betaFeedback

WorksPhysics and Astronomy

Minimization of micromotion for nanoparticles in a Paul trap

Abstract

When a charged particle in a Paul trap is displaced from the node of the AC trapping field, excess micromotion arises as an undesired effect. Excess micromotion heats the particle, limits the precision with which the particle can be localised, and acts as a decoherence channel in quantum mechanical experiments. However, thus far there is no standard procedure for micromotion compensation with mesoscopic particles. Here, we experimentally demonstrate three different methods for minimizing the micromotion of a nanoparticle in a linear Paul trap along three axes. The most precise method allows us to nullify the stray field to within 2.9 V/m, which is comparable to reported values in trapped-ion experiments.

Background

Micromotion in Paul traps

Paul traps[1] are a well-developed platform for trapping charged particles: electrons[2], single atomic ions and crystals thereof[3], molecular ions[4, 5, 6, 7], highly charged ions[8], nanoparticles[9, 10, 11, 12], and microparticles[13, 14]. A particle in a Paul trap exhibits both secular motion and micromotion[15]; excess micromotion arises when the stray field, which we define as the total DC field at the node of the AC trapping field, is nonzero. Excess micromotion heats the particle, prevents high-precision localisation of the particle, and limits coherence in quantum mechanical experiments. In order to nullify the stray field and thus the excess micromotion, DC voltages are applied to compensation electrodes. Methods for detecting micromotion with trapped ions are based on a range of techniques, including resolved sidebands[16, 17, 18], photon correlations[15, 16], parametric excitation[16, 19, 20, 21, 22] and RF power variation[23, 24]. The choice of technique depends on the context[16].

In recent years, Paul traps have emerged as a leading platform for levitodynamical experiments with nano- and microparticles[9, 10, 11, 12, 13, 14]. Excess micromotion introduces the same problems for levitodynamical experiments as it does for trapped-ion experiments[25, 27, 28, 29] [26]. The micromotion detection methods developed for trapped ions are not in general directly transferable, due to the absence of a spectroscopic transition in mesoscopic particles. Thus far, there is no standard procedure for minimizing micromotion within the levitodynamical community, although we note preliminary studies on the topic[9, 25, 30, 31, 32].

Here, we present and characterize three methods to minimize nanoparticle micromotion along three axes in linear Paul traps. The tickling method, based on the response of the particle to a resonant modulation of the trapping potential, yields the highest compensation precision and is the most convenient to implement in the laboratory. The radial stray field after using this method is (0.0±2.9) V m−1(0.0 \pm2.9)\ \mathrm{V\,m^{-1}}. We use this method to characterize different contributions to the stray field in our experimental system. We then implement the tickling method in two additional systems, demonstrating the broad applicability of this method for micromotion compensation in a levitodynamics context.

Micromotion occurs in periodically driven systems such as Paul traps. A Paul trap uses AC and DC electric fields to confine a charged particle dynamically. The position components of the particle’s center of mass ui(t)u_i(t) for i∈{x,y,z}i \in\{x, y, z\} satisfy the Mathieu equation[15]. The lowest-order solution, valid for Mathieu parameters ∣ai∣≪1|a_i| \ll1, qi2≪1q_i^2 \ll1, is

ui(t)=[u0i+u1icos⁡(ωit+Φi)][1+qi2cos⁡(Ωt)],(1)u_i(t) = [u_{0i} + u_{1i}\cos(\omega_i t + \Phi_i)]\left[1 + \frac{q_i}{2}\cos(\Omega t)\right], \tag*{(1)}

where u0iu_{0i} is a component of the equilibrium displacement from the AC field minimum u0u_0, u1iu_{1i} and ωi\omega_i are the amplitude and frequency of the secular motion, Φi\Phi_i is a phase dependent on the initial conditions, and Ω\Omega is the frequency of the AC field. The secular frequencies can be approximated as ωi≈Ω2ai+12qi2\omega_i \approx\frac{\Omega}{2}\sqrt{a_i + \frac{1}{2}q_i^2}.

Equation (1) contains a secular motion term at ωi\omega_i with amplitude u1iu_{1i}, two intrinsic micromotion terms at Ω±ωi\Omega\pm\omega_i with amplitude 14u1iqi\frac{1}{4}u_{1i}q_i, and an excess micromotion term at Ω\Omega with amplitude 12u0iqi\frac{1}{2}u_{0i}q_i. The amplitude of the intrinsic micromotion can be reduced by cooling the secular mode or by lowering the qiq_i parameter, e.g., by increasing Ω\Omega. However, both u1iu_{1i} and qiq_i are strictly nonzero, and therefore intrinsic micromotion is always present.

Excess micromotion is the subject of this work. For a given set of trap and particle parameters (fixed qiq_i), u0iu_{0i} needs to be minimized to minimize the amplitude of excess micromotion along the ii-axis. This is done by counteracting the stray field component that causes u0i≠0u_{0i} \ne0.

Furthermore, if in the linear Paul trap shown in Figure 1 we consider a nonzero phase shift Φac≪1\Phi_{\mathrm{ac}} \ll1 between the AC driven electrodes, an additional excess micromotion term −qi4r0αΦacsin⁡(Ωt)-\frac{q_i}{4}r_0\alpha\Phi_{\mathrm{ac}}\sin(\Omega t) is present in ui(t)u_i(t), where r0r_0 is a characteristic length, and α\alpha is a geometric factor[15]. Such a phase shift arises through experimental imperfections but is typically negligible, as we assume in this work.

Figure 1. Schematic of a linear Paul trap. One grounded electrode pair and one AC-driven electrode pair constitute the radial trapping electrodes. These electrodes are shown in blue, endcaps in orange, and compensation electrodes in pink and red. We displace a particle in three dimensions by varying the DC potentials ϕ~top\tilde{\phi}_{\mathrm{top}}, ϕ~bottom\tilde{\phi}_{\mathrm{bottom}}, ϕ~x′\tilde{\phi}_{x'}, ϕ~y′\tilde{\phi}_{y'}. Left: Side projection of the trap onto the z−x′z-x' plane. The distance between the surfaces of the top and bottom endcaps is 2z02z_0. Right: Top projection of the trap onto the x−yx-y plane, including compensation electrodes. The distance between the surfaces of opposing radial trapping electrodes is 2r02r_0. The amplitude of the applied AC potential between the pairs of radial trapping electrodes is VrfV_{\mathrm{rf}}. We include a phase shift Φac\Phi_{\mathrm{ac}} between the potentials at the two AC-driven electrodes, assumed in this work to be negligible.

Excess micromotion minimization in trapped-ion systems

Here we outline common methods for micromotion minimization in the trapped-ion community[15]. These serve as a comparison point to the methods introduced in this work for micromotion minimization for nanoparticles. For brevity, we label them Methods I–IV.

Method I is the resolved sideband method. The amplitude of excess micromotion is determined by measuring the amplitude of the micromotion sideband relative to the carrier through spectroscopy of the atomic transition[17]. Some reported stray field components[33] after compensation with this method are (0±10) V m−1(0 \pm10)\ \mathrm{V\,m}^{-1}[18] and (0.6±0.6) V m−1(0.6 \pm0.6)\ \mathrm{V\,m}^{-1}[16].

Method II is the photon-correlation or cross-correlation method[15]. The amplitude of excess micromotion and the phase with respect to the AC drive are extracted from the time delays between detected fluorescence and zero crossings of the drive. Reference[16] reported a stray field component after compensation with this method of (0.09±0.09) V m−1(0.09 \pm0.09)\ \mathrm{V\,m}^{-1}.

Method III is the parametric resonance or tickling method[16, 19, 20, 21, 22]. The amplitude of excess micromotion is extracted from the fluorescence response to a modulation of the drive at ωm=2ωi/n\omega_m = 2\omega_i/n for n∈Nn \in\mathbb{N}[19]. Method III allows for micromotion compensation along three axes with just one laser. Some reported stray field components after compensation with this method are (0±6) V m−1(0 \pm6)\ \mathrm{V\,m}^{-1}[34] and (0.0±0.3) V m−1(0.0 \pm0.3)\ \mathrm{V\,m}^{-1}[16].

Other methods to minimize micromotion in trapped-ion systems exist, such as RF power variation[23, 24], using ultracold clouds of neutral atoms[35], using high finesse cavities[36], or analyzing the ion’s trajectory[37]. Combinations of the above methods are also used in situations in which laser access is restricted[34]. Micromotion minimization can be sped up by using machine learning[38].

Excess micromotion minimization for nanoparticles

In this section we introduce three methods that were used to detect and subsequently compensate for micromotion in our system. The methods described here are based on detecting the micromotion amplitude directly (Method 1), measuring the phase of the micromotion (Method 2) and mapping the trapping field via tickling (Method 3), and are deliberately indexed such as to show similarity to the trapped-ion Methods I–III.

Experimental setup

The linear Paul trap used in this work is shown schematically in Figure 1. The distance between the surfaces of opposing AC-driven electrodes is 2r0=1.8 mm2r_0 = 1.8\,\mathrm{mm}, and the distance between the surfaces of the endcaps is 2z0=3.2 mm2z_0 = 3.2\,\mathrm{mm}. The trap axis (zz-axis) is vertical. We trapped a silica sphere with a nominal diameter of 300 nm, at a pressure of around 10−210^{-2} mbar, with a drive frequency Ω/2π=7\Omega/2\pi= 7 kHz and with a peak-to-peak amplitude 2Vrf=4502V_{\mathrm{rf}} = 450 V. The mass of the particle was not determined. A voltage of 70 V was applied to both endcaps before any deliberate displacement of the particle; this value was high enough to render the axial displacement due to gravity negligible. These parameters led to eigenfrequencies (ωz,ωx,ωy)/2π=(610,798,886)(\omega_z,\omega_x,\omega_y)/2\pi= (610,798,886) Hz. All data were taken with the same particle, using these trap parameters, and without center-of-mass cooling, unless explicitly stated.

The particle was displaced in the radial plane by applying DC voltages (ϕ~x′,ϕ~y′)(\tilde{\phi}_{x'},\tilde{\phi}_{y'}) to two pairs of compensation electrodes (see Figure 1). These potentials had a negligible effect on the trap stiffness in any direction. The particle was displaced along the trap axis by changing the relative endcap potential ϕ~z=ϕ~bottom−ϕ~top\tilde{\phi}_z = \tilde{\phi}_{\mathrm{bottom}} - \tilde{\phi}_{\mathrm{top}}, where ϕ~bottom\tilde{\phi}_{\mathrm{bottom}} and ϕ~top\tilde{\phi}_{\mathrm{top}} were the potentials applied to the bottom and top endcaps (see Appendix A). The DC fields that resulted from applying these potentials were close to orthogonal (see Appendix B). Experimentally, the values (ϕx′,ϕy′,ϕbottom,ϕtop)(\phi_{x'},\phi_{y'},\phi_{\mathrm{bottom}},\phi_{\mathrm{top}}) were the setpoints of the DC source (EHS 8220n, iseg Spezialelektronik GmbH); these voltages were subsequently stepped down by a factor of approximately 1/61/6 at a low-pass filtering stage before the electrodes, such that ϕ~k≈ϕk/6\tilde{\phi}_k \approx\phi_k/6 for every setpoint ϕk\phi_k.

The particle’s motion was detected optically, via three separate confocal detection systems that collect light scattered from the particle[39]. Each system was optimized to be most sensitive to motion along only one of the x′x',y′y' and z′z' axes, which form a basis rotated 45∘45^\circ about the trap axis relative to the eigenbasis (see Figure 1). The detection systems were not calibrated, so measured motional amplitudes are given in decibels relative to a value of 1 V2/Hz1\ \mathrm{V}^2/\mathrm{Hz}. A typical power spectral density (PSD) of the detected signal before micromotion compensation is shown in Figure 2.

Figure 2. A typical PSD before micromotion compensation, obtained from the detection system most sensitive to the x′x' projection of the particle’s motion. Motional peaks are visible at ωi\omega_i, Ω\Omega, Ω±ωi\Omega\pm\omega_i, i∈(x,y)i \in(x,y). The peak at Ω+ωx\Omega+\omega_x is barely distinguishable from the noise floor, and the peak at 4141 Hz is electronic noise.

Method 1: Direct amplitude detection

In Method 1, we extract the amplitude Ai′∈{x′,y′,z′}A_{i'}\in\{x',y',z'\} of the excess micromotion peak at Ω\Omega from the measured PSD; see, for example, Figure 2, in which Ax′=−32A_{x'}=-32 dB. Due to the quadrupole symmetry in the trap’s radial plane, Ax′A_{x'} is minimized at u0y′≡u⃗0⋅y^′=0u_{0y'}\equiv\vec{u}_0\cdot\hat{y}'=0; analogously, Ay′A_{y'} is minimized at u0x′≡u⃗0⋅x^′=0u_{0x'}\equiv\vec{u}_0\cdot\hat{x}'=0. The radial micromotion minimum is extracted from measurements of Ax′A_{x'} and Ay′A_{y'} for different (ϕx′,ϕy′)(\phi_{x'},\phi_{y'}) pairs, keeping ϕz′\phi_{z'} constant. The profile of Ax′+Ay′A_{x'}+A_{y'} is expected to resemble that of ∣Ex′∣+∣Ey′∣|E_{x'}|+|E_{y'}|, where Ei′E_{i'} is the projection of the AC field along the i′i' axis. The profile of ∣Ex′∣+∣Ey′∣|E_{x'}|+|E_{y'}| for an ideal quadrupole field is shown in Figure 3(a). The axial micromotion minimum (at u0z′≡u⃗0⋅z^′=0u_{0z'}\equiv\vec{u}_0\cdot\hat{z}'=0) is extracted from a measurement of Az′A_{z'} for different ϕz′\phi_{z'}, keeping (ϕx′,ϕy′)(\phi_{x'},\phi_{y'}) constant.

Figure 3. Simulations of an ideal quadrupole field E⃗\vec{E} generated by the radial trapping electrodes in Figure 1. The simulations are centered at the node for a given time; displacement of a trapped particle from the node causes excess micromotion. Equipotential lines are shown in white. (a) Sum of the absolute values of the projections Ex′≡E⃗⋅x^′E_{x'} \equiv\vec{E}\cdot\hat{x}' and Ey′≡E⃗⋅y^′E_{y'} \equiv\vec{E}\cdot\hat{y}'. (b) Sum of the signs of the projections Ex′E_{x'} and Ey′E_{y'}. (c) Sum of the absolute values of the projections Ex≡E⃗⋅x^E_x \equiv\vec{E}\cdot\hat{x} and Ey≡E⃗⋅y^E_y \equiv\vec{E}\cdot\hat{y}.

For micromotion minimization, Method 1 is intuitive and easy to implement along one axis. However, minimization along three axes requires three detection systems, each sensitive to motion along one axis, for the best results. Additionally, drifts in the detection sensitivity can skew the results obtained via Method 1.

Method 2: Direct phase detection

In Method 2, we extract the phase Φi′∈{x′,y′,z′}\Phi_{i'}\in\{x',y',z'\} of the micromotion relative to the AC drive. Due to the quadrupole symmetry in the trap’s radial plane, Φx′\Phi_{x'} undergoes a π\pi shift at u0y′=0u_{0y'}=0; analogously, Φy′\Phi_{y'} undergoes a π\pi shift at u0x′=0u_{0x'}=0. The radial micromotion minimum is extracted from measurements of Φx′\Phi_{x'} and Φy′\Phi_{y'} for different (ϕx′,ϕy′)(\phi_{x'},\phi_{y'}) pairs, keeping ϕz′\phi_{z'} constant. The profile of (Φx′+Φy′)(mod2π)(\Phi_{x'}+\Phi_{y'})\pmod{2\pi} is expected to resemble that of sgn⁡Ex′+sgn⁡Ey′\operatorname{sgn} E_{x'}+\operatorname{sgn} E_{y'}. The profile of sgn⁡Ex′+sgn⁡Ey′\operatorname{sgn} E_{x'}+\operatorname{sgn} E_{y'} for an ideal quadrupole field is shown in Figure 3(b). The axial micromotion minimum (at u0z′=0u_{0z'}=0) is extracted from a measurement of Φz′\Phi_{z'} for different ϕz′\phi_{z'}, keeping (ϕx′,ϕy′)(\phi_{x'},\phi_{y'}) constant.

For micromotion minimization, Method 2 does not require the detection to be calibrated for any drifts. However, as for Method 1, this method requires three detection systems, each sensitive to motion along one axis, for best results.

Method 3: Resonant tickling

In Method 3, we modulate the amplitude of the AC trap drive. Resonant excitation of mode i∈{x,y,z}i\in\{x,y,z\} occurs when the modulation frequency Ωtickle,i\Omega_{\mathrm{tickle},i} is equal to Ω±ωi\Omega\pm\omega_i (see Appendix § 1.1.3) and there is a nonzero projection of the AC field along i^\hat{i}. The amplitude of the AC potential is modulated according to

Vrf(t)=Vrf(0)(1+∑ihisin⁡(Ωtickle,it)),(2)V_{\mathrm{rf}}(t)=V_{\mathrm{rf}(0)}\left(1+\sum_i h_i\sin(\Omega_{\mathrm{tickle},i}t)\right), \tag*{(2)}

where hi≪1h_i\ll1 is the modulation depth at Ωtickle,i\Omega_{\mathrm{tickle},i}. The potential applied at the AC-driven electrodes is then Vrf(t)sin⁡(Ωt)V_{\mathrm{rf}}(t)\sin(\Omega t). Equation (2) shows that, while not strictly necessary, Method 3 allows for simultaneous tickling of multiple modes.

From the measured PSD, we extract the amplitude UiU_i of the secular-motion peak at ωi\omega_i; see, for example, Figure 2, where Ux=−25U_x=-25 dB. Due to the quadrupole symmetry in the trap’s radial plane, UxU_x is minimized during tickling of the xx-mode at u0x=0u_{0x}=0; analogously, UyU_y is minimized during tickling of the yy-mode at u0y=0u_{0y}=0. The radial micromotion minimum is extracted from measurements of UxU_x and UyU_y for different (ϕx′,ϕy′)(\phi_{x'},\phi_{y'}) pairs while tickling the xx and yy modes simultaneously and keeping ϕz′\phi_{z'} constant. The profile of Ux+UyU_x+U_y is expected to resemble that of ∣Ex∣+∣Ey∣|E_x|+|E_y|, where EiE_i is the projection of the AC field along the ii axis. The profile of ∣Ex∣+∣Ey∣|E_x|+|E_y| for an ideal quadrupole field is shown in Figure 3(c). The axial micromotion minimum (at u0z′=0u_{0z'}=0) is extracted from a measurement of UzU_z for different ϕz′\phi_{z'} while tickling the zz mode and keeping (ϕx′,ϕy′)(\phi_{x'},\phi_{y'}) constant.

Method 3 has three advantages over Methods 1 and 2. First, its precision is limited by the thermal noise rather than our detection noise floor. Second, optimal compensation along three axes can be achieved with a single detection system. Third, Method 3 relies on the ωi\omega_i peaks instead of the Ω\Omega peak, which separates the micromotion amplitude information for each axis.

Experimental demonstration of minimization methods

Finding the radial micromotion minimum

Figure 4 is the main result of this work. It shows data from applying Methods 1–3 to find the micromotion minimum in the radial plane.

Figure 4. Data from applying Methods 1–3 to find the micromotion minimum in the radial plane. The setpoints ϕx′\phi_{x'} and ϕy′\phi_{y'} are shown on the horizontal and vertical axes, and their ranges in (a–c) correspond to scanning an area of approximately (0.014r0)2(0.014r_0)^2. (a) Method 1: sum of the excess micromotion amplitudes Ax′A_{x'} and Ay′A_{y'}. The extracted micromotion minimum is shown in red. (b) Method 2: sum of the excess micromotion phases Φx′\Phi_{x'} and Φy′\Phi_{y'}. The extracted micromotion minimum is shown in green. (c) Method 3: sum of the secular-motion amplitudes UxU_x and UyU_y during simultaneous tickling of the xx and yy modes. The extracted micromotion minimum is shown in blue. (d) Radial excess micromotion minima found via Methods 1–3.

Figure 4(a) shows the sum Ax′+Ay′A_{x'}+A_{y'} as a function of ϕx′\phi_{x'} and ϕy′\phi_{y'}. The profile we obtained is consistent with that of Figure 3(a). In the dark region, the measured micromotion amplitudes are at the detection noise floor. However, sub-noise precision in the micromotion minimum can be obtained. Here, we extracted the minimum (ϕx′,ϕy′)=(−215±9,203±20)(\phi_{x'},\phi_{y'})=(-215\pm9,203\pm20) V. Appendix § 1.1.4 provides more information on sub-noise precision and describes how the minima were obtained for this plot and others. Figure 4(b) shows the sum Φx′+Φy′\Phi_{x'}+\Phi_{y'} as a function of ϕx′\phi_{x'} and ϕy′\phi_{y'}. The profile we obtained is consistent with that of Figure 3(b). Near the phase boundaries, the micromotion can no longer be resolved, so the measured phase becomes noisy. Again, sub-noise precision in the micromotion minimum can be obtained (see Appendix § 1.1.4). Here, we extracted the minimum (ϕx′,ϕy′)=(−215±7,202±9) V(\phi_{x'},\phi_{y'})=(-215\pm7,202\pm9)\ \mathrm{V}.

Figure 4(c) shows the sum Ux+UyU_x+U_y as a function of ϕx′\phi_{x'} and ϕy′\phi_{y'} with simultaneous tickling at Ω−ωx\Omega-\omega_x and Ω−ωy\Omega-\omega_y. The tickling depths were hx,hy≈1%h_x,h_y\approx1\%. The profile we obtained is consistent with that of Figure 3(c). In the dark region, the excitation of both secular modes is minimized. The data in Figure 4(c) appear less noisy than for the other methods, due to the reasons we discussed above in presenting Method 3. Here, we extracted the minimum (ϕx′,ϕy′)=(−219±5,198±7) V(\phi_{x'},\phi_{y'})=(-219\pm5,198\pm7)\ \mathrm{V}.

The optimal ϕx′\phi_{x'}, ϕy′\phi_{y'} pairs extracted from the three methods are shown together in Figure 4(d). Details of how these voltages are found from the data are given in Appendix § 1.1.4. The values obtained from all methods are in agreement. Figure 4(d) also shows the precision of the three methods; Method 3 is seen to be the most precise, followed by Method 2 and then Method 1.

To allow for direct comparison with other systems, three metrics for the micromotion minimization precision are shown in Table I (see Appendix E). The smallest uncertainty of the residual stray field is found for Method 3 at 2.9 V m−12.9\ \mathrm{V\,m^{-1}}, which is comparable to values found for trapped ions^{18,23,34}. For this reason, Method 3 has since been used in two other Paul traps in our laboratory, for nanoparticles of different sizes and materials (see Appendix F).

MethodLocalization precision / μm\mu\mathrm{m}Residual excess micromotion amplitude / nm\mathrm{nm}Residual stray field uncertainty / V m−1\mathrm{V\,m^{-1}}
10.61087.2
20.581044
30.571032.9

Table I. Precision metrics for the radial data presented in Figure 4. For each metric, we report the largest value across the two dimensions x′x' and y′y'.

Finding the axial micromotion minimum

We now turn to finding the micromotion minimum along the trap axis. As the amplitude of axial micromotion was much smaller than the amplitude of radial micromotion, we increased the peak-to-peak amplitude of the drive to 960 V960\ \mathrm{V}, leading to eigenfrequencies (ωz,ωx,ωy)/2π=(622,2112,2368) Hz(\omega_z,\omega_x,\omega_y)/2\pi=(622,2112,2368)\ \mathrm{Hz}. For each method, we do not extract the minimum from a fit, since we make no assumptions about the functional form of each profile in Figure 5. Instead, we identify a set of data points that corresponds to the minimum, shown in red.

Figure 5. Data from applying Methods 1–3 to find the micromotion minimum along the zz axis. The setpoint ϕz′\phi_{z'} is shown on the horizontal axes, and its range corresponds to approximately 180 μm180\,\mu\mathrm{m}. The extracted micromotion minima are shown as red-shaded regions. (a) Method 1: the excess micromotion amplitude Az′A_{z'}. (b) Method 2: the excess micromotion phase Φz′\Phi_{z'}. (c) Method 3: the secular-motion amplitude UzU_z during tickling of the zz mode.

Figure 5(a) shows Az′A_{z'} as a function of ϕz′\phi_{z'}. We see a V-shaped profile with a broad minimum. All measured amplitudes were within the linear response range of our detection system. The red points correspond to amplitudes within 3 dB3\ \mathrm{dB} fluctuations of the minimum value. From the mean and standard deviation of these points, we extracted the minimum ϕz′=(−23±24) V\phi_{z'}=(-23\pm24)\ \mathrm{V}.

Figure 5(b) shows Φz′\Phi_{z'} as a function of ϕz′\phi_{z'}. We see two plateaus separated by a region where the measured phase is noisy. In order to extract the phase jump Δ\Delta between the plateaus, we fit a sigmoid function; the obtained value Δ=(0.83±0.03)π\Delta=(0.83\pm0.03)\pi is lower than the expected value of π\pi. The red points are all those at least 0.2Δ0.2\Delta above the lower plateau and at least 0.2Δ0.2\Delta below the upper plateau. From the mean and standard deviation of these points, we extracted the minimum ϕz′=(−16±23)\phi_z' = (-16 \pm23) V.

Figure 5(c) shows UzU_z as a function of ϕz′\phi_z'. We see a profile like the Aries glyph, with a sharp minimum. The drop in UzU_z for values ∣ϕz′∣≳80|\phi_z'| \gtrsim80 V is due to both a decreased sensitivity and a nonlinear response in the detection system. The red points correspond to amplitudes within 3 dB fluctuations of the minimum value. From the mean of these points and the scan resolution, we extracted the minimum ϕz′=(−20±2)\phi_z' = (-20 \pm2) V.

The minima obtained from all methods are in agreement. We highlight that individual and vertical slices of the datasets added together in Figure 4(a–Figure 4c) (see Appendix § 1.1.4) have the same structure as the profiles (constant ϕx′,ϕy′\phi_x',\phi_y') shown in Figure 5(a–Figure 5c), which is expected. The detected amplitudes of excess micromotion, AiA_i, reach the noise floor in Figure 4(a) and 5(a). This places an upper bound Φac≤2u0ir0α≈1×10−4\Phi_{\mathrm{ac}} \leq\frac{2u_{0i}}{r_0\alpha} \approx1 \times10^{-4} for typical values of α\alpha[15], which validates the assumption that Φac\Phi_{\mathrm{ac}} is negligible. As in Figure 4, Method 3 is seen in Figure 5 to be the most precise.

Figure 6. Radial minimum obtained via Method 3 for different ϕz′\phi_z'. The setpoints ϕx′\phi_x' and ϕy′\phi_y' are shown on the horizontal and vertical axes, and the setpoint ϕz′\phi_z' is shown on the color-bar. The range of ϕz′\phi_z' corresponds to approximately 90 μ\mum.

Characterization of the stray field

In this section, we characterize contributions to the stray field in our system by studying changes in the radial micromotion minimum. Only data obtained via Method 3 are presented, due to its higher precision and simpler practical implementation.

Figure 6 shows the radial minimum for different ϕz′\phi_z'. For ∣ϕz′∣<50|\phi_z'| < 50 V, we cannot identify any change in the radial component of the stray field. For ∣ϕz′∣>50|\phi_z'| > 50 V, there is a 2σ2\sigma variation in the minimum. The two endcaps of our trap are radially displaced from one another by approximately one endcap radius (0.25 mm), which we suspect caused the observed variation.

Figures 7(a) and 7(b) show the y′y' and x′x' components of the radial minimum as a function of the time since the particle was loaded. Eleven days after loading the particle, we connected a pressure gauge to the vacuum chamber; we attribute the jumps in the figures to a change in the local charge environment caused by the gauge[30]. Before and after the charging event, the compensation voltages required to nullify the stray field are seen to relax towards stable values. We interpret this as a slow stabilization of the local charge environment[40]. Around one week after the charging event, the stray field reached a steady state. All of the data in this work, with the exception of those shown in Figure 7(a) and 7(b), were acquired only after reaching the steady state.

Figure 7. The ϕy′\phi_{y'} and ϕx′\phi_{x'} components of the radial minimum, obtained via Method 3, as a function of (a, b) the time since the particle was loaded; (c, d) ϕtop\phi_{\mathrm{top}}, for ϕz′=0\phi_{z'} = 0; (e, f) the Mathieu parameter qyq_y. Eleven days after loading the particle, there was a charging event, indicated in red in (a) and (b).

Figures 7(c) and 7(d) show the y′y' and x′x' components of the radial minimum as a function of ϕtop\phi_{\mathrm{top}}, for ϕz′=0\phi_z' = 0. The value of ϕx′\phi_x' remained constant, whereas the value of ϕy′\phi_y' increased significantly as ϕtop\phi_{\mathrm{top}} was increased. Based on this observation, we hypothesize that the endcap axis is radially displaced from the AC node by Δx′≈0\Delta x' \approx0, Δy′>0\Delta y' > 0.

Figures 7(e) and 7(f) show the y′y' and x′x' components of the radial minimum as a function of qyq_y, which had the largest absolute value of the Mathieu parameters {qx,qy,qz}\{q_x,q_y,q_z\}. We increased qyq_y by increasing the amplitude of the AC drive; the range of 0.2<qy<0.90.2 < q_y < 0.9 corresponds to a peak-to-peak AC drive amplitude in the range (360 to 870) V. The values of both ϕx′\phi_x' and ϕy′\phi_y' remained constant, as expected, since changing qiq_i should not change the stray field.

Conclusion and Outlook

Motivated by the lack of a standard procedure, we have introduced and demonstrated three methods for the detection and subsequent compensation of micromotion for nanoparticles in a Paul trap. The methods are based on direct amplitude detection, direct phase detection, and resonant tickling. We have shown that all three methods are in agreement and allow for micromotion minimization along three axes.

The tickling method was the most precise and is also the most convenient, since it requires only one detection system. This method can be used in any Paul trap in which the thermal amplitudes of the trapped particle’s secular modes can be resolved. Via this method, we have characterized the stray field in our system and nullified its radial component to within 2.9 V m−1^{-1}, which is comparable to values reported for trapped-ion systems.

We anticipate that a higher precision in each method presented here can be achieved by increasing the detection efficiency, fitting each profile, cooling the center-of-mass motion, and reducing the pressure in the vacuum chamber. Furthermore, we have not examined how to use these methods to minimize micromotion efficiently.

The methods presented here provide a basis for developing a standard procedure for micromotion minimization within the levitodynamics community.

The datasets and analysis code are available at https://doi.org/10.5281/zenodo.21620140.

Acknowledgments

This research was funded in part by the Austrian Science Fund (FWF) [10.55776/COE1, 10.55776/I5540, 10.55776/ESP258]. For Open Access purposes, the authors have applied a CC BY public copyright license to any author accepted manuscript version arising from this submission.

Axial displacement at constant stiffness

In order to displace the particle along the trap axis, we need to apply different potentials to each endcap. However, arbitrary endcap potentials will in general change the axial stiffness, that is, the second derivative of the axial trapping potential. For our system, we saw that a constant geometric mean ϕend=ϕtopϕbottom\phi_{\mathrm{end}} = \sqrt{\phi_{\mathrm{top}}\phi_{\mathrm{bottom}}} maintains a constant stiffness. The value of ϕend\phi_{\mathrm{end}} was seen to stay constant over a displacement ϕz′\phi_{z'} in the range (−0.44ϕend,0.44ϕend)(-0.44\phi_{\mathrm{end}}, 0.44\phi_{\mathrm{end}}) at a pressure of 10−210^{-2} mbar, where the linewidth was approximately 10 Hz.

Note that when the trap axis is vertical, as in our case, the value ϕend\phi_{\mathrm{end}} should be chosen such that the axial confinement is not significantly altered by gravity. We do this by increasing ϕend\phi_{\mathrm{end}} for ϕz′=0\phi_{z'} = 0 until the particle is no longer vertically displaced.

Voltage-to-position transformation

Figure 8 shows the particle’s position r⃗=(x′,y′,z′)\vec{r} = (x', y', z') as a function of ϕ⃗=(ϕx′,ϕy′,ϕz′)\vec{\phi} = (\phi_{x'}, \phi_{y'}, \phi_{z'}), which we assume to satisfy r⃗=Aϕ⃗+r⃗0\vec{r} = \mathbf{A}\vec{\phi} + \vec{r}_0. The position was extracted from two orthogonal cameras, and the origin r⃗=0⃗\vec{r} = \vec{0} was arbitrarily chosen. Error bars in the position arise from uncertainties in the camera calibrations.

Figure 8. Particle’s spatial coordinates x′x' (top row), y′y' (middle row), z′z' (bottom row) as a function of each setpoint ϕx′\phi_{x'} (left column), ϕy′\phi_{y'} (middle column), ϕz′\phi_{z'} (right column). Data are shown in blue, and linear fits are shown in red. The origin (x′,y′,z′)=(0,0,0)(x',y',z')=(0,0,0) was arbitrarily chosen, and in each column the other setpoints were held at 0 V.

From the data shown in Figure 8, we extract r⃗0=(72.9±2.5,68.6±2.3,206±7) μm\vec{r}_0 = (72.9 \pm2.5, 68.6 \pm2.3, 206 \pm7)\ \mu\mathrm{m} and reconstruct the matrix

A=(20.7±1.82.1±2.031±304.3±1.817.5±1.823±281±62±6(5.0±0.8)×102) nm V−1\mathbf{A} = \begin{pmatrix} 20.7 \pm1.8 & 2.1 \pm2.0 & 31 \pm30 \\ 4.3 \pm1.8 & 17.5 \pm1.8 & 23 \pm28 \\ 1 \pm6 & 2 \pm6 & (5.0 \pm0.8) \times10^{2} \end{pmatrix}\ \mathrm{nm}\ \mathrm{V}^{-1}

via Aij=∂ri∂ϕjA_{ij} = \frac{\partial r_i}{\partial\phi_j}. The particle is significantly more sensitive to ϕz′\phi_{z'} than to other components of ϕ⃗\vec{\phi}, as expected from finite element simulations. This transformation is dependent on trap parameters (see, for example, Eqn. 16 of [15]), and A\mathbf{A} was obtained under the default parameters used in this work.

Tickling frequencies

Here we discuss the excitation of the particle due to different tickling frequencies Ωtickle\Omega_{\mathrm{tickle}}. By considering the force from the AC field as a perturbation to harmonic particle dynamics, it can be shown that Eqn. (2) gives rise to leading-order force terms at frequencies Ω±Ωtickle\Omega\pm\Omega_{\mathrm{tickle}} in the one-dimensional case. Hence a modulation at Ωtickle=Ω−ωi\Omega_{\mathrm{tickle}} = \Omega- \omega_i leads to a force term at ωi\omega_i. The resulting excitation is stronger than that from the force terms at Ω±ωi\Omega\pm\omega_i arising from Ωtickle=ωi\Omega_{\mathrm{tickle}} = \omega_i. For this reason, we chose Ωtickle=Ω−ωi\Omega_{\mathrm{tickle}} = \Omega- \omega_i. We note, however, that as in trapped-ion experiments, excitation was experimentally observed in our system for Ωtickle=2ωi/n\Omega_{\mathrm{tickle}} = 2\omega_i/n, n∈{1,2,3}n \in\{1, 2, 3\}, and we anticipate that Ωtickle=Ω−2ωi/n\Omega_{\mathrm{tickle}} = \Omega- 2\omega_i/n could lead to similar behavior.

Extraction of the minimum in the radial plane

The plots shown in Figure 4 are presented as sums of two different two-dimensional profiles. For example, Figure 4(a) shows Ax′(ϕx′,ϕy′)+Ay′(ϕx′,ϕy′)A_x'(\phi_{x'}, \phi_{y'}) + A_{y'}(\phi_{x'}, \phi_{y'}). One of these two-dimensional profiles, Ax′(ϕx′,ϕy′)A_{x'}(\phi_{x'}, \phi_{y'}), is shown in Figure 9. We find the minima in horizontal (constant ϕy′\phi_{y'}) or vertical (constant ϕx′\phi_{x'}) slices. The minimum for each vertical slice in Figure 9 is shown in red and is extracted from a profile similar to that shown in Figure 5(a). We then fit the minima with a straight line. Finally, we repeat this procedure for Ay′(ϕx′,ϕy′)A_{y'}(\phi_{x'}, \phi_{y'}) and find the intersection of the two fitted lines, achieving sub-noise precision in the radial micromotion minimum. This example is given for Method 1, but the procedure is analogous for Methods 2 and 3.

Figure 9. Data from applying Method 1 to find the micromotion minimum along the y′y' axis. The setpoints ϕx′\phi_{x'} and ϕy′\phi_{y'} are shown on the horizontal and vertical axes, and the excess micromotion amplitude Ax′A_{x'} is shown in the color bar.

Precision metrics

Here we explain how the values in Table I were obtained. For each method, we extract the uncertainty Δϕi′\Delta\phi_{i'} in each compensation voltage required to nullify the stray field. From the Method 1 data in Figure 4(a), for example, we extracted Δϕx′=9\Delta\phi_{x'}=9 V and Δϕy′=20\Delta\phi_{y'}=20 V.

In the first column in Table I, the localization precision is Δu0i′=Δ(A⋅ϕ⃗)i′\Delta u_{0i'}=\Delta(\mathbf{A}\cdot\vec{\phi})_{i'}, where A\mathbf{A} is defined in Appendix B. In the second column, the residual excess micromotion amplitude is 12u0iqi\frac{1}{2}u_{0i}q_i (see Eqn. (1)). As the value for u0iu_{0i}, we use the localization precision Δu0i′\Delta u_{0i'}. As the value for qiq_i, we use 2ωi2Ω\frac{2\omega_i\sqrt{2}}{\Omega} since our radial Mathieu parameters satisfy ∣ai∣≪qi2/2\lvert a_i\rvert\ll q_i^2/2. In the third column, the residual stray field uncertainty is given by ηi′Δϕi′\eta_{i'}\Delta\phi_{i'}, where ηi′\eta_{i'} is the field component at the trap center along the i′i' direction per unit ϕi′\phi_{i'}. From finite-element simulations of our trap, we obtain the values (ηx′,ηy′,ηz′)=(0.5,0.4,6.6) m−1(\eta_{x'},\eta_{y'},\eta_{z'})=(0.5,0.4,6.6)\ \mathrm{m}^{-1}.

Method 3 in different Paul traps

Figure 10 shows data from applying Method 3 to find the micromotion minimum in the radial plane for a different Paul trap than that described in the main text. The particle was a cluster of several nanodiamonds of 100 nm diameter (the mass was not determined), trapped in a linear blade-style Paul trap driven at Ω/2π=15\Omega/2\pi=15 kHz, with a peak-to-peak amplitude of 571 V and with 810 V applied to both endcaps. The modulation depth at Ωtickle,i=ωi\Omega_{\mathrm{tickle},i} = \omega_i was 0.45 %. After minimizing the micromotion, we estimated a residual excess micromotion amplitude of ∼1 nm\sim1\,\mathrm{nm}.

Figure 10. Data from applying Method 3 to find the micromotion minimum in the radial plane for a different Paul trap than that described in the main text. The setpoints ϕx′\phi_x' and ϕy′\phi_y' are shown on the horizontal and vertical axes, and the sum of the secular-motion amplitudes UxU_x and UyU_y during simultaneous tickling of the xx and yy modes are shown in the color bar.

In a third Paul trap in our laboratory, similar results were also obtained with a silica nanoparticle 300 nm in diameter.

References

  1. [1]W. Paul, Rev. Mod. Phys. 62, 531 (1990).
  2. [2]C. Matthiesen, Q. Yu, J. Guo, A. M. Alonso, and H. Häffner, Phys. Rev. X 11, 011019 (2021).
  3. [3]D. Leibfried, R. Blatt, C. Monroe, and D. Wineland, Rev. Mod. Phys. 75, 281 (2003).
  4. [4]W. G. Rellergert, S. T. Sullivan, S. J. Schowalter, S. Kotchigova, K. Chen, and E. R. Hudson, Nature 495, 490 (2013).
  5. [5]C.-w. Chou, C. Kurz, D. B. Hume, P. N. Plessow, D. R. Leibfradt, and D. Leibrfried, Nature 545, 203 (2017).
  6. [6]M. Sinhal, Z. Meir, K. Najafian, G. Hegi, and S. Willitsch, Science 367, 1213 (2020).
  7. [7]Z. Wu, T. Duka, M. Isaza-Monsalve, M. Kautzky, V. Švarc, A. Turci, R. Nardi, M. Gronowski, M. Tomza, B. J. Furey, and P. Schindler, “Infrared absorption spectroscopy of a single polyatomic molecular ion,” (2025), arXiv:2511.19687 [quant-ph].DOI
  8. [8]L. Schömöger, O. O. Versolato, M. Schwarz, M. Köhnen, A. Windberger, B. Piest, S. Feuchtbeiner, J. Pedregosa-Gutierrez, T. Leopold, P. Micke, A. K. Hansen, T. M. Baumann, M. Drewsen, J. Ullrich, P. O. Schmidt, and J. R. C. López-Urrutia, Science 347, 1233 (2015).
  9. [9]E. Bonvin, L. Devaud, M. Rossi, A. Militaru, L. Dania, D. S.
  10. [10]D. S. Bykov, P. Mestres, L. Dania, L. Schömger, and T. E. Northup, Appl. Phys. Lett. 115, 034101 (2019).
  11. [11]G. F. M. Tomassi, D. Veldhuizen, B. Melo, D. Candoli, A. Riera-Campeny, O. Romero-Isart, N. Meyer, and R. Quidant, Physical Review Research 8, L012026 (2026).
  12. [12]A. Pontin, N. P. Bullier, M. Toroš, and P. F. Barker, Phys. Rev. Res. 2, 023349 (2020).
  13. [13]T. Delord, P. Huillyer, L. Nicolas, and G. Hétet, Nature 580, 56 (2020).
  14. [14]M. Message, F. Cerisola, J. D. Pritchett, K. O’Flynn, Y. Ren, M. Rashid, J. Anders, and J. Millen, Phys. Rev. Lett. 135, 217101 (2025).
  15. [15]D. J. Berkeland, J. D. Miller, J. C. Bergquist, W. M. Itano, and D. J. Wineland, Journal of Applied Physics 83, 5025 (1998).
  16. [16]J. Keller, H. L. Partner, T. Burgermeister, and T. E. Mehlstäubler, Journal of Applied Physics 118, 104501 (2015).
  17. [17]C. J. B. Goham and J. W. Britton, AIP Advances 12, 115315 (2022).
  18. [18]S. Charles Doret, J. M. Amini, K. Wright, C. Volin, T. Killian, A. Ozakin, D. Denison, H. Hayden, C.-S. Pai, R. E. Slusher, and A. W. Harter, New Journal of Physics 14, 073012 (2012).
  19. [19]Y. Ibaraki, U. Tanaka, and S. Urabe, Applied Physics B 105, 219 (2011).
  20. [20]C. W. Hogle, A. D. Burch, J. D. Sterk, M. N. H. Chow, M. Ivory, D. S. Lobser, P. Maunz, J. Van Der Wall, C. G. Yale, S. M. Clark, D. Stick, and M. C. Revelle, Frontiers in Quantum Science and Technology 3, 1352800 (2024).
  21. [21]M. Drewsen, A. Mortensen, R. Martinussen, P. Staanum, and J. L. Sørensen, Physical Review Letters 93, 243201 (2004).
  22. [22]S. M. Clark, D. Lobser, M. C. Revelle, C. G. Yale, D. Bossert, A. D. Burch, M. N. Chow, C. W. Hogle, M. Ivory, J. Pehr, B. Salzbrener, D. Stick, W. Sweatt, J. M. Wilson, E. Winrow, and P. Maunz, IEEE Transactions on Quantum Engineering 2, 1 (2021).
  23. [23]R. Saito, K. Saito, and T. Mukaiyama, Journal of Applied Physics 129, 124302 (2021).
  24. [24]R. Saito and T. Mukaiyama, AIP Advances 15, 115119 (2025).
  25. [25]L. Dania, K. Heidegger, D. S. Bykov, G. Cerchiari, G. Araneda, and T. E. Northup, Phys. Rev. Lett. 129, 013601 (2022). B. A. Stickler, K. Hornberger, and M. S. Kim, Nature Reviews Physics 3, 589 (2021).
  26. [27]C. Gonzalez-Ballestero, M. Aspelmeyer, L. Novotny, R. Quidant, and O. Romero-Isart, Science 374, 168 (2021).
  27. [28]S. Gupta, B. Faulend, D. S. Bykov, T. E. Northup, and C. Gonzalez-Ballestero, “Quantum theory of electrically levitated nanoparticle-ion systems: Molecular dynamics and sympathetic cooling,” (2026), arXiv:2511.21495 [quant-ph].arxiv.org/abs/2511.21495
  28. [29]M. Schut and C. Scarani, “Proposal for macroscopic delocalisation of a large mass in a rf trap,” (2026), arXiv:2509.17081 [quant-ph].arxiv.org/abs/2509.17081
  29. [30]L. Dania, D. S. Bykov, M. Knoll, P. Mestres, and T. E. Northup, Phys. Rev. Research 3, 013018 (2021).
  30. [31]T. W. Penny, A Low-Noise Electrically Levitated Oscillator for Investigating Fundamental Physics, Ph.D. thesis, University College London (2022).
  31. [32]M. Cavigelli, Toward Time-Domain Brillouin Scattering in Levitation, Ph.D. thesis, ETH Zurich (2025).DOI
  32. [33]Different authors choose to present the same information in different ways. When highlighting values reported in other works, we have kept the style from the source where appropriate. While the notation of (0 ± ΔE) conveys a clear message, we want to highlight that the measured quantity is the uncertainty ΔE, with the assumption that the residual stray field is zero measured to a precision given by this uncertainty.
  33. [34]S. Tanaka, K. Masuda, Y. Akimoto, K. Koda, Y. Ibaraki, and S. Urabe, Applied Physics B 107, 907 (2012).
  34. [35]S. Harter, A. Krü̈kow, A. Brunner, and J. Hecker Denschlag, Applied Physics Letters 102, 22115 (2013).
  35. [36]B. L. Chauh, N. C. Lewty, R. Cazan, and M. D. Barrett, Optics Express 21, 10632 (2013).
  36. [37]T. F. Gloger, P. Kaufmann, D. Kaufmann, M. T. Baig, T. Collath, M. Johannang, and C. Wunderlich, Physical Review A 92, 043421 (2015).
  37. [38]Y. Liu, Q.-f. Lao, P.-f. Lu, X.-x. Rao, H. Wu, T. Liu, K.-x. Wang, Z. Wang, M.-s. Li, F. Zhu, and L. Luo, Applied Physics Letters 119, 134002 (2021).
  38. [39]L. Dania, D. S. Bykov, F. Goschin, M. Teller, A. Kassid, and T. E. Northup, Phys. Rev. Lett. 132, 133602 (2024).
  39. [40]The data in Figs. 7(a) and 7(b) suggest the stray field primarily changed along the y-axis. This directionality, as well as the relaxation mechanism of the stray field, are beyond the scope of this work.

Paper details

vlnc:W_gt4wnabj
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

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