Implementation of an adaptive-optics assisted isoSTED nanoscope

Abstract

In stimulated emission depletion (STED) nanoscopy, high 3D resolution requires harnessing a 4Pi architecture using two opposing objectives. Here, we provide the step-by-step process for the construction and alignment of a 4Pi-STED nanoscope, commonly referred to as an ‘isoSTED nanoscope’. The procedure guides interested researchers through the assembly of the optomechanical components, the configuration of the electronic and control devices, the alignment of the optical beam path and the assessment of the instrument’s performance. The protocol offers a detailed roadmap for constructing an isoSTED nanoscope with adaptive optics in about 12 months and is designed for users with expertise in optical instrumentation builds. Once the instrument is finely calibrated, researchers can expect to achieve 3D biological images with isotropic sub-50-nm resolution in thick samples ≤35 μm in depth.

Introduction

Stimulated emission depletion (STED) nanoscopy, stemming from innovations first described in 1994¹, stands as a pivotal technique that has redefined fluorescence imaging. Its first experimental demonstration in 1999² and subsequent application in biological samples in 2000³ marked a turning point by showing that the diffraction limit of far-field light microscopy can be broken. Operating in a targeted switching mode, STED nanoscopy confines fluorescence to sub-diffraction-sized regions by depleting surrounding excited fluorophores via stimulated emission.

Originally limited to 2D imaging, technical advances over the past approximately 15 years have enabled STED nanoscopy development into 3D imaging. In particular, 3D STED nanoscopy uses a so-called ‘4Pi geometry’ in which light illuminating the sample coherently through two opposing objectives facilitates a sharp central depletion minimum in the axial direction, affording a remarkable 33-nanometer z-resolution. This breakthrough, termed ‘STED-4Pi microscopy’^4,5 laid the groundwork for the subsequent development of isoSTED nanoscopy⁶, which demonstrated 3D isotropic resolution by combining fluorescence confinement in all three dimensions. isoSTED nanoscopy yields an effective point spread function (PSF) 1,500 times smaller in volume than that of conventional confocal microscopy, achieving a striking 20–50-nm isotropic resolution. isoSTED nanoscopy has been demonstrated in material science applications for imaging self-assembled nanostructures⁷ and in cell biology, revealing subcellular structures in both fixed cells^8,9 and living cells¹⁰.

However, despite its remarkable resolution, the original isoSTED implementation has limitations, particularly in the form of axially shifted side-lobes in its effective PSF that can generate axially repetitive artifacts. The adaptive-optics (AO)-assisted isoSTED nanoscope described in this protocol, hereafter referred to exclusively, addresses these artifacts by introducing moderate defocus on the axial depletion beam¹¹ coupled with AO corrections as discussed in the previous publication from our group. Defocused axial depletion foci from the two opposing objectives generate partial destructive interference, leading to non-zero intensity at the locations of side lobes in the previously developed STED-4Pi microscopy. The fluorescence is thus completely quenched except near the focal plane. These improvements have enabled isoSTED imaging in thick specimens¹², extending the range of samples successfully imaged to include brain tissue sections and Drosophila egg chambers. Imaging of these latter samples is inherently more challenging, primarily because of the heightened optical aberrations.

Although isoSTED nanoscopy has emerged as a potent imaging tool, its complexity requires considerable expertise across multiple disciplines for both construction and operation. In a bid to democratize access to this instrument, our protocol encapsulates a comprehensive guide enabling researchers to construct and maintain an isoSTED nanoscope. We estimate that a skilled builder focusing full time on this project would take about 12 months to create the described instrument. In addition, although we provide the blueprint based on hardware and software currently available, some components may become obsolete in the future. In such cases, they can be substituted with suitable alternatives without affecting the performance of the system. We therefore encourage readers to contact this Protocol’s authors for updates and questions before or during the building process.

Conceptual description of the AO-isoSTED setup

An AO-assisted isoSTED nanoscope features two opposing objectives in a vertically oriented 4Pi interference cavity. Similar to other STED modalities, in isoSTED nanoscopy, the fluorescence is restricted in three dimensions to a volume smaller than the diffraction limit by depleting excited fluorophores in the vicinity of the excitation volume using stimulated emission. To achieve this, the depletion laser generates a hollow, sphere-shaped focus to superimpose on the excitation focus (Fig. 1). To achieve uniform compression of the fluorescence emission volume, two depletion patterns (i.e., STED_xy for lateral and STED_z for axial depletion) with a common focal zero are combined. Specifically, the STED_xy pattern is generated by the constructive interference of two ring-shaped foci, whereas the STED_z beams interfere destructively along the z-axis in the focal plane by tightly focused Gaussian-like beams. The incoherent summation of these two patterns creates a ‘zero’-intensity minimum at the center surrounded by a high-intensity crest in all directions.

Fig. 1 |. PSFs in AO-isoSTED nanoscopy in the axial (xz) plane.

From left to right: STED_xy, STED_z, and the combined depletion (red) and effective (green) PSFs. The small hollow arrows beside the large arrows denote the polarization state of the depletion beam when generating the respective PSFs. Specifically, the STED_xy PSF is created through constructive interference of the two counterpropagating beams, whereas the STED_z PSF is formed through destructive interference.

For optimal performance, it is crucial to maintain stable interference patterns throughout the imaging procedure. Addressing and mitigating potential disruptive factors, notably, aberrations, is imperative. Because the isoSTED nanoscope images deeper into specimens, refractive index variations can lead to aberrations that make imaging impossible. To facilitate super-resolution imaging of thick specimens, the isoSTED nanoscope implements a sensorless AO architecture, aimed at correcting aberrations induced by the specimens.

Optical design overview

To simplify the description, we use abbreviations for all the components (Supplementary Table S1). The AO-isoSTED nanoscope uses two opposing 100×/1.35 numerical aperture silicone-oil immersion objectives, designated as ‘OBJ1’ and ‘OBJ2’ in Fig. 2, installed on a vertical optical breadboard. Stable interference across the entire field of view (FOV) relies on precise alignment of the two objectives with nearly identical magnification (±0.3%). A quarter-wave plate (QWP) (QWP4 and QWP5) set at 45° with respect to the vertical breadboard plane is positioned immediately behind each objective (QWP4 for OBJ1 and QWP5 for OBJ2). The QWPs convert the linearly polarized incident beams to circularly polarized ones, minimizing excitation and depletion selectivity regarding fluorophore dipole orientation and reducing polarization-induced background¹² in the recorded images. The back pupil plane of each objective is conjugated to a deformable mirror (DM) (DM1 for OBJ1 and DM2 for OBJ2) using a 4f telecentric system (L10, L11 for OBJ1 and L12, L13 for OBJ2). The DMs function as the main devices for correcting aberrations, particularly those induced by the specimen. To ensure that the reflection angles of the beams stay within acceptable limits for the DMs (±15°), a pair of mirrors folds the beam path inside the 4f telecentric system (M11, M12 for DM1 and M13, M14 for DM2). Notably, M11 is mounted on a single-axis piezo translation stage to adjust the optical path difference for the zero-phase retardance of interference.

Fig. 2 |. Optical beam path overview.

AOM, acousto-optic modulator; AOTF, acousto-optic tunable filter; APD, avalanche photodiode; C, camera; CL, cylindrical lens; D, D-shape mirror; DI, dichroic mirror; DM, deformable mirror; F, filter; GM, galvanometer mirror; HWP, half-wave plate; L, lens; LH, laser head; M, mirror; MF, multimode fiber; OBJ, objective; PBS, polarizing beam splitter cube; PMT, photomultiplier tube; QWP, quarter-wave plate; RM, resonant mirror; SLM, spatial light modulator; W, BK-7 glass wedges/optical flat.

Before being reflected by DM1, laser light passes through a stationary BK7 optical flat (W1) and a half-wave plate (HWP) (HWP4). In the other arm, a BK7 wedge pair (W2), with one wedge mounted on a translation stage, replaces the optical flat, and the HWP is absent. A polarizing beam splitter (PBS) cube (PBS2) splits the laser light into these two arms. This PBS, along with all the previously described components, constitutes the core part of the isoSTED nanoscope known as the ‘4Pi interference cavity’. Adjusting the optical path length difference between the upper and lower arms in this cavity is possible either by translating M11 or by adjusting the variable wedge in W2. Ideally, the optical path length difference is zero, eliminating any phase retardance for interference.

The coalescent illumination beams, comprising both depletion and excitation, enter the 4Pi interference cavity through the bottom-right side of PBS2. A white-light laser (LH2) provides the 78-MHz pulsed excitation light. Immediately after the laser, an acousto-optic tunable filter (AOTF) selects for the desired excitation components at wavelengths of 650, 590 and 488 nm. After compensation for the temporal chromatic dispersion within a pulse (for details, see Temporal chromatic dispersion compensation) and beam size expansion (using L15 and L16), the excitation beam is merged with the depletion beam path via a dichroic mirror (DI1). A pair of wave plates (HWP6 and QWP7) tunes the polarization orientation of the excitation beam to 45° relative to that of the depletion beam so that the entire excitation beam can propagate through the upper arm in the 4Pi interference cavity.

A pulsed 775-nm wavelength laser (LH1) operating at 78 MHz, with a pulse duration of about 0.9 ns and synchronized to the excitation laser, serves as the depletion source. The beam intensity is tunable via the acousto-optic modulator (AOM) mounted after LH1. To mitigate interference between the two polarization components later in the common focus between the objectives (Fig. 1), one polarization component of the laser is delayed with respect to the orthogonally polarized component via a polarization-separation module (for details, see Polarization-separation module alignment). After beam size expansion using a 4f telecentric system (L1 and L2), the beam emerging from the polarization-separation module illuminates a spatial light modulator (SLM). A double-pass configuration¹³ encodes a vortex and a moderate defocus phase pattern (hologram) onto the two orthogonally oriented polarization components of the depletion beam. These two holograms are optically conjugated to each other and are further imaged onto a 16-kHz resonant mirror (RM) through two additional 4f telecentric systems (L4 and L5 for the first 4f; L6 and L7 for the second). A mirror (M9) is placed between these two 4f telecentric systems to fold the beam path.

After merging, both the depletion and excitation beams illuminate the scanner, where the RM scans the combined laser beams along the fast-scanning axis (x), while two synchronized, scanning galvanometer mirrors (GMs) (GM1 and GM2) scan the beams along the slow-scanning axis (y). Collectively, the two GMs behave as a single scan mirror, allowing the three scanning mirrors to act as a fast, dual-axis scanning system conjugated to the back pupil planes of the objectives. Imaging the RM onto the DMs in the 4Pi interference cavity via a 4f telecentric system (L8 and L9) guarantees the conjugation. A mirror (M10) between L8 and L9 folds the beam path from the optical table to the vertical breadboard plane. After this mirror, HWP3 rotates the two polarization orientations of the incident depletion beam to ±45° with respect to the polarization axis of PBS2, enabling precise 50/50 power allocation of both polarization components between the two arms of the 4Pi interference cavity. Simultaneously, because of its incident 45° linear polarization, the excitation beam is rotated to 0° by HWP3 and is therefore coupled into the upper arm only. HWP4 further rotates the polarization of the beams in the upper arm for subsequent interference in the combined focus.

The sample is mounted at the common focal plane of the objectives. Fluorescence from the sample is collected by both objectives, combined at PBS2, de-scanned by the scanner and then separated from the common beam path via a custom-made, 5-mm-thick quad-band-pass dichroic mirror (DI2). Two additional dichroic mirrors (DI5 and DI6) split the fluorescence into three detection bands for multi-color imaging. For each band, two filters of the same type (F4, F5 and F6) reject stray excitation and depletion laser light. The filters in each detection band are separated by about 50 mm and mounted at a small angle relative to each other to avoid interference between them. Fluorescence in each detection band is coupled into a multimode fiber acting as a confocal pinhole, with a pinhole size of about 0.8 Airy units. Each fiber is connected to a single-photon counting avalanche photodiode (APD) (APD1, APD2 and APD3).

The upper-right side of PBS2 is not used for acquiring STED fluorescence imaging; instead, it serves as the outlet for the intense depletion laser after passing through the 4Pi interference cavity. A photomultiplier tube (PMT), roughly conjugated to the objective focal plane, detects the non-de-scanned photons without going through a confocal detection pinhole. This allows the instrument to acquire illumination PSFs of the excitation or depletion laser foci without the influence of the detection PSF of the optical system¹⁴. This channel is used to assess the laser foci for aberration corrections.

The optical components installed at the back of the vertical breadboard are used to lock the positions of both objectives. A near-IR laser (LH3) is collimated by a lens (L20) and directed toward the objectives by a dichroic mirror (DI3). When the two objectives share the focal plane, the laser beam passing sequentially through OBJ1 and OJB2 is collimated. This beam is then reflected by DI4 and focused by L21 onto a camera (C) via a cylindrical lens (CL), intentionally introducing astigmatism to allow for easy detection of the collimation status.

The isoSTED nanoscope function requires meticulous optical engineering and precise alignment, especially aligning multiple planes to be conjugated and co-centered, including: (i) the apertures of the objectives, (ii) the DMs, (iii) the RM and (iv) the two circular holograms on the SLM for STED_xy and STED_z beams, respectively. Moreover, the conjugation of the detection pinhole with the common focus of the objectives in three dimensions is crucial. Detailed optical design specifications generated using Zemax (Ansys) can be accessed at the Zenodo repository (see Data availability).

Mechanical design overview

The mechanical design (Fig. 3a) prioritizes: (i) mechanical stability, (ii) easy maintenance, (iii) precise alignment and (iv) simple use. For these purposes, the optical system was initially designed using Zemax, and then the mechanical structure was crafted around it using SolidWorks to secure all optical components. According to this design blueprint, the system was assembled using commercially available optomechanical elements as well as custom-designed, precision-machined parts to allow for precise alignment and minimize drift and vibrations. This strategy has been used to build working systems at Yale University and ShanghaiTech University. We provide an overview of the mechanical design in the following paragraphs. For more specific information regarding individual components and their assembly, we refer the reader to Equipment setup.

Fig. 3 |. Mechanical design overview.

a, CAD rendering of the isoSTED nanoscope. b, Top view. c, Side view from the left showing the stage stack. d, Side view from the front showing the 4Pi interference cavity.

The mechanical design of the instrument is composed of two main bodies: components arranged horizontally on an optical table and those on a vertical optical breadboard. The components on the optical table can be categorized into depletion, excitation and detection paths, as shown in Fig. 3b. The components on the vertical breadboard can be categorized into a 4Pi interference cavity beam path and a so-called tower assembly, which holds the sample and objectives, as shown in Fig. 3c,d. The vertical beam path configuration allows biological samples to be positioned in a conventional horizontal orientation, in which the sample is sandwiched by two objectives in the direction of gravity.

The key part among the optics built horizontally on the optical table is a double-pass SLM configuration. The SLM is held by an xy-linear stage and a custom mount that can be rotated along the z-axis. The oblique 4f telecentric system is placed on a custom rail and a custom board, which defines the SLM’s angle. The remaining optics on the optical table are easily adjustable because they are mounted by optical cage and rail systems. However, a primary design constraint is that the height of all components should be centered on the midpoint of the SLM’s active area. In addition, within the depletion beam path, custom mounts are designed to hold translation lens mounts perpendicular to the optical axis and kinematic mirror mounts at a 45° angle. This design ensures precision, because the alignment accuracy of the depletion optics is crucial for achieving high-quality STED imaging.

The vertical optical breadboard is mounted to the optical table with two pillar legs. The breadboard has tapped holes for dowel pins based on the optical design, and the optical components attached to it are held by custom mounts that are screwed down to the breadboard. Therefore, the components can be fixed in the designated position with high precision. This approach also allows high repeatability for the removal and reinstallation of components. The key movable part on the breadboard is the PBS. The tilt and azimuthal angles of the PBS are controlled by a motorized goniometer and a motorized rotation stage, respectively. In the upper arm of the 4Pi interference cavity, a piezo actuator serves for translating the attached mirror at sub-nanometer resolution, thereby optimizing the phase difference of interference at the common focal plane by tuning the optical path length difference between the upper and the lower arms. In addition, two DMs for aberration correction are aligned to be co-centered with both the SLM and each objective back pupil. The second section of the vertical configuration is the stage stack in the tower assembly. In this assembly, the upper objective is translated along the optical axis through a long-range piezo linear stage, enabling nanometer-level precision in positioning, while affording a 26-mm travel range to accommodate easy sample insertion and removal. The lower objective is mounted on an xy piezo stage. In combination, these two piezo actuators allow for the relative alignment of the two objectives in all three dimensions. The sample is secured atop a pair of piezo stages, comprising an xy stage and a z stage. This assembly is supported by three high-resolution linear actuators that provide coarse sample z-positioning and tip/tilt leveling. The alignment strategy for this stage stack involves positioning the sample in the focal plane of the lower objective, followed by positioning the upper objective to ensure the convergence of both objectives onto the same focal plane. Lateral alignment is then achieved by optimizing the xy position of the lower objective. After precisely aligning both objectives, their positions are optically locked during the imaging process (see Focus-lock module).

Detailed mechanical drawings and computer-aided design (CAD) models are provided in the Data availability statement. Figure 4 shows the optical beam path overlaid on the mechanical layout of the instrument. This model is drawn to scale and intended to show the relationship between the optical components and their position within the instrument.

Fig. 4 |. Labeled diagram of the isoSTED optical beam path.

Note that the vertical optical breadboard shown on the top left is rotated in this diagram to a horizontal orientation for display purposes. FA, fiber adapter; LS, linear stage; N, nudger; R, optical rail; S, screw; SH, shutter; T, tube; TS, motorized translation stage. The remaining labels and light path colors are the same as those presented in Fig. 2.

Control design overview

The isoSTED instrument control is centered around a single Microsoft Windows-based personal computer (PC) for data collection, synchronization and device control (Fig. 5). Specifications for the instrument control PC are discussed in Computer and data storage. Notably, the computer motherboard must have at least two peripheral component interconnect express (PCIe) slots for the use of two National Instruments (NI) field programable gate array (FPGA) data acquisition (DAQ) boards that form the core of the image-acquisition hardware. Specifically, an NI PCIe-7852R DAQ FPGA board is used for data collection, and an NI PCIe-7841R DAQ FGPA board is used for waveform generation and synchronization. The details and connections of these FPGA boards are discussed in FPGA control. The control and image acquisition are coordinated through an instrument control interface programmed in the LabVIEW environment (available at the Zenodo repository listed in Code availability) and specifically tailored for controlling the isoSTED nanoscope. Modular control is used for the SLM, DM, synchronization, time gating and focus lock, and additional device control is integrated into a single, easy-to-navigate, unified user interface accessible to non-experts.

Fig. 5 |. Control design of the isoSTED nanoscope.

AO, analog output; C1, controller of the white-light excitation laser (NKT, SuperK EXTREME EXR-20); C2, controller of the 775-nm depletion laser (MPB Communications, cat. no. PFL-80-3000-775-B1R); C3, controller of the AOTF (Quanta Tech, cat. no. AOTFnC-400.650-TN); C4, controller of the AOM (AA Opto Electronic, cat. no. MTS40-A3-750.850); C5, controller of the SLM (Hamamatsu, cat. no. X10468-02); C6, controller of the motorized linear stage (Thorlabs, cat. no. MTS25/M-Z8); C7, controller of the vernier micrometers (Newport, cat. no. SM-25); C8, controller of the piezo actuator (PI, cat. no. P-841.10); C9 and C10, controller of the DM (Boston Micromachines, cat. no. Multi-DM5.5); C11, controller of the objective mount (PI, cat. no. N-565.260); C12, chassis of the E-621 modules; C13, controller of the linear actuator (PI, cat. no. M-227.10); C14, controller of the xy piezo stage (PI, cat. no. U751.24); C15, controller of the linear stage pair (Applied Scientific Instrumentation, cat. no. LS-50); D1, fanout line driver; D2 and D3, driver of the galvanometer mirror (CTI, cat. no. dynAXIS XS); D4, driver of the RM (EOPC, cat. no. SC-30), DB1–DB3, delay board (Opsero, custom designed); DIO, digital input/output; MO1 and MO2, monitor; USB, universal serial bus.

Applications

The isoSTED nanoscope has achieved isotropic 3D sub-50-nm resolution in various biological samples. One-color isoSTED imaging has shown up to 3-μm-thick images of subcellular structures, including microtubules, synaptonemal complexes and mitochondria, demonstrating high-resolution isotropic renderings of individually labeled targets¹². Two-color isoSTED imaging has shown relationships between structures such as mitochondria and endoplasmic reticulum¹², Golgi apparatus and Coatomer protein complex I¹², as well as coilin and survival of motor neuron proteins in a Cajal body¹⁵. Implementing adaptive optics into isoSTED nanoscopy enabled two-color imaging in tissues like Drosophila egg chamber ring canals and 30-μm-thick mouse hippocampal brain sections¹².

Advantages and limitations

isoSTED nanoscopy has so far achieved the highest 3D isotropic resolution among all STED technologies, thus providing an appealing tool for imaging biological structures. Like other far-field, super-resolution fluorescence microscopes, isoSTED can image biological structures with the high specificity enabled by immunofluorescence labeling. Similar to other STED approaches, isoSTED offers super-resolution imaging without post-processing and straightforward multi-color imaging. Two-color STED imaging using the same depletion laser is particularly suitable for biomolecule co-localization studies, because the two channels are intrinsically free of chromatic mismatch. Thanks to detection pinholes, which suppress out-of-focus fluorescence, isoSTED is better suited for thick-tissue imaging than other 3D super-resolution microscopy techniques, such as single-molecule localization microscopy¹⁶ or structured illumination microscopy¹⁷. Using AO, the isoSTED instrument can correct aberrations induced by the system as well as the sample to enable optimal imaging conditions even in thick and complex tissues.

However, the use of isoSTED nanoscopes has been limited by the complexity of the optical apparatus. Installation and operation of the instrument necessitate mastery of specialized alignment techniques and complex control software. Samples require sub-diffraction-limit gold nanoparticles and/or fluorescent beads added before imaging to enable aberration correction by the sensorless AO, which can constrain or complicate sample preparation. As for all STED nanoscopy techniques, specific fluorescent dyes are required for optimal image resolution, limiting the choice of reagents. In addition, the use of high-numerical aperture objectives requires refractive-index matching during sample preparation, adding yet another layer of complexity to the experimental protocol. These instrument and sample challenges underscore the need for a detailed protocol, such as this, to successfully navigate and use isoSTED’s capabilities.

In summary, this nanoscope offers exceptional 3D resolution below 50 nm isotropically in a wide range of samples. The guidance provided in this protocol makes this complex technology more accessible.

Procedure overview

Building an isoSTED nanoscope requires meticulous attention to detail. Although the order of some steps is flexible, many are interdependent and require a linear process.

Before starting this protocol, thoroughly reviewing the recommendations provided in Experimental design and Equipment setup is highly recommended.

The experimental setup begins with the installation of software (Steps 1–4), followed by the assembly of all optomechanical components (Steps 5 and 6) and all electronic connections (Steps 7 and 8). Subsequently, the depletion laser is mounted on the optical table to initiate the alignment of its beam path (Steps 9–41). Given the critical function of the depletion laser, its optical path serves as the reference for aligning the entire 4Pi interference cavity. Upon successful depletion path alignment, the excitation light path alignment is performed (Steps 42–62). Notably, the excitation and depletion beams should travel horizontally at the same height across the horizontal board referring to the center of the SLM. More precise tuning is available at each specific site to meet the designed requirement.

After combination of the excitation and depletion beams, the 4Pi interference cavity components are mounted on the front side of the vertical breadboard, and the 4Pi interference cavity is aligned (Steps 63–118). Once this is accomplished, the focus-lock module is installed on the back side of the same vertical breadboard (Steps 119–125). For optimal focus-lock results, a sample is inserted between the two opposing objectives during this process. The sample-preparation procedure is described in Steps 126–141.

Next, using this test sample, the detection path is aligned (Steps 142–157), which is the final step in setting up the hardware. Subsequent procedures, including adjustments to the pulse gate/delay module and the AO module settings (Steps 158–176), are performed exclusively within the control software interface.

Because the isoSTED nanoscope is a laser scanning microscope, the scanner (i.e., the combination of RM and GMs) needs to be calibrated to avoid image distortions and quantify the magnification of the instrument (Steps 177–193). Finally, both fluorescent beads and biological samples are imaged to assess the achievable resolution and evaluate the performance of the instrument under typical imaging conditions, respectively (Steps 194–204).

General beam alignment

To facilitate easier construction of the isoSTED nanoscope, we have implemented a rail system designed to be a fixed reference for beam direction. Dovetail rails are frequently used for assisting the beam alignment. An alignment camera that can be moved along the rail is used to monitor the laser beam’s position along this path. Notably, three common beam alignment procedures are regularly used while building the isoSTED nanoscope and are detailed below:

1. Box 1: Single-mirror beam direction alignment

2. Box 2: Dual-mirror beam path alignment

3. Box 3: 4f telecentric system alignment

BOX 1. Single-mirror beam direction alignment.

1. Mount the alignment camera on the rail near the beam source. Secure it in place and mark the current beam position with the cursor as the initial target position in the camera software.

2. Move the alignment camera along the rail in the beam propagation direction. If the beam is tilted relative to the rail axis, the beam position on the camera will be shifted off the cursor position.

3. Adjust the angle of the mirror to move the beam to the cursor position.

4. Move the alignment camera back to the near end of the rail (towards the mirror) and check whether the current beam position is on target to the cursor position.

5. If not, update the cursor position to the current beam position. Repeat steps 1–5 until the beam position does not change on the camera anymore when translating the camera along the rail.

   ▲ CAUTION The procedure in Box 1 aligns the beam direction along the rail axis while the beam position has been determined by the beam position when it reaches the mirror. Box 2 describes a procedure designed to align both.

BOX 2. Dual-mirror beam path alignment.

1. Position the alignment camera at the near end of the rail toward the light source. Secure it in place. This is the ideal position that the beam should target on the camera with the cursor in the software.

2. Adjust the angle of the first mirror (Mirror 1 in the drawing below) to move the beam to the target position.

3. Move the camera to the far end of the rail.

4. If the beam deviates from the target position, adjust the angle of the second mirror (Mirror 2 in the drawing below) to compensate for the observed shift.

5. Move the camera back to the near end of the rail.

6. If the beam position is shifted off the target position on the camera, repeat steps 1–6 until the beam position is at the target position on the camera along the rail.

   ▲ CAUTION It is unlikely that the manufacturer precisely centered the camera chip in its case. The ideal position that the beam should target is therefore not necessarily the center pixel of the chip. This position needs to be determined independently (i.e., by an iris on the optical mount).

BOX 3. 4f telecentric system alignment.

The alignment of three factors is imperative for each 4f telecentric system: (i) the collinearity of the optical axes of both lenses, (ii) the distance between the two lenses adjusted to achieve beam collimation and (iii) the beam centered on both lenses. Note that the drawing emphasizes the beam center tracing and ignores the focusing effect on the beam size by the lenses. The following alignment procedure, suitable for systems in which both lenses are mounted in xy translation mounts, is designed to achieve all three targets:

1. Before aligning the 4f telecentric system, the incident beam should have been aligned along the rail. Place the camera at the far end of the rail, well past the second lens position, and secure it in place. With the two lenses removed, mark the center of the beam on the camera as the target position using the cursor in the camera software.

2. Insert the first lens (Lens 1) and verify that the beam is incident to the lens at a normal angle. The beam position on the camera would be shifted off the target position if the center of Lens 1 is off that of the incident beam.

3. Translate Lens 1 in the direction perpendicular to the optical axis until the beam is centered on the target position.

4. Insert the second lens (Lens 2) on the rail at the distance to the first lens approximately the sum of the focal lengths of Lens 1 and Lens 2 (f1 + f2). Verify that the lens mount is oriented at a normal angle with respect to the outgoing beam. The beam position on the camera would be shifted off the target position if the center of Lens 2 is off that of the incident beam.

5. Translate Lens 2 in the direction perpendicular to the optical axis until the beam is centered on the target position.

6. Adjust the distance between the two lenses until the outgoing beam is collimated. Beam collimation can be confirmed either by checking with a shear plate for coherent (depletion) beams or by checking if the beam size remains unchanged when the camera is translated along the propagation direction.

   ▲ CAUTION To precisely locate the beam center, we use a top-hat 0/π hologram projected on the SLM. Through Fresnel diffraction, this hologram generates a distinct spot at the beam’s center when the beam is perfectly aligned with the hologram center. This spot’s size is adjustable and can be minimized at any distance along the beam’s path by altering the hologram’s inner radius. Critically, this spot always appears, regardless of the beam’s collimation, serving as a reliable and easily recognizable marker for the beam’s center, even with lenses present. To leverage this effect for lens alignment, one first records the spot’s position with a camera placed roughly at the focal plane of the lens. Upon lens insertion, any observed shift in the spot’s position necessitates a compensatory translation of the lens perpendicular to the beam propagation direction. When the spot shows no positional change after passing through the lens anymore, the beam and lens are co-centered.

   ▲ CAUTION In our isoSTED design, the angle of the lenses is inherently fixed and non-adjustable. To ensure precise alignment, we have implemented a dedicated datum plane within the lens mount (Supplementary Fig. S1). This datum plane serves as a reference to define the orientation of the lens mount, effectively determining the alignment of the lens within the optical system. This design approach guarantees stability and accuracy in the positioning of each lens, contributing to the overall reliability and performance of the isoSTED nanoscope.

   ▲ CAUTION To assess the collimation of the depletion beam in the isoSTED nanoscope, we use a shear plate interferometer. However, this approach is not suitable for evaluating the excitation beam originating from the white-light laser due to its limited coherence length. In lieu of interferometry, the collimation of the excitation beam is gauged by measuring the beam size. A perfectly collimated beam is expected to exhibit minimal dispersion with distance, rendering beam size measurements a practical and reliable method for assessing the collimation of the excitation beam in the isoSTED nanoscope.

These methods underscore the importance of precise beam control and alignment procedures in the construction process of the isoSTED nanoscope, ensuring optimal performance and reliability. For the other beam alignment for which the precision requirements are not discussed explicitly, a ±0.5-mm precision suffices in general.

▲ CRITICAL Repeatability in mounting the alignment camera on the rail affects the alignment’s precision. Make sure to mount the camera on the rail properly during the process. Accuracy in identifying the beam center affects the alignment precision as well. The methods in Boxes 1 and 2 are compatible only with collimated or slightly focused beams and may not work well with tightly focused laser beams.

AOM and AOTF alignment

The goal for aligning an AOM or AOTF is to optimize its diffraction efficiency, specifically maximizing the power of the output beam in the +1 order. To enhance this efficiency, the incident beam collimation, polarization and incident beam angle need to be aligned, and the frequency and amplitude of the radio frequency signal need to be adjusted.

Polarization-separation module alignment

Two beams with orthogonal linear polarization, co-aligned as STED_xy and STED_z depletion beams, respectively, are constructed with a single 775-nm pulsed laser. To ensure that these two depletion beams do not interfere with each other and create unpredictable intensity variations in the combined depletion PSF (Fig. 1), they need to be made incoherent. This means that their waves are not synchronized in phase, so their combined effect at the common focus of the two objectives is simply the sum of their individual intensities. To avoid interference between the STED_xy and STED_z beams, a polarization-separation module, inspired by the layout of a Michelson interferometer (Fig. 6), has been introduced. In this module, one polarization component of the laser undergoes a controlled delay relative to its orthogonally polarized counterpart. The optical path length difference between the two arms of the module is carefully optimized to introduce a critical delay: it must be longer than the coherence length of the laser pulses, thereby avoiding interference between the two beams, and sufficiently small so that both depletion pulses arrive shortly after the excitation pulse to allow for efficient depletion (see Pulse gate/delay modules). This synchronization feature is pivotal because it significantly influences depletion effects and, consequently, resolution. For our system, a suitable construction is when one arm is 12.5-mm longer than the other, corresponding to a 25-mm difference in optical path length or a temporal delay of 0.83 ns. At this difference, interference between the two polarization components is negligible, while depletion is still highly efficient.

Fig. 6 |. Polarization-separation module.

This module consists of an optical cage system centered around PBS1 and is used to split and delay the STED_xy and STED_z depletion beams. HWP1, HWP2, QWP1 and QWP2 are mounted in rotation mounts. The kinematic mirror mounts (for M3 and M4) can slide along the cage rod, allowing for easy adjustment of the lengths of two arms. The red arrows indicate the propagation direction of the depletion beam.

SLM alignment

A single reflective SLM is used to generate two adjacent circular holographic grating patterns, which modulate the STED_xy and STED_z beams, respectively. Because it is a polarization-selective device, the SLM acts only on light polarized in one specific direction. In our case, this means it modulates light that is linearly polarized parallel to the optical table (p-polarization), while having virtually no effect on light polarized perpendicular to it (s-polarization). A virtual 4f telecentric system is established by a single lens and a mirror at its focal plane. The mirror reflects the light back into the lens so that the 4f assembly projects the first holographic pattern to the second one. The depletion beam, coming from the polarization-separation module, consists of two incoherent beams with orthogonally linear polarization. Upon the first reflection off the SLM, one of the beams with p-polarization undergoes 0–2π vortex phase modulation programmed in the SLM, rendering it into the STED_xy beam. The reflected beams go through the virtual 4f telecentric system and project to the second holographic pattern. Because the beams pass a QWP (QWP3) twice, the polarization of both beams is rotated by 90°. At the second reflection on the SLM, the STED_xy beam is therefore unaffected, whereas the other, previously unaffected beam experiences the SLM modulation for the axial depletion (STED_z). In the isoSTED mode, the STED_z beam is a normal Gaussian beam profile modulated by a moderate defocus phase pattern to reduce side lobes in the microscope’s PSF¹¹.

Two factors in SLM alignment are critical for achieving high image quality in the isoSTED nanoscope later. First, the beams need to illuminate the SLM holograms uniformly (Fig. 7a–c), because non-uniform illumination causes the focus to tilt along the z-axis, causing asymmetry of the PSF (Fig. 7d–f). Second, the 4f telecentric system needs to be aligned for accurate image conjugation between two holograms so that both holograms can simultaneously be conjugated to the objective back pupils.

Fig. 7 |. Effects of illumination uniformity on depletion laser PSFs.

a, Uniform illumination on the SLM. The hologram displayed on the SLM is a top-hat shaped 0/π pattern (bottom-right). This specific holographic configuration was selected because of its heightened sensitivity to aberrations, thereby amplifying the distortion within the aberrated PSF. b, xz | _{y = 0} cross section of PSF under uniform illumination. c, Cross sections at different depths (z) in b. d, Non-uniform illumination on the SLM. The hologram added on the SLM is a top-hat shaped 0/π pattern (bottom-right). e, xz | _{y = 0} cross section of the PSF under non-uniform illumination. f, Cross sections at different depths (z) in e. Each figure shown is normalized to its own peak intensity. Scale bars: 0.5 μm.

Scanner alignment

To ensure uniform illumination across the entire FOV, the scanner is strategically located at an intermediate conjugation plane positioned between the SLM and the back pupils of the objectives (Fig. 2). It features one RM for fast-axis (x) scanning, complemented by two GMs operating synchronously to scan along the slow-axis (y). The two GMs establish a virtual tilt plane seamlessly coinciding with the resonant scanning plane. Because the RM is placed in a plane conjugated to the back pupil of the objectives, this arrangement allows for 2D beam tilting in the back pupil of the objectives (Fig. 8a). To achieve this, the rotation angles of the first (α) and the second (β) galvanometer mirrors satisfy the following relationship: αL₁ = −β (L₁ + L₂), where L₁ and L₂ denote the distance from the RM to the first GM and between the first and the second GMs, respectively. The negative sign in the equation indicates that the two GMs rotate in opposite directions during operation. Specifically, when, L₁ = L₂, α = −2β (Supplementary Fig. S2). In addition, synchronization between the RM scanning and laser blanking is important (Fig. 8b and Supplementary Figs. S3 and S4): as the RM oscillates, the focal point scans along the x-axis in a sinusoidal fashion within the focal plane (Fig. 8c). To mitigate image distortion, only the relatively linear region in one scan direction, corresponding to one-third of the period of the sinusoidal scan motion, is used for imaging. The remaining difference from the ideal linear scan motion would still cause distortions in the image, and a non-uniform sampling period, in which pixel dwell times are shorter when the scanner is moving faster, is therefore implemented to correct it. Furthermore, to reduce photobleaching, the synchronization signal from the RM also serves as the timing reference for laser blanking to turn off the laser illumination when no image data are collected (Fig. 8b).

Fig. 8 |. Beam-scanning concept of the isoSTED nanoscope.

a, Scanning mirror module (including one RM and two GMs). b, Synchronization and laser pulse trigger signals. c, Focus position in three dimensions during imaging. From top to bottom: the position of the focus in the x-, y- and z-directions. Note the different time scales of the axes. In the top row of c, the position of the focus in the x-direction is a sinusoidal curve, with an amplitude positively related to the control voltage of the RM. The synchronization signal (the periodic square wave in b) originates from the RM, and both have an identical period of 62.5 μs, corresponding to a 16-kHz scanning frequency. Because of internal circuit delays, the focus is not precisely at the x = 0 position at the rising edge of the synchronization signal. To avoid image distortion, only the relatively linear region of one scan direction is used for imaging. This interval, corresponding to 1/3 of the sinusoidal scan period (20.833 μs), is marked as the blue ‘Forward scan’ region. The time delay from the rising edge of the synchronization signal to the start of imaging is termed the ‘Trigger delay for imaging’. The linear region in the reverse scan direction, marked as the green ‘Reverse scan’, is not used for imaging, to simplify system synchronization. The time interval between these two regions is exactly 10.416 μs (1/6 of the synchronization period). The laser operates in slave mode relative to the RM, being turned on after a specific time interval following each rising edge of the synchronization signal. This interval is the ‘Trigger delay for laser’ shown in b. The laser must be active for the entire ‘Forward scan’ period. To minimize overexposure and prevent photobleaching, the laser is not kept on continuously. The optimal solution is to trigger the laser for a duration slightly longer than the imaging period, with imaging starting immediately after the laser turns on. In the middle and bottom rows of c, the waveforms of the focus y- and z-positions are identical to those of the control signals for the GMs and the piezo z-axis stage, respectively. Sync., synchronization.

Although recording data during the forward scan is straightforward, image formation with scanning in both directions requires reversing the backward scanning in which pixels are recorded. Consequently, specialized read-write buffers must be used to invert the image data. To simplify system synchronization, we have chosen to collect data only during the forward scan and not use the backward scan for imaging.

4Pi interference cavity alignment

In the 4Pi interference cavity setup depicted in Figs. 2 and 4, PBS2 is used to generate a counterpropagating pair of beams. To ensure correct alignment, the GMs and RM need to be powered on when aligning the post-scanner beam paths, especially the 4Pi interference cavity. Moreover, as at other locations in the isoSTED nanoscope, the setup incorporates two rails (R5 and R6 in Fig. 4), installed along the upper and lower arms of the 4Pi interference cavity. These rails serve as fixed references for the direction of the beams, ensuring easier reproducibility of the beam alignment.

Before the PBS2, the HWP3 rotates the polarization of the incoming depletion beam to ±45° relative to the polarization axis of the PBS2. This adjustment allows for precise power allocation at a 50:50 ratio between the two cavity arms. Concurrently, the polarization of the excitation beam, already polarized at 45° upon incidence at HWP3, is rotated to be parallel to the vertical breadboard. Consequently, this beam is exclusively coupled into the upper arm of the 4Pi interference cavity. In the upper arm, HWP4 further rotates the polarization of the depletion beams to match those of the lower arm, facilitating subsequent interference at the common focus. To minimize phase delay discrepancies between the two arms, M11 and W2 are translated within the upper and the lower arm, respectively. In addition, both arms of the interference cavity have a DM integrated in the plane conjugated to the back pupil of the respective objective. The DMs (DM1 and DM2) enable aberration correction and manipulation of the PSF, as detailed in DM alignment and Aberration correction. To ensure circular polarization of all beams at the focus, QWP4 and QWP5 are positioned right in front of the objectives. These components collectively contribute to the precise manipulation and control of beam polarization and phase, essential for optimal performance within the 4Pi interference cavity.

DM alignment

The alignment of the DMs includes both axial conjugation and lateral co-centering with designated optical elements. By conjugation with the RM of the scanner, the DMs in this system are at the same time conjugated to the objective back pupils and the SLM holograms. Designated positions at the end of rails R5 and R6 (rails in Fig. 4) denote where the DM surfaces should approximately reside (Fig. 9a,b). Mounting the alignment camera at the designed position allows verifying the SLM images before installing and fine-tuning the DMs. After offline calibration (for details, see SLM and DM calibration), the two DMs are installed at their designated positions on the vertical optical breadboard, roughly co-centered and conjugated with the SLM holograms. The DM mount provides the capability for axial (z) translation as well as lateral (xy) translation and tip/tilt angle adjustment, allowing for more precise conjugation and co-centering.

Fig. 9 |. DM conjugation and co-centering alignment.

a, Upper arm. b, Lower arm. Blue arrows indicate the reference positions for assembly of the alignment camera. The rails R5 and R6 should be installed in accordance with the CAD model, and the end surface of the rail carrier should coincide with the edge of the rails. The alignment camera is placed at the desired DM surface with this assembly. c–f, The alignment camera is installed at the objective lens mount when DMs are installed for aligning its conjugation (c,d) and co-centering to SLM holograms (e,f). c,d, Images of the DM surface at the back aperture of the objective; when unconjugated, a grid pattern is visible (c), which disappears upon conjugation (d). e,f, Images of both the DM and SLM at the position beyond the back aperture of the objective. To highlight their corresponding centers, a top-hat and a cross pattern are loaded to the SLM and DM, respectively. An asymmetric image is evident when the co-centering relationship is not satisfied (e), whereas symmetry is restored upon achieving co-centering (f). Scale bars, 1 mm.

Each DM is conjugated to the back pupil plane of its corresponding objective through a 4f telecentric system. A successful conjugation is indicated by the absence of grid patterns stemming from DM membrane surface diffraction at the objective’s pupil plane (Fig. 9c,d). For co-centering, a cross-shaped and a top-hat shaped 0/π phase pattern are loaded to the DM and SLM, respectively. The presence of an asymmetric PSF indicates a lack of co-centering alignment (Fig. 9e,f).

Aberration correction

Careful adjustment of the AO elements is crucial to guarantee good imaging performance of the system. The following steps outline the process: (i) correction of system aberrations, (ii) compensation for sample-induced aberrations, (iii) objective alignment, (iv) enabling the focus-lock function (for details, refer to Focus-lock module), (v) translation to the ROI and (vi) fine compensation of local sample-induced aberrations.

Step (i) is required only for the initial system setup and takes about 2 d. To correct sample-induced aberrations, steps (ii)–(vi) need to be performed for every sample. The duration of this process depends on the sample type. A typical full correction requires about 1 h and involves sequential optimization of Zernike modes 5–11 (astigmatism, coma, trefoil and first-order spherical aberration) and mode 22 (second-order spherical aberration). However, in the case of thick tissue samples, additional higher-order aberration modes should be corrected.

In our setup, we did not use a wave front sensor and opted instead for image-based aberration correction to optimize the Zernike or 4Pi coefficients¹⁸ of the back pupil(s). This approach involves a systematic procedure in which a series of images are captured while introducing varying degrees of specific aberration modes via the DMs or SLM. For each set of introduced aberrations, an image is acquired. We then use an image-quality metric, such as brightness or contrast, to quantitatively assess the fidelity of each acquired image. Once optimized, the Zernike coefficients can be applied during real-time isoSTED imaging without further delay. However, because local correction requires repeated sampling of a confined region, photobleaching may occur. If photobleaching is observed, the adjacent region can be imaged using the same Zernike coefficients without further correction.

Following previously established procedures¹⁹, our correction algorithm plots for each specific aberration mode determined image-quality metric values as a function of the applied aberration mode amplitudes. Fitting a (Gaussian) curve to these data, the algorithm then determines the optimal amplitude for this aberration mode that maximizes image quality. In particular, in thick samples, sample stability is critical for optimal aberration correction, because any vibration or movement during correction may reduce the accuracy of the fitting curve. This iterative process is completed for each of the aberration (Zernike) modes under consideration. This approach does not need a complex wave front sensing apparatus and offers an efficient and tractable means of aberration correction. By leveraging image-based techniques in tandem with systematic optimization algorithms, we achieve targeted aberration correction, thereby enhancing the overall imaging performance of the isoSTED nanoscope. Moreover, we anticipate that image quality could be further improved by incorporating an advanced AO technique²⁰.

Image matching of upper and lower objectives

This step minimizes the lateral offset between the FOVs of the upper and lower objectives. To achieve the interference that is essential for isoSTED nanoscopy, the laser foci produced by the two objectives must overlap to a high degree. This alignment is achieved by translating the lower objective using the two-axis piezo stage. Although this procedure is regularly applied during the later routine isoSTED operation, particularly when transitioning to a new sample, the initial setup often shows such strong misalignment between the objectives that it surpasses the travel range of the piezo stage. To address this, a coarse adjustment of the two FOVs is required by adjusting the threaded-rod pushers underneath the piezo stage with a hex key (Supplementary Fig. S5). During this adjustment, two beam spots near the PMT detector that are on the top-right side of PBS2 serve as the reference points (Fig. 2). These spots originate from the depletion beams split by PBS2 and passing through the cavity in clockwise and counter-clockwise direction, respectively. They can be observed with a camera or directly by eye using a paper screen. Adjustment is considered complete when these two spots align in position.

Excitation laser path alignment

When aligning the excitation laser path, it is imperative to carefully adjust it to be colinear with the depletion beam. In addition, critical elements include: (i) a filter, more specifically an AOTF in our isoSTED configuration, that selects the desired excitation wavelengths from the white-light laser beam for multicolor fluorescence excitation; (ii) temporal chromatic dispersion correction for the selected wavelengths within a white-light laser pulse to synchronize excitation with the depletion laser pulses and thereby achieve maximum stimulated depletion for all used excitation windows (for details, refer to Temporal chromatic dispersion compensation); (iii) optimal beam size magnification to adequately fill the back aperture of the objective, thereby maximizing excitation intensity; and (iv), analogous to the depletion beam, careful attention to the polarization of the excitation beam for appropriate power allocation between the 4Pi interference cavity arms.

Detection beam path alignment

The detection light path involves de-scanned collection of the fluorescence through a multi-band dichroic mirror that separates it from the common light path with excitation/depletion lasers. Subsequently, the fluorescence light is split into three spectral windows by additional dichroic mirrors (DI5 and DI6 in Fig. 2), and each of the three resulting beams is focused by a lens and coupled into a multi-mode fiber for collection by an APD. For the best collection efficiency, each fiber tip is conjugated to the common focus of the objectives in three dimensions, and the focused fluorescence is coupled to the fiber to the maximal efficiency.

Data analysis

In contrast to single-molecule localization microscopy or structured illumination microscopy, isoSTED nanoscopy does not need post-processing for image reconstruction. Nevertheless, data analysis is still of the essence for resolution quantification. In instances in which the sample exhibits sparse, small structural features, such as when imaging fluorescent beads, the resolution is assessed from the spot size by fitting a Lorentzian model function to the data (code available at the Zenodo repository; see Code availability). We caution against directly extrapolating system resolution from biological samples using Lorentzian fitting, because the intricate nature of such specimens may dramatically compromise accuracy. Instead, for a more comprehensive evaluation of attained resolution in images, Fourier ring correlation (FRC) methods²¹, nested loop fitting²² or decorrelation²³ prove to be viable analytical approaches. Notably, the reported resolution is a holistic measure that reflects both the effective PSF and the structure size of the sample. Therefore, for accurate resolution quantification, we recommend using small structures (ideally <20 nm). A comprehensive quantification of our isoSTED nanoscope’s resolution, performed using both fluorescent beads and various biological samples, has been extensively detailed in our previous work¹².

FPGA control

Signal generation and data collection are centered around two NI FPGA DAQ boards housed in a single PC running the LabVIEW software environment in Microsoft Windows. Specifically, an NI PCIe-7852R FPGA board is used for data collection and overall synchronization, and an NI PCIe-7841R FPGA board is used for waveform generation and laser power control and operates in response to triggering from the aforementioned 7852R FPGA board. This necessitates using a PC with a minimum of two PCIe slots on the motherboard (for specific requirements on the computer, refer to Computer and data storage and Nanoscope control computer). An overview of the key hardware connections for image acquisition is presented in Fig. 10. The two FPGA boards compose the core of the image collection hardware and are physically linked (internally with a real-time system integration (RTSI) cable) to share trigger signals. The 7852R FPGA board coordinates data collection from the APDs for light detection with the scan motion of the 16-kHz RM. The synchronization signal from the 16-kHz RM triggers waveform generation from the 7841R FPGA board for controlling GMs, as well as AOM and AOTF for the depletion and excitation beam blanking, respectively. Although these two FPGA boards sit inside the control PC, once running, they operate autonomously by executing code that does not rely on any timing or input information from the PC. The high data-collection speed required to work with a 16-kHz mirror prevent the use of software-timed data acquisition or on-the-fly adjustments. The 7852R FPGA board collects photon time traces from each APD (color channel) with a constant time interval. Because the RM moves sinusoidally, the sampling in space varies. To compensate for this effect, the time trace is processed by re-binning it into evenly spaced pixels (linearization) in space (rather than time). This process happens in real-time on the 7852R FPGA board. Furthermore, the speed of the 16-kHz RM results in pixel dwell times on the order of tens of nanoseconds, and in a single scan motion of the RM, very few photons are collected. Scanning a single line multiple times and accumulating the recorded signal is therefore necessary to achieve reasonable photon numbers. This accumulation happens in real time on the 7852R FPGA board. After a single line has been scanned a user-set number of times, the accumulated signal is sent by the 7852R FPGA board to the PC where it is concatenated by the LabVIEW program to the current image and displayed. The process continues until either the live image capture is stopped by the user, or the image capture ends after a set number of frames.

Fig. 10 |. FPGA-based DAQ key connections.

Two FPGA-based DAQ boards act as the core of the instrument-control and image-acquisition hardware. Key connections for image acquisition are presented here. Briefly, the 16-kHz resonance mirror synchronization signal acts as the master clock for image acquisition and is routed into the 7841R FPGA board, where it triggers waveform output to the two GMs and is used as the reference signal for blanking the excitation and depletion beams. The 7841R board internally passes the 16-kHz synchronization signal to the 7852R board, which triggers data collection from the user-selected APDs or the PMT. The 7841R FPGA board also controls the sample piezo z-position and opens and closes shutters in the cavity beam path.

Synchronization and timing

Several carefully timed events must be coordinated for successful isoSTED imaging. They are synchronized to two master timing signals that run independently but in combination govern STED data collection. First, the 16-kHz RM acts as the master clock for image data collection. This signal regulates when the FPGA boards check for photon arrivals from the APDs/PMT, blank the AOM/AOTF and trigger waveform generation for GM control. Second, the white-light laser (about 78-MHz repetition rate) acts as the master clock for laser pulse synchronization and detector gating. The electronic synchronization signal from the white-light laser triggers pulse generation for the depletion laser and triggers the detector gating boards for gated photon detection.

For image data collection, rising edges from the APDs, each representing a detected photon, must be registered to the current scan position of the RM scanning a line of the sample. The 16-kHz RM synchronization output transitions between high (+5 V) and low (0 V) each time the mirror changes direction. When the 7841R board detects a rising edge from the RM synchronization signal, marking the start of a new line scan in the forward direction (Fig. 8), the signal is passed to the 7852R FPGA board. Once this rising edge signal is detected, the 7852R FPGA board waits for 5.208 μs before an approximately 20-μs photon time course with 5-ns resolution from each APD is collected (Fig. 10), corresponding to data collection only from the central approximately one-third portion of the period in the scan motion for image formation. This limitation ensures that data are collected only over the relatively linear portion of the mirror scan. During the data collection time course of the 7852R FPGA board, it records the time of each APD output rising edge. Each APD output is checked every 5 ns. The deviation of the scanner motion from an ideal linear scan pattern results in the regular 5-ns temporal binning corresponding to an unevenly binned sampling pattern in space. To correct this, the temporal bins of the photon time course are re-binned to create evenly spaced pixels in sample space. Detector gating and laser synchronization are set independently from photon detection using dedicated electronics described below.

For optimal STED resolution, the depletion and excitation laser pulses must arrive in close (≪1 ns) succession at the sample. In the system presented here, the white-light laser cannot be externally triggered and, therefore, has to act as the master (laser) clock. The electronic trigger output signal from the built-in photodiode of the white-light laser is converted and split into three lines to act as the input, or reference signal, for three custom pulse gate/delay modules designed and built by Opsero Electronic Design. Each pulse gate/delay module has two inputs: one for the white-light laser reference signal and the other for an APD detector signal. The modules are used for two separate functions: they can (i) introduce a user-defined delay between a trigger input and a trigger output pulse, and (ii) they can be used to pass a connected APD signal only for a user-defined time window after a trigger input. For the first function, the pulse gate/delay modules can electronically delay the input signal from the white-light laser up to 20 ns in 10-ps steps. The delayed trigger signal is passed to the depletion laser via a signal translator, allowing their pulse arrival time at the sample to be matched with the white-light laser pulse’s arrival. In practice, the timing between the depletion and excitation lasers is optimized by continuous imaging of beads until the best resolution (if a vortex hologram has been applied to the SLM) or nearly complete depletion (if the SLM is displaying only a flat and gratings) has been achieved. The second function of the pulse gate/delay modules allows them additionally to be used for gating the signal from the APDs. Specifically, when an electronic pulse is detected, the pulse gate/delay module waits for a programmable time period (the ‘delay’) before allowing the APD signal to pass to the output and on to the connected 7852R FPGA board. The delay between the white-light synchronization pulse and the start of the detection window can be adjusted between 0.1 and 9.8 ns in 10-ps steps. Similarly, the detection window width can also be adjusted within the same range with the same precision. For window-adjustment purposes, the detection window width can be set to the minimal value, and the delay relative to the excitation pulse is scanned until the strongest fluorescence signal is observed. This synchronizes the detection window with the excitation pulse. The window delay is then adjusted to be about 1 ns after the excitation pulse to allow the depletion pulse to act before fluorescence is detected for the best resolution, and the window width is opened to several nanoseconds to cover the fluorescence lifetime for a full detection.

Experimental design

Personnel

At least one full-time experienced microscope developer will need about 12 months to build an isoSTED using this protocol.

Budget

Building an isoSTED nanoscope will require ≥US $400,000. Major high-cost components include the following: actuator stages, about US $80,000 in total; an excitation laser, about US $60,000; a depletion laser, about US $28,000; DMs, about US $49,000; objectives, about US $22,000; an SLM, about US $15,000; and FPGA boards, about US $10,000. In addition, approximately US $22,000, US $29,000 and US $49,000 are required for custom parts, optomechanical parts and optical elements, respectively.

Purchasing and delivery

We highly recommend purchasing parts far ahead of time. It takes in our experience 1–2 months to collect quotations and place orders. Some critical system components require up to 6 months, in extreme cases even longer, to be delivered after placing purchase orders and should therefore be ordered as early as possible. These parts include in our experience the deformable mirrors, actuator stages, the SLM, lasers, FPGA boards and objectives.

The pair of objectives used in the system needs to match magnification within 0.3% to ensure sufficient image overlap throughout the FOV. Therefore, we strongly recommend asking the vendor to provide a pair of magnification-matched objectives. After delivery, we recommend imaging a Ronchi ruling with micrometer scale for a magnification comparison. The same region should be measured using each objective on a widefield microscope stand. Fourier analysis can be used to measure the average distance between the line spacings.

Optical table and device setup

A minimum size of 1 × 1.5 m² is required for the optical table. The optical table top should have a honeycomb core structure and dampers. The table legs should be equipped with automatic leveling vibration isolators. After the optical table has been installed, verify its leveling and check that the vibration-isolation legs are operating properly by pushing the optical table top up and down to confirm an adequate range of motion. This quick verification process should be repeated on a daily to weekly basis, especially during system installation, because some vibration isolators require, in our experience, regular tuning.

Because the nanoscope presented here includes many devices with controllers and power supplies, we recommend installing an overhead shelf or a 19-inch rack for the controllers and power supplies, rather than placing them on the optical table. Devices featuring fans should under no circumstances be placed on the optical table during the final phases of system installation and later operation. Cables connecting these devices to the optical instrument need to be attached firmly to the table surface or other heavy, rigid structures to isolate the optical instrument from vibrations emerging from the vibration sources.

Laboratory recommendations

To achieve the best performance, it is required to install the isoSTED nanoscope in a dark, temperature-controlled (temperature stable within ±0.5 °C over extended periods of time, including nights and weekends) and low-vibration (Vibration Criteria curve VC-C or better) laboratory. Thus, we recommend a laboratory space without windows located on the ground level or basement. In addition, an isoSTED nanoscope requires minimal environmental vibrations to preserve the axial resolution enhanced by light interference, and thus the laboratory space must be located away from heavy machinery and foot traffic. We strongly recommend that builders carefully consider air conditioning systems to prevent direct airflow on the system, because this can also cause fluctuations in the interference pattern. The laboratory space should be large enough for the optical table, a computer desk, a chair and either a 19-inch rack system or an overhead shelf for placing controllers and power supplies along with sufficient space for easy personnel access from at least three sides of the optical table. We recommend having an uninterruptible power supply system as a backup power option for unexpected power outages. The laboratory recommendations, including a floor plan example, are listed in detail as Supplementary Table S2.

Machine shop selection

The system presented in this protocol includes many custom-made parts that require a tolerance of ±0.1 mm, an angular tolerance of ±0.5°, a surface finish of 1.6 μm and anodization. We recommend choosing a nearby machine shop with quick turnaround times, because in our experience some errors are inevitable, and faster return of revised parts will save time.

Alignment dowel pins

Alignment dowel pins are used to assemble the custom parts in their designated positions. Using the dowel pins ensures that parts can be positioned precisely and reproducibly. Thereby, the alignment dowel pins should fit with a small tolerance; for example, the diameter of holes should be between 4.025 and 4.050 mm for a 4-mm-diameter dowel pin. In many cases, the diameters of holes are reduced after anodization. Therefore, we recommend that builders use a hand-reamer to enlarge the holes as needed. Dowel pins should be selected to be longer than the depth of the hole (manually confirm the hole depth before inserting any dowel pin) to prevent pins from disappearing in a hole, in which case it is extremely difficult to retrieve them.

Laser safety

Setting up and using an isoSTED nanoscope as described in this protocol requires careful consideration of laser safety. This system uses high-power lasers (>3 W), posing a significant safety hazard. In particular, the 775-nm depletion laser can appear dim to the human eye because of its long wavelength. This may lead to an underestimation of the associated exposure risks. Therefore, collaboration with local laser safety personnel is essential during system construction and operation, and adherence to established safety protocols is paramount. In addition, maintaining a fully enclosed system and ensuring laser interlocks on either the enclosure or laboratory door is crucial to prevent accidental exposure. During setup and alignment, extra precautions must be taken due to temporarily open beam paths.

Alignment camera

In the alignment procedure, the beam needs to be frequently inspected at specific locations of the beam path. For instance, to confirm the conjugation and co-centering of the SLM and a DM, reflections from both device surfaces need to be imaged in the position of the corresponding objective’s back pupil plane. Although these tasks can in general be executed by eye with a screen made, for example, of paper, using an inexpensive camera is highly advantageous because it allows the researcher to magnify, quantify and record the beam profile at the inspected location.

Many complementary metal-oxide-semiconductor monochrome universal serial bus (USB) camera models can be used as an alignment camera. The main criterion is that the camera should be compact enough to fit into the optical path without colliding with existing optical elements. In addition, the position of the sensor within the camera body should be known so that the sensor can be placed at the correct location dictated by the optical design of the instrument’s beam path.

In this protocol, commercially available cameras (UI-1240LE) from IDS were used. To minimize the camera size, we replaced the camera sensor housing with a custom-designed mount that ensured the mechanical centering of the sensor and established a known distance between the sensor surface and the mounting flange. The CAD file of this mount is available at the Zenodo repository in Data availability.

Nanoscope enclosure

The enclosure surrounding the isoSTED nanoscope is an integral part of the system, serving several crucial functions. It effectively blocks ambient light. Moreover, it has a vital role in eliminating the impact of air fluctuations on the system’s stability. In addition, it is thermally insulated and accommodates heating pads and temperature sensors that can be used optionally to maintain the entire system at a constant temperature. This stabilizes the system thermally, thereby minimizing thermal drift, and makes the system, in principle, compatible with live-cell imaging by allowing heating to 37 °C.

This protocol does not provide a specific design for the nanoscope enclosure. An appropriate enclosure will depend on the dimensions of the optical table and the arrangement of the nanoscope body, electronic cables and drivers on or surrounding the table. When designing an enclosure, it is essential to avoid mechanical resonances that could exacerbate air vibrations.

Computer and data storage

A single computer is sufficient for instrument control, data collection and processing. A typical 3D stack is <100 MB in size, and a typical experiment generates 1–5 GB of data. Two solid state drives (SSDs) should be installed for fast data write/read access (e.g., a 1-TB SSD drive for the Microsoft Windows Operating System and software installation and a 2-TB SSD for data storage. A unique feature is that two FPGA boards for fast data acquisition and device control are required and need to be connected to the PCIe slots on the motherboard. A 14-port USB hub is connected to the computer’s internal USB port to expand the device connections. A 12-core Intel Core i7-7800X CPU at 3.5 GHz is installed as the computer’s central processing unit. A video graphics card is installed for graphics processing.

FPGA

The control architecture for the entire nanoscope system revolves around two FPGA boards (Fig. 11). They are connected to the PC via PCIe slots on the motherboard. Supplementary Table S3 lists all active connectors on each FPGA board in our nanoscope, along with their corresponding functions. In brief, the analog signals from the first FPGA board (NI, PCIe-7841R) are used to actuate the piezo stages, scan the GMs and RM and adjust the laser beam intensity by modulating the transmittance of the AOM and AOTF. Digital outputs from the same FPGA board are used for shuttering and AOM/AOTF blanking. These devices also provide status-monitoring signals as feedback to this FPGA board. The second FPGA board (NI, PCIe-7852R) receives and counts signal pulses from the APDs and transmits the count numbers to the nanoscope control software, which generates the fluorescence image. The two FPGA boards are synchronized via an RTSI bus.

Fig. 11 |. Simplified data flow, control flow and light path.

DVI, digital visual interface.

Pulse gate/delay modules

The application mechanism of the pulse gate/delay module is detailed in Synchronization and timing and recapitulated in Fig. 12a–c. To achieve optimal performance, it is critical to optimize the three parameters in the LabVIEW control software (i.e., ‘window width’, ‘window delay’ and ‘laser delay’). The optimal setting is characterized by imaging a fluorescent bead sample with a single objective. First, the optimal ‘window width’ is set at approximately the spontaneous fluorescence lifetime (a few nanoseconds) to collect the most fluorescence signal without adding excessive noise. Second, the optimal detection ‘window delay’ for STED imaging is set about 1 ns after the maximal fluorescence detection is achieved with the depletion laser off (Fig. 12d). The additional delay avoids detection of ‘unwanted’ fluorescence from the peripheral excitation volume before the stimulated emission is completed. Last, one of the pulse gate/delay modules is used to activate the depletion laser at a user-defined delay. The optimal ‘laser delay’ is set at the minimized fluorescence, indicating the most effective fluorescence depletion, when the depletion PSF is switched to a Gaussian pattern (Fig. 12e). Once all the three parameters are optimized, only the fluorescence from the fluorophores within the central region of the depletion focal spot is detected, leading to improved resolution and image contrast (Supplementary Fig. S6).

Fig. 12 |. Time-gated detection.

a, Principle of time-gated detection in the isoSTED nanoscope. This diagram shows the time-correlated fluorescence photon counting on the detector under the time sequence of excitation pulse, depletion pulse and signal detection. The time-gated detection is characterized by the depletion laser delay (T_l), the detection window delay (T_d) and the width of the detection window (ΔT). At the center of the ring-shaped depletion PSF, the intensity is zero, and the fluorophore lifetime remains unchanged (‘wanted’ fluorescence, orange line). At positions that coincide with the intensity maxima of the depletion PSF, the fluorophore lifetime is dramatically reduced (‘unwanted’ fluorescence, orange dashed line). By recording the arrival time of photons when imaging, it is thus possible to increase image resolution by constructing the final image only from photons that arrive after a certain time; that is, early photons from the periphery of the effective PSF are excluded from recording. b, Internal logic of the circuitry (namely, ‘pulse gate/delay module’) used to implement the time-gated detection. c, Pulse trains at each connector on the pulse gate/delay module. d, Typical fluorescence photon counting as a function of the detection window delay setting (T_d′) on the pulse gate/delay module. The optimized fluorescence collection for STED imaging is set about 1 ns after the peak (highlighted by a red arrow). e, Typical fluorescence photon counting as a function of the depletion laser delay setting (T_l′) on the pulse gate/delay module. To quantify the depletion effect, the depletion PSF is switched to a Gaussian pattern. The minimal fluorescence (highlighted by a red arrow) indicates the optimal delay set for the depletion laser. The optimal values of T_d′ and T_l′ need to be calibrated for each system.

Refractive index considerations

To guarantee optimal resolution across a depth of up to tens of micrometers within the sample, it is imperative to minimize the refractive index (n) mismatch between the sample, its mounting medium and the objective immersion medium. The isoSTED nanoscope uses two silicon-oil immersion objectives (n = 1.406 (ref.24)), which provide a good compromise for a large variety of samples ranging from fluorescent beads and cell and tissue samples in ProLong Diamond antifade mountant (Invitrogen, n = 1.47 under coverslip²⁵) to potential live-cell applications (n = 1.37–1.39)²⁶. In addition, the DMs in the 4Pi interference cavity compensate for residual aberrations.

Multi-color imaging

Multi-color STED imaging is achieved with three excitation beams, which have wavelengths of 488, 590 and 650 nm, respectively, and one depletion beam at 775 nm. To separate the illumination and fluorescence detection beam paths, a multi-band dichroic mirror is used (Fig. 13a). In the detection path, two long-pass dichroic mirrors separate three channels (Fig. 13b). Generally, dyes with well-separated spectra are used (e.g., ATTO647N or STAR635P for the red channel and ATTO594 for the orange channel). In addition, three-color STED imaging can be achieved using a dye with a wide emission spectrum (e.g., STAR 520SXP) because its fluorescent signal can be detected simultaneously by the two/three APDs (see Supplementary Fig. S7 for details) and depleted by the same 775-nm laser.

Fig. 13 |. Spectrum configuration for multi-color imaging.

a, Transmission spectrum of DI2 (black solid line), laser wavelengths (vertical lines labeled with ‘488’, ‘590’, ‘650’ and ‘775’). b, Transmission spectra of DI5 (orange solid line) and DI6 (red solid line) in the detection path. Both figures show additionally the three emission filter spectra of the APDs as shaded areas.

Focus-lock module

In isoSTED nanoscopy, optimal resolution can be achieved only when the two opposing objectives are consistently aligned to a shared focal point termed the ‘focal zero’. Maintaining objective alignment during imaging is critical and requires lateral (xy) and axial (z) synchronization. When the PSFs fail to overlap laterally, the interference pattern that is necessary for this technique cannot be created. Axial misalignment of the objectives results in a shift/distortion of the interference fringes, thereby degrading the imaging quality.

To maintain objective alignment within ±5-nm tolerance during imaging sessions, our nanoscope incorporates a focus-lock module (Fig. 14). Within this setup, a continuous-wave laser operating at 980 nm is coupled into the upper arm of the 4Pi interference cavity by DI3. The laser beam is then focused by the upper objective and collected again by the lower objective. Subsequently, DI4 separates the beam from the imaging path. The focus-lock laser beam passes a 400-mm-focal-length lens (L21) followed by a weak CL to introduce astigmatism before being projected to a camera to observe the resulting cross-shaped astigmatic focal spot. Throughout the imaging process, both the position and the shape, more specifically, the aspect ratio, of the resulting focal spot are recorded and monitored. The spot’s positional changes indicate the relative x and y alignment of the objectives, while changes in its aspect ratio indicate drift in the z-direction. When changes are detected, the piezo objective stages are translated accordingly for compensation using a proportional-integral-derivative feedback control loop programmed in LabVIEW.

Fig. 14 |. Overview of the focus-lock module.

An NIR laser (LH3) is launched and collimated by L20. The beam is sequentially reflected by M20, M21, M22 and DI3. The beam is then focused and collimated by OBJ1 and OBJ2, respectively, and directed to the back panel by DI4, M23 and M24. L21 focuses the beam on a camera (C) via M25 and a filter (F7), while CL deliberately introduces astigmatism into it.

This focus-lock module facilitates prolonged imaging sessions exceeding 3 h. The alignment laser has negligible impact on the sample or image thanks to its low laser power (<5 μW) and its 980-nm wavelength being far longer than the laser and emission spectra used for sample imaging.

Temporal chromatic dispersion compensation

In the context of white-light laser pulses, temporal chromatic dispersion describes the phenomenon in which the light at different wavelengths within the pulse spreads out in time as the pulse propagates. It originates from the wavelength-dependent refractive index of various optical elements in the beam path, which causes the light at different wavelengths to travel at different speeds. In our isoSTED nanoscope, the temporal mismatch between pulses of different excitation wavelengths is primarily caused by the white-light laser itself and can be as long as several hundred picoseconds (Fig. 15a). Because the maximum depletion effect is achieved only when an excitation pulse is immediately followed by the depletion pulse illuminating the sample, dispersion needs to be compensated to match the delays between the depletion laser pulse and the excitation laser pulses for the three different color channels.

Fig. 15 |. Temporal chromatic dispersion of the white-light laser and its compensation strategies.

a, Temporal chromatic dispersion of the white-light laser. The original data were measured by the laser manufacturer (NKT Photonics). b, Schematic and principle of our temporal chromatic dispersion compensator. The incident excitation light with linear polarization passes completely through the PBS after an HWP adjusts its direction of polarization. The three main components of the beam with wavelengths of 650, 590 and 488 nm, are respectively reflected on three adjacent filters, which produce different relative time delays due to the length of the optical path. The polarization of the reflected beam is rotated 90° as it passes through a QWP twice and finally exits the module from the reflecting path of the PBS. In our design, all filters are 1 mm in thickness.

To correct for the color-dependent temporal mismatch, we introduced a compensation module (Fig. 15b) to the excitation beam path. The linearly polarized excitation beam, which consists of three chromatically dispersed wavelength bands selected by the AOTF, enters this module through a PBS. Three long-pass filters (Supplementary Fig. S8 and Supplementary Table S4), each reflecting one of the three wavelength bands, are placed sequentially behind the PBS. The spatial distance between adjacent filters introduces a temporal delay to ensure that the light with different wavelengths excites the sample simultaneously. The beam passes twice through a QWP oriented at 45°. As a result, the polarization direction of the reflected beam is rotated by 90° so that it is reflected by the PBS on its way out.

Resolution measurement

To evaluate the resolution of the isoSTED system, sub-resolution-sized objects are imaged, and their full width at half maximum (FWHM) values are measured. We recommend using either 100-nm crimson fluorescent microspheres or DNA origami structures. After acquiring the bead images, intensity profiles are fitted with a Gaussian or Lorentzian function for confocal imaging or STED imaging, respectively, using MATLAB analysis software (code available at the Zenodo repository listed in Code availability). The FWHM of the effective PSF, or the resolution, can then be estimated as (Supplementary Note S1)

$${\text{FWHM}}_{\text{PSF}}=\sqrt{{\text{FWHM}}_{\text{image}}^2-{\text{FWHM}}_{\text{bead}}^2}\text{.}$$

With a spherical model for the beads, ${\text{FWHM}}_{\text{bead}}=\sqrt{3}d/2$, and d is the diameter of the beads.

When assessing the resolution of a biological specimen, such as tissue, Fourier ring correlation is calculated with two images that are consecutively acquired in an identical FOV. This analysis is performed using an open-source ImageJ plugin. We additionally recommend considering the ‘nested loop fitting’ approach²² that allows users to assess the resolution from tubular-shaped structures in biological samples (e.g., immunolabeled microtubules).

Besides resolution quantification, users can perform cross-validation using conventional STED or confocal microscopes to verify system performance. In our isoSTED nanoscope, conventional STED imaging can be achieved by imaging with the upper objective only (open the upper shutter while closing the lower shutter), and confocal images are collected by turning the depletion laser off. Fluorescent beads or well-characterized biological structures such as microtubules in fixed cells are suitable as standard samples.

Data format and image visualization

Similar to other confocal and STED microscopes, raw images from our isoSTED nanoscope do not require image post-processing before displaying and quantification. As the sample moves incrementally along the z-axis, our nanoscope can acquire snapshot images of each optical section. These optically sectioned images are later compiled and stacked, directly forming a 3D representation of the sample.

To store the raw image data, we use the ‘.dbl’ format, a binary file containing data stacks of up to four dimensions. This format includes essential header information about the physical dimensions of the sampled volume. Access to these files is available through our nanoscope control software or via Imspector. Moreover, our custom MATLAB code (available at the Zenodo repository listed in Code availability) easily converts these files to the more universal ‘TIFF’ format without compromising information integrity.

After data conversion, numerous software options can facilitate visualization. A good open-source option is, for example, FluoRender²⁷, which was used to visualize the 3D datasets in this protocol.

Materials

This section details all materials required to build an isoSTED nanoscope, along with a step-by-step guide for setting up these components. A complete material list, including reagents, tools, control computer, software and drivers, can be found in Supplementary Table S5. The section also spans a wide range of procedures, from the preparation of fluorescent beads and biological samples to the calibration of adaptive optics devices, the assembly of optomechanical components and software installation. Furthermore, this section provides guidance on key calibration and alignment procedures to ensure optimal system performance.

Reagents

• Precision #1.5H glass coverslips, 25-mm diameter (Thorlabs, cat. no. CG15XH)

• 39-mm-diameter anodized aluminum sample holder with 25.3-mm-diameter, 500-μm-deep pocket (custom made; for a technical drawing, see FAB-P0009 (available at the Zenodo repository listed in the Data availability section))

• 100-nm crimson beads (625/645 nm; Thermo Fisher, custom order)

• 100-nm red beads (580/605 nm; Thermo Fisher, cat. no. F8801)

• 100-nm yellow-green beads (505/515 nm; Thermo Fisher, cat. no. F8803)

• 23-nm-diameter fluorescent beads, DNA origami conjugated to ATTO-647N (GATTA-Beads R; GATTAquant DNA Nanotechnologies)

• Biotinylated BSA solution (1 mg ml⁻¹ in PBS; GATTA-Beads R kit; GATTAquant DNA Nanotechnologies)

• Neutravidin solution (1 mg ml⁻¹ in PBS; GATTA-Beads R kit; GATTAquant DNA Nanotechnologies)

• Gold nanoparticles (150-nm diameter; Sigma-Aldrich, cat. no. 742058–25ML)

• ProLong Diamond antifade mountant (Invitrogen, cat. no. P36970)

• Picodent TwinSil (two-component silicone adhesive; Picodent, cat. no. 1300 1000)

• Poly-L-lysine (Sigma-Aldrich, cat. no. P4707)

Equipment

Tools

• Mechanical tools: hex key set, metric screw driver set

• Cage-alignment plates (e.g., Thorlabs, cat. nos. SCPA1 and CPA1)

• Dovetail rails (e.g., Thorlabs, cat. no. RLA450/M)

• Shim spacers (e.g. stainless steel slotted shim; McMaster-Carr, cat. no. 9722K812)

• Oscilloscope (e.g.,Tektronix, cat. no. MDO3054)

• Optical power meter digital console (e.g., Thorlabs, cat. no. PM100D) and slim photodiode power sensors (e.g., <500 mW; Thorlabs, cat. no. S130C) or thermal power sensor head (e.g., 2 mW–10 W; Thorlabs, cat. no. S425C)

• Multimeter (e.g., IDEAL, cat. no. 61–360)

• Absolute height gauge (e.g., digital absolute (ABS) height gauge; Mitutoyo, cat. no. 570–244)

• Level (e.g., 1.25-inch circular bullseye level; Thorlabs, cat. no. LVL01)

• Micrometer (e.g., CD-6-inch CSX; Mitutoyo)

• Laser safety goggles (e.g., certified laser safety glasses; Thorlabs, cat. no. LG2)

• IR laser card (e.g., VRC2 visible/near IR (NIR) detector card; Thorlabs)

• Shearing interferometers (e.g., Thorlabs, cat. no. SI254)

• Alignment camera (e.g., IDS, cat. no. UI-1240LE)

• UV lamp (e.g., 365 nm; UVL-18 EL; Analytik Jena)

• UV-curing optical adhesive (e.g., Thorlabs, cat. no. NOA81)

• DC servo controller/driver (Newport, CONEX-CC)

• Neutral-density filter (e.g., Thorlabs, cat. no. NE20A)

• Laserline cleanup filters (Semrock, cat. nos. LL01-647-25, FF01-591/6-25 and LL01-488-25)

• Linear polarizer (e.g., Thorlabs, cat. no. GT15-A)

• A 1-inch-diameter, 40-mm-focal-length optical lens assisting scan mirror alignment (e.g., Thorlabs, cat. no. LA1422-B)

Optomechanical components

The system is composed of numerous commercial and customized optical and optomechanical components. A full part list is available in Supplementary Table S6, and a SolidWorks model of the system assembly is available at the Zenodo repository listed in Data availability.

Nanoscope control computer

• Operating system: Windows 10 Pro, 64 bits

• CPU: 12-core Intel(R) Core (TM) i7-7800X CPU at 3.5 GHz

• Random access memory: DDR4, 32 GB

• Graphics processing unit: NVIDIA GeForce GTX 1060 6G

• Hard drive: Samsung SSD 970 EVO 1 TB and WDC WDS200T2B0A-00SM50 2 TB

• Motherboard: ASUSTeK WS X299 SAGE

Nanoscope control software

• National Instruments, LabVIEW 2018 32-bit device drivers, Virtual Instrument Software Architecture (VISA), Measurement & Automation Explorer (MAX), Vision Development Module and FPGA Development Module

• Boston Micromachines, LinkUiMulti, DM Shapes

• MPB Communications, Pulsed Fiber Laser Control

• NKT Photonics, CONTROL

• Physik Instrumente (PI), PIMikroMove

• Thorlabs, Kinesis

• Applied Scientific Instrumentation (ASI), ASI Console

• IDS Imaging Development Systems, Camera Manager and uEye Cockpit

• Quanta-Tech (AA Opto-Electronic), MPDSnC

• Opsero, LUFA (Lightweight USB Framework for AVRs) VirtualSerial.inf

• Hamamatsu Photonics, liquid crystal on silicon (LCOS)-SLM Control

Data visualization software

• MathWorks, MATLAB 2018a 64-bit or later

• University of Utah, FluoRender 2.22.3 or later

• Abberior Instruments, Imspector 0.13.11588 or later

• National Institutes of Health, ImageJ 2.3.0/1.54b or later

Reagent setup

Fluorescent bead samples

Fluorescent bead samples are required during instrument alignment and resolution measurement steps. DNA origami beads labeled with ATTO 647N with a diameter of 23 nm are particularly recommended for the resolution test. In our experimental setup, fluorescent and gold nanoparticles were deposited on a 25-mm-diameter round coverslip, which was subsequently covered with another 25-mm-diameter round coverslip using the ProLong Diamond antifade mountant. This sandwiched configuration ensures optimal imaging conditions. The assembled coverslips were securely positioned within a custom aluminum sample holder and firmly secured using two-component silicone adhesive.

Biological test samples

ATTO 647N-labeled microtubule samples are recommended for testing the imaging performance of the system. The immunolabeling process for isoSTED nanoscopy closely mirrors that for other STED nanoscopy applications and has been previously described (e.g., in Microtubules in ref. 12). For isoSTED imaging, we recommend additionally depositing both 150-nm-diameter gold nanoparticles and 100-nm-diameter fluorescent beads on the opposing coverslip to evaluate and modify system alignment before imaging the biological sample. Start preparing the coverslip with bead samples by mixing gold nanoparticles and fluorescent beads at a 1:10,000 (vol/vol) dilution. Add 25 μl of poly-L-lysine to the coverslip, allow it to sit for 15 min and then aspirate poly-L-lysine. Add 25 μl of the bead mixture, allow it to sit for 15 min and then aspirate the bead mixture. Place the coverslip with beads in the sample holder and add a drop of curing mountant. Place the coverslip with the cells upside down on the coverslip in the sample holder. Gently blot the excess liquid from the coverslip sandwich using a lint-free cleaning tissue. Seal the coverslips and fixate them to the sample holder using two-component silicone adhesive.

Equipment setup

SLM and DM calibration

For aberration correction (Fig. 16a–c), we use two DMs (Fig. 16d) in the 4Pi interference cavity and an SLM (Fig. 16e) in the depletion beam path to achieve a sensorless AO architecture. To maximize the performance of AO, both SLM and DMs require calibration.

Fig. 16 |. AO in isoSTED nanoscopy.

a–c, Principle of aberration correction in the isoSTED nanoscope; depletion and effective PSFs with and without sample-induced aberrations are shown as a and b, respectively. c, The DMs in the interference cavity introduce phase patterns that counteract aberrations, restoring nearly perfect optical performance. d,e, Schematic of AO devices. Specifically, in a DM (d), the shape of the front reflection membrane (mirror layer) is controlled by discrete actuators. When a voltage is put on an electrode, the actuator layer above it is attracted, and it pulls the mirror layer down. Changing the shape of the mirror layer enables us to control the wavefront of the reflected beam (highlighted in yellow). In an SLM (e), as liquid crystals are birefringent, applying a voltage to the cell changes the effective refractive index seen by the incident wave, and thus the phase retardation of the reflected wave (highlighted in yellow). Panels a–c adapted from ref. 12, Springer Nature America Inc.

The DM features a light reflective membrane that can be deformed dynamically by an array of software-controllable actuators. DM calibration is vitally important to characterize how the reflective surface of the mirror reacts to the pokes of the actuators. Each DM is calibrated offline using an external Twyman-Green interferometer in which the DM is systematically controlled and analyzed by custom code programmed in Python, as described in detail in ref. 28. In brief, tilting the flat mirror in the interferometer’s reference arm allows for the acquisition of the phase induced by the DM. This phase is derived through Fourier fringe analysis and phase unwrapping. The pivotal task of DM calibration involves computing the control matrix, denoted as H. This matrix acts as the guiding map, connecting a vector u—comprising the voltages assigned to each actuator of the DM—to its corresponding vector z, containing the coefficients derived from the Zernike analysis of the phase. The computation of matrix H involves gathering a series of paired input and output vectors—u and z, respectively—followed by solving a least-squares problem. This method allows for the derivation of H, unraveling the intricate relationship between actuator voltages and the resulting Zernike phase coefficients.

Similarly, the SLM calibration is critical to determine the phase response, corresponding to the gray level of each pixel of the SLM, and construct a comprehensive look-up table. Amidst various calibration approaches, we have chosen diffractogram analysis²⁹ for its distinct advantages. Although offline calibration is also possible, the implementation of diffractogram analysis facilitates convenient in situ calibration of the SLM by placing an additional camera immediately after the SLM. During diffractogram analysis, a binary hologram is displayed on the SLM under the illumination of a linearly polarized beam. The camera captures the diffracted beam pattern produced by the hologram on the SLM. By examining the intensity variations between the diffraction peaks and valleys, alongside any fringe shifts within the pattern, we can ascertain the phase discrepancies across distinct binary regions of the SLM, as elaborated in ref. 29. The principle of diffraction establishes a link between phase disparities and the corresponding gray levels of the pixels, enhancing our understanding of pixel-specific responses.

Notably, the control matrix of the DM and the look-up table of the SLM may be available from the manufacturers, but we highly advocate for calibrating these devices just before integrating them into the nanoscope. The nuances of the local electronic environment can potentially affect the response of the pixels/actuators.

Optomechanical assembly

In general, the instrument setup follows the CAD model. Assemble all parts including custom-manufactured mechanical components, stages, optical components, commercial optomechanical components, light sources, dowel pins, SLM and DMs by following the CAD model.

• We highly recommend manually checking the dimensions of every custom-machined mechanical component before integrating it into the system (ideally already at the time of delivery). In our experience, manufacturing errors are unavoidable and can cause serious optical alignment issues if not detected and corrected early in the process.

• Pay close attention to details, especially regarding optical components such as lens mounting depth in holders, as well as the orientation of lenses, filters and retarders.

• The xy piezo stage holding OBJ2 may have significant vibration issues in closed loop mode when a heavy objective is installed. Adjusting the gain settings in the controller can solve this problem. We suggest consulting the manual of the piezo stage or inquiring with the manufacturer about the specific tuning procedure to accommodate the objective and objective mount’s mass on the stage.

PBS cube installation

The PBS cube is glued to a custom part (FAB-IS0002, custom-designed) by using a UV-curing optical adhesive with UV lamp illumination. Because the PBS is a central component of the nanoscope, it is essential to position the PBS accurately and securely glue it in place.

Translation stage tower installation

The tower assembly is detailed in the CAD model. A pair of ASI stages is mounted on the two pillar legs. They hold the assembly of the sample stages and the objectives, facilitating axial translations of this assembly as a whole. This allows operators to change the optical path difference between the upper and lower arms of the 4Pi interference cavity over a long range without changing the alignment between the objectives and the sample. Precise alignment is enabled by the ASI stages’ two operational modes: synchronous movement and independent stage adjustment.

For initial assembly, we advise securing one pillar leg to the customized base plate while slightly loosening the other. Next, translate the ASI stage that can be moved independently until it reaches the same height as the other stage, verified using a height gauge. Once the heights are aligned, switch to synchronous movement of the stages.

Next, move the ASI stages up and down to the highest and lowest positions that the stages can reach without colliding with other components. Adjust the position of the loose supporting leg until the stages move smoothly for the complete travel range and fix the loose supporting leg in place by tightening the screws connecting it to the optical table.

Finally, move the ASI stages vertically to the position indicated in the CAD model. Note that the ASI stages are rarely moved after installation, underscoring the importance of proper alignment during the setup phase.

Sample stage leveling

The sample is mounted on a leveled stage, where the xy and z sample stage stack are held by three linear actuators positioned beneath. Coarse translation along the z-axis is achieved by simultaneously adjusting all three actuators, while individual manipulation of the actuators enables tip and tilt motions of the stage stack. A bubble level gauge, such as the LVL01 from Thorlabs, can be used to guide the leveling process. Each actuator is individually adjusted until the sample stage stack is oriented horizontally. Finally, the three linear actuators are simultaneously translated along the z-axis to position the sample stage to be approximately aligned with the position in the CAD model.

QWP setup in the 4Pi interference cavity

In the 4Pi interference cavity setup, described in detail below, precise tuning of the QWPs near the objectives is paramount. The QWPs have a crucial role converting linearly polarized depletion beams into circularly polarized ones to produce beam profiles essential for generating ring-shaped depletion foci. In addition, circular polarization of the excitation light ensures uniform excitation of fluorophores with different dipole orientations. The orientation of the QWPs dictates whether the depletion beams’ polarization becomes left-handed or right-handed circular polarization. The choice between these orientations is critical for isoSTED nanoscopy, in which the circular polarization direction must align with the rotational direction of the vortex phase pattern imprinted on the depletion beam. This alignment is necessary to prevent high laser intensity residue at the center of the ring-shaped beam, which reduces fluorescence signals and negatively affects the resolution³⁰. Right-handed circular polarization corresponds to an anti-clockwise vortex phase pattern, while left-handed circular polarization serves the same purpose but matches the opposite direction for the vortex phase pattern.

Device drivers and software installation

Before initiating the alignment protocol, connect all computer-controlled devices to the computer and install the manufacturer-supplied drivers. Confirm that control software and devices are functional. The necessary software packages for installation include:

• Boston Micromachines, LinkUiMulti and DM Shapes for DM control

• NKT Photonics, CONTROL for white-light laser control

• MPB, Pulsed Fiber Laser Control for depletion laser control

• PI, PIMikroMove for piezo, sample, objective Z and DC-Mike control

• Thorlabs, Kinesis for BK7 wedge motor controlASI, ASI Console for LS-50 linear translation stage control

• IDS Imaging, uEye camera drivers and camera manager for alignment camera control

• Opsero, LUFA VirtualSerial.inf file for pulse gate/delay module

• AA Opto-Electronic, MPDSnC for AOTF control

• Hamamatsu Photonics, LCOS-SLM Controller

To operate the nanoscope control software (Supplementary Fig. S9), ensure that various software packages from NI are installed on the computer. These are:

• LabVIEW 2018 32-bit development environment

• All-in-one package of NI device drivers (version January 2019) including NI-DAQmx, NI MAX, NI-VISA, NI Vision, NI-RIO and NI-Serial

• NI Vision Development Module

• NI FPGA Development Module

To visualize the data, we further need:

• MathWorks, MATLAB 2018a 64-bit or later

• University of Utah, FluoRender 2.22.3 or later

• Abberior Instruments, Imspector 0.13.11588 or later

• National Institutes of Health, ImageJ 2.3.0/1.54b or later

Procedure

▲ CRITICAL Refer to Fig. 4 for component names and the system structures unless noted otherwise in the procedure. A list of component name abbreviations is available in Supplementary Table S1 in alphabetical order.

▲ CRITICAL For critical components like AOM, AOTF, SLM and DM, we provide a quick reference table, as shown in Supplementary Table S7, to check whether their alignment is successful.

Software installation and setup

● TIMING 1 d

• 1Install the LabVIEW development environment and all necessary add-ons as detailed in Nanoscope control software.
• 2Connect all computer-controlled devices such as stages, lasers and FPGA boards to the computer. Install the necessary manufacturer-supplied drivers and control software. Confirm the proper functionality of all devices to ensure that they are working correctly.
• 3Install custom nanoscope control software.
• 4Press the ‘Run’ button of the FPGA VI to initiate the compilation process. Follow the provided instructions to execute the Xilinx compiler, generating the bitfiles for the FPGA boards.

Optomechanical assembly

● TIMING 1 week

• 5To ensure accurate installation of the mechanical components and proper fixation of screws, refer to the SolidWorks model provided at the Zenodo repository listed in Data availability. During installation, pay attention to mounts equipped with pins designed to be installed on the vertical breadboard. These pins aid in securing the components in their designed positions and ensure optimal functionality of the setup. Once the mechanical components are installed, and screws are properly fixed, verify alignment and stability to ensure the successful assembly of the apparatus.
• 6Ensure that all optical components, including lenses, mirrors, dichroic mirrors, waveplates, etc., are securely fixed in their respective mounts either mechanically or using suitable adhesive. However, avoid tightening screws that directly exert force on optical elements, especially mirrors, too much, because that will deform these elements and can cause severe optical aberrations. Pay close attention to the orientation of lenses and filters, because they are often not reversible. Properly fix these components in their mounts to maintain the desired functionality. Install all the assemblies in accordance with the instructions provided in Equipment setup.

Electronics assembly

● TIMING 1 week

• 7Connect cables between the FPGA connectors and their corresponding breakout boxes.
• 8Connect all cables and wires from the electronic controlled devices, including stages, APDs, PMT, RM, GMs, AOM and AOTF, to their respective terminals in the breakout boxes. Refer to Supplementary Table S3 for the FPGA wire configurations.

AOM setting and alignment

● TIMING 2 h

• 9Read Procedure overview, especially the AOM and AOTF alignment subsection.
• 10Install LH1 and measure its power using a laser power meter to confirm that the specified output laser power is available. (>2 W of average power from the depletion laser source is required.) Confirm that the beam exits LH1 at the specified beam height of the SLM screen center.
• 11To adjust M1, temporarily install an optical rail and, using M1, align the laser beam to run parallel to the rail, by following the general beam alignment procedure described in General beam alignment.
• 12Mount the AOM on an angular alignment stage (Newport, cat. no. 9071-M) and add posts and spacers as needed, so that the laser beam passes through the centers of the entrance and exit windows of the AOM. Turn on the AOM and set the frequency and amplitude using the manufacturer-provided software (MPDSnC) until a diffracted beam appears at the +1 order.
• 13Adjust the angle of the AOM using the angular alignment stage to optimize the power of the +1 order diffracted beam.
• 14Use the manufacturer-provided software (MPDSnC) to tune the drive frequency of the AOM to maximize the output power in the +1 order diffracted beam.
• 15Repeat the optimization of Steps 13 and 14 to maximize the power of the +1 order diffracted spot, which should surpass 90% of the input power.

Polarization-separation module alignment

● TIMING 4 h

• 16Use M2 to reflect the +1 order diffracted light beam emerging from the AOM and align it according to the layout depicted in Figs. 2 and 4. For this purpose, temporarily install a rail and align the beam according to General beam alignment.
• 17Install the assembled polarization-separation module (Fig. 6). Use a cage alignment plate (e.g., Thorlabs, cat. no. CPA1) to align the polarization-separation module along the beam. The input beam and the transmitted beam through PBS1 should hit the center of the cage alignment plate when placed in front of PBS1 and M3.
• 18Align the reflection arm of the polarization-separation module (used for the STED_xy beam) by rotating PBS1 so that the reflected beam hits the center of a cage alignment plate placed behind QWP2.
• 19Place the cage alignment plate behind HWP2. Block the transmitted light behind QWP1 and adjust M4 so that the reflected beam hits the center of the cage alignment plate.
• 20Rotate QWP2 to maximize the output power at the exit of the polarization-separation module.
• 21To align the transmission arm (used for the STED_z beam), block the beam behind QWP2 and adjust M3 to reflect the beam to hit the center of the cage alignment plate. Rotate QWP1 to maximize the output power at the exit of the polarization-separation module.
• 22Rotate the angle of HWP1 to allocate the power for the STED_z to STED_xy beams to a ratio of 1:5.5. In general, for simplicity and to avoid confusion, keep using a single depletion beam (e.g., STED_xy beam) and block the other beam at the respective arm of the polarization-separation module for the following system alignment. Switch beams or apply both only when necessary.
• 23Rotate the angle of HWP2 to set the polarization of the STED_xy beam to horizontal orientation. The STED_z beam will automatically be polarized perpendicularly to the STED_xy beam. The polarization orientation can be checked by measuring the laser power behind a polarizer and will be validated by the modulation of the SLM once aligned.

SLM installation and alignment

● TIMING 1 d

• 24Adjust the mirrors M5 and M6 to align the beam along the center of R1 at the height of the SLM hologram center. Follow the method in Box 3 to align the L1-L2 4f telecentric system for beam expansion.
• 25Temporarily install a rail at the designated position of the D-shaped mirror (D) before installing D. Mount an alignment camera on the temporary rail at the designated D position to align the beam by following Box 1. Adjust the tip and tilt angles of M7 and translate it along R1 until the beam is aligned to the center of the rail.
• 26Remove the temporary rail and install D in the designated position. Translate M7 along R1 to have the beam centered on D if its edges cut off the beam.
• 27Temporarily install a rail at the designated position of the SLM along the designated beam direction reflected by D. Mount the alignment camera on the rail. Translate D using N1 and adjust its tip and tilt angles until the beam reflected by D propagates parallel to the rail. The tip and tilt angles of D are adjusted in this step, while its translation will be adjusted later.

  ▲ CRITICAL There are two dowel pins located at the bottom of the D mount (FAB-IS0050, custom-designed). These pins serve to restrict the movement of D to the propagation direction of the incident beam. Ensure that these pins are in place before translating D.

• 28Remove the temporary rail and mount the SLM on the base plate. Use a ruler to roughly place the SLM screen center at the focus of L3. Apply a 0/π top-hat hologram pattern (300 pixels in diameter) along with the blazed gratings (set ‘Tip’ = 7 on the ‘SLM’ panel in the LabVIEW-based control software) on the SLM. Ensure that the hologram is positioned at the center of the SLM (‘Holo 1’ coordinates: x = 0, y = 0 on the ‘SLM’ panel, as shown in Supplementary Fig. S10a). Slightly loosen the screws that secure the D mount. Translate D until the hologram is uniformly illuminated, checked visually by a paper screen held into the beam after it has been reflected at the SLM. Re-tighten the screws.

  ▲ CRITICAL To minimize the stray light outside the objective apertures of the nanoscope (represented here by the size of the hologram pattern), apply another blazed grating pattern to the SLM area outside the hologram(s) (e.g., ‘Bkg gradient X’ = 40, ‘Bkg gradient Y’ = 40 on the ‘SLM’ panel; Supplementary Fig. S10).

• 29Temporarily replace R2 with a long rail (e.g., Thorlabs, cat. no. RLA300/M). Mount an alignment camera on this rail to check the beam. Turn on the ‘Top Hat’ hologram with ‘Top hat radius’ set to ~0.2 to highlight a clear hologram center imprinted on the beam. Adjust the tip and tilt angles of the SLM by rotating it on its mount and using spacer shims to fine-tune the direction of the reflected beam until it runs parallel to the rail. Alternate the illumination beam between STED_xy and STED_z beams by blocking either the reflected or transmitted arm of the polarization-separation module and adjust HWP2 until only the STED_xy but not the STED_z beam is modulated by the applied hologram (Supplementary Fig. S10a).
• 30Replace the temporary rail with R2 and install QWP3 and the L3 mount without the lens. Add an iris to the L3 mount and place an alignment camera after QWP3 for beam monitoring. Adjust the linear stage underneath the SLM in the direction perpendicular to the incident beam to center the top-hat pattern through the iris. Translate D for uniform illumination of the hologram on the SLM (Supplementary Fig. S10a). Mark the current beam center on the camera in the uEye software. Install the lens L3 in the mount and tune its position to maintain the beam position on the camera.
• 31Mount M8 on R2 at the focal plane of L3. Adjust the angle of M8 to reflect the beam back along the incident beam path through the same iris on L3.

  ▲ CRITICAL To determine the focal position of L3, speckle analysis³¹ may be applied.

• 32Activate both holograms in the LabVIEW program and arrange them side by side as displayed in Supplementary Fig. S10b. Set the coordinates of the two hologram centers to (−200, 0) and (200, 0), respectively. Apply blazed gratings with identical periods but in opposite directions to the holograms (i.e., for the left hologram, ‘Tip’ = +7, and for the right hologram, ‘Tip’ = −7, on the ‘SLM’ panel).
• 33Translate D to uniformly illuminate the left hologram on the SLM. Ideally, with the well-aligned 4f system on R2 (consisting of passing twice through L3), the reflected beam from M8, after passing through L3, should illuminate the right hologram with both holograms now being co-centered, ready for further tuning later.
• 34Mount an alignment camera on R3 to monitor the beams. Illuminate only with the STED_xy beam (modulated by the left hologram). Rotate QWP3 while sequentially adding top-hat patterns to either of the two holograms until the beam modulation by the right hologram is minimized, aiming for as little modulation as possible. Then, illuminate with both STED_xy and STED_z beams. Ideally, the two hologram patterns are co-centered; if not, adjust the angles of M8 to correct it.
• 35Move the alignment camera along R3 to verify that the beam is parallel to R3 by following the instructions in Box 1. If a beam direction mismatch is detected, tune the angles of D to align the beam direction with R3 and adjust M7 to ensure a uniform illumination of the holograms. Install the L4 lens mount with an iris, without the lens, in front of the camera to verify that the hologram center passes through the iris center.
• 36Mount L4 and L5 on R3. Use a ruler to set L4 at a distance equal to its focal length from the right hologram on the SLM. Align this 4f telecentric system by following Box 3. At the polarization-separation module, block the beam modulated by the left hologram (STED_xy beam) and illuminate only with the beam modulated by the right hologram (STED_z beam). Install an alignment camera after L5 (approximately at the location of M9) and securely position the camera on R3 where a clear image of the right hologram pattern is visible (see Supplementary Video S1 for an example).
• 37Illuminate only with the STED_xy beam modulated by the left hologram. Check the hologram image on the camera to see whether it is in focus. Adjust N3 to move R2 along the rail direction to ensure a clear image on the camera.

  ▲ CRITICAL Do not change the angle of R2 while tuning its position. This can be achieved by pushing R2 against the corresponding edge of the assembly plate (FAB-IS0042).

• 38Keep the camera at its current position and verify the following factors: (i) ensure that a clear image of both holograms is observed by switching between the two depletion beams (optimized by N3), (ii) check for co-centering of the two holograms (optimized by M8) and (iii) verify illumination uniformity across the holograms (optimized by M7). Optimize these factors if necessary. Once all these conditions are met, secure all components and avoid any further changes.
• 39Remove the alignment camera and install M9 in its place. Adjust the angles of M9 to reflect the beam along R4. Move M9 along R3 to align the reflected beam with R4.
• 40Align the L6-L7 4f telecentric system by following Box 3.
• 41Install an iris in the intermediate focal plane between L6 and L7 as a special filter to block stray light.

Excitation path alignment

● TIMING 3 d

• 42Read Procedure overview, especially the Excitation laser path alignment subsection.
• 43Install the laser head for excitation (LH2) and set the beam height the same as the depletion laser.
• 44Mount the AOTF on the angular alignment stage (Newport, cat. no. 9071-M) and add posts and spacers to match with the laser beam height.
• 45Reflect the laser beam with M15 to pass through the centers of the entrance and exit windows of the AOTF.
• 46Temporarily insert a filter stack, consisting of a neutral-density filter (e.g., Thorlabs, cat. no. NE20A), a laserline cleanup filter (e.g., Semrock, cat. no. LL01-647-25) and a linear polarizer (e.g., Thorlabs, cat. no. GT15-A), between LH2 and AOTF. Ensure that this filter stack does not alter the beam direction and that the orientation of the polarizer matches with the polarization modulated by AOTF.
• 47In the MPDSnC software, enable Channel #1 and designate the operational wavelength of the AOTF as 650 nm.

  ▲ CRITICAL In MPDSnC, only the driving frequency, not the wavelength, is directly adjustable. Refer to the manufacturer-provided calibration sheet to find the driving frequency matching the wavelength of interest.

• 48Adjust the angle of the AOTF using the angular alignment stage on which it is mounted to maximize the power of the +1 order diffracted beam with the 650-nm laser.
• 49In MPDSnC, optimize the driving amplitude of the AOTF, aiming to maximize the output power of the 650-nm laser component.

  ▲ CRITICAL The largest driving amplitude does not necessarily result in the highest power output.

• 50Repeat Steps 48 and 49 to maximize the power of the +1 order diffracted spot.
• 51Replace the laserline cleanup filter with appropriate filters for the other two desired excitation wavelengths, such as Semrock FF01–591/6–25 for 590 nm and Semrock LL01-488-25 for 488 nm. Configure the corresponding driving frequencies in Channels #2 and #3 in MPDSnC. Activate each channel sequentially while deactivating the others. Adjust the driving frequency and amplitude for each channel to maximize the power of the +1 order diffracted beam for each wavelength.
• 52Remove the cleanup filters for subsequent alignment and imaging applications. Use the 650-nm laser as the default for the remaining alignment procedures.
• 53Use M16 to direct the beam path toward R7.
• 54Install the assembled cage system of the temporal chromatic dispersion compensator (Fig. 15b). Align the cage system along the beam using the cage alignment plate (e.g., Thorlabs, cat. no. SCPA1). The laser beam should hit the center of the cage alignment plate in front of PBS3 and behind QWP6.
• 55Adjust the angles of PBS3 to direct the reflected beam going through the hole in the center of the alignment plate on the cage system. Rotate HWP5 to maximize the power of the beam transmitted through PBS3.
• 56Install F1, F2 and F3 according to the CAD model.

  ▲ CRITICAL To minimize temporal chromatic dispersion effects, position filters F2 and F3 44.8 and 66.1 mm from F1, respectively. The optimal positions may vary depending on the laser dispersion. For further details, refer to Temporal chromatic dispersion compensation.

• 57Adjust the angles of F1, F2 and F3 so that their corresponding reflected beams are re-combined to the same path through the center of the alignment plate on the cage system. Note that the adjustment of F1, F2 and F3 should be using 488-, 590- and 650-nm lasers from the AOTF channels, respectively.
• 58Rotate QWP6 to maximize the power of the beam reflected by PBS3.
• 59Install M17, L15, L16, QWP7 and M18 according to the layout in Fig. 4. Align the L15-L16 4f telecentric system and ensure that the beam travels along the direction of the rail.
• 60Adjust M18 and DI1 to couple the excitation beam into the depletion beam path. Switch between the three different excitation beams and ensure that all three beams travel the same path as the two depletion beams along R4 after DI1. If not, adjust F2 and F3.

  ▲ CRITICAL Tuning the angles of DI1 affects not only the reflected excitation beam path, but also slightly affects the transmitted depletion beam path.

• 61HWP6 and QWP7 remain to be tuned after aligning the 4Pi interference cavity (refer to 4Pi interference cavity alignment) to maximize the excitation power in the upper arm of the cavity.
• 62Keep using the depletion beam for the following alignment procedure unless specified otherwise.

Scanner installation and alignment

● TIMING 1 d

• 63Ensure that all three excitation beams and both depletion beams travel the same path along R4 after DI1 using the checking method in Step 60.
• 64Temporarily mount a dovetail rail in the propagation direction of the beam emerging from DI2 at the designated position of the scan mirrors. Adjust the angles of DI2 to align the beam direction along the rail.

  ▲ CRITICAL Ensure that the beam reflected by DI2 is horizontal to the baseboard; otherwise, the fluorescence detection light path, which shares the same beam path but in the reversed direction, will experience large beam height changes over its long beam path.

• 65Remove the temporary rail from Step 64. Mount the RM and two GMs (GM1 and GM2) in the custom-built mount on translational stage LS2.

  ▲ CRITICAL Before installing RM, remove its protective case to prevent obstruction of the light beam. Position the stage at its designated location on the base plate. Translate LS2 so that the beam hits the center of RM.

• 66Temporarily adjust the positions of GM1 and GM2 to ensure that the beam reflected by RM is not obstructed by either of the GMs.
• 67Temporarily mount a dovetail rail on the base plate in the propagation direction of the beam emerging from GM2 at the designated position of L8 and M10. Loosen the screw that secures RM and adjust RM until the reflected beam travels horizontally along the rail. Once aligned, tighten the screw to secure RM in its holder. Translate the holder of RM in the direction perpendicular to the vertical breadboard within the scanner mount until the beam is ~80 mm to the vertical breadboard.
• 68Temporarily install a lens with a 40-mm focal length (e.g., Thorlabs, cat. no. LA1422-B) in the L8 lens mount. Increase the RM scanning amplitude to ~20 μm in the LabVIEW control program. Translate the alignment camera along the rail to find the position where the beam image shows minimal scanning motion (i.e., no beam image elongation induced by scanning). Activate the top-hat pattern on the SLM in the LabVIEW control program. Adjust the height of the RM mount using shims until the beam center shows minimal blur during scanning, indicating that the RM scanning axis is aligned with the center of the beam hologram applied by the SLM.
• 69Set the amplitude of the RM control voltage to 0. Translate the alignment camera on the rail toward the far end to monitor the position of the beam reflected by GM2.

  ▲ CRITICAL Keep the RM and the GMs turned on and maintain the control voltage setting for all subsequent alignment procedures.

• 70Adjust the angles of GM1 and GM2 according to the instructions in Box 2 until the beam travels along the rail after passing both GMs.
• 71Loosen the screws (Supplementary Fig. S11) securing the RM mount. Translate the RM mount in the direction of the incident beam until the beam hits the center of GM2. Once aligned, tighten the screws to secure the RM mount in its new position. Repeat Step 70 as needed.
• 72Mount M10 at the designated position without mounting L8. Install R5 on the vertical breadboard and mount an alignment camera assembly (Supplementary Fig. S12) on R5. Adjust the angles of M10 until the beam propagates along R5. Secure the alignment camera at the designated position of the DM surface (Fig. 9) and mark the beam center on the camera image.
• 73Mount L9. This might cause a shift of the beam position on the camera image. Mark the new beam position as the target and compare the beam positions before and after L9 installation. If there is a shift perpendicular to the vertical breadboard, loosen the screws securing the pillar legs and translate the breadboard in the same direction for compensation. If there is a vertical shift, translate M10 along the propagation direction of the beam emerging from L8. Repeat Steps 70–72 as needed until inserting L9 does not alter the beam center position on the camera anymore.
• 74Keep the alignment camera at the designated DM surface position for subsequent alignment until otherwise stated. Mount L8 with two pins. Translate its position in the direction perpendicular to the incident beam from the scanner and use shims for beam height compensation as needed until the insertion of L8 does not change the beam center position marked in Step 73 on the camera.

PBS installation and angle alignment

● TIMING 3 h

• 75Install the rotation stage (Newport, cat. no. M-RS65) on the vertical optical breadboard.
• 76Mount the goniometer spacer (FAB-IS0001, custom designed) with the rotation stage M-RS65.
• 77Mount the goniometer (Newport, cat. no. M-GON40-U) on FAB-IS0001 and ensure that the orientation of the goniometer matches the CAD model.
• 78Mount motorized linear actuators (Newport, cat. no. CONEX-TRA12CC) as the actuators of the rotation stage M-RS65 and goniometer (Newport, cat. no. M-GON40-U).
• 79Place the goniometer cube holder (FAB-IS0002, custom designed) on a flat work surface.

  ▲ CRITICAL The bottom of the cube holder is not flat. To level it, insert a 2-mm-thick spacer underneath it.

• 80Apply one drop of UV-curing adhesive to the surface of the cube holder and place PBS2 (CVI Laser Optics, cat. no. PBSH-450-1300-100) on top, ensuring that it is positioned so that the side surfaces of PBS2 attach to the datum plane on the cube holder (Supplementary Fig. S1).
• 81Place the cube holder ~30 cm away from the UV lamp. Turn on the UV lamp and irradiate the optical adhesive for 15 min.
• 82Mount the cube holder on the goniometer (Newport, cat. no. M-GON40-U).

  ▲ CRITICAL Confirm that the UV adhesive is fully cured to avoid PBS2 from detaching when mounted on the vertical breadboard.

• 83Glue the BK-7 glass block on its holder (FAB-IS0010, custom designed) using optical adhesive (see Steps 80 and 81) and mount the holder at the designated position on the vertical optical breadboard.
• 84Use the alignment camera to confirm that the beam propagates along R5.
• 85Mount the motorized stage (Thorlabs, cat. no. MTS25/M-Z8) on the vertical optical breadboard.
• 86Glue the BK-7 glass wedges to the corresponding holders using optical adhesive (see Steps 80 and 81) and then mount the holders on the vertical breadboard.
• 87Rotate HWP5 until most of the laser power is reflected by PBS2.
• 88Use the DC Servo Controller/Driver (Newport, CONEX-CC) to tune the tilt and azimuthal angles of PBS2 until the reflected beam propagates parallel to R6.

Power allocation wave plate alignment

● TIMING 30 min

• 89Mount HWP3 in the rotation mount in front of L9.
• 90Rotate HWP3 and measure the depletion beam power transmitted and reflected by PBS2 until the power allocation equals 50:50.
• 91Lock the rotation mount using the locking mechanism.

DM, scanners and SLM conjugation alignment

● TIMING 2 h

• 92Set the RM scanning range to ~20 μm in the LabVIEW control program. If the beam image appears elongated in the RM scanning direction as a result of the scan motion, the RM surface is not conjugated to the camera plane. Activate the ‘Top Hat’ pattern on the SLM (Supplementary Fig. S10b) and mark the beam center position as the target in the image with the camera mounted at the designated DM1 surface position (Fig. 9). Translate DI2 along R4 until the beam image is not elongated during scanning anymore to conjugate the RM scanning plane and the camera plane.

  ▲ CRITICAL Translating DI2 may shift the beam position on the camera; correct the beam shift back to the target by translating LS2.

• 93Adjust the conjugation between the RM surface and the hologram on the SLM using the camera mounted at the designated DM1 surface position. Activate the hologram by ‘Holo ON’ on the SLM (Supplementary Fig. S10b) and set the control voltages of the scanners to 0. Adjust the L4-L5 4f telecentric system along R3 to achieve a sharp hologram image on the camera. Use the piezo pusher (FAB P0013 and FAB P0014, custom designed) to gently move the lens mounts of the L4-L5 4f telecentric system. Loosen the screws of the lens mounts and push the two lenses in the same direction by the same distance using identical spacers.
• 94Use the ‘Live’ image acquisition mode in the nanoscope control software (Supplementary Fig. S9). This functionality enables automatic renewal of imaging within the same FOV, facilitating real-time observation and adjustment. Set the scan range to ~20 μm for both directions and confirm that RM scanning is conjugated to the camera plane. The GM1 and GM2 mirrors scan synchronously. Optimize the ‘Galvanometer Scale Factor’ in the ‘Initialize Waveforms V4.vi’ LabVIEW file to minimize the scanning motion of the hologram image on the camera, as demonstrated in Supplementary Video S1. This value is crucial for establishing a virtual scanning plane for the GMs that coincides with the scanning plane of the RM. The optimized alignment is shown in Supplementary Video S2.

4Pi interference cavity alignment

● TIMING 1 week

• 95Mount HWP4, W1, W2, SH1, SH2, DM1 and DM2 on the vertical optical breadboard in the 4Pi interference cavity according to the CAD model.
• 96Secure the half-inch lens mount (FAB-IS0012 and FAB-IS0013, custom designed for L10 and L12, respectively) at the designated position on the breadboard. Install a Ø0.5-inch target (e.g., Thorlabs, cat. no. P1000HW) in the lens mount to take the place of the lens for alignment purpose.
• 97Adjust the tip/tilt angles of DM1 until the hologram center precisely propagates through the center hole of the target. Then, remove the target from the lens mount.
• 98Glue the Ø1-inch mirror (M11) on the actuator (PI, cat. no. P-841.10) with UV glue and install the actuator on the custom-designed mount (FAB-IS0026).
• 99Mount M12, DI3 and QWP4.
• 100Remove OBJ1 and install the alignment camera with an iris in front of it on the upper objective mount (FAB-P0007, custom designed). Translate the upper objective stage (PI, cat. no. N-565.260) to monitor the beam with the camera. Adjust the angles of M11 and M12 until the beam hologram stays centered through the iris along the stage translation.
• 101Mount L10 in the half-inch lens mount (FAB-IS0013, custom designed).
• 102Mark the beam hologram center on the camera. Mount L11, then translate it parallel to the breadboard and use shims to adjust its height until the beam hologram center is back to the marked position.
• 103Translate the upper objective stage until the DM1 surface is imaged on the camera. This is verified by powering off the SLM and checking the beam pattern on the camera, as illustrated in Fig. 9c,d.
• 104Power on the SLM and activate the holograms. Adjust DM1 axially by adjusting its mount (Newport, cat. no. LP-2A) until the holograms appear sharp on the camera. Verify the conjugation by observing minimal beam scan motion on the camera, as illustrated in Supplementary Videos S1 and S2. Repeat Steps 103 and 104 as necessary until the conjugations of DM surface, the virtual scanning plane and SLM hologram are achieved at the same camera plane.
• 105The axial translation of DM1 in Steps 103 and 104 affects the co-centering of the beam (hologram) center to L10. To compensate for the misalignment, remove lenses L10 and L11 and repeat Steps 96–102.
• 106Flip the alignment camera board in its mount and mount it at the lower objective mount, so that the camera is facing up to monitor the upper beam. Activate the ‘Top Hat’ pattern on the SLM. Launch the ‘DM Shapes’ software from Boston Micromachines to load the cross shape to DM1. Adjust the position of DM1 using LP-2A (Newport) to align the DM1 surface co-centered with the SLM hologram, as depicted in Fig. 9e,f.

  ▲ CRITICAL The cross pattern applied on DM is invisible at the conjugated plane but becomes visible far beyond the conjugation. For this reason, the camera is mounted in the position of the lower, not the upper, objective mount.

• 107Close SH1. Align the lower arm from DM2 to OBJ2 using the method described above for aligning the upper arm of the 4Pi interference cavity (Steps 95–106). Mount the camera on the upper objective mount and have the camera facing downward to monitor the beam. Use the translation path of the upper objective stage as the reference to which the beam direction of the lower arm is aligned. Mount the following components and align them accordingly: M13, L12, SH2, DM2, M14, L13, DI4 and QWP5.
• 108Temporarily remove QWP4 and QWP5. Open both SH1 and SH2. Position a power meter (e.g., Thorlabs, cat. no. PM130D) behind the top-right side of PBS2. Rotate HWP4 to minimize the measured power value.
• 109Re-mount QWP4 and QWP5. Adjust their orientations iteratively until the power measured by the power meter is maximized.
• 110Remove the power meter. Extend R6 toward the top-right direction. Mount an alignment camera assembly on the extended rail.
• 111Use the alignment camera to locate the beams. If two spots are observable, adjust M13 until the spots coincide. Then, translate the camera to the far end of the rail and overlap the spots again by adjusting M14. Repeat this step for several rounds until both spots are consistently overlapping along the rail. Once completed, remove the camera assembly and the extended rail.
• 112Rotate HWP3 in front of PBS2 to achieve an approximate 50:50 power allocation between the two arms of the 4Pi interference cavity.

Objective alignment

● TIMING 3 d

• 113Mount both objectives in their corresponding lens mounts.

  ▲ CRITICAL The objectives for isoSTED nanoscope are Olympus UPLSAPO100XS silicone oil objectives, which feature correction collars to compensate for spherical aberrations induced by the cover glass. Adjust this ring as needed. In our case, it is calibrated for 170-μm-thick cover glasses.

• 114Apply a drop of silicone oil (Olympus, cat. no. SIL300CS) to OBJ2. Place the gold nanoparticle sample between the two objectives on the sample holder.

  ▲ CRITICAL Refer to Test sample preparation for the detailed preparation procedure. Because this process is time intensive, and the sample must be ready for the objective alignment phase, we strongly recommend initiating preparation ≥3 d in advance.

• 115Install L14 and mount the PMT at the designed position. Add an iris and a neutral-density filter (e.g., Thorlabs, cat. no. ND20A) in front of the PMT to reduce background noise.
• 116Activate the blazed-grating hologram on the SLM.
• 117Close SH1. Use the PMT to image the sample with OBJ2. Adjust the z position of the sample until the beads are in focus.

  ◆ TROUBLESHOOTING

• 118Apply a drop of silicone oil to the top cover glass. Open SH1 and close SH2. Lower OBJ1 using the upper objective stage piezo to get the same gold nanoparticles in focus.

  ▲ CRITICAL If there is no overlap between the FOVs of OBJ1 and OBJ2, adjust the adjustment screws (Newport, cat. no. AJS100–0.5H-NL) underneath the lower objective stage to coarsely overlap the FOVs. Repeat Step 107 to align the lower arm between DM2 and OBJ2 to the new lateral position of OBJ2 until the FOVs of both OBJs overlap again. Additional fine adjustments to achieve perfect FOV overlap can be made by translating the piezo stage (PI, cat. no. P-612.2SL).

Focus-lock module alignment

● TIMING 4 h

• 119Install L20 in the xy translation mount by following the CAD model as shown in Fig. 14. Adjust its axial position to collimate the 980-nm NIR laser beam of the laser diode LH3.
• 120Merge the NIR beam with the other beams via DI3 in the top arm of the 4Pi interference cavity. Adjust M20 and M21 so that the 980-nm beam hits the back aperture center of the upper objective.
• 121Let the NIR beam propagate through both objectives and separate it from the common beam path by DI4 in the lower arm of the 4Pi interference cavity.
• 122Focus the NIR beam on the camera C using L21. Adjust M24 and M25 until the beam hits the center of the camera.
• 123Insert the CL between C and L21 to introduce astigmatism to the focus so that the focus is elliptical when the focal planes of the two objectives no longer coincide.
• 124Translate the lower objective stage in both axes’ directions with the nanoscope control software and observe the direction that the focus is moving in the camera image. Rotate C until the focus moves horizontally and vertically in the image when translating the lower objective stage.
• 125Activate the focus-lock module in the nanoscope control software. Once alignment of the two objectives is achieved, click the ‘Locked’ button to register the position. Subsequently, the ‘Drift’ indicator turns yellow when the two objectives are misaligned from the locked position. The ‘Control Option’ button located below the ‘Locked’ button can toggle between ‘Manual’ and ‘Automatic’ misalignment-compensation modes.

Test sample preparation

● TIMING 3 d

• 126Clean a 25-mm-diameter round precision cover glass (Thorlabs, cat. no. CG15XH) in 1 M KOH in a sonic bath for 15 min.
• 127Rinse the cover glass three times with filtered water.
• 128Rinse the cover glass in 100% ethanol.
• 129Let the cover glass dry at room temperature.
• 130Apply 200 μl of 1 mg ml⁻¹ BSA-biotin solution diluted in PBS to the cover glass and let it sit for 5 min.
• 131Remove the solution and rinse three times with PBS.
• 132Apply 200 μl of 1 mg ml⁻¹ neutravidin solution diluted in PBS. Let it sit on the cover glass for 5 min.
• 133Remove the solution and rinse three times with 10 mM MgCl₂ solution diluted in PBS.

  ▲ CRITICAL This step is required only for using the GattaQuant beads, but not with other fluorescent beads described below.

• 134Prepare bead solution containing gold nanoparticles (Sigma-Aldrich, cat. no. 742058–25-ML, 150-nm diameter), 1:8 dilution; 10 mM MgCl₂, final concentration; GattaQuant Bead R (Atto 647N conjugated, 23-nm diameter), 1:4 dilution.

  ▲ CRITICAL We recommend making four samples, one with GattaQuant beads (for resolution measurement), as described, and, for alignment purposes, one each with crimson beads, red beads and yellow-green beads replacing the GattaQuant beads. For the latter three, no MgCl₂ is required, and we recommend dilutions of 1:10,000. If stored in the dark at room temperature with ProLong Diamond mountant, these samples can be used for a few months.

• 135Apply 20 μL of the bead solution to the cover glass and let it sit on the glass for 5 min.
• 136Remove the solution. Do not rinse.
• 137Place the glass beads side up in the customized sample holder.
• 138Add a drop of ProLong Diamond antifade mountant on the beads side of the cover glass.
• 139Apply another 25-mm round cover glass on top of the beads side of the prepared cover glass and then press it down gently.
• 140Allow the sample to sit in the dark at room temperature for 48 h.
• 141Secure the cover glass sandwich to the customized sample holder using two-component silicone adhesive.

Detection path alignment

● TIMING 1 d

• 142Turn on LH1. Turn on LH2 and set its output power to 15%. In the nanoscope control software, set the 650-nm line approximately to the maximum.
• 143Rotate HWP3 to achieve a roughly equal allocation of the excitation laser power between the upper and lower arms of the 4Pi interference cavity.
• 144Insert a sample of crimson beads and gold nanoparticles between the two objectives. Image the gold nanoparticles with the PMT. Toggle between illuminating the sample with either the excitation or the depletion beam at power levels that create sufficient signal. In the nanoscope control software, optimize the ‘Tip’, ‘Tilt’, and ‘Defocus’ coefficients of the holograms displayed on the SLM to ensure the co-centered overlap of the depletion PSFs with the excitation ones in three dimensions.
• 145Image the beads with the upper and lower objectives sequentially to ensure that the two objectives image the same FOV. Replace the neutral density (ND) filter in front of the PMT with the band-pass filter to detect fluorescence from the crimson beads.
• 146Activate the focus-lock module to maintain locked objective positions throughout the entire alignment procedure. This ensures that the objectives remain in their relative positions, providing stable and consistent imaging conditions.
• 147Block the PMT with a cap to prevent overexposure.
• 148Switch the illumination beam from the excitation beam to the STED_xy depletion beam. Set the depletion beam power to maximum.

  ▲ CRITICAL Because a small portion of the depletion laser power can leak into the detection module due to the imperfect reflection of DI2 at 775 nm (Fig. 13), we can use this leaked light for the alignment of the detection path.

• 149Install the three rails (R8, R9 and R10) for the detection module on the base plate. Each rail corresponds to one detection channel.

  ▲ CRITICAL Because the thick glasses of DI5 and DI6 result in a noticeable shift in the lateral direction when the beam passes through them, it is important to align the channels in the following order: (i) APD1 for the green channel, (ii) APD2 for the orange channel and (iii) APD3 for the far-red channel.

• 150Open both SH1 and SH2.
• 151Mount DI5 on R8. Translate the DI5 mirror mount to roughly align the beam to hit the center of the front surface of DI5. Adjust the angle of the mount to ensure that the beam is reflected parallel to rail R8. Translate the mount to center the reflected beam above R8.

  ▲ CRITICAL It is possible that two or more spots are observed with the alignment camera, because the uncoated surface of the dichroic mirror can create ghost reflections. In this case, use only the brightest spot as the reference for alignment.

• 152Mount L17 on R8. Translate L17 in two dimensions to align its center with the beam (analog to Box 3).
• 153Connect the fiber head adapter (Thorlabs, cat. no. SM1FC) with the fiber (Thorlabs, cat. nos. FG105LCA and FT05SS) and mount it on R8. Position a camera close to the other end of the fiber. Translate the adapter in three dimensions to maximize the brightness of the spot observed on the camera.
• 154Turn off the depletion beam and turn on the excitation beam. Confirm that the sample is still in focus by imaging crimson beads with the PMT channel. If not, adjust the sample stage position. Repeat the brightness maximization of Step 153.
• 155Mount the two band-pass filters F4 in a lens tube (Thorlabs, cat. no. SM1S35), maintaining a gap of ~50 mm between them to avoid possible interference. Screw the lens tube to the front of the fiber head adapter mount. This will reduce environmental light detection.
• 156Remove the camera and connect the fiber end with APD3. Image the crimson sample with APD3 while gradually increasing the excitation laser power until the beads can be detected by the APD. Translate the fiber head adapter in three dimensions to maximize the photon count.

  ▲ CRITICAL Start from low laser power and increase gradually to avoid saturation and potential damage of the APDs.

• 157Perform the same procedure for the orange (APD2) and far-red (APD3) detection channels using the corresponding optomechanical components and samples with yellow-green and red beads (Thermo Fisher, cat. nos. F8803 and F8801), respectively.

Setting up pulse gate/delay modules

● TIMING 1 h

• 158Insert a fluorescent bead sample between two objectives. Configure the system to use only one objective for imaging by opening one shutter (SH1 or SH2) while closing the other.

  ▲ CRITICAL The system comprises three pulse gate/delay modules, each associated with a specific detection channel. Therefore, the selection of fluorescent beads should be based on their emission wavelengths, aligning with the passing band of the detection channel to be set up.

• 159Turn on the excitation beam for imaging and move the beads into focus. On the ‘Timing’ control panel in the nanoscope control software, set ‘Window Width (ns)’ to a small value (~0.1 ns). In continuous imaging mode, maximize the peak intensity observed in the collected images by adjusting the ‘Window Delay (ns)’.
• 160Activate the blazed-grating hologram on the SLM. Turn on both excitation and depletion beams for imaging. Set ‘Window Width (ns)’ to its maximum value at 10. Minimize the fluorescence signal observed in the collected images by adjusting the ‘Laser Delay (ns)’ value.

  ▲ CRITICAL It requires >60-mW depletion laser power measured at the back aperture of the objective to observe a substantial STED effect on crimson beams.

• 161Set the ‘Window Width (ns)’ value to 5. Activate the ‘Vortex’ hologram on the SLM control panel of the nanoscope control software. Turn on both excitation and depletion beams for imaging. Fine-tune ‘Laser Delay (ns)’ and ‘Window Delay (ns)’ values (±0.5 ns) as needed for the visually judged best resolution and contrast. The optimized ‘Window Delay (ns)’ value is expected to be ~1-ns larger than the value found in Step 159.
• 162Adjust the ‘Window Width (ns)’ value to maximize the contrast observed in the collected images.

  ▲ CRITICAL The optimal ‘Window Width’ value is expected to be roughly the fluorescence lifetime of the imaged fluorophore.

• 163Save these settings for future use.

  ▲ CRITICAL The optimal combination of values depends on the spectral properties of the imaged fluorophores and therefore can vary when switching to a differently labeled sample.

• 164Repeat this procedure for the other two channels.

Setting up and operating the AO module

● TIMING 2 d

• 165Read Aberration correction for instructions.

  ▲ CRITICAL The following steps describe the initial set up of the AO module, first correcting for system aberrations and then for sample-induced aberrations. Some of these steps will be revisited during daily operation of the instrument, whereas others are required only during the initial setup. In addition, we describe the procedure for the correction of aberrations in thick samples.

• 166To correct the system aberrations caused by imperfections of the optical components and their alignment, turn on LH2 for excitation and image 100-nm-diameter crimson fluorescent beads as the target (for details, refer to Initial test for sample imaging).
• 167Confirm that the objective correction collar is set to match the thickness of the cover glass.
• 168Rotate HWP6 to re-allocate a portion of the excitation laser power to the lower arm of the 4Pi interference cavity.
• 169Perform image-based aberration correction to optimize the Zernike coefficients of DM1 (‘Upper DM’ tab) and DM2 (‘Lower DM’ tab) by opening only one shutter at a time. To achieve this, open the ‘Deformable Mirrors’ control panel and click the ‘Scan Modes’ button. In the subsequent popup panel, choose a single Zernike mode (e.g., #7 for ‘horizontal coma’) for optimization. The nanoscope control software will automatically apply a series of different incremental corrections for the coefficient of the selected Zernike mode while recording the corresponding value of the metric (e.g., image brightness) at each step. The control software will subsequently fit a Gaussian or quadratic function to the measured values of the metric. The optimal correction value is determined as the value that maximizes the fitted function. Repeat this measurement for each required Zernike mode (typically, #5 to #11, and #22). Save the settings of the Zernike mode coefficients.

  ▲ CRITICAL Correcting all the typical Zernike modes once is good enough theoretically, while repeating more cycles can correct residual aberrations when it is not good enough.

• 170Rotate HWP6 back to its original orientation and turn off LH2 after finishing this step.
• 171Turn on LH1 for depletion. Display a ‘Top Hat’-shaped hologram on the SLM. Proceed with imaging the 150-nm-diameter gold nanoparticles using the xz scanning mode. Manually adjust each Zernike coefficient via the ‘SLM’ panel within the nanoscope control software until achieving a PSF with a central minimum approaching zero intensity and resembling the one shown in Fig. 7b,c.

  ▲ CRITICAL Correcting system aberrations (Steps 166–171) is not a routine operation. It should be performed only during the initial system implementation or when any key optical components are re-aligned.

• 172When imaging thick samples that induce large amounts of aberrations caused by the refractive index mismatch between sample, embedding medium and immersion liquid (silicone oil), proceed as follows. Image 150-nm-diameter gold nanoparticles attached to the bottom coverslip of the thick sample using the depletion beam for illumination and the PMT for detection. Repeat Steps 166–169 to optimize the Zernike coefficients for each DM individually by maximizing the brightness as the metric. To reduce noise contributions, apply a background pixel threshold by changing the ‘metric’ setting to ‘Threshold Mean’ in the pop-up panel. Adjust the ‘Lower Limit’ and ‘Upper Limit’ values to define the desired background threshold. These specific threshold values should be determined empirically on the basis of the signal-to-noise ratio observed in the acquired images.
• 173In preparation for minimizing the remaining aberrations, align and lock the objectives (for details, refer to Objective alignment and Focus-lock module alignment).
• 174Turn on both LH1 for excitation and LH2 for depletion. Translate the sample to the ROI. Activate the ‘Vortex’ hologram on the SLM for 2D STED imaging. Choose a weak depletion power (~10% of the power level used for ordinary imaging) to limit photobleaching.
• 175Compensate for the local sample-induced aberrations by simultaneously adjusting both DMs. ‘Scan Modes’ for automated DM adjustment is used in general, whereas manual adjustment is more reliable when the image signal-to-noise ratio is <2. Proceed to optimize the 4Pi mode coefficients¹⁸ sequentially, using the image brightness as the metric for assessment. This iterative process is repeated as needed to ensure effective correction of aberrations induced by the sample.

  ▲ CRITICAL This step is optional when imaging relatively thin cell samples, particularly if the imaging volume is close to the coverslip. However, for thick tissue samples, it becomes essential to compensate for local sample-induced aberrations to enable high-quality isoSTED imaging.

  ▲ CRITICAL For manual compensation of sample-induced aberrations in isoSTED imaging, initiate live imaging in the nanoscope control software. Under the ‘Upper DM’ tab, introduce a small value (e.g., +0.1) into the second column of a chosen Zernike mode row (e.g., #7 for ‘horizontal coma’). Repeat this adjustment with the identical value in the corresponding cell under the ‘Lower DM’ tab. Iterate with varying values to maximize photon counts. Subsequently, apply a small positive offset (e.g., +0.1) to the ‘Upper DM’ value and a corresponding negative offset (e.g., −0.1) to the ‘Lower DM’ value, again optimizing for maximum photon counts. Repeat this process for all relevant Zernike modes. Perform two to three sequential optimization cycles encompassing all necessary 4Pi modes to achieve optimal correction. Note that three specific 4Pi modes¹⁸ ($Q_1^{-1},Q_2^{+1}$, and $Q_3^{+1}$) induce image shifts rather than aberrations and thus require no adjustment in this step.

• 176To establish bias values for 3D imaging of thick samples, begin with using the 2D isoSTED mode with a weak depletion power, set at ~10% of the power used for standard imaging. Next, optimize the 4Pi mode with the ‘z Comp’ panel in the nanoscope control software (Supplementary Fig. S9), including ‘Piston’ for fringe-shift, ‘Defocus’ and ‘Spherical’ for the first-order spherical aberration, to manipulate the interference fringes underneath the PSF envelope until the effective PSF size is unchanged (see more details in ref. 12 and its extended data figure 3). Repeat this measurement every 1 μm in the z direction. Once a series of bias values corresponding to different depths is obtained, calculate the step size on the basis of linear regression analysis applied to these discrete points. Activate the bias-setting function and set the step sizes in the nanoscope control software.

  ▲ CRITICAL Imaging large 3D volumes (>1 μm in the z direction) in thick tissues (>30 μm) necessitates this step.

Initial test for sample imaging

● TIMING 2 h

▲ CRITICAL The first time that the steps in this section will be executed, the image size (scan range) has not been calibrated yet. The calibration steps are described in Image size (scan range) calibration.

• 177Install a sample to be imaged between two objectives.
• 178Load the settings of pulse gate/delay modules from previously saved data.
• 179Align the objectives by following the procedure outlined in Objective alignment.
• 180Compensate for the sample-induced aberrations by following the procedure outlined in Steps 169–173.

  ◆ TROUBLESHOOTING

• 181Lock the positions of the objectives by following the procedure outlined in Focus-lock module alignment.
• 182Use the ‘Image Acquisition Parameters’ panel within the nanoscope control software to preliminarily image the sample at relatively low laser power. Translate the sample to the ROI. Configure the imaging parameters (e.g., scan mode, image dimensions, step size and line accumulation values) to assess imaging quality and performance without significantly bleaching it. Repeat the process outlined in Steps 169–173 for local aberration compensation.

  ◆ TROUBLESHOOTING

• 183Set the depletion laser power above 60 mW (measured at the back aperture of the objective) for high-resolution STED imaging. Typical parameters for high-quality isoSTED images are 10 × 10 μm image dimensions, 1,024 × 1,024 pixels and 100 line accumulations.

  ◆ TROUBLESHOOTING

• 184Enable the preferred detection channel(s) by selecting them in the ‘Detectors’ panel and activate the corresponding laser(s) via the ‘Laser Controls’ panel. Click the ‘Live’ button to initiate live imaging mode.
• 185To record data for later analysis, set the desired file path and click the ‘Capture’ button.

Image size (scan range) calibration

● TIMING 2 d

• 186Mount the Ronchi ruling (Edmund Optics, cat. no. 38–566; 600 line pairs mm⁻¹; period length T = 1.667 μm). Open SH1 and close SH2 to use OBJ1 for imaging the sample with the PMT. Illuminate the sample with the STED_xy beam configuration. Acquire images at a format of 1,024 × 1,024 pixels.
• 187Calibrate the RM first. Carefully rotate the gratings to align the periodic structures parallel with the scanning direction of RM.

  ▲ CRITICAL Scan a large range (e.g., 20 μm) to determine and adjust the orientation of the gratings as precisely as possible.

• 188Apply different control voltages (V) to RM to acquire images of the Ronchi ruling grating at varying sizes of the FOV. Plot the values as shown in Fig. 17a to depict the relationship between the control voltage and the corresponding true image size measured by the Ronchi ruling.

  ▲ CRITICAL The correlation between the control voltage and the image size becomes increasingly imprecise for smaller image sizes (<5 μm).

• 189Fit the raw data using a linear function y = ax + b. In the example shown in Fig. 17a, a = 0.146 and b = −0.1441.
• 190Update the fitting parameters (a and b) in the ‘Re Mirror um to Volts.vi’ in the folder ‘Control Software’ → ‘Host Vis’ → ‘General Sub Vis’.
• 191In preparation for calibrating the GMs, rotate the Ronchi ruling by 90°.

  ▲ CRITICAL We opted to use a second Ronchi ruling grating (Edmund Optics, cat. no. 66–346; 240 line pairs mm⁻¹), because the first one could not be mounted properly after undergoing a 90° rotation).

• 192Repeat Steps 187 and 188 to obtain the fitting parameters for the GMs. In the examples shown in Fig. 17b, fitting with the function y = ax + b yields a = 0.04481 and b = −0.01072.
• 193Update these parameters in the sub-VI called ‘Galvanometers um to Volts.vi’ in the folder ‘Control Software’ → ‘Host Vis’ → ‘General Sub Vis’.

Fig. 17 |. Image size calibration as a function of control voltage on scanning mirrors.

a, RM. b, GM. The image size is measured with Ronchi ruling gratings, and linear curve fitting is applied to identify the control parameters in the LabVIEW program.

Resolution measurement (bead samples)

● TIMING 1 d

• 194Acquire Gatta quant bead images by referring to section ‘Initial test for sample imaging’. Create a new folder and transfer the raw images with DBL file extension.
• 195Open the analysis software named ‘exe.m’ written in MATLAB, located within the folder ‘resolution-quantification-beads’ (code available at the Zenodo repository listed in Code availability).
• 196Type ‘ROIr’, ‘pixelSize’ and ‘beadr’ in lines # 8, 9, and 10, respectively. ‘ROIr’ should be appropriate to crop a single bead. ‘pixelSize’ and ‘beadr’ represent the dimension of each pixel size and the diameter of the bead, respectively.
• 197Run ‘exe.m’ and select the folder that includes bead data.
• 198When an image is opened, proceed to click the center of each bead in the image. Press the ‘Enter’ key to finalize the selection.
• 199Repeat Step 195 for all data.
• 200The software will automatically export the resolution after the analysis.

Resolution measurement (biological samples)

● TIMING 1 d

• 201Save two different images with the same FOV sequentially.
• 202Open ImageJ (National Institutes of Health) and open two saved images to perform Fourier ring correlation.
• 203Run the plugin by following Plugins → BIOP → Image Analysis → FRC → FRC calculation.
• 204Select two images and ‘Fixed 1/7’ for resolution criteria²¹. (Optionally, the method provided in ref. 22 is recommended for resolution measurements with tubular biological structures).

User operation

After completing the above steps, the isoSTED nanoscope should be properly aligned and calibrated and ready for biological applications. Users primarily operate the instrument through the nanoscope control software (the user interface is shown in Supplementary Fig. S9). Steps 177–185 in Initial test for sample imaging describe the standard operation procedure that users should follow to image their samples. Each time placing a new sample between the OBJs for imaging, it is recommended to evaluate the aberrations by imaging the point-like objects (e.g., gold nanopaticles or fluorescent beads) included in the sample and correct the aberrations with the AO devices controlled through the nanoscope control software. The aberration-correction operations may take 15–45 min depending on how severe the aberrations are. The operations of AO devices for aberration correction require the operator to be familiar with the optical aberration patterns in the primary Zernike modes. We recommend an optical engineer or anyone with relevant training to be involved in the operation.

To obtain images that fully take advantage of the optical resolution of the isoSTED nanoscope, the pixel size should be smaller than half of the expected resolution, in accordance with the Nyquist sampling theorem. The ‘Pixel Size (nm)’ is determined by the ‘Image Format’ and ‘Image Dimension (μm)’ settings in the ‘Image Acquisition Parameters’ control panel. In 3D z-stack imaging mode, it is additionally required to set the ‘Z Steps Size (μm)’ to a value smaller than half of the expected axial resolution. isoSTED nanoscope produces images at super resolution as soon as the image acquisition is complete. The image acquisition time (in seconds) of a single frame is determined by (Image line number × Line accumulations)/Fast Axis Frequency (Hz). Typically, the image acquisition time is a few (<10) seconds for a single frame and a few minutes for z-stack imaging depending on the total frames acquired.

Troubleshooting

In complex optomechanical systems such as this isoSTED nanoscope, challenges are manifold and range from suboptimal resolution to a non-functional nanoscope. Table 1 lists commonplace issues, but it is impossible to anticipate every potential problem. We invite our readers to contact us should they run into obstacles.

Table 1 |.

Troubleshooting table

Step	Problem	Possible reason	Solution
117	Very low or no signal levels	The shutters in front of the DMs are closed	Open the shutters
		The detection window width is set to 0	Set the detection window width to >3 ns
		Poorly labeled sample	Check the sample with a different microscope
		Lack of immersion oil	Add immersion oil until the droplet fills the gap between the objective and the specimen. Make sure that no air bubble is introduced in the oil
		Poor cable connections	Check the electronic connections. A frequently overlooked place is the RTSI bus between the two FPGA boards
180	Heavily distorted PSF	Malfunction of DM(s)	Regular checks on the membrane are encouraged. A quick check can be done by inspecting the beam reflected off the DM when the depletion laser is on. If one or two actuators are damaged, the illumination becomes non-uniform, which is observable directly by eye or by using a camera. A thorough inspection is possible by re-calibrating the DM offline using a Twyman-Green interferometer (for details, see SLM and DM calibration)
		Beam misalignment	Misalignment of any of the beam paths disrupts the conjugation between key devices (SLM, RM and DMs) with the back pupil of one or both objectives. A rapid assessment can be conducted by substituting the objective with a camera and capturing an image of its back pupil. Small misalignments can potentially be addressed by directly shifting the hologram pattern on the SLM. Larger misalignments require manual correction
182	High background	Open the instrument enclosure	Close the enclosure during imaging
		Incorrect polarization states of the depletion beam in the 4Pi interference cavity	Rotate the wave plates inside the 4Pi interference cavity to allow the depletion beam to leave the cavity from the empty (top-right) side of PBS2 after passing through both objectives12
	Low fluorescence intensity	Aberrated PSFs	If only the depletion PSF is affected, use the SLM for correcting the systematic aberrations in the depletion beam path. In cases in which the excitation PSF is also affected, opt for the use of DMs
		The sample is out of focus	Translate the sample until it is in focus
183	The STED imaging mode photobleaches the sample, whereas the confocal imaging mode is normal	Sample labeling used dyes that are sensitive to photobleaching	Make a new sample with fresh dyes that have been reported to work well with STED imaging
	The STED imaging mode quenches fluorescence signals, causing images to be extremely dim and poorly resolved	The depletion laser’s beam quality is poor, with too high intensity at the center, or the depletion laser focus is spatially mismatched with the excitation beam focus	Use a sub-diffraction bead sample to evaluate the depletion beam quality; adjust the SLM holograms to optimize the depletion laser beam quality and its spatial overlap with the excitation laser focus
	The STED imaging mode saturates the detector in the images of fluorescent samples that are mixed with gold nanoparticles	Gold nanoparticles in the imaging area scatter laser light, leading to detector saturation	Use the PMT detector to observe the gold nanoparticles and move the sample XY stage to an area without gold nanoparticles

Timing

We provide an approximate time frame below for essential tasks during setup. Notably, the actual duration may vary depending upon part availability, as well as the experience of the personnel involved. Given the complex nature of the system, a generous allotment of time is advised for troubleshooting and potential repetition of tasks and to correct for errors in machined components or replace damaged elements. With careful planning and a focused effort, a collaboration between a biophysicist or optical engineer for instrument setup and biologists for specimens, isoSTED images can be acquired within 12 months. Upon gaining proficiency in constructing the nanoscope, coupled with the availability of all required components, the instrument can be set up in about 2–3 months.

Reagent setup and sample preparation: 1 h for fluorescent bead samples, 1–2 d for cell samples and 1 week for tissue samples

Equipment setup and room preparation: 1 week to 2 years, based on our experience, contingent upon facility considerations. This duration is highly variable and is intricately tied to the availability of suitable space and responsiveness of the local facilities department to fix problems.

Equipment setup and lead times on components: certain pivotal components, such as paired objective lenses and DMs often entail lead times of several months. Early ordering of these components is advisable to circumvent potential delays.

Equipment setup, software installation and familiarization: 1 week for control and data collection software; 3 d for other custom code.

Steps 1–4, software installation and setup: 1 d

Steps 5–6, optomechanical assembly: 1 week

Steps 7–8, electronics assembly: 1 week

Steps 9–15, AOM setting and alignment: 2 h

Steps 16–23, polarization-separation module alignment: 4 h

Steps 24–41, SLM installation and alignment: 1 d

Steps 42–62, excitation path alignment: 3 d

Steps 63–74, scanner installation and alignment: 1 d

Steps 75–88, PBS installation and angle alignment: 3 h

Steps 89–91, power allocation wave plate alignment: 30 min

Steps 92–94, DM, scanner and SLM conjugation alignment: 2 h

Steps 95–112, 4Pi interference cavity alignment: 1 week

Steps 113–118, objective alignment: 3 d

Steps 119–125, focus-lock module alignment: 4 h

Steps 126–141, test sample preparation: 3 d

Steps 142–157, detection path alignment: 1 d

Steps 158–164, setting up pulse gate/delay modules: 1 h

Steps 165–176, setting up and operating the AO module: 2 d

Steps 177–185, initial test for sample imaging: 2 h

Steps 186–193, image size (scan range) calibration: 2 d

Steps 194–200, resolution measurement (bead samples): 1 d

Steps 201–204, resolution measurement (biological samples): 1 d

Anticipated results

The example presented here illustrates the expected performance of the system. When it is fully functional, the isoSTED nanoscope enables users to image fine subcellular structures in single- or multi-color mode in cells and in tissues (e.g., the Golgi apparatus in HeLa cells; Fig. 18a–c). If a sufficiently high 3D resolution is reached, the layered structure of the Golgi apparatus becomes clearly visible, and the cross sections at different depth are distinct (Fig. 18d–f).

Fig. 18 |. Example isoSTED images of coat protein complex I (COPI) vesicles (magenta) and Golgi apparatus (green) in a HeLa cell visualized by immunolabeling βCOP with ATTO 647N and GM130 with ATTO 594.

a, Projection image in the z direction. b,c, Cross sections in the xz (c) and yz (d) planes at the positions denoted by the dashed lines in a. d–f, xy cross sections at depths (z) of 250 nm (d), 500 nm (e) and 750 nm (f). Scale bar, 1 μm.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41596-026-01365-7.

Key points.

• The adaptive-optics-assisted isoSTED nanoscope has two vertically aligned and opposing objectives, where the fluorescence can be detected within a 3D volume smaller than that of the diffraction limit by depleting the excited fluorophores localized within the excitation volume using stimulated emission.

• The isoSTED approach is advantageous for imaging thick-tissue samples compared to other 3D super-resolution microscopy techniques, such as single-molecule localization microscopy or structured illumination microscopy.

Data availability

The CAD file repository built in SolidWorks (Dassault Systèmes SE), an Excel file named ‘Image Data.xlsx’ that details the imaging parameters and plot data and a zip file named ‘Fig. 18 Raw Data.zip’, which is the raw dataset used for Fig. 18, can be found at the shared Zenodo repository: https://doi.org/10.5281/zenodo.18442076.

The complete material list, including reagents, tools, control computer, drivers and other software, are available in Supplementary Table S5. The complete part list is available in Supplementary Table S6.

Additional relevant information is available by reasonable request from the corresponding authors X.H. and J.B. Source data are provided with this paper.

Code availability

The control and data collection software built in LabVIEW (National Instruments) and the source code for functions like data transformation and resolution quantification are available at the Zenodo repository (https://doi.org/10.5281/zenodo.18442076). The source code for SLM calibration²⁹ is available from the Github link (https://github.com/Hao-Laboratory/Diffraction-based-SLM-calibration) and the source code for DM calibration²⁸ is found at https://github.com/jacopoantonello/dmlib.