🔭 astrophysics

Updating the SCExAO/CHARIS polarimetric calibration following the Nasmyth beam-switcher upgrade

This paper presents an updated Mueller matrix polarimetric calibration model for the Subaru/SCExAO/CHARIS system following the Nasmyth beam-switcher upgrade, which improves accuracy by adopting an elliptical retarder model for the image derotator, characterizes new polarization effects from the YJH50 dichroic and tertiary mirror diattenuation, and provides an open-source Python package (pyPolCal) to facilitate future recalibrations.

Original authors: Thomas McIntosh, Manxuan Zhang, Briley L. Lewis, Miles Lucas, Maxwell A. Millar-Blanchaer, Jaren Ashcraft, Kyohoon Ahn, Jeffrey Chilcote, Thayne Currie, Vincent Deo, Yoshiyuki Doi, Tyler Groff, Olivie

Published 2026-07-28

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Thomas McIntosh, Manxuan Zhang, Briley L. Lewis, Miles Lucas, Maxwell A. Millar-Blanchaer, Jaren Ashcraft, Kyohoon Ahn, Jeffrey Chilcote, Thayne Currie, Vincent Deo, Yoshiyuki Doi, Tyler Groff, Olivier Guyon, Takashi Hattori, Tomoyuki Kudo, Kellen Lawson, Julien Lozi, Yosuke Minowa, Yuhei Takagi, Rob G. Van Holstein, Sebastien Vievard

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to take a picture of a faint, glowing dust cloud swirling around a distant star. The problem is that the star itself is a blindingly bright flashlight, making the dust cloud impossible to see. Astronomers have a clever trick to solve this: they use a technique called "polarimetric differential imaging." Think of starlight as a crowd of people walking in a straight line, all facing the same way. When this light bounces off tiny dust grains in the cloud, it gets "scrambled" and starts spinning or wiggling in a specific direction. This change is called polarization. By using a special camera that only sees light spinning in that specific direction, astronomers can filter out the straight-line starlight and make the dust cloud pop into view.

However, building a camera that sees these subtle spins is like trying to listen to a whisper in a hurricane. Every mirror, lens, and piece of glass the light passes through on its way to the camera can accidentally twist or scramble the spin, adding "noise" to the signal. To fix this, scientists build a mathematical map called a Mueller matrix. Think of this map as a recipe book that tells the computer exactly how every single piece of the telescope messes with the light, so it can mathematically "un-mess" the data and reveal the true picture. Without this map, the measurements of the dust clouds would be wrong, and we couldn't learn about how planets are born.

The Paper's Story: Updating the Recipe Book

This paper is about updating that recipe book for a very powerful telescope system called SCExAO/CHARIS, which sits on top of the Subaru Telescope in Hawaii. The team, led by Thomas McIntosh and colleagues, realized that the telescope had recently undergone some major renovations, specifically the installation of a new "beam-switcher" (a device that directs light to different instruments) and a new set of mirrors for a near-infrared sensor. They knew that adding new glass and mirrors to the path of the light was like adding new ingredients to a cake recipe; the old instructions wouldn't work anymore. They needed to re-measure how the light behaved to ensure their pictures of dust clouds remained accurate.

The New "Twist" in the Tale
The biggest surprise the team found was in a part of the telescope called the image derotator. You can think of the derotator as a rotating platform that keeps the image steady as the Earth spins. In previous years, scientists treated this platform like a simple "linear retarder"—a device that just shifts the light's spin in a straight line. But the new data showed that this simple model was failing. The derotator was actually acting like an elliptical retarder, a more complex device that twists the light in a spiral shape.

To fix this, the authors swapped their old, simple model for a more complex "elliptical" one. They found that this new model fit the data much better, reducing errors significantly. If they had stuck with the old model, their measurements of the dust clouds would have been slightly off, especially for targets that are highly polarized.

The New Parts: Good News and Bad News
The team also investigated the new hardware they installed:

The YJH50 Dichroic: This is a special mirror that splits light, sending some to a sensor and some to the science camera. The team found that this new mirror was causing a tiny bit of "crosstalk" (unwanted mixing of signals) in the J-band (a specific color of infrared light). They couldn't fully model this effect with their current tools, so they had to admit that for very bright, polarized targets in this specific color range, the accuracy drops slightly.
The Nasmyth Beam-Switcher (NBS): This new device directs light to the camera. The team discovered that, surprisingly, it doesn't scramble the light much at all. Its only real effect is to flip the sign of the polarization (like turning a left-handed glove inside out to look like a right-handed one). This is a very clean, predictable change that is easy to correct for.

The Mirror That Got a Makeover
The team also looked at the telescope's tertiary mirror (M3), which reflects light into the instrument. They noticed that since the last calibration, this mirror had become more "diattenuating," meaning it absorbs or blocks one type of light polarization more than the other. They suspect this happened because the mirror's coating changed over time or was recently recoated. Because of this, they had to update the mathematical model for the mirror's behavior. They noted that since the mirror was recoated after they took their data, their current numbers are just an estimate, and the mirror will need to be measured again soon.

How Good is the New Model?
The authors calculated how accurate their new map is. For a target that is 1% polarized (a very faint signal), their model is accurate to within 0.02% to 0.12% in measuring the strength of the polarization, and within 0.5° to 3.2° in measuring the angle of the polarization. They point out that the biggest source of error isn't the math itself, but the "noise" in the data they took from the sky.

The Gift to the Community
Finally, the authors didn't just keep these new rules to themselves. They packaged all their code into an open-source tool called pyPolCal. This is like giving every other astronomer a free, updated instruction manual and a set of tools so they can easily re-calibrate the telescope whenever new parts are added or old ones wear out. They emphasize that because telescope optics change over time (due to dust, oxidation, or recoating), regular re-calibration is necessary to keep the science accurate.

In short, this paper is a maintenance update for a high-tech camera. The team found that the camera's internal spinning parts had changed their behavior, added a new mirror that was mostly harmless, and updated the recipe for the main telescope mirror. By doing so, they ensured that when astronomers look at the dusty nurseries of new planets, the picture they see is as clear and true as possible.

Technical Summary: Updating the SCExAO/CHARIS Polarimetric Calibration Following the Nasmyth Beam-Switcher Upgrade

Problem Statement
Quantitative polarimetry with the Subaru Coronagraphic High Angular Resolution Imaging Spectrograph (CHARIS) requires precise correction for instrumental polarization effects introduced by the optical path. These effects are modeled using a Mueller matrix to reconstruct the incident Stokes vector of astronomical targets. Since the last major calibration (Ref. 24), significant hardware changes have occurred, including the 2025 installation of the Nasmyth Beam-Switcher (NBS) and new dichroic beamsplitters (specifically the YJH50) for the near-infrared wavefront sensor (NIRWFS). Additionally, the tertiary mirror (M3) of the Subaru telescope underwent re-coating. These changes, combined with potential temporal degradation of existing optics, necessitated an updated calibration model to preserve the instrument's capability for measuring dust grain properties in circumstellar disks and other high-contrast imaging science cases. Furthermore, previous calibration efforts lacked publicly available source code and a streamlined process, hindering regular re-calibration.

Methodology
The authors developed an updated Mueller matrix model and the accompanying open-source Python package, pyPolCal, to facilitate future calibrations. The methodology involved:

Data Acquisition:
- Internal Calibrations: Data were collected using a polarized internal source (100% $-Q$ ) and an unpolarized internal source. Datasets were acquired before and after the NBS installation, with and without the YJH50 dichroic inserted.
- On-Sky Calibration: An unpolarized standard star (HD293396) was observed post-NBS installation to calibrate the diattenuation of the tertiary mirror (M3).
Modeling Framework:
- The optical path was modeled as a chain of Mueller matrices and rotation matrices ( $T(\Theta)$ ).
- Key Model Updates:
  - Derotator: The model for the image derotator (k-mirror) was upgraded from a linear retarder to an elliptical retarder ( $M_{ER}$ ), introducing three retardance parameters ( $\Delta_H, \Delta_{45}, \Delta_R$ ) per wavelength bin.
  - NBS: Modeled as a diagonal matrix $\text{diag}(1, -1, 1, -1)$ , representing a sign flip in $Q$ and $V$ components, derived from ray tracing.
  - YJH50 Dichroic: Assigned an identity matrix, as attempts to fit retardance or diattenuation models failed to reduce residuals.
  - M3: Modeled as a diattenuator-retarder using physical models based on the dielectric functions of silver, with parameters fitted globally across wavelengths.
- Fitting Strategy: Parameters were fitted using scipy.optimize and emcee. Global parameters (e.g., HWP layer thicknesses, misalignment angles) were fitted across all wavelength bins simultaneously, while retardance and modulation efficiency were fitted per wavelength.
Error Analysis: Polarimetric accuracy was calculated using the standard error of the mean of residuals ( $\bar{s}_x$ ), distinguishing between errors scaling with polarized intensity ( $s_{rel}$ ) and total intensity ( $s_{abs}$ ).

Key Contributions

Model Revision: The identification that an elliptical retarder model for the image derotator significantly improves fit quality over the previously used linear retarder model, particularly in the presence of new optical components.
Hardware Characterization:
- Determined that the NBS primarily induces a sign flip in $Q$ and $V$ with minimal other polarization effects.
- Identified that the YJH50 dichroic introduces faint crosstalk in the J-band that cannot be characterized by the current component Mueller matrix modeling, leading to reduced accuracy for highly polarized targets in this band.
- Found that the diattenuation of the tertiary mirror (M3) has increased since the previous calibration, likely due to aging or re-coating.
Software Release: The release of pyPolCal, an open-source package containing the fitting code, data reduction scripts, and tutorial notebooks, enabling straightforward re-calibration after future optical changes (e.g., mirror recoatings).

Results

Accuracy: For a 1% polarized target, the updated model achieves a polarimetric accuracy in the Degree of Linear Polarization (DoLP) ranging from 0.02% to 0.12%, and in the Angle of Linear Polarization (AoLP) from 0.5° to 3.2°.
Limitations: The primary limitation on accuracy is the noise in the on-sky calibration data, which drives the absolute error ( $s_{abs}$ ). For highly polarized targets (>10%), the J-band accuracy is further degraded by the unmodeled crosstalk from the YJH50 dichroic.
Efficiency: The polarimetric efficiency is strongly wavelength-dependent and minimized at derotator angles of approximately 45° and 135°, where linear polarization is converted to circular polarization (which is lost in the linear-only detection scheme).
M3 Changes: The fitted diattenuation of M3 is significantly higher than in previous calibrations, and the physical model parameters differ, indicating a change in the mirror's polarization properties over time.

Significance and Claims
The paper claims that this model update is essential for maintaining the quantitative polarimetric capabilities of CHARIS following the Nasmyth beam-switcher upgrade in October 2025. The authors emphasize that the optical path's polarization effects are not static; the shift from a linear to an elliptical retarder model for the derotator and the increased diattenuation of M3 demonstrate that components change over time. Consequently, the authors argue that regular re-calibration is necessary. The provision of pyPolCal is presented as a critical step toward making this regular calibration feasible and transparent. The authors explicitly state that the current accuracy estimates are provisional pending new calibration data, particularly noting that the M3 re-coating in December 2025 will likely alter the diattenuation model further, requiring future updates to the parameters.

More like this