---
title: Bichromatic Tweezers in Quantum Control
url: https://www.emergentmind.com/topics/bichromatic-tweezers
type: topic
---

# Bichromatic Tweezers in Quantum Control

Bichromatic tweezers are optical-tweezer configurations that employ two distinct optical frequencies, but the term has two technically different uses in current literature. In neutral-atom qudit work, it denotes a two-color trapping potential engineered so that the combined scalar and tensor light shifts satisfy magic conditions, as proposed for qudits encoded in the \(5s5p\,{}^{3}P_2\) manifold of \({}^{87}\mathrm{Sr}\) [2601.16328]. In molecular tweezer work, the same phrase can instead denote bichromatic operation within an optical tweezer, where one optical transition provides cooling and a second provides fluorescence for background-free detection, as demonstrated for single CaF molecules [2208.12159]. The shared feature is spectral division of labor across two optical frequencies; the physical objective, however, depends on context.

## 1. Terminological scope and conceptual distinction

The literature represented here uses “bichromatic tweezers” in two non-equivalent senses. In one case, the tweezer potential itself is bichromatic: two co-propagating optical fields form a composite trap whose net polarizability is engineered. In the other, the trap is monochromatic but the operation performed inside it is bichromatic: one transition cools while another produces collected fluorescence.

| Context | Bichromatic element | Primary objective |
|---|---|---|
| \({}^{87}\mathrm{Sr}\) qudits | Two co-propagating tweezer wavelengths | Scalar magic trapping and tensor-shift cancellation |
| CaF molecules | Two internal optical transitions in a tweezer array | Background-free imaging with simultaneous cooling |

This distinction is consequential because the dominant Hamiltonian and the dominant error channels are different. For \({}^{87}\mathrm{Sr}\), the central problem is the \(m_F\)-dependent tensor AC Stark structure of a \(J=2\) manifold, which dephases qudits unless the tensor polarizability is suppressed [2601.16328]. For CaF, the central problem is that molecules in shallow tweezers require continuous cooling during imaging, so using the same wavelength for cooling and fluorescence creates detector background; bichromatic operation separates those roles spectrally [2208.12159].

## 2. Bichromatic optical trapping for \({}^{87}\mathrm{Sr}\) qudits

The \({}^{87}\mathrm{Sr}\) proposal is formulated for a ten-dimensional qudit architecture using the nuclear-spin manifold \(I=9/2\). The computational manifolds are chosen as \(g \equiv 5s^2\,{}^{1}S_0\), with \(\ket{0}_g=\ket{g,m_I=-9/2},\dots,\ket{9}_g=\ket{g,m_I=9/2}\), and \(e \equiv 5s5p\,{}^{3}P_2, F=9/2\), with \(\ket{0}_e=\ket{e,m_F=-9/2},\dots,\ket{9}_e=\ket{e,m_F=9/2}\) [2601.16328]. The attraction of the \(5s5p\,{}^{3}P_2\) manifold is its much larger magnetic moment: the ratio of \(g\)-factors is approximately \(1.1\times 10^3\), which makes rf control in \({}^{3}P_2\) much faster.

That advantage is paired with a coherence problem. In \(5s^2\,{}^{1}S_0\), \(J=0\), so tensor light shifts are essentially absent and nuclear-spin coherence is naturally protected. In \(5s5p\,{}^{3}P_2\), \(J=2\), the optical potential acquires a substantial tensor component. For a qudit encoded across many \(\ket{F,m_F}\) states, this tensor Stark effect produces an \(m_F\)-dependent energy landscape, so different qudit basis states accumulate different phases in the trap. The proposal targets exactly this dephasing mechanism.

The paper defines a bichromatic tweezer as a trapping potential formed by two co-propagating optical tweezers of different wavelengths \(\lambda_1\) and \(\lambda_2\), with equal waist \(w_0\), and independently controlled optical powers \(P_1,P_2\), parameterized as
\[
P_{1,2}=(1\pm x)P.
\]
The wavelengths are chosen so that their scalar polarizabilities can be combined to satisfy a magic condition between \(g\) and \(e\), while their tensor polarizabilities in \(e\) are of opposite sign and similar magnitude. The specific physical goal is to suppress differential light shifts across all magnetic sublevels of \(5s5p\,{}^{3}P_2\), enabling scalar magic conditions between the ground state and \(5s5p\,{}^{3}P_2\), and tensor magic conditions for qudits encoded within the metastable manifold [2601.16328].

A central point is that ordinary monochromatic magic trapping is not sufficient for qudits encoded in hyperfine states with \(J\neq 0\). The usual strategy of operating at the tensor magic angle alone becomes impractical because it requires either very large magnetic fields or extremely precise angle control. The bichromatic proposal shifts the strategy from overpowering tensor shifts with bias field to canceling them optically.

## 3. Light-shift structure, magic conditions, and tensor-angle robustness

The \({}^{87}\mathrm{Sr}\) analysis starts from a coordinate-invariant light-shift Hamiltonian decomposed into scalar, vector, and tensor parts. For linear polarization, the vector term is ideally absent, leaving the tensor contribution as the dominant source of qudit dephasing [2601.16328]. The field amplitude at the beam center is
\[
E_0(r=0)=\sqrt{\frac{4P}{\pi c\epsilon_0 w_0^2}},
\]
and for the ideal linearly polarized case the polarization vector is taken as
\[
\boldsymbol{\epsilon}(\beta)=[\sin\beta,\,0,\,\cos\beta],
\]
where \(\beta\) is the angle between the polarization direction and the quantization axis.

The bichromatic construction is expressed through effective scalar and tensor polarizabilities,
\[
\bar \alpha_{i}^{s}(\lambda_1, \lambda_2, x) = (1+x)\alpha_i^s(\lambda_1) + (1-x)\alpha_i^s(\lambda_2)
\]
and
\[
\bar \alpha_{i}^{t}(\lambda_1, \lambda_2, x) = (1+x)\alpha_i^t(\lambda_1) + (1-x)\alpha_i^t(\lambda_2).
\]
The scalar magic condition becomes
\[
\bar \alpha_{e}^{s}(\lambda_1, \lambda_2, x_0)=\bar \alpha_{g}^{s}(\lambda_1, \lambda_2, x_0),
\]
while the tensor magic condition is
\[
\bar \alpha_{e}^{t}(\lambda_1, \lambda_2, x_0)=\bar \alpha_{g}^{t}(\lambda_1, \lambda_2, x_0).
\]
Because \({}^{1}S_0\) has negligible tensor polarizability, the working condition is effectively
\[
\bar \alpha_e^t(\lambda_1,\lambda_2,x_0)\approx 0.
\]

This cancellation mechanism depends on the two colors having tensor polarizabilities of opposite sign. If one color gives positive curvature versus \(m_F\) and the other gives negative curvature of similar size, then the net \(m_F\)-dependent curvature can be nearly zero at the appropriate intensity ratio. The paper explicitly links the possibility of sign change in the tensor polarizability to the \(6j\)-symbol factor and the detuning factor in the hyperfine polarizability sums [2601.16328].

The tensor magic angle is
\[
\beta_0=54.7^\circ,
\]
corresponding to the familiar condition \(3\cos^2\beta-1=0\). In the high-field diagonal limit, the tensor shift reduces to a form proportional to
\[
\alpha^t \,(3\cos^2\beta-1)\,\bigl[3m_F^2-F(F+1)\bigr].
\]
The new feature of the bichromatic scheme is that it suppresses \(\alpha^t\) itself, so operation near \(\beta_0\) becomes robust rather than fragile. The residual angular sensitivity is summarized as
\[
\Bigg | \frac{ \partial E_{\text{LS}}}{\partial \beta} \Bigg | \propto I_0\sin(2\beta)\times \tilde{\alpha}_{i}^{t},
\]
with \(\tilde{\alpha}_{i}^{t}=\alpha_i^t(\lambda)\) for monochromatic trapping and \(\tilde{\alpha}_{i}^{t}=\bar{\alpha}_{i}^{t}(\lambda_1,\lambda_2,x)\) for bichromatic trapping. Reducing \(\bar\alpha^t\) therefore directly reduces sensitivity to angle errors.

The paper argues that, with opposite-sign tensor shifts from the two colors, a tuned power ratio \(x_0\), and operation near \(\beta_0=54.7^\circ\), the remaining light shift becomes second-order sensitive to power and angle fluctuations, and off-diagonal terms of \(H_{\text{LS}}\) become negligible in practical magnetic fields [2601.16328].

## 4. Wavelength pairs, operating parameters, and coherence protection in \({}^{87}\mathrm{Sr}\)

Two concrete bichromatic pairs are proposed. The proof-of-principle pair is \(891.5\,\mathrm{nm}\) and \(518.0\,\mathrm{nm}\), described as two suitable scalar magic wavelengths for \(g\) and \(e\). The more experimentally motivated pair is \(813.5\,\mathrm{nm}\) and \(521.3\,\mathrm{nm}\), where \(813.46504\,\mathrm{nm}\) is used as the calculated \({}^{1}S_0\leftrightarrow{}^{3}P_2\) magic wavelength from the model, and the shorter wavelength is chosen because its tensor polarizability in \({}^{3}P_2\) has the opposite sign to that at \(813.5\,\mathrm{nm}\) [2601.16328].

| Wavelength pair | \(x_0\) and example powers | Trap parameters |
|---|---|---|
| \(891.5\,\mathrm{nm} + 518.0\,\mathrm{nm}\) | \(x_0=0.2573\); \(P_1=1.257\,\mathrm{mW}\), \(P_2=0.743\,\mathrm{mW}\) for \(P=1\,\mathrm{mW}\) | Trap depth \(144\,\mu\mathrm{K}\) |
| \(813.5\,\mathrm{nm} + 521.3\,\mathrm{nm}\) | \(x_0=7.467\times 10^{-4}\); \(P_1=1.001\,\mathrm{mW}\), \(P_2=0.999\,\mathrm{mW}\) | Trap depth \(164.56\,\mu\mathrm{K}\); \(\omega_r=53.33\,\mathrm{kHz}\), \(\omega_a=5.09\,\mathrm{kHz}\) |

The \(813.5/521.3\) configuration is especially notable because \(813.5\,\mathrm{nm}\) can remain the readout or detection tweezer wavelength. In this arrangement, the readout stage is achieved simply by extinguishing the \(521.3\,\mathrm{nm}\) beam. The appendix attributes the opposite tensor signs to different dominant intermediate-state contributions: at \(813.5\,\mathrm{nm}\), the \({}^{3}P_2\) tensor polarizability is dominated mainly by the \(5s6s\,{}^{3}S_1\) contribution, while near \(521.3\,\mathrm{nm}\) the strong \(5p^2\,{}^{3}P_2\) matrix element dominates the sign and magnitude [2601.16328].

For target qudit fidelity \(\mathcal F=0.999\) at \(\delta\beta=1^\circ\), the required fractional power uncertainties are given as \(\delta P_1/P_1=4.2\times10^{-3}\) and \(\delta P_2/P_2=7.1\times10^{-3}\) for \(891.5/518.0\,\mathrm{nm}\), and \(\delta P_1/P_1=4.2\times10^{-3}\) and \(\delta P_2/P_2=4.3\times10^{-3}\) for \(813.5/521.3\,\mathrm{nm}\). A wavelength uncertainty of about \(100\,\mathrm{MHz}\) for the short-wavelength beam near \(500\,\mathrm{nm}\) is stated not to materially affect the target fidelity.

Coherence protection is quantified through a Hilbert-Schmidt fidelity,
\[
\mathcal{F} = \frac{1}{d(d+1)} \Bigg( \left|\Tr\!\left( e^{i H_{\text{Target}}t} e^{-i H_{\text{Realized}}t} \right)\right|^2 + d \Bigg),
\]
with \(d=10\), \(H_{\text{Target}}=H_{\text{Zeeman}}\), and \(H_{\text{Realized}}=H_{\text{LS}}+H_{\text{Zeeman}}\). The operational target is \(\mathcal F=0.999\) over a \(1\,\mathrm{ms}\) evolution time. The proposal identifies \(B\sim 1\text{--}5\,\mathrm{G}\) as the practical perturbative regime, with \(B\lesssim 5\,\mathrm{G}\) especially favorable. Larger fields eventually introduce appreciable quadratic Zeeman curvature,
\[
\Delta^2_B(m_F,B)=(188.7\,\mathrm{Hz/G^2})\,m_F^2\,B^2.
\]

The contrast with monochromatic trapping is sharp. In a monochromatic \({}^{3}P_2\) tweezer, the paper quotes a tensor light-shift scale around \(0.3\,\mathrm{MHz}\) for a \(1\,\mathrm{mW}\), \(w_0=1.0\,\mu\mathrm{m}\) trap at \(813.5\,\mathrm{nm}\), large enough to cause rapid qudit dephasing. It further states that monochromatic magic-angle trapping would require fields of order \(10^3\,\mathrm{G}\) and angular precision \(\delta\beta=2\times10^{-5}\,\mathrm{rad}\), whereas bichromatic cancellation makes \(\delta\beta=2\times10^{-2}\,\mathrm{rad}\approx 1^\circ\) viable [2601.16328].

## 5. Bichromatic imaging inside molecular optical tweezers

In the CaF experiment, “bichromatic” refers not to a two-color trapping potential but to a two-transition imaging protocol implemented inside a monochromatic optical tweezer array [2208.12159]. Single CaF molecules in the electronic ground state \(X^2\Sigma(v=0)\) are trapped in a linear array of 20 optical tweezers formed at \(781\,\mathrm{nm}\), with waist \(w_0=720(14)\,\mathrm{nm}\), ground-state trap depth \(V=k_B\times1.28(11)\,\mathrm{mK}\), and microscope objective NA \(0.65\).

The bichromatic method uses the \(X^2\Sigma(v=0,N=1)\rightarrow A^2\Pi_{1/2}(v=0,J=1/2,+)\) transition at \(606\,\mathrm{nm}\) for \(\Lambda\)-cooling, and the \(X^2\Sigma(v=0,N=1,F=0)\rightarrow B^2\Sigma(v=0,N=0)\) transition at \(531\,\mathrm{nm}\) for fluorescence generation. The camera path includes dichroics that reject the \(606\,\mathrm{nm}\) cooling light and pass the \(531\,\mathrm{nm}\) fluorescence, producing background-free imaging. This scheme addresses a molecular-specific difficulty: molecules in shallow tweezers cannot scatter many photons without simultaneous cooling, so using one transition for both cooling and fluorescence leads to severe stray-light background.

The cycling structure is supported by vibrational repumpers on
\[
X^2\Sigma(v=1,2,3,N=1)\rightarrow A^2\Pi_{1/2}(v=0,1,2,J=1/2,+),
\]
and by a rotational repumper on
\[
X^2\Sigma(v=0,N=3)\rightarrow B^2\Sigma(v=0,N=2).
\]
A major bichromatic-specific loss channel is the two-photon decay \(B\rightarrow A\rightarrow X\), which populates opposite-parity rotational states \(X^2\Sigma(v=0,N=0,2)\). The measured \(B\rightarrow A\) branching ratio is \(7.9(16)\times10^{-5}\), from which the authors infer \(|d_{AB}| = 0.30(3)\, e a_0\) [2208.12159].

The tweezer light strongly perturbs the excited states. Although the \(781\,\mathrm{nm}\) trap is far detuned from optical transitions out of \(X^2\Sigma(v=0)\), it lies relatively near the \(B\!-\!C\) transitions. At \(781.1\,\mathrm{nm}\), the inferred peak differential AC Stark shift at the tweezer center is
\[
\Delta_{XB}=h\times78(2)\,\mathrm{MHz},
\]
substantially smaller than the quoted theoretical expectation \(\Delta^{\mathrm{th}}_{XB}\approx h\times600\,\mathrm{MHz}\). By scanning the tweezer wavelength around \(781\,\mathrm{nm}\), the experiment extracts \(|d_{BC}| = 2.0(1)\,e a_0\), smaller than the theoretical prediction \(|d_{BC}^{\mathrm{th}}| = 7.33\,e a_0\). At \(782.0\,\mathrm{nm}\), minimal dependence of \(\Delta_{XB}\) on intensity is observed, indicating a near-magic or near-cancellation point for the \(X\) and \(B\) polarizabilities [2208.12159].

Imaging performance is defined through histogram-based state discrimination. With
\[
\epsilon(p)=p\,\epsilon_{10} + (1-p)\,\epsilon_{01},
\]
the imaging fidelity is
\[
f = 1-\epsilon_{\mathrm{opt}}(0.5).
\]
The measured value is \(f=0.977(2)\), or \(97.7(2)\%\), with optimal threshold \(\theta_{\mathrm{opt}}\approx5100\). The non-destructive detection fidelity \(f_{\mathrm{ND}}\), measured using a \(10\,\mathrm{ms}\) first bichromatic image followed by a \(30\,\mathrm{ms}\) verification image, reaches \(95.5(6)\%\) at \(15\%\) data rejection. The overall collection or detection efficiency is \(0.049(2)\); at optimal parameters for a \(30\,\mathrm{ms}\) image, approximately 35 photons are collected per molecule against approximately 0.3 background photons per tweezer region. At fixed \(X\!-\!B\) intensity \(I=12(1)\,\mathrm{mW/cm^2}\), the imaging lifetime is \(\tau=148(14)\,\mathrm{ms}\) and the \(X\!-\!B\) scattering rate is \(\Gamma_{\mathrm{sc}}=1.95(4)\times10^4\,\mathrm{s^{-1}}\) [2208.12159].

## 6. Error channels, limitations, and broader significance

The two branches of the subject have different limitation profiles. In \({}^{87}\mathrm{Sr}\), the main limitations include quadratic Zeeman curvature, Raman and Rayleigh scattering, photoionization near \(521\,\mathrm{nm}\), and vector shifts from high-NA focusing [2601.16328]. Raman scattering within \({}^{3}P_2\) and to other metastable \({}^{3}P_J\) states remains present at roughly \(10^{-2}\) to \(10^{-1}\) photons/s for individual channels. Rayleigh scattering is around a few photons/s, while the effective decoherence contribution is much smaller, around \(10^{-2}\)–\(10^{-1}\) photons/s depending on \(m_F\) and wavelength pair. For the \(813.5/521.3\,\mathrm{nm}\) pair, extrapolation from prior work yields an estimated two-photon photoionization rate of about \(3.8\,\mathrm{s^{-1}}\) for the \(521.3\,\mathrm{nm}\) contribution. Polarization gradients from high-NA focusing can create vector light shifts of order \(\pm1\,\mathrm{kHz}\) per \(m_F\) at the edge of the motional wavepacket, though these are small at the trap center and may be mitigated by dynamical decoupling.

In CaF imaging, the limitation profile is dominated by state-changing loss and trap-dressed excited-state physics rather than by photon shot noise alone [2208.12159]. The figure of merit
\[
\eta=\Gamma_{\mathrm{sc}}\tau
\]
is measured to be \(2.7(2)\times10^3\) at the operating point and saturates to \(\eta\approx3\times10^3\) at high \(X\!-\!B\) power, compared with \(\eta\approx40\times10^3\) for single-color \(\Lambda\)-imaging. The excess bichromatic loss rate,
\[
\gamma_{\mathrm{excess}}=\gamma(I)-\gamma(0),
\]
grows linearly with \(X\!-\!B\) intensity. Heating is ruled out as the dominant explanation by thermometry: after \(100\,\mathrm{ms}\), the temperature is \(123(20)\,\mu\mathrm{K}\) after \(\Lambda\)-cooling only and \(90(20)\,\mu\mathrm{K}\) after bichromatic imaging, while \(X\!-\!B\) light without \(\Lambda\)-cooling heats the molecules to \(310(30)\,\mu\mathrm{K}\) in \(10\,\mathrm{ms}\). The identified parity-changing loss plus \(C\)-state admixture loss account for only about \(\sim30\%\) of the observed excess loss away from the dispersive \(B\!-\!C(v=3)\) features, leaving open additional mechanisms such as \(C\)-state predissociation, photoionization, or off-resonant excitation to high-lying Rydberg states.

A common misconception is to treat “bichromatic tweezers” as a single standardized method. The available literature instead shows two separate methodological families. In \({}^{87}\mathrm{Sr}\), bichromaticity is used to decouple scalar and tensor trapping requirements in a high-spin hyperfine manifold, thereby supporting qudit quantum computing, quantum sensing, quantum simulation, and array assembly with coherence preservation [2601.16328]. In CaF, bichromaticity spectrally separates cooling and readout, enabling small-scale rearrangeable molecular tweezer arrays with background-free detection relevant to dipolar quantum simulation, molecular quantum information, and precision measurement [2208.12159].

Taken together, these works show that the unifying principle of bichromatic tweezer methods is not a single hardware template but controlled spectral partitioning of functions that cannot be satisfied simultaneously by one optical frequency. In neutral atoms, the partition is between scalar and tensor light-shift engineering. In molecules, it is between cooling and fluorescence detection.

Source: https://www.emergentmind.com/topics/bichromatic-tweezers