrease in their respective tolerances. This potentially mitigates the downside of higher-order HG modes for their suffering from excessive misalignment and mode-mismatch induced power losses.

Introduction

The sensitivity of all leading gravitational wave detectors such as Advanced LIGO and Advanced Virgo [1, 2, 3] at signal frequencies around 100 Hz is limited by the thermal noise of the test-mass optics. This is also likely to be the case for next generation detectors such as Cosmic Explorer and Einstein Telescope [4, 5]. To reduce this noise and thus obtain better detector sensitivity, the idea of replacing the currently used fundamental Gaussian laser beam with higher-order optical modes such as Laguerre-Gauss (LG) modes or Hermite-Gauss (HG) modes has been proposed. Higher-order modes with more uniform intensity distributions than the fundamental Gaussian beam can better “average out” the effects of this thermal noise [6, 7]. Research into the potential use of higher-order LG3,3_{3,3} mode carried out with numerical simulations and tabletop experiments [8, 9] has shown that the surface figure imperfections present even in state-of-the-art mirrors will cause significant impurity and losses for the LG3,3_{3,3} mode in realistic, high finesse cavities [10, 11, 12]. However, it has also been shown [13, 14] that higher-order HG modes such as HG3,3_{3,3} mode can be made almost as robust as the currently used HG0,0_{0,0} against mirror surface deformations with the deliberate addition of astigmatism to the test masses.

A. Jones et al. [15] and the authors of this paper [16] have shown however the misalignment and mode mismatch induced power coupling losses increase with the mode order. As a result, higher-order HG modes are more sensitive to beam distortions such as misalignment and mode mismatch than the fundamental mode. However, in this paper we show that higher-order HG modes also have stronger alignment and mode mismatch sensing signals and thus better sensing and control capabilities. This is essential for the implementation of higher order HG modes in future gravitational wave detectors for their thermal noise benefits, as well as in many other areas that utilize the beneficial higher order spatial laser modes, such as in multimode optical quantum information systems [17] and a variety of microscopy-related systems for high resolution imaging [18]. We demonstrate this result by calculating the so-called relative sensing gain, which is defined as the error signal ratio of higher-order mode compared against the fundamental mode. In this paper we will investigate both analytically and through FINESSE [19, 20, 21, 22] simulations the alignment and mode mismatch sensing gains as functions of HG mode index for a single symmetric Fabry–Perot cavity in different detection schemes.

We consider the traditional heterodyne wavefront sensing (WFS) scheme [23, 24] and the more recently proposed radio-frequency jitter modulation sensing (RFJ) scheme [25] for the alignment sensing, and the mode-converter sensing (MCS) scheme [26] and radio-frequency lens modulation sensing (RFL) scheme [25, 27, 28, 29] for mode mismatch sensing, as shown in Figure 1. Specifically, WFS uses two split photodetectors separated by π2\frac{\pi}{2} radians Gouy phase for the two orthogonal misalignment degrees of freedom per axis, namely tilt and lateral displacement. RFJ for alignment sensing on the other hand makes use of an electro-optic beam deflector (EOBD) to impose RF jitter sidebands separated from the carrier frequency by the higher-order mode difference frequency of the optical cavity. Demodulating the beat signal between the RF jitter modulation-induced offset mode sidebands and the misalignment-induced carrier frequency offset modes on a single element photodiode in orthogonal demodulation phases (i.e. separated by π2\frac{\pi}{2} radians) results in linear error signals for both tilt and translation of the input beam with respect to the cavity axis. For MCS we use an astigmatic mode-converter and two 45°-rotated quadrant photodetectors separated by π4\frac{\pi}{4} radians Gouy phase to sense the two orthogonal mode mismatch degrees of freedom, namely waist size and waist position mismatch. RFL for the mode mismatch sensing on the other hand uses an electro-optic lens device (EOL) to produce RF defocus sidebands separated from the carrier frequency by twice the higher-order mode difference frequency of the optical cavity to simultaneously extract the full mode mismatch error signals in the orthogonal demodulation phases from a single photodiode. We show that in all the aforementioned sensing schemes, higher-order HG modes always have stronger sensing signals compared against the fundamental mode. In particular, RFJ/L schemes show an increase in the sensing gains and the signal to shot noise ratio that exactly matches the decrease in the corresponding tolerance [15, 16], which potentially mitigates the downside of higher-order HG modes for their suffering from excessive misalignment and mode-mismatch induced power losses in sensing-noise-limited interferometers. We also conduct the corresponding Finesse simulations for comparison against the analytical results for the sensing gains. The Finesse results agree extremely well with the analytical models.

Figure 1. Illustration of various alignment and mode mismatch sensing schemes considered in this paper. The traditional sensing schemes are on the top row, and the more recently proposed jitter-based sensing schemes are shown on the bottom.

This paper is structured as follows: In § 1.1 and III we present a step by step theoretical derivation of the alignment and mode mismatch sensing signals in symmetric cavities for arbitrary HG modes, respectively. We also include a comparison with the corresponding Finesse simulation results. We report our conclusions and discussions in § 2.1.

Alignment Sensing

In the following two subsections, we will calculate the alignment sensing error signals in the traditional wavefront sensing scheme (WFS) [23, 24] and the more recently proposed radio-frequency jitter modulation sensing scheme (RFJ) [25] with an arbitrary higher-order HG mode as the carrier.

The general expression for a HG mode propagating along the zz axis is [30]

Un,m(x,y,z)=Un(x,z)Um(y,z)(1)\mathcal{U}_{n,m}(x,y,z)=\mathcal{U}_{n}(x,z)\mathcal{U}_{m}(y,z) \tag*{(1)}

with

Un(x,z)=(2π)1/4(exp⁡(i(2n+1)Ψ(z))2nn!w(z))1/2×Hn(2xw(z))exp⁡(−ikx22Rc(z)−x2w2(z)),(2)\mathcal{U}_{n}(x,z)=\left(\frac{2}{\pi}\right)^{1/4}\left(\frac{\exp(i(2n+1)\Psi(z))}{2^n n!w(z)}\right)^{1/2}\times H_n\left(\frac{\sqrt{2}x}{w(z)}\right)\exp\left(-i\frac{kx^2}{2R_c(z)}-\frac{x^2}{w^2(z)}\right), \tag*{(2)}

where Ψ(z)=arctan⁡(z−z0zR)\Psi(z)=\arctan\left(\frac{z-z_0}{z_R}\right) is the Gouy phase with zR=πw02λz_R=\frac{\pi w_0^2}{\lambda} being the Rayleigh range. kk is the wavenumber, λ\lambda is the wavelength, w(z)w(z) is the beam radius and Rc(z)R_c(z) is the wavefront radius of curvature. They are related to the beam waist size w0w_0 and beam waist position z0z_0 via

w(z)=w01+(z−z0zR)2(3)w(z)=w_0\sqrt{1+\left(\frac{z-z_0}{z_R}\right)^2} \tag*{(3)}
Rc(z)=z−z0+zR2z−z0R_c(z)=z-z_0+\frac{z_R^2}{z-z_0}

The single mode electric field before entering the cavity can thus be written as

E(x,y,z)=E0Un,m(x,y,z)ei(ωt−kz)(4)E(x,y,z)=E_0\mathcal{U}_{n,m}(x,y,z)e^{i(\omega t-kz)} \tag*{(4)}

where E0E_0 is the initial field amplitude and ω\omega is the field angular frequency.

For alignment sensing, as illustrated in the top panels of Figure 2, since HG modes are separable in xx and yy, and therefore the sensing gains for misalignment in one axis are independent of the mode index in the orthogonal axis, one can always consider the single-axis behaviour, such as a misalignment in the xzxz plane for the HGn,0\mathrm{HG}_{n,0} mode, without loss of generality [16]. By symmetry the same arguments can be applied to the misalignment in the yzyz plane for the HG0,m\mathrm{HG}_{0,m} mode and HGn,m\mathrm{HG}_{n,m} modes in general.

Figure 2. Illustration of misalignment (top) and mode mismatch (bottom) between the input laser mode (red) and the cavity eigenmode (grey). Top, the left panel shows lateral translation of δx0\delta x_0 and the right panel shows tilt of α\alpha. Bottom, the left panel shows the waist size mismatch of amount δw0\delta w_0 and the right shows the waist position mismatch of amount δz\delta z.

In this Section we consider a generic HGn,0\mathrm{HG}_{n,0} mode with the misalignment of the beam occurring in the xzxz plane at the center of a symmetric cavity, which we make coincident with the origin of our coordinate system, i.e. z0=0z_0=0. We can thus simplify the initial beam as

E(x,z)=E0Un(x,z)eiωt(5)E(x,z)=E_0\mathcal{U}_{n}(x,z)e^{i\omega t} \tag*{(5)}

Alignment Sensing: Wavefront Sensing (WFS)

In WFS scheme, we make use of two quadrant photodetectors in reflection of the cavity away from the cavity waist by π\pi and π/2\pi/2 Gouy phases for the tilt and lateral offset degrees of freedom, respectively.

WFS: tilt

Let us look at the tilt degree of freedom first. We apply a phase modulation with modulation index mm at a frequency Ω\Omega to the carrier field in the HGn,0\mathrm{HG}_{n,0} mode. Keeping the first order sidebands only, the beam becomes

E=E0Uneiωt(1+im2(e−iΩt+eiΩt))(6)E=E_0\mathcal{U}_n e^{i\omega t}\left(1+i\frac{m}{2}\left(e^{-i\Omega t}+e^{i\Omega t}\right)\right) \tag*{(6)}

According to the results in paper [16], upon application of a misalignment angle α\alpha between the input beam axis and the cavity optical axis about the cavity waist, the field at the cavity input mirror, up to the first order, becomes

Un′≈Un+iαΘ(n+1Un+1e−iΨ+nUn−1eiΨ)(7)\mathcal{U}'_n\approx\mathcal{U}_n+i\frac{\alpha}{\Theta}\left(\sqrt{n+1}\mathcal{U}_{n+1}e^{-i\Psi}+\sqrt{n}\mathcal{U}_{n-1}e^{i\Psi}\right) \tag*{(7)}

where Θ=λπw0\Theta=\frac{\lambda}{\pi w_0} is the far-field divergence angle and Ψ\Psi is the accumulated Gouy phase from the input mirror to the cavity waist. The input modulated beam from Eq. (6), now reads

E=E0(Un+iαΘ(n+1Un+1e−iΨ+nUn−1eiΨ))eiωt×(1+im2(e−iΩt+eiΩt))(8)\begin{aligned} E=E_0\left(\mathcal{U}_n+i\frac{\alpha}{\Theta}\left(\sqrt{n+1}\mathcal{U}_{n+1}e^{-i\Psi}+\sqrt{n}\mathcal{U}_{n-1}e^{i\Psi}\right)\right)e^{i\omega t} \\ &\times\left(1+i\frac{m}{2}\left(e^{-i\Omega t}+e^{i\Omega t}\right)\right) \tag*{(8)} \end{aligned}

In order to detect an alignment error signal in reflection of the cavity, we must consider the reflected field from the cavity EreflE_{\mathrm{refl}}. In reflection, each term in Eq. (8) is multiplied by the cavity complex reflectance function F(ω,n)F(\omega,n), where nn is the mode index. The complex reflectance function is given by

F(ω,n)=r1−r2t12exp⁡(−i(Dωc+(n+1)Ψrt))1−r1r2exp⁡(−i(Dωc+(n+1)Ψrt))(9)F(\omega,n)=r_1-\frac{r_2t_1^2\exp\left(-i\left(D\frac{\omega}{c}+(n+1)\Psi_{\mathrm{rt}}\right)\right)}{1-r_1r_2\exp\left(-i\left(D\frac{\omega}{c}+(n+1)\Psi_{\mathrm{rt}}\right)\right)} \tag*{(9)}

where r1r_1, t1t_1, r2r_2, and t2t_2 are the amplitude reflectivities and transmissivities of the ITM and ETM, respectively; DD is the exact round-trip length of the cavity, and Ψrt\Psi_{\mathrm{rt}} the Gouy phase accumulated on one round-trip path inside the cavity. In general the cavity complex reflectance function F(ω,n)F(\omega,n) is complicated. However, for a high-finesse, completely over-coupled cavity that is geometrically stable, all reflectivity coefficients are real, and either equal to 11 for non-resonant field components, or −1-1 for resonant field components. For example, the cavities in Advanced LIGO have similar properties and so we can make the assumptions throughout to make the analytical results more intuitively understandable. As a result, we have F(ω,n)=−1F(\omega,n)=-1 while the reflectance functions for the non-resonant field components, such as F(ω,n±1)F(\omega,n\pm1), F(ω±Ω,n)F(\omega\pm\Omega,n), and F(ω±Ω,n±1)F(\omega\pm\Omega,n\pm1), are all 11.

The reflected field EreflE_{\mathrm{refl}} with the above assumption is

Erefl=(−Un+im2(UneiΩt+Une−iΩt)−iαΘ(Un+1n+1+Un−1n)+αm2Θ(Un+1n+1eiΩt+Un+1n+1e−iΩt+Un−1neiΩt+Un−1ne−iΩt))⋅E0eiωt(10)E_{\mathrm{refl}}=\left(-\mathcal{U}_n+\frac{im}{2}\left(\mathcal{U}_ne^{i\Omega t}+\mathcal{U}_ne^{-i\Omega t}\right)-\frac{i\alpha}{\Theta}\left(\mathcal{U}_{n+1}\sqrt{n+1}+\mathcal{U}_{n-1}\sqrt{n}\right)+\frac{\alpha m}{2\Theta}\left(\mathcal{U}_{n+1}\sqrt{n+1}e^{i\Omega t}+\mathcal{U}_{n+1}\sqrt{n+1}e^{-i\Omega t}+\mathcal{U}_{n-1}\sqrt{n}e^{i\Omega t}+\mathcal{U}_{n-1}\sqrt{n}e^{-i\Omega t}\right)\right)\cdot E_0e^{i\omega t} \tag*{(10)}

where we have set the accumulated Gouy phase from the cavity waist to the QPD to be π\pi. This produces an extra factor of eiπ⋅(+1)=−1e^{i\pi\cdot(+1)}=-1 for the adjacent upper and lower modes scattered from the original mode, as the result of tilt.

The photocurrent produced by a split photodetector in reflection from the cavity is given by

IPD=(∫0∞dx−∫−∞0dx)Erefl⋅Erefl∗(11)I_{\mathrm{PD}}=\left(\int_{0}^{\infty}dx-\int_{-\infty}^{0}dx\right)E_{\mathrm{refl}}\cdot E_{\mathrm{refl}}^{*} \tag*{(11)}

for an ideal split photodetector, assuming an appropriate responsivity of 1 A/W. EreflE_{\mathrm{refl}} and Erefl∗E_{\mathrm{refl}}^{*} have 9 terms each, so there are 81 terms to evaluate in the product EreflErefl∗E_{\mathrm{refl}}E_{\mathrm{refl}}^{*}. However, most terms will have no contribution to the signal. In particular, only terms that are odd in xx have contributions to the split photodetector signal. We also know only terms that oscillate with the modulation frequency Ω\Omega contribute to the photodetector signal after demodulation at Ω\Omega and low-pass filtering. The non-vanishing photocurrent is then

IPD=−2E02αΘm(eiΩt+e−iΩt)(n+1βn,n+1+nβn,n−1)(12)I_{\mathrm{PD}}=-2E_{0}^{2}\frac{\alpha}{\Theta}m\left(e^{i\Omega t}+e^{-i\Omega t}\right)\left(\sqrt{n+1}\beta_{n,n+1}+\sqrt{n}\beta_{n,n-1}\right) \tag*{(12)}

where βn,n+1\beta_{n,n+1} are the beat coefficients

βn,n+1=(∫0∞dx−∫−∞0dx)Un⋅Un+1(13)\beta_{n,n+1}=\left(\int_{0}^{\infty}dx-\int_{-\infty}^{0}dx\right)\mathcal{U}_{n}\cdot\mathcal{U}_{n+1} \tag*{(13)}

In general evaluating the beat coefficients is complicated, but as nn goes large βn,n+1\beta_{n,n+1} approaches a constant value around 0.64. For further details see the appendix IV.

We may assume the photocurrent signal is converted to a voltage signal by an appropriate trans-impedance stage with a gain of 1 V/A. Demodulating the beat signal with a phase ϕ=0\phi=0 (a condition often known as in-phase or I-phase demodulation) and removing signal components at 2Ω2\Omega with a low-pass filter, we obtain the following expression for the tilt error signal measured at QPD1_1:

VWFSΩ,QPD1=−2E02αΘm(n+1βn,n+1+nβn,n−1)(14)V_{\mathrm{WFS}}^{\Omega,\mathrm{QPD}_{1}}=-2E_{0}^{2}\frac{\alpha}{\Theta}m\left(\sqrt{n+1}\beta_{n,n+1}+\sqrt{n}\beta_{n,n-1}\right) \tag*{(14)}

We define the relative sensing gain Σn\Sigma_n as the ratio of the error signal for the higher-order mode HGn,0_{n,0} and the fundamental mode HG0,0_{0,0}

Σn=(n+1βn,n+1+nβn,n−1)/β0,1(15)\Sigma_n=\left(\sqrt{n+1}\beta_{n,n+1}+\sqrt{n}\beta_{n,n-1}\right)/\beta_{0,1} \tag*{(15)}

The sensing gain increases with the mode index.

WFS: translation.

Now let us look at the lateral offset degree of freedom. According to the results in paper [16], the beam after the lateral displacement with respect to the cavity optical axis, at the cavity input mirror becomes

Uδx0(x,z)≈Un+δx0w0(n+1Un+1e−iΨ−nUn−1eiΨ)(16)\mathcal{U}^{\delta x_{0}}(x,z)\approx\mathcal{U}_{n}+\frac{\delta x_{0}}{w_{0}}\left(\sqrt{n+1}\mathcal{U}_{n+1}e^{-i\Psi}-\sqrt{n}\mathcal{U}_{n-1}e^{i\Psi}\right) \tag*{(16)}

where δx0\delta x_{0} is the lateral displacement along the xx direction. The difference for this path compared earlier is that the accumulated Gouy phase from the cavity waist to the QPD is set to π/2\pi/2 instead. This produces an extra factor of eiπ2(±1)=±ie^{i\frac{\pi}{2}(\pm1)}=\pm i for the upper and lower adjacent modes scattered from the original mode. We thus have

Uδx0(x,z)≈Un+iδx0w0(n+1Un+1e−iΨ+nUn−1eiΨ)(17)\mathcal{U}^{\delta x_{0}}(x,z)\approx\mathcal{U}_{n}+i\frac{\delta x_{0}}{w_{0}}\left(\sqrt{n+1}\mathcal{U}_{n+1}e^{-i\Psi}+\sqrt{n}\mathcal{U}_{n-1}e^{i\Psi}\right) \tag*{(17)}

Comparing against Eq. (7) (with an additional factor of -1 for the adjacent upper and lower scattered modes from the π\pi Gouy phase), we see that the alignment signal calculation for tilt and lateral displacement are essentially the same if one changes the expansion parameter Θθ\frac{\Theta}{\theta} to −δx0w0-\frac{\delta x_{0}}{w_{0}}. As a result, we can write down directly the lateral translation error signal from Eq. (14):

VWFSΩ,QPD2=2E02δx0w0m(n+1βn,n+1+nβn,n−1)(18)V_{\mathrm{WFS}}^{\Omega,\mathrm{QPD}_{2}}=2E_{0}^{2}\frac{\delta x_{0}}{w_{0}}m\left(\sqrt{n+1}\beta_{n,n+1}+\sqrt{n}\beta_{n,n-1}\right) \tag*{(18)}

with the demodulation phase set to zero as well. We also obtain the same sensing gain in this translation path as the tilt case in Eq. (15).

As shown in Figure 5 in the appendix IV, the beat coefficients converge to around 0.64 as the mode order gets large. We thus can approximate our exact results in Eqs. (14) and (18) by replacing the complicated higher-order beat coefficients (i.e. excluding β0,1\beta_{0,1}) with a constant value 0.64. After the approximation explained above we obtain

Figure 5. Beat coefficients βn,n+1\beta_{n,n+1} on a split photodiode in blue. The two straight dashed lines connecting the four end points intersect at the one point with its y coordinate being around 0.64.

Σn≈0.64⋅(n+1+n)/β0,1(19)\Sigma_n\approx0.64\cdot\left(\sqrt{n+1}+\sqrt{n}\right)/\beta_{0,1} \tag*{(19)}

As shown in Figure 3, the approximation is quite good except for lower mode HG0,0_{0,0}, which is due to the fact that for lower mode index the beat coefficients, β0,1\beta_{0,1} for example, are quite different from our approximation 0.64. We also conduct the corresponding Finesse simulation in black in Figure 3. They agree exceptionally well with the exact analytical results in yellow.

Figure 3. Relative alignment sensing gain Σn\Sigma_n in WFS and RFJ schemes for HGn,0_{n,0} modes. The analytical sensing gains in Eqs. (15) and (29) and the approximation in Eq. (19) are included. The corresponding FINESSE numerical results are also included for comparison.

We thus conclude that the alignment sensing gain for higher-order mode HGn,0\mathrm{HG}_{n,0} in this traditional WFS scheme scales approximately as n+n+1\sqrt{n}+\sqrt{n+1}. The calculation is done assuming misalignment in the xzxz plane for a generic HGn,0\mathrm{HG}_{n,0} (i.e. 2D), but one can easily generalize it to an arbitrary 3D case since HG modes are separable in xx and yy axes and any off-axis misalignment can be resolved into a tilt in each axis.

B Alignment Sensing: Radio-Frequency Jitter sensing (RFJ)

For the alignment sensing with the RF jitter modulation scheme, we use an electro-optic beam deflector (EOBD) to impose RF jitter modulation sidebands separated from the carrier frequency by the higher-order mode difference frequency of the optical cavity. The scheme involves demodulating the beat signal between the RF jitter modulation-induced offset mode sidebands and the misalignment-induced carrier frequency offset modes on a single-element photodetector in orthogonal demodulation phases. This produces simultaneously linear error signals for the two orthogonal misalignment degrees of freedom. With a single-element photodetector, the beat coefficients between say HGn,m\mathrm{HG}_{n,m} and HGn′,m′\mathrm{HG}_{n',m'} modes will simply be the Kronecker delta functions

β(n,m),(n′,m′)≡∫−∞∞∫−∞∞dxdy Unm⋅Un′m′=δn,n′⋅δm,m′(20)\beta_{(n,m),(n',m')} \equiv\int_{-\infty}^{\infty}\int_{-\infty}^{\infty} dxdy\, U_{nm}\cdot U_{n'm'} = \delta_{n,n'}\cdot\delta_{m,m'} \tag*{(20)}

We will show explicitly with orthogonal demodulation phases this beat signal provides simultaneously linear error signals for both tilt and translation of the two axes.

Similar to the WFS scheme, we consider misalignment in the xzxz plane for a generic HGn,0\mathrm{HG}_{n,0} beam propagating along the zz axis. The initial field in Eq. (5) becomes

U(x,z)=E0Un(x,z)eiωt(21)\mathcal{U}(x,z)=E_0\mathcal{U}_n(x,z)e^{i\omega t} \tag*{(21)}

The field after the electro-optic beam deflector modulator (EOBD) becomes [25]

UEOBD≈E0Uneiωt+imα2ΘmE0(n+1Un+1e−iΨ1+nUn−1eiΨ1)(ei(ω+Ω)t+ei(ω−Ω)t)(22)\mathcal{U}^{\mathrm{EOBD}}\approx E_0\mathcal{U}_n e^{i\omega t}+i\frac{m_\alpha}{2\Theta_m}E_0\left(\sqrt{n+1}\mathcal{U}_{n+1}e^{-i\Psi_1}+\sqrt{n}\mathcal{U}_{n-1}e^{i\Psi_1}\right)\left(e^{i(\omega+\Omega)t}+e^{i(\omega-\Omega)t}\right) \tag*{(22)}

where mαm_\alpha is the RF jitter modulation depth, Θm\Theta_m is the far-field beam divergence angle at the tilt-modulator location, and Ψ1\Psi_1 is the accumulated Gouy phase from the modulator to the cavity waist.

RFJ: tilt

Let us look at the tilt degree of freedom. Upon applying a tilt of angle α\alpha between the beam coming out of the EOBD and the cavity optical axis at the cavity waist, Eq. (22) becomes

Uα≈E0Uneiωt+iαΘcE0(n+1Un+1e−iΨ2+nUn−1eiΨ2)+imα2Θm[n+1(Un+1+iαΘc(n+2Un+2e−iΨ2+n+1UneiΨ2))e−iΨ1+n(Un−1+iαΘc(nUne−iΨ2+n−1Un−2eiΨ2))eiΨ1](ei(ω+Ω)t+ei(ω−Ω)t)(23)\begin{aligned} \mathcal{U}^{\alpha}\approx{}&E_0\mathcal{U}_n e^{i\omega t}+i\frac{\alpha}{\Theta_c}E_0\left(\sqrt{n+1}\mathcal{U}_{n+1}e^{-i\Psi_2}+\sqrt{n}\mathcal{U}_{n-1}e^{i\Psi_2}\right)\\ &+i\frac{m_\alpha}{2\Theta_m}\left[\sqrt{n+1}\left(\mathcal{U}_{n+1}+i\frac{\alpha}{\Theta_c}\left(\sqrt{n+2}\mathcal{U}_{n+2}e^{-i\Psi_2}+\sqrt{n+1}\mathcal{U}_n e^{i\Psi_2}\right)\right)e^{-i\Psi_1}\right.\\ &\left.\qquad+\sqrt{n}\left(\mathcal{U}_{n-1}+i\frac{\alpha}{\Theta_c}\left(\sqrt{n}\mathcal{U}_n e^{-i\Psi_2}+\sqrt{n-1}\mathcal{U}_{n-2}e^{i\Psi_2}\right)\right)e^{i\Psi_1}\right]\left(e^{i(\omega+\Omega)t}+e^{i(\omega-\Omega)t}\right) \tag*{(23)} \end{aligned}

where Ψ2\Psi_2 is the accumulated Gouy phase from the cavity waist to the PD. We can safely ignore Ψ2\Psi_2 from now on since single-element photodetectors can only detect the beat between the same mode sidebands (such as HGn+1,0\mathrm{HG}_{n+1,0} and HGn+1,0\mathrm{HG}_{n+1,0}, HGn−1,0\mathrm{HG}_{n-1,0} and HGn−1,0\mathrm{HG}_{n-1,0}) assuming the aperture of the PD is much larger than the beam size. These same mode sidebands have the same accumulated Gouy phase Ψ2\Psi_2. This overall phase factor however has no contribution to the photocurrent E⋅E∗E\cdot E^* as it cancels. Later on we will also see the extra accumulated Gouy phase Ψ1\Psi_1 for the sideband modes can also be ‘absorbed’ into an overall demodulation phase and does not concern us.

In order to extract the error signals in reflection of the cavity, we need to multiply each term in Eq. (23) with a suitable reflectance function F(ω,n)F(\omega,n) of the cavity to obtain EreflE_{\mathrm{refl}}. In general the cavity reflectance function F(ω,n)F(\omega,n) in

Eq. 9 is complex. However, in the high-finesse, completely over-coupled cavity case, all resonant components have the reflectance function of -1 whereas non-resonant field components have the reflectance function of 1. RFJ scheme relies on modulating at the higher-order mode difference frequency, which makes the upper mode in the upper sideband and the lower mode in the lower sideband resonant in the optical cavity. As a result, we have F(ω,n)=F(ω+Ω,n+1)=F(ω−Ω,n−1)=−1F(\omega,n)=F(\omega+\Omega,n+1)=F(\omega-\Omega,n-1)=-1 while the rest reflectance values for the non-resonant field components, such as F(ω,n±1)F(\omega,n\pm1), are all 1. The photocurrent signal on the single-element photodetector, which is defined as

IPD=∫−∞∞dx Erefl⋅Erefl∗(24)I_{\mathrm{PD}}=\int_{-\infty}^{\infty} dx\, E_{\mathrm{refl}}\cdot E_{\mathrm{refl}}^* \tag*{(24)}

becomes

IPD=mααΘmΘcE02(2n+1)(eiΨ1eiΩt+e−iΨ1e−iΩt)(25)I_{\mathrm{PD}}=\frac{m_{\alpha}\alpha}{\Theta_m\Theta_c}E_0^2(2n+1)\left(e^{i\Psi_1}e^{i\Omega t}+e^{-i\Psi_1}e^{-i\Omega t}\right) \tag*{(25)}

where we again have assumed a responsivity of 1 A/W, and used the orthonormal condition on a single-element photodiode in Eq. (20)

∫−∞∞dx Un⋅Un′=δn,n′(26)\int_{-\infty}^{\infty}dx\,\mathcal{U}_n\cdot\mathcal{U}_{n'}=\delta_{n,n'} \tag*{(26)}

Now let us perform the mathematical operations equivalent to the demodulation process. With demodulation, we multiply the above photocurrent signal (after being converted to a voltage signal by a trans-impedance stage with a gain of 1 V/W1\,\mathrm{V/W}) by a local oscillator cos⁡(Ωt+δϕ)=12(ei(Ωt+δϕ)+e−i(Ωt+δϕ))\cos(\Omega t+\delta\phi)=\frac{1}{2}\left(e^{i(\Omega t+\delta\phi)}+e^{-i(\Omega t+\delta\phi)}\right) and then extract the DC terms with a low-pass filtering.

VRFJΩ,Q=IPD×cos⁡(Ωt+δϕ)=mααE02(2n+1)(eiΨ1eiΩt+e−iΨ1e−iΩt)ΘmΘc×ei(Ωt+δϕ)+e−i(Ωt+δϕ)2=mααE02(2n+1)2ΘmΘc(ei(Ψ1−δϕ)+ei(−Ψ1+δϕ))+(terms in 2Ω)(27)\begin{aligned} V_{\mathrm{RFJ}}^{\Omega,Q}=I_{\mathrm{PD}}\times\cos(\Omega t+\delta\phi) \\ &=\frac{m_{\alpha}\alpha E_0^2(2n+1)\left(e^{i\Psi_1}e^{i\Omega t}+e^{-i\Psi_1}e^{-i\Omega t}\right)}{\Theta_m\Theta_c}\times\frac{e^{i(\Omega t+\delta\phi)}+e^{-i(\Omega t+\delta\phi)}}{2} \\ &=\frac{m_{\alpha}\alpha E_0^2(2n+1)}{2\Theta_m\Theta_c}\left(e^{i(\Psi_1-\delta\phi)}+e^{i(-\Psi_1+\delta\phi)}\right)+\text{(terms in }2\Omega\text{)} \tag*{(27)} \end{aligned}

The DC terms can be maximized by setting the overall demodulation phase offset δϕ\delta\phi to be Ψ1\Psi_1. We have thus seen that the difference in the accumulated Gouy phases in the upper and lower sidebands does not concern us here if we adjust the overall demodulation phase correspondingly (and we know how much exactly we should adjust). Demodulating with demodulation phase set to be Ψ1\Psi_1 (i.e. I-phase demodulation), we obtain the error signal

VRFJΩ,I=mααΘmΘcE02(2n+1)(28)V_{\mathrm{RFJ}}^{\Omega,I}=\frac{m_{\alpha}\alpha}{\Theta_m\Theta_c}E_0^2(2n+1) \tag*{(28)}

The relative sensing gain then is simply

Σn=2n+1(29)\Sigma_n=2n+1 \tag*{(29)}
RFJ: lateral offset

Now let us consider the lateral offset error signal. Upon applying a lateral displacement of amount δx0\delta x_0, Eq. (22) becomes [16]

Uδx0≈E0Uneiωt−δx0w0E0eiωt(n+1Un+1e−iΨ2−nUn−1eiΨ2)+imα2ΘmE0[n+1(Un+1−δx0w0(n+2Un+2e−iΨ2−n+1UneiΨ2))e−iΨ1+n(Un−1−δx0w0(nUne−iΨ2−n−1Un−2eiΨ2))eiΨ1](ei(ω+Ω)t+ei(ω−Ω)t)(30)\mathcal{U}^{\delta x_0}\approx E_0\mathcal{U}_n e^{i\omega t}-\frac{\delta x_0}{w_0}E_0e^{i\omega t}\left(\sqrt{n+1}\mathcal{U}_{n+1}e^{-i\Psi_2}-\sqrt{n}\mathcal{U}_{n-1}e^{i\Psi_2}\right)+i\frac{m_{\alpha}}{2\Theta_m}E_0\left[\sqrt{n+1}\left(\mathcal{U}_{n+1}-\frac{\delta x_0}{w_0}\left(\sqrt{n+2}\mathcal{U}_{n+2}e^{-i\Psi_2}-\sqrt{n+1}\mathcal{U}_n e^{i\Psi_2}\right)\right)e^{-i\Psi_1}+\sqrt{n}\left(\mathcal{U}_{n-1}-\frac{\delta x_0}{w_0}\left(\sqrt{n}\mathcal{U}_n e^{-i\Psi_2}-\sqrt{n-1}\mathcal{U}_{n-2}e^{i\Psi_2}\right)\right)e^{i\Psi_1}\right]\left(e^{i(\omega+\Omega)t}+e^{i(\omega-\Omega)t}\right) \tag*{(30)}

We can again ignore Ψ2\Psi_2 since the field components that have contributions to the single-element photodetector signal have the same accumulated Gouy phase Ψ2\Psi_2. This overall phase factor drops out and has no contribution to the photocurrent. Using the fact that F(ω,n)=F(ω+Ω,n+1)=F(ω−Ω,n−1)=−1F(\omega,n)=F(\omega+\Omega,n+1)=F(\omega-\Omega,n-1)=-1, and F=1F=1 for all other terms for our high-finesse, completely over-coupled cavity, the single-element PD signal, as defined in Eq. (24), becomes

IPD=imαδx0Θmw0E02(2n+1)(eiΨ1eiΩt−e−iΨ1e−iΩt)(31)I_{\mathrm{PD}}=\frac{im_{\alpha}\delta x_0}{\Theta_m w_0}E_0^2(2n+1)\left(e^{i\Psi_1}e^{i\Omega t}-e^{-i\Psi_1}e^{-i\Omega t}\right) \tag*{(31)}

where we have used the orthonormal condition on a single-element photodiode in Eq. (26). For the demodulation process, similar to Eq. (27), we can adjust the overall demodulation phase to be Ψ1+π/2\Psi_1+\pi/2 (a condition often known as quadrature-phase or Q-phase demodulation) to extract the optimal error signal

VRFJΩ,Q=−mαδx0Θmw0E02(2n+1)(32)V_{\mathrm{RFJ}}^{\Omega,Q}=-\frac{m_{\alpha}\delta x_0}{\Theta_m w_0}E_0^2(2n+1) \tag*{(32)}

We thus see we can extract the two orthogonal alignment sensing error signals in Eqs. (28) and (32) simultaneously with a single element photodetector with orthogonal demodulation phases Ψ1\Psi_1 and Ψ1+π/2\Psi_1+\pi/2 respectively. The resulting error signals are proportional to each other, and they scale as 2n+12n+1 with nn being the mode order of the carrier HGn,0\mathrm{HG}_{n,0}, see Figure 3. This linear dependence in nn comes from the fact that the “effective” modulation depth for the EOBD in Eq. (22) scales roughly as n\sqrt{n}. Beating it with the scattered modes caused by the static misalignment (whose amplitude also scales roughly as n\sqrt{n} as shown in Eq. (23)) results in the linear dependence for the RFJ sensing gain here. We also conduct the corresponding FINESSE simulation. As shown in blue in Figure 3, they agree extremely well with the analytical results in orange.

We thus see that the alignment sensing error signals in RFJ scheme increase linearly with respect to the mode index, as shown in Figure 3. This increase is even faster than the scaling relation in the traditional WFS technique in Eq. (14), due to the extra factor of n\sqrt{n} from the effective modulation depth of the EOBD for the jitter-based scheme.

Mode-Mismatch Sensing

For the mode mismatch sensing, as illustrated in the bottom panels of Figure 2, we can’t reduce the problem to a single-axis case, so we will have to consider both xx and yy axes at the same time [16]. In this Section, we will calculate the mode mismatch sensing error signals with an arbitrary higher-order HG mode HGn,m\mathrm{HG}_{n,m} as the carrier in both the mode-converter sensing scheme (MCS) the more recently proposed radio-frequency lens modulation sensing scheme (RFL). For the sake of compactness and simplicity, we will consider the case where the mode mismatch occurs at the center of a symmetric cavity, which we make coincident with the origin of our coordinate system, i.e. z0=0z_0 = 0. As a result, we can simplify the initial beam in Eq. (4) as

E(x,y,z)=E0Un,m(x,y,z)eiωt(33)E(x,y,z)=E_0\mathcal{U}_{n,m}(x,y,z)e^{i\omega t} \tag*{(33)}

where Un,m(x,y,z)\mathcal{U}_{n,m}(x,y,z) is the transverse function representing the HGn,m\mathrm{HG}_{n,m} mode in Eq. (1).

Mode-Mismatch Sensing: Mode-converter Sensing (MCS)

For the MCS scheme, we make use of a π/2\pi/2 mode converter and two 45∘45^\circ-rotated quadrant photodetectors away from the cavity waist by π/4\pi/4 and π/2\pi/2 Gouy phases for the two orthogonal waist size and waist position mismatch degrees of freedom, respectively. This scheme was first introduced by Magaña-Sandoval et al. [26] in the case of HG0,0\mathrm{HG}_{0,0} mode mismatch sensing. In this section, however, we are going to extend this scheme to a generic HGn,m\mathrm{HG}_{n,m} mode mismatch sensing. Similar to the alignment error signal calculation, we start with applying a phase modulation with modulation index mm at a frequency Ω\Omega to the carrier HGn,m\mathrm{HG}_{n,m} mode. Keeping only the first-order sidebands, the field becomes

E=E0Un,meiωt(1+im2(e−iΩt+eiΩt))(34)E=E_0\mathcal{U}_{n,m}e^{i\omega t}\left(1+i\frac{m}{2}\left(e^{-i\Omega t}+e^{i\Omega t}\right)\right) \tag*{(34)}
MCS: waist size mismatch

Let us consider the waist size mismatch degree of freedom first. Upon application of a waist size mismatch between the input beam and the cavity eigenmode, the input beam becomes [16]

Un,mϵ(x,y,z)≈Un,m−iϵ2(AnUn+2,m+BnUn−2,m+AmUn,m+2+BmUn,m−2)(35)\mathcal{U}^{\epsilon}_{n,m}(x,y,z)\approx\mathcal{U}_{n,m}-i\frac{\epsilon}{2}\left(A_n\mathcal{U}_{n+2,m}+B_n\mathcal{U}_{n-2,m}+A_m\mathcal{U}_{n,m+2}+B_m\mathcal{U}_{n,m-2}\right) \tag*{(35)}

where ϵ=ww0−1\epsilon=\frac{w}{w_0}-1 is the relative waist size mismatch, and An=(n+2)(n+1)A_n=\sqrt{(n+2)(n+1)} and Bn=n(n−1)B_n=\sqrt{n(n-1)}. We have also used the fact that the accumulated Gouy phase from the cavity center to the QPD for each mode order is π/4\pi/4 in the above equation. This introduces a factor of eiπ4(±2)=±ie^{i\frac{\pi}{4}(\pm2)}=\pm i for the upper and lower adjacent modes scattered from the original mode by two mode order as the result of mode mismatch. As illustrated above, in the mode mismatch sensing with WFS scheme we use a π/2\pi/2 mode-converter, which causes an additional π/2\pi/2 Gouy phase accumulation and thus a factor of eimπ/2=ie^{im\pi/2}=i for each mode order in the focusing axis (yy axis) while the nonfocusing axis (xx axis) experiences normal Gouy phase accumulation [26]. The above mode mismatched beam after passing through the mode converter becomes

Un,mϵ(x,y,z)≈im(Un,m−iϵ2(AnUn+2,m+BnUn−2,m−AmUn,m+2−BmUn,m−2))(36)\mathcal{U}^{\epsilon}_{n,m}(x,y,z)\approx i^m\left(\mathcal{U}_{n,m}-i\frac{\epsilon}{2}\left(A_n\mathcal{U}_{n+2,m}+B_n\mathcal{U}_{n-2,m}-A_m\mathcal{U}_{n,m+2}-B_m\mathcal{U}_{n,m-2}\right)\right) \tag*{(36)}

where every mode order in the yy direction accumulates one extra factor of eiπ/2=ie^{i\pi/2}=i due to the π/2\pi/2 mode-converter. Similar to the alignment sensing case, we multiply each term in the incoming beam with the corresponding cavity reflectance function to get the reflected beam EreflE_{\mathrm{refl}}. After making the assumption about a high-finesse and completely over-coupled cavity so that only F(ω,n+m)=−1F(\omega,n+m)=-1 while F=1F=1 for all other terms to simplify our analytical results, we obtain

Erefl=E0imeiωt(−Un,m+im2(Un,meiΩt+Un,me−iΩt)−iϵ2(Un+2,mAn−Un,m+2Am+Un−2,mBn−Un,m−2Bm)+ϵm4(Un+2,mAneiΩt+Un+2,mAne−iΩt−Un,m+2AmeiΩt−Un,m+2Ame−iΩt+Un−2,mBneiΩt+Un−2,mBne−iΩt−Un,m−2BmeiΩt−Un,m−2Bme−iΩt))(37)\begin{aligned} E_{\mathrm{refl}}={}&E_0i^m e^{i\omega t}\Bigl(-\mathcal{U}_{n,m}+\frac{im}{2}\bigl(\mathcal{U}_{n,m}e^{i\Omega t}+\mathcal{U}_{n,m}e^{-i\Omega t}\bigr)-\frac{i\epsilon}{2}\bigl(\mathcal{U}_{n+2,m}A_n-\mathcal{U}_{n,m+2}A_m+\mathcal{U}_{n-2,m}B_n-\mathcal{U}_{n,m-2}B_m\bigr)\\ &+\frac{\epsilon m}{4}\bigl(\mathcal{U}_{n+2,m}A_ne^{i\Omega t}+\mathcal{U}_{n+2,m}A_ne^{-i\Omega t}-\mathcal{U}_{n,m+2}A_me^{i\Omega t}-\mathcal{U}_{n,m+2}A_me^{-i\Omega t}+\mathcal{U}_{n-2,m}B_ne^{i\Omega t}+\mathcal{U}_{n-2,m}B_ne^{-i\Omega t}-\mathcal{U}_{n,m-2}B_me^{i\Omega t}-\mathcal{U}_{n,m-2}B_me^{-i\Omega t}\bigr)\Bigr) \tag*{(37)} \end{aligned}

EreflE_{\mathrm{refl}} and Erefl∗E_{\mathrm{refl}}^* have 15 terms each, so there are 225 terms to evaluate in the product EreflErefl∗E_{\mathrm{refl}}E_{\mathrm{refl}}^* for the photocurrent. However, only terms that are odd with respect to y=±xy=\pm x have contributions to the 45∘45^\circ-rotated QPD that we are using. We also know only terms that oscillate with the modulation frequency Ω\Omega have contribution to the photodetector signal after demodulation at the RF modulation frequency Ω\Omega and low-pass filtering. The photocurrent is then

IQPD=E02ϵm(eiΩt+e−iΩt)(−Anβ(n,m),(n+2,m)+Amβ(n,m),(n,m+2)−Bnβ(n,m),(n−2,m)+Bmβ(n,m),(n,m−2))(38)I_{\mathrm{QPD}}=E_0^2\epsilon m\left(e^{i\Omega t}+e^{-i\Omega t}\right)\left(-A_n\beta_{(n,m),(n+2,m)}+A_m\beta_{(n,m),(n,m+2)}-B_n\beta_{(n,m),(n-2,m)}+B_m\beta_{(n,m),(n,m-2)}\right) \tag*{(38)}

where β(n,m),(n′,m′)\beta_{(n,m),(n',m')} are the beat coefficient at a 45∘45^\circ-rotated quadrant photodetector

β(n,m),(n′,m′)=(∫0∞dy∫−yydx+∫−∞0dy∫y−ydx−∫0∞dx∫−xxdy−∫−∞0dx∫x−xdy)Un,m⋅Un′,m′(39)\beta_{(n,m),(n',m')} = \left(\int_{0}^{\infty} dy \int_{-y}^{y} dx + \int_{-\infty}^{0} dy \int_{y}^{-y} dx - \int_{0}^{\infty} dx \int_{-x}^{x} dy - \int_{-\infty}^{0} dx \int_{x}^{-x} dy\right) \mathcal{U}_{n,m}\cdot\mathcal{U}_{n',m'} \tag*{(39)}

One can easily confirm that the beat coefficient β(n,m),(n′,m′)\beta_{(n,m),(n',m')} satisfies

β(n,m),(n′,m′)=β(n′,m′),(n,m)=−β(m,n),(m′,n′)(40)\beta_{(n,m),(n',m')} = \beta_{(n',m'),(n,m)} = -\beta_{(m,n),(m',n')} \tag*{(40)}

Demodulating at phase ϕ=0\phi=0 we obtain the I-phase signal

VMCSΩ,QPD1=E02ϵm(−Anβ(n,m),(n+2,m)+Amβ(n,m),(n,m+2)−Bnβ(n,m),(n−2,m)+Bmβ(n,m),(n,m−2))(41)V_{\mathrm{MCS}}^{\Omega,\mathrm{QPD1}} = E_0^2\epsilon m\left(-A_n\beta_{(n,m),(n+2,m)} + A_m\beta_{(n,m),(n,m+2)} - B_n\beta_{(n,m),(n-2,m)} + B_m\beta_{(n,m),(n,m-2)}\right) \tag*{(41)}

It is sometimes common to use the Rayleigh range zR=πw02λ≈zR0+2πw02λ⋅ϵz_R = \frac{\pi w_0^2}{\lambda} \approx z_{R0} + \frac{2\pi w_0^2}{\lambda}\cdot\epsilon to characterize the amount of waist size mismatch. With the Rayleigh range mismatch δzR=2πw02λ⋅ϵ\delta z_R = \frac{2\pi w_0^2}{\lambda}\cdot\epsilon as the expansion parameter, we can rewrite the error signal as

VMCSΩ,QPD1=E02λm2πw02δzR(−Anβ(n,m),(n+2,m)+Amβ(n,m),(n,m+2)−Bnβ(n,m),(n−2,m)+Bmβ(n,m),(n,m−2))(42)V_{\mathrm{MCS}}^{\Omega,\mathrm{QPD1}} = E_0^2\frac{\lambda m}{2\pi w_0^2}\delta z_R\left(-A_n\beta_{(n,m),(n+2,m)} + A_m\beta_{(n,m),(n,m+2)} - B_n\beta_{(n,m),(n-2,m)} + B_m\beta_{(n,m),(n,m-2)}\right) \tag*{(42)}

We define the relative sensing gain Ωn,m\Omega_{n,m} as the ratio of the error signal for the higher-order mode HGn,m\mathrm{HG}_{n,m} and the fundamental mode HG0,0\mathrm{HG}_{0,0}

Ωn,m=−Anβ(n,m),(n+2,m)+Amβ(n,m),(n,m+2)−Bnβ(n,m),(n−2,m)+Bmβ(n,m),(n,m−2)22β(0,0),(0,2)(43)\Omega_{n,m} = \frac{-A_n\beta_{(n,m),(n+2,m)} + A_m\beta_{(n,m),(n,m+2)} - B_n\beta_{(n,m),(n-2,m)} + B_m\beta_{(n,m),(n,m-2)}}{2\sqrt{2}\beta_{(0,0),(0,2)}} \tag*{(43)}

To make it more precise, let us consider the case where n=mn=m, i.e. we are considering symmetric higher-order HG modes such as HG0,0\mathrm{HG}_{0,0}, HG1,1\mathrm{HG}_{1,1}, HG2,2\mathrm{HG}_{2,2}, HG3,3\mathrm{HG}_{3,3}, etc. As a result, using the properties of β(n,m),(n′,m′)\beta_{(n,m),(n',m')} in Eq. (40), we can simplify the relative sensing gain to

Ωn,n=Anβ(n,n),(n,n+2)+Bnβ(n,n),(n,n−2)2β(0,0),(0,2)(44)\Omega_{n,n} = \frac{A_n\beta_{(n,n),(n,n+2)} + B_n\beta_{(n,n),(n,n-2)}}{\sqrt{2}\beta_{(0,0),(0,2)}} \tag*{(44)}

We thus have an increasing mode mismatch sensing gain for higher-order modes.

MCS: waist position mismatch

Now let’s look at the waist position mismatch error signal. The beam after the waist position displacement δz0\delta z_0, according to the results in paper [16], is

Un,mδz0(x,y,z)≈Un,m−iλδz04πw02(AnUn+2,m+BnUn−2,m+AmUn,m+2+BmUn,m−2−(Cn+Cm)⋅Un,m)(45)\mathcal{U}_{n,m}^{\delta z_0}(x,y,z) \approx\mathcal{U}_{n,m} - i\frac{\lambda\delta z_0}{4\pi w_0^2}\left(A_n\mathcal{U}_{n+2,m} + B_n\mathcal{U}_{n-2,m} + A_m\mathcal{U}_{n,m+2} + B_m\mathcal{U}_{n,m-2} - (C_n+C_m)\cdot\mathcal{U}_{n,m}\right) \tag*{(45)}

where Cn=2n+1C_n=2n+1, and we have used the fact that the accumulated Gouy phase from the cavity waist to the QPD in this path for each mode order is π/2\pi/2. And this brings an extra factor of eiπ2(±2)=−1e^{i\frac{\pi}{2}(\pm2)}=-1 for the offset modes. After the π/2\pi/2 mode-converter, the beam becomes

Un,mδz0(x,y,z)=im(Un,m−iλδz04πw02(AnUn+2,m+BnUn−2,m−AmUn,m+2−BmUn,m−2−(Cn+Cm)⋅Un,m))(46)\mathcal{U}_{n,m}^{\delta z_0}(x,y,z) = i^m\left(\mathcal{U}_{n,m} - i\frac{\lambda\delta z_0}{4\pi w_0^2}\left(A_n\mathcal{U}_{n+2,m} + B_n\mathcal{U}_{n-2,m} - A_m\mathcal{U}_{n,m+2} - B_m\mathcal{U}_{n,m-2} - (C_n+C_m)\cdot\mathcal{U}_{n,m}\right)\right) \tag*{(46)}

Comparing Eqs. (36) and (46) we see that we only need to track two changes going from the waist size mismatch error signal in Eq. (41) to the unknown waist position mismatch error signal we are calculating: (a) we need to replace the Un,m\mathcal{U}_{n,m} coefficient from 1 to (1+i12(Cn+Cm))(1+i\frac{1}{2}(C_n+C_m)); (b) we need to replace the mode mismatch parameter ϵ\epsilon with λδz02πw02\frac{\lambda\delta z_0}{2\pi w_0^2}. The additional term resulting from Replacement (a) is proportional to Un,m\mathcal{U}_{n,m} and thus has to beat with the upper or lower modes, such as Un±2,m\mathcal{U}_{n\pm2,m} and Un,m±2\mathcal{U}_{n,m\pm2} to have a non-vanishing beat signal. This additional beat signal however is of second order in ϵ\epsilon and thus does not affect the linear error signal, as can be seen from Eq. (37). As a result, Replacement (a) has zero net contribution to the photocurrent and subsequently the error signal up to the linear order. Now to obtain the waist position mismatch error signal, one simply needs to make the Replacement (b), which results in

VMCSΩ,QPD2=E02λδz02πw02m(−Anβ(n,m),(n+2,m)+Amβ(n,m),(n,m+2)−Bnβ(n,m),(n−2,m)+Bmβ(n,m),(n,m−2))(47)V_{\mathrm{MCS}}^{\Omega,\mathrm{QPD2}} = E_0^2\frac{\lambda\delta z_0}{2\pi w_0^2}m\left(-A_n\beta_{(n,m),(n+2,m)} + A_m\beta_{(n,m),(n,m+2)} - B_n\beta_{(n,m),(n-2,m)} + B_m\beta_{(n,m),(n,m-2)}\right) \tag*{(47)}

which interestingly has the same slopes as the waist size mismatch error signal in Eq. (42). And we have the same sensing gain as in the waist size sensing case in Eq. (43).

Similar to the alignment WFS case, we want to approximate our exact results in Eqs. (42) and (47) to have a better sense of the scaling relation of the sensing gain with respect to the carrier’s mode order. As in the alignment sensing case, we can simplify Eq. (44) by replacing the complicated higher-order beat coefficients (excluding β(0,0),(0,2)\beta_{(0,0),(0,2)}) with a constant value of 0.41. For more details see IV.

Ωn,n≈0.41⋅(n+1)(n+2)+n(n−1)2β(0,0),(0,2)(48)\Omega_{n,n} \approx0.41\cdot\frac{\sqrt{(n+1)(n+2)}+\sqrt{n(n-1)}}{\sqrt{2}\beta_{(0,0),(0,2)}} \tag*{(48)}

where we have used the definitions for AnA_n and BnB_n.

This approximation is also quite close to the exact results, as shown in Figure 4. We thus conclude that the mode mismatch sensing gain for higher order HGn,n\mathrm{HG}_{n,n} modes in this mode-converter and QPD sensing scheme scales approximately as (n+1)(n+2)+n(n−1)\sqrt{(n+1)(n+2)}+\sqrt{n(n-1)}. We also conducted the corresponding Finesse simulation. As shown in black in Figure 4, they agree extremely well with the exact analytical results in yellow.

Figure 4. Relative mode mismatch sensing gain Ωn,n\Omega_{n,n} in MCS and RFL sensing schemes for symmetric higher-order HGn,n_{n,n} modes. The analytical sensing gains in Eqs. (44) and (57) and the approximated sensing gain in Eq. (48) are shown. The corresponding Finesse simulation results are included as well for comparison.

Mode-Mismatch Sensing: Radio-Frequency Lens Sensing (RFL)

For the RFL mode mismatch sensing scheme, we use an electro-optic lens (EOL) device to impose RF lens modulation sidebands separated from the carrier frequency by twice the higher-order mode difference frequency of the optical cavity, since we need the second order upper modes and lower modes generated from mode mismatch to resonate in the cavity. The scheme requires demodulating the beat signal between the RF lens modulation induced offset mode sidebands and the mode-mismatch-induced carrier frequency offset modes on a single element photodiode in the orthogonal demodulation phases. This can produce simultaneously linear error signals for the waist position and waist size mismatch between the input beam and the cavity eigenmode.

Similar to the mode-converter mode mismatch sensing scheme, we consider a generic HGn,m\mathrm{HG}_{n,m} mode as the carrier in Eq. (33)

U(x,y,z)=E0Un,m(x,y,z)eiωt(49)\mathcal{U}(x,y,z)=E_{0}\mathcal{U}_{n,m}(x,y,z)e^{i\omega t} \tag*{(49)}

We pass the carrier through an EOL device to modulate the curvature of the beam SS. If we consider a sinusoidal oscillation in the wavefront curvature of the beam

S=mScos⁡(Ωt)(50)S=m_{S}\cos(\Omega t) \tag*{(50)}

where mSm_{S} is the RF lens modulation depth. The input beam after the EOL modulator becomes [16]

UEOL≈E0Un,meiωt−ikw02mS16E0(AnUn+2,me−2iΨ1+BnUn−2,me2iΨ1+AmUn,m+2e−2iΨ1+BmUn,m−2e2iΨ1+(Cn+Cm)⋅Un,m)(ei(ω+Ω)t+ei(ω−Ω)t)(51)\begin{aligned} \mathcal{U}^{\mathrm{EOL}} \approx{}& E_{0}\mathcal{U}_{n,m}e^{i\omega t}-i\frac{kw_{0}^{2}m_{S}}{16}E_{0}\bigl(A_{n}\mathcal{U}_{n+2,m}e^{-2i\Psi_{1}}+B_{n}\mathcal{U}_{n-2,m}e^{2i\Psi_{1}}\\ &+A_{m}\mathcal{U}_{n,m+2}e^{-2i\Psi_{1}}+B_{m}\mathcal{U}_{n,m-2}e^{2i\Psi_{1}}+(C_{n}+C_{m})\cdot\mathcal{U}_{n,m}\bigr)\left(e^{i(\omega+\Omega)t}+e^{i(\omega-\Omega)t}\right) \tag*{(51)} \end{aligned}

keeping only the first order terms, where An=(n+2)(n+1)A_{n}=\sqrt{(n+2)(n+1)} and Bn=n(n−1)B_{n}=\sqrt{n(n-1)}.

RFL: waist size mismatch

Now let us consider the RFL waist size mismatch sensing error signal. Upon application of waist size mismatch ϵ=ww0−1\epsilon=\frac{w}{w_{0}}-1, Eq. (51) becomes

Uϵ≈E0Un,meiωt+E0ϵ2(AnUn+2,m−BnUn−2,m+AmUn,m+2−BmUn,m−2)eiωt−ikw02mS16E0(An(Un+2,m+ϵ2(An+2Un+4,m−Bn+2Un,m+AmUn+2,m+2−BmUn+2,m−2))e−2iΨ1+Bn(Un−2,m+ϵ2(An−2Un,m−Bn−2Un−4,m+AmUn−2,m+2−BmUn−2,m−2))e2iΨ1+Am(Un,m+2+ϵ2(AnUn+2,m+2−BnUn−2,m+2+Am+2Un,m+4−Bm+2Un,m))e−2iΨ1+Bm(Un,m−2+ϵ2(AnUn+2,m−2−BnUn−2,m−2+Am−2Un,m−Bm−2Un,m−4))e2iΨ1+(Cn+Cm)(Un,m+ϵ2(AnUn+2,m−BnUn−2,m+AmUn,m+2−BmUn,m−2)))(ei(ω+Ω)t+ei(ω−Ω)t)(52)\begin{aligned} \mathcal{U}^{\epsilon}\approx{}&E_{0}\mathcal{U}_{n,m}e^{i\omega t}+E_{0}\frac{\epsilon}{2}\left(A_{n}\mathcal{U}_{n+2,m}-B_{n}\mathcal{U}_{n-2,m}+A_{m}\mathcal{U}_{n,m+2}-B_{m}\mathcal{U}_{n,m-2}\right)e^{i\omega t}\\ &-i\frac{kw_{0}^{2}m_{S}}{16}E_{0}\Bigl(A_{n}\bigl(\mathcal{U}_{n+2,m}+\frac{\epsilon}{2}(A_{n+2}\mathcal{U}_{n+4,m}-B_{n+2}\mathcal{U}_{n,m}+A_{m}\mathcal{U}_{n+2,m+2}-B_{m}\mathcal{U}_{n+2,m-2})\bigr)e^{-2i\Psi_{1}}\\ &+B_{n}\bigl(\mathcal{U}_{n-2,m}+\frac{\epsilon}{2}(A_{n-2}\mathcal{U}_{n,m}-B_{n-2}\mathcal{U}_{n-4,m}+A_{m}\mathcal{U}_{n-2,m+2}-B_{m}\mathcal{U}_{n-2,m-2})\bigr)e^{2i\Psi_{1}}\\ &+A_{m}\bigl(\mathcal{U}_{n,m+2}+\frac{\epsilon}{2}(A_{n}\mathcal{U}_{n+2,m+2}-B_{n}\mathcal{U}_{n-2,m+2}+A_{m+2}\mathcal{U}_{n,m+4}-B_{m+2}\mathcal{U}_{n,m})\bigr)e^{-2i\Psi_{1}}\\ &+B_{m}\bigl(\mathcal{U}_{n,m-2}+\frac{\epsilon}{2}(A_{n}\mathcal{U}_{n+2,m-2}-B_{n}\mathcal{U}_{n-2,m-2}+A_{m-2}\mathcal{U}_{n,m}-B_{m-2}\mathcal{U}_{n,m-4})\bigr)e^{2i\Psi_{1}}\\ &+(C_{n}+C_{m})\bigl(\mathcal{U}_{n,m}+\frac{\epsilon}{2}(A_{n}\mathcal{U}_{n+2,m}-B_{n}\mathcal{U}_{n-2,m}+A_{m}\mathcal{U}_{n,m+2}-B_{m}\mathcal{U}_{n,m-2})\bigr)\Bigr)\\ &\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\left(e^{i(\omega+\Omega)t}+e^{i(\omega-\Omega)t}\right) \tag*{(52)} \end{aligned}

where similar to RFJ we have omitted the accumulated Gouy phase difference Ψ2\Psi_{2} from the cavity waist to the single element PD because it does not affect the PD signal. The photocurrent on the single-element photodetector is defined as

IPD=∫−∞∞∫−∞∞dx dy Erefl⋅Erefl∗(53)I_{\mathrm{PD}}=\int_{-\infty}^{\infty}\int_{-\infty}^{\infty}dx\,dy\,E_{\mathrm{refl}}\cdot E_{\mathrm{refl}}^{*} \tag*{(53)}

where the reflected field EreflE_{\mathrm{refl}} is obtained by multiplying each term in Eq. (52) with a suitable reflectance function F(ω,n)F(\omega,n). Using the fact that F(ω,n+m)=F(ω+Ω,n+m−2)=F(ω−Ω,n+m−2)=−1F(\omega,n+m)=F(\omega+\Omega,n+m-2)=F(\omega-\Omega,n+m-2)=-1, and F=1F=1 for all other terms for completely over-coupled cavity, the photocurrent becomes

IPD=ikw02mSϵE028(n2+m2+n+m+2)(e2iΨ1eiΩt−e−2iΨ1e−iΩt)(54)I_{\mathrm{PD}}=\frac{ikw_{0}^{2}m_{S}\epsilon E_{0}^{2}}{8}\left(n^{2}+m^{2}+n+m+2\right)\left(e^{2i\Psi_{1}}e^{i\Omega t}-e^{-2i\Psi_{1}}e^{-i\Omega t}\right) \tag*{(54)}

where we have used the orthonormal condition in Eq. (20).

Similar to the RFJ sensing scheme in Eq. (27), we can adjust the overall demodulation phase to be 2Ψ1+π/22\Psi_{1}+\pi/2 (not Ψ1+π/2\Psi_{1}+\pi/2) to eliminate the effect of accumulated Gouy phase difference between different sideband modes and maximize the demodulated signal. This Q-phase error signal is

VRFLΩ,Q=−kw02mS8E02(n2+m2+n+m+2)(55)V_{\mathrm{RFL}}^{\Omega,\mathrm{Q}}=-\frac{k w_{0}^{2}m_{\mathrm{S}}}{8}E_{0}^{2}\left(n^{2}+m^{2}+n+m+2\right) \tag*{(55)}

Using the Rayleigh range mismatch δzR=2πw02λϵ\delta z_{\mathrm{R}}=\frac{2\pi w_{0}^{2}}{\lambda}\epsilon as the expansion parameter, we can rewrite the error signal as

VRFLΩ,Q=−mSE028δzR(n2+m2+n+m+2)(56)V_{\mathrm{RFL}}^{\Omega,\mathrm{Q}}=-\frac{m_{\mathrm{S}}E_{0}^{2}}{8}\delta z_{\mathrm{R}}\left(n^{2}+m^{2}+n+m+2\right) \tag*{(56)}

Thus, the relative waist size sensing gain for HGn,n\mathrm{HG}_{n,n} mode in the RFL scheme is

Ωn,n=n2+n+1(57)\Omega_{n,n}=n^{2}+n+1 \tag*{(57)}
RFL: waist position mismatch

Upon application of a waist position mismatch δz0\delta z_{0} between the input beam and the cavity eigenmode at the cavity waist, the input beam in Eq. (51) becomes

Uδz0≈E0Un,meiωt+E0iλδz04πw02eiωt(AnUn+2,m+BnUn−2,m+AmUn,m+2+BmUn,m−2+(Cn+Cm)Un,m)−ikw02mS16E0(AnUn+2,m+iλδz04πw02(An+2Un+4,m+Bn+2Un,m+AmUn+2,m+2+BmUn+2,m−2+(Cn+2+Cm)Un+2,m)e−2iΨ1+Bn(Un−2,m+iλδz04πw02(An−2Un,m+Bn−2Un−4,m+AmUn−2,m+2+BmUn−2,m−2+(Cn−2+Cm)Un−2,m)e2iΨ1+Am(Un,m+2+iλδz04πw02(AnUn+2,m+2+BnUn−2,m+2+Am+2Un,m+4+Bm+2Un,m+(Cn+Cm+2)Un,m+2)e−2iΨ1+Bm(Un,m−2+iλδz04πw02(AnUn+2,m−2+BnUn−2,m−2+Am−2Un,m+Bm−2Un,m−4+(Cn+Cm−2)Un,m−2)e2iΨ1)+(Cn+Cm)(Un,m+iλδz04πw02(AnUn+2,m+BnUn−2,m+AmUn,m+2+BmUn,m−2+(Cn+Cm)Un,m))))(ei(ω+Ω)t+ei(ω−Ω)t)(58)\begin{aligned} \mathcal{U}^{\delta z_{0}} \approx{}& E_{0}\mathcal{U}_{n,m}e^{i\omega t}+E_{0}\frac{i\lambda\delta z_{0}}{4\pi w_{0}^{2}}e^{i\omega t}(A_{n}\mathcal{U}_{n+2,m}+B_{n}\mathcal{U}_{n-2,m}+A_{m}\mathcal{U}_{n,m+2}+B_{m}\mathcal{U}_{n,m-2}+(C_{n}+C_{m})\mathcal{U}_{n,m})\\ &-i\frac{k w_{0}^{2}m_{\mathrm{S}}}{16}E_{0}(A_{n}\mathcal{U}_{n+2,m}+\frac{i\lambda\delta z_{0}}{4\pi w_{0}^{2}}(A_{n+2}\mathcal{U}_{n+4,m}+B_{n+2}\mathcal{U}_{n,m}+A_{m}\mathcal{U}_{n+2,m+2}+B_{m}\mathcal{U}_{n+2,m-2}+(C_{n+2}+C_{m})\mathcal{U}_{n+2,m})e^{-2i\Psi_{1}}+B_{n}(\mathcal{U}_{n-2,m}+\frac{i\lambda\delta z_{0}}{4\pi w_{0}^{2}}(A_{n-2}\mathcal{U}_{n,m}+B_{n-2}\mathcal{U}_{n-4,m}+A_{m}\mathcal{U}_{n-2,m+2}+B_{m}\mathcal{U}_{n-2,m-2}+(C_{n-2}+C_{m})\mathcal{U}_{n-2,m})e^{2i\Psi_{1}}\\ &+A_{m}(\mathcal{U}_{n,m+2}+\frac{i\lambda\delta z_{0}}{4\pi w_{0}^{2}}(A_{n}\mathcal{U}_{n+2,m+2}+B_{n}\mathcal{U}_{n-2,m+2}+A_{m+2}\mathcal{U}_{n,m+4}+B_{m+2}\mathcal{U}_{n,m}+(C_{n}+C_{m+2})\mathcal{U}_{n,m+2})e^{-2i\Psi_{1}}\\ &+B_{m}(\mathcal{U}_{n,m-2}+\frac{i\lambda\delta z_{0}}{4\pi w_{0}^{2}}(A_{n}\mathcal{U}_{n+2,m-2}+B_{n}\mathcal{U}_{n-2,m-2}+A_{m-2}\mathcal{U}_{n,m}+B_{m-2}\mathcal{U}_{n,m-4}+(C_{n}+C_{m-2})\mathcal{U}_{n,m-2})e^{2i\Psi_{1}})\\ &+(C_{n}+C_{m})(\mathcal{U}_{n,m}+\frac{i\lambda\delta z_{0}}{4\pi w_{0}^{2}}(A_{n}\mathcal{U}_{n+2,m}+B_{n}\mathcal{U}_{n-2,m}+A_{m}\mathcal{U}_{n,m+2}+B_{m}\mathcal{U}_{n,m-2}+(C_{n}+C_{m})\mathcal{U}_{n,m}))))(e^{i(\omega+\Omega)t}+e^{i(\omega-\Omega)t}) \tag*{(58)} \end{aligned}

where similarly we have omitted the accumulated Gouy phase difference for different sidebands Ψ2\Psi_{2}. We can similarly simplify the reflected field EreflE_{\mathrm{refl}} using the fact that F(ω,n+m)=F(ω+Ω,n+m+2)=F(ω−Ω,n+m−2)=−1F(\omega,n+m)=F(\omega+\Omega,n+m+2)=F(\omega-\Omega,n+m-2)=-1, and F=1F=1 for all other terms for completely over-coupled cavity. The PD signal on a single-element PD becomes

IPD=−mSE028δz0(e2iΨ1eiΩt+e−2iΨ1e−iΩt)(n2+m2+n+m+2)(59)I_{\mathrm{PD}}=-\frac{m_{\mathrm{S}}E_{0}^{2}}{8}\delta z_{0}\left(e^{2i\Psi_{1}}e^{i\Omega t}+e^{-2i\Psi_{1}}e^{-i\Omega t}\right)\left(n^{2}+m^{2}+n+m+2\right) \tag*{(59)}

Similarly, we can eliminate the effect of the accumulated Gouy phase difference between different sideband modes by adjusting the overall demodulation phase to be 2Ψ12\Psi_{1} (not Ψ1\Psi_{1}). The result I-phase signal is

VRFLΩ,I=−mSE028δz0(n2+m2+n+m+2)(60)V_{\mathrm{RFL}}^{\Omega,\mathrm{I}}=-\frac{m_{\mathrm{S}}E_{0}^{2}}{8}\delta z_{0}\left(n^{2}+m^{2}+n+m+2\right) \tag*{(60)}

which interestingly has the same slope as the waist size mismatch error signal in Eq. (56).

We have thus seen, similar to the alignment sensing with the RFJ scheme, we can extract the two orthogonal mode mismatch sensing error signals in Eqs. (60) and (55) with a single element photodetector in the orthogonal demodulation phases. The resulting error signals in RFL scheme scale quadratically with respect to the mode order, as shown in Figure 4. And this

Conclusion

quadratic dependence comes from the fact that the “effective” modulation depth for the EOL in Eq. (51) scales roughly as nn. Beating it with the scattered modes caused by the static mode mismatch (whose amplitude also scales roughly as nn) results in the quadratic dependence for the RFL sensing gains, which scales even faster than the MCS technique.

We have offered a detailed analytical derivation for the first time for the alignment and mode mismatch sensing error signals with a generic higher-order HG mode as the carrier. Specifically, we have investigated the traditional wavefront sensing scheme (WFS) and the more recently proposed radio-frequency jitter modulation sensing scheme (RFJ) in § 1.1 for the alignment sensing; and for the mode mismatch sensing, we have studied the mode-converter sensing scheme (MCS) and the radio-frequency lens modulation sensing scheme (RFL) Section III. We have seen analytically for instance the necessity for the π/2\pi/2 accumulated Gouy phase difference between the two split photodetectors for the alignment WFS, and the necessity for the π/2\pi/2 mode-converter and the π/4\pi/4 accumulated Gouy phase difference between the two 45∘45^{\circ} rotated QPDs for the mode mismatch sensing. We have also conducted the corresponding Finesse simulations for the alignment and mode mismatch sensing for higher-order HG modes, and the resultant relative sensing gains agree extremely well with the respective exact analytical results, as shown in Figure 3 and Figure 4.

We have shown that in all schemes the alignment and mode mismatch sensing signals are stronger for higher-order Hermite-Gauss modes, as summarised in Table I for the sensing gains. On the other hand, the shot noise at the photodetector in all of the analyzed sensing schemes is independent of the mode indices of the beam. Therefore the alignment or mode mismatch signal to shot noise ratio increases with mode indices exactly as derived for the sensing gains. In a shot noise-limited sensing regime, and a sensing noise-limited control loop, increasing mode index is therefore seen to reduce the residual alignment or mode mismatch error. This goes at least some way to nullify the downside of increasing mode indices in terms of the reduced alignment and mode mismatch tolerances as described in Refs. [15, 16]. In particular, the increases in the sensing gain with RFJ/L scheme exactly match the decreases in the corresponding tolerances, as shown in the last two rows in Table I. This result shows that for these schemes the downside of using higher-order HG modes with respect to their suffering from excessive misalignment and mode mismatch-induced power losses could be eliminated.

AlignmentMode mismatch
Traditional schemesWFS: n+n+1\sqrt{n}+\sqrt{n+1}MCS: n(n−1)+(n+1)(n+2)\sqrt{n(n-1)}+\sqrt{(n+1)(n+2)}
Beam jitter based schemesRFJ: 2n+12n+1RFL: n2+n+1n^2+n+1
Power loss2n+12n+1n2+n+1n^2+n+1

Table I. Summary of the alignment and mode mismatch sensing gains, and induced power coupling losses in Ref. [16].

Acknowledgments

This work was supported by National Science Foundation grants PHY-1806461 and PHY-2012021.

Appendix: Beat coefficients

Some example beat coefficients βn,n+1\beta_{n,n+1} on a split photodetectors in Eq. (13) and β(n,n),(n,n+2)\beta_{(n,n),(n,n+2)} and β(n,n),(n,n−2)\beta_{(n,n),(n,n-2)} on a quadrant photodetectors in Eq. (39) are shown in Figure 5 and Figure 6. We can see that they both asymptotically approach to some fixed value around 0.64 and 0.41 respectively as nn goes large, which are determined by linearly extrapolating from the end points and finding the intercepts. As a result, we can approximate the exact sensing gains in WFS and MCS in Eqs. (15) and (44) and simply the scaling relations by replacing the complicated higher-order beat coefficients with their rough asymptotic values, i.e. 0.64 for βn,n+1\beta_{n,n+1} and 0.41 for β(n,n),(n,n+2)\beta_{(n,n),(n,n+2)} and β(n,n),(n,n−2)\beta_{(n,n),(n,n-2)}.

Figure 6. Beat coefficients β(n,n),(n,n+2)\beta_{(n,n),(n,n+2)} in red and β(n,n),(n,n−2)\beta_{(n,n),(n,n-2)} in yellow on a 45 degree rotated quadrant photodiode. They both converge to around 0.41 as nn goes large. Notice that β(n,n),(n,n−2)\beta_{(n,n),(n,n-2)} have no definition when n<2n<2, as shown in the yellow line.

References

  1. [1]J. Aasi and et al., Advanced ligo, Classical and Quantum Gravity 32, 074001 (2015).
  2. [2]A. Buikema and et al., Sensitivity and performance of the advanced ligo detectors in the third observing run, Phys. Rev. D 102, 062003 (2020).DOI
  3. [3]F. Acernese et al., Advanced Virgo: a second-generation interferometric gravitational wave detector, Classical and Quantum Gravity 32, 024001 (2014).DOI
  4. [4]M. Evans and et al., A horizon study for cosmic explorer: Science, observatories, and community (2021).arxiv.org/abs/2109.09882
  5. [5]M. Punturo and et al., The Einstein Telescope: a third-generation gravitational wave observatory, Classical and Quantum Gravity 27, 194002 (2010).
  6. [6]B. Mours, E. Tournefier, and J.-Y. Vinet, Thermal noise reduction in interferometric gravitational wave antennas: using higher order TEM modes, Classical and Quantum Gravity 23, 5777 (2006).DOI
  7. [7]J.-Y. Vinet, Reducing thermal effects in mirrors of advanced gravitational wave interferometric detectors, Classical and Quantum Gravity 24, 3897 (2007).DOI
  8. [8]S. Chelkowski, S. Hild, and A. Freise, Prospects of higher-order laguerre-gauss modes in future gravitational wave detectors, Phys. Rev. D 79, 122002 (2009).DOI
  9. [9]P. Fulda, K. Kokeyama, S. Chelkowski, and A. Freise, Experimental demonstration of higher-order laguerre-gauss mode interferometry, Phys. Rev. D 82, 012002 (2010).DOI
  10. [10]C. Bond, P. Fulda, L. Carbone, K. Kokeyama, and F. Andreas, Higher order laguerre-gauss mode degeneracy in realistic, high finesse cavities, Phys. Rev. D 84, 102002 (2011).DOI
  11. [11]T. Hong, J. Miller, H. Yamamoto, Y. Chen, and R. Adhikari, Effects of mirror aberrations on laguerre-gaussian beams in interferometric gravitational-wave detectors, Phys. Rev. D 84, 102001 (2011).DOI
  12. [12]B. Sorazu, P. J. Fulda, B. W. Barr, A. S. Bell, C. Bond, L. Carbone, A. Freise, S. Hild, S. H. Huttner, J. Macarthur, and K. A. Strain, Experimental test of higher-order laguerre–gauss modes in the 10 m glasgow prototype interferometer, Classical and Quantum Gravity 30, 035004 (2013).DOI
  13. [13]L. Tao, A. C. Green, and P. Fulda, Higher-order hermite-gauss modes as a robust flat beam in interferometric gravitational wave detectors (2020), arXiv:2010.04338 [astro-ph.IM].DOI
  14. [14]S. Ast, S. Di Pace, J. P. Millo, M. Pichot, M. Turconi, N. Christensen, and W. Chaibi, Higher-order hermite-gauss modes for gravitational waves detection, Phys. Rev. D 103, 042008 (2021).DOI
  15. [15]A. W. Jones and A. Freise, Increased sensitivity of higher-order laser beams to mode mismatches, Optics Letters 45, 5876 (2020).DOI
  16. [16]L. Tao, J. K. Kelley-Derzon, A. C. Green, and P. Fulda, Power coupling losses for misaligned and mode-mismatched higher-order hermite-gauss modes, Opt. Lett. 46, 2694 (2021).DOI
  17. [17]M. Lassen, V. Delaubert, J. Janousek, K. Wagner, H.-A. Bachor, P. K. Lam, N. Treps, P. Buchhave, C. Fabre, and C. C. Harb, Tools for multimode quantum information: Modulation, detection, and spatial quantum correlations, Phys. Rev. Lett. 98, 083602 (2007).pubmed.ncbi.nlm.nih.gov/17359098
  18. [18]N. Uribe-Patarroyo, A. Fraine, D. S. Simon, O. M. Minaeva, and A. V. Sergienko, Object identification using correlated orbital angular momentum states, Phys. Rev. Lett. 110, 043601 (2013).
  19. [19]A. Freise, G. Heinzel, H. Lück, R. Schilling, B. Willke, and K. Danzmann, Frequency-domain interferometer simulation with higher-order spatial modes, Classical and Quantum Gravity 21, S1067 (2004).DOI
  20. [20]D. D. Brown and A. Freise, FINESSE, http://www.gwoptics.org/finesse (2014), The software and source code is available at http://www.gwoptics.org/finesse.
  21. [21]D. Brown and A. Freise, Pykat (2017), http://www.gwoptics.org/pykat.
  22. [22]D. D. Brown, P. Jones, S. Rowlinson, A. Freise, S. Leavey, A. C. Green, and D. Toyra, PyKat: Python package for modelling precision optical interferometers (2020), arXiv e-print 2004.06270, Submitted to SoftwareX, arXiv:2004.06270.arxiv.org/abs/2004.06270
  23. [23]E. Morrison, B. J. Meers, D. I. Robertson, and H. Ward, Automatic alignment of optical interferometers, Appl. Opt. 33, 5041 (1994).DOI
  24. [24]E. Morrison, B. J. Meers, D. I. Robertson, and H. Ward, Experimental demonstration of an automatic alignment system for optical interferometers, Appl. Opt. 33, 5037 (1994).DOI
  25. [25]P. Fulda, D. Voss, C. Mueller, L. F. Ortega, G. Ciani, G. Mueller, and D. B. Tanner, Alignment sensing for optical cavities using radio-frequency jitter modulation, Appl. Opt. 56, 3879 (2017).DOI
  26. [26]E. Magaña-Sandoval, T. Vo, D. Vander-Hyde, J. R. Sanders, and S. W. Ballmer, Sensing optical cavity mismatch with a mode-converter and quadrant photodiode, Physical Review D 100, 101103 (2019).DOI
  27. [27]A. A. Ciobanu, D. D. Brown, P. J. Veitch, and D. J. Ottaway, Mode matching error signals using radio-frequency beam shape modulation, Appl. Opt. 59, 9884 (2020).DOI
  28. [28]G. Ciani, M. Bazzano, J.-P. Zendri, L. Conti, A. Grimaldi, and M. Valentini, Test of lg10 rf high order mode sensing technique for laser-cavity mode-matching, LSC-Virgo-Kagra Meeting September (2020).
  29. [29]M. Diaz-Ortiz and P. Fulda, An electro-optic lens concept for mode mismatch sensing, LSC-Virgo-Kagra Meeting September (2020).
  30. [30]C. Bond, D. Brown, A. Freise, and K. A. Strain, Interferometer techniques for gravitational-wave detection, Living Reviews in Relativity 19, 10.1007/s41114-016-0002-8 (2017).

Paper details

vlnc:W_n4dbf5zn
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents