Abstract
The K-band (2.0-2.4 μm) is one of the most important wavebands for ground-based infrared observations, effectively penetrating interstellar dust and observing cool astrophysical objects and high-redshift galaxies. To address the long-standing gap and strong demand for near-infrared astronomical observations in China, this paper reports the astronomical observation results of a domestically developed mercury cadmium telluride (HgCdTe) infrared focal plane camera on the Sun Yat-sen University 80 cm infrared telescope. Located at the Lenghu Astronomical Observatory in Qinghai Province, the telescope is equipped with an HgCdTe infrared focal plane camera developed by the University of Science and Technology of China. Its chip is composed of three 640×512 HgCdTe infrared focal plane array detectors with a pixel pitch of 15 μm, developed by the Shanghai Institute of Technical Physics, Chinese Academy of Sciences. Test results show that the 5σ limiting magnitude is 15.3 mag (Vega system) in a single 20-second exposure, reaching the international level of the Two Micron All-Sky Survey (2MASS), with a photometric accuracy of up to 12 mmag for bright sources. After stacking for 30 minutes and 2 hours, the limiting magnitudes reach 17.5 mag and 18.2 mag, respectively. These results demonstrate that the system can conduct K-band time-domain astronomical observations at different detection depths and temporal resolutions. This achievement marks China ′s first successful astronomical K-band scientific imaging observation using a domestically developed HgCdTe infrared focal plane camera, providing a solid and reliable independent observation platform for infrared astronomy research in China.
0 Introduction
The near-infrared band usually refers to the wavelength range of 1–5 μm, which connects the visible light and the longer-wavelength mid-infrared band. The near-infrared band has a longer wavelength than the visible light, which can effectively penetrate interstellar dust, thereby observing targets such as high extinction regions, star-forming regions, protoplanetary disks and active galactic nuclei that are blocked in the optical band. Many low-temperature celestial bodies(such as late-type giants and brown dwarfs) emit near-infrared radiation that is stronger than that in the optical band, so near-infrared observation has a natural advantage; in particular, the absolute optical correction of the K band is relatively stable and can be used as a simple luminosity index [1]. In addition, the expansion of the universe has shifted the spectrum of distant celestial bodies to the red end, making near-infrared observation crucial for studying the early universe and high-redshift galaxies. For example, in recent years, near-infrared observations by the James Webb Space Telescope(JWST) have discovered and confirmed several galaxies with z> 10 and even z ≈ 14, highlighting the transformative role of space near-infrared observation in the study of the early universe [2]. Currently, the near-infrared band is an important part of astronomical observation and has been widely used in the study of star formation and the Milky Way nucleus.
As an important atmospheric window for ground-based near-infrared observation, the K-band is the longest band for ground-based infrared telescopes to achieve deep imaging observations. Its center wavelength is about 2.2 μm(the commonly used bandwidth is 2.0–2.4 μm, which varies depending on the filter system). Figure 1 shows the theoretical atmospheric transmittance of the cold lake in winter and the response curve of the K-band filter of Mauna Kea Observatories(MKO) [3]. Compared with the other two bands of near-infrared—the J-band(~1.25 μm) and the H-band(~1.65 μm)—the longer wavelength of the K-band makes it more penetrating to dust and can receive thermal radiation from hot dust or relatively cold celestial bodies, which has advantages in some research directions. For example, a study on the period-luminosity relationship of Cepheid variable stars in the Large Magellanic Cloud(LMC) found that the overall dispersion of the infrared K-band is the smallest [4], which is particularly suitable for standard candle studies(especially high extinction regions). For low-temperature celestial bodies, the radiation peak of brown dwarfs with a temperature of 1000–1500 K is exactly in the K-band range. For exoplanets, the edge dimming effect is weaker in the long-wavelength band, which is beneficial for the accurate determination of radius and orbital inclination. However, the overall brightness of the sky background in the K-band is higher [5], and it is easily affected by the thermal radiation of the environment and the instrument itself, so there are high requirements for the cooling of the system.
Fig.
1
Theoretical atmospheric transmittance and MKO K-band filter response curves in Lenghu during winter (atmospheric transmittance was calculated using PWV=1.6 mm and airmass=1.0, and the filter response curves did not consider the influence of transmittance) [3].
The history of K-band astronomical observation can be traced back to the late 1960s. The Two-Micron Sky Survey(TMSS) project first used a liquid nitrogen-cooled lead sulfide(PbS) photodetector on a 62-in telescope to achieve a survey observation with a center wavelength of 2.2 μm [6]. However, the real imaging capability emerged after the maturity of two-dimensional infrared detector array technology, especially the application of mercury cadmium telluride(HgCdTe) material. This gradually enabled large ground-based telescopes such as the United Kingdom Infrared Telescope(UKIRT), Keck Telescope, and Very Large Telescope(VLT) to obtain high-precision imaging and adaptive optics capabilities in the K-band [7]. For example, in 1998, UKIRT replaced the small array indium antimonide(InSb) camera IRCAM with a large array of mercury cadmium telluride camera UFTI, which greatly improved the field of view, resolution and sensitivity [8]. The 2MASS project completed a full-sky survey in the J, H and Ks(~2.16 μm) bands. In space missions, the Hubble Space Telescope(HST), JWST, and Euclid space telescope are also equipped with mercury cadmium telluride infrared focal plane detectors to achieve near-infrared observations [8]. Internationally, the most widely used detectors are the H waii-2RG(H2RG) series detectors developed by Teledyne Imaging Sensors in the United States. The array size is 2048×2048, the dark current is less than 0.05 e-/s/pix, and the readout noise is less than 20e- [9]. JWST and Euclid are equipped with H2RG detectors [10-11]. The future Roman space telescope and ELT telescope are planned to be equipped with H4RG detectors, with an array size of 4096×4096.
As China's infrared observation started late and lacked the impetus of large-scale astronomical projects, the development of near-infrared cameras in China has been slow. At present, high-performance near-infrared cameras in the world are not open to import into China. In 1986, the Xinglong Observatory of the National Astronomical Observatories of the Chinese Academy of Sciences installed a 1.26 m infrared telescope on the top of a mountain at an altitude of 960 m and used an indium antimonide photometer for infrared band observation [12]. Later, due to the lack of infrared array detectors, it was changed to optical observation, which left China's infrared astronomical observation blank for a long time. At present, the few infrared cameras in China are mainly indium gallium arsenide(InGaAs) cameras [13], whose cutoff wavelength is usually around 1.7 μm, which can not meet the needs of H and K band observation. However, large-scale national projects such as the China Space Station Telescope(CSST) have effectively promoted the development of near-infrared detectors in China. The CSST program is equipped with eight mercury cadmium telluride infrared detectors with a cutoff wavelength of 2 μm to conduct photometry and seamless spectroscopic observations in the J and H bands [14]. Its initial results have been applied to the Sheshan 1.56 m telescope of the Shanghai Astronomical Observatory of the Chinese Academy of Sciences [15]. In 2024, the Shanghai Institute of Technical Physics of the Chinese Academy of Sciences developed a mercury cadmium telluride infrared focal plane detector with a cutoff wavelength of 2.5 μm for astronomical observation to meet the needs of K-band astronomical observation. The team of the University of Science and Technology of China used this mercury cadmium telluride detector to complete the development of the first K-band mosaic camera in China [16]. The camera was used for actual astronomical observations on the 80 cm infrared telescope of Sun Yat-sen University, achieving a breakthrough in K-band astronomical observation in China from 0 to 1.
This article reports the test results of the first scientific-grade near-infrared K-band imaging observation achieved by the 80 cm infrared telescope of Sun Yat-sen University using domestically produced equipment. First, the basic design parameters of the telescope and camera are introduced; then, the data processing flow is explained, and relevant performance and preliminary results are demonstrated; finally, the article summarizes and looks forward to the future scientific applications of the telescope, as well as the expected performance of this mercury cadmium telluride infrared focal plane camera when paired with a larger aperture telescope.
1 Equipment
1.1 Sun Yat-sen university 80 cm infrared telescope
The 80 cm infrared telescope of Sun Yat-sen University is a new generation of infrared telescope in China. It is located at the Lenghu Astronomical Observatory in Qinghai(see Figure 1). The E area of Saishiteng Mountain, at an altitude of 4100 m, has an excellent observation environment and excellent seeing [17].
Fig.
2
Sun Yat-sen University 80 cm Infrared Telescope.
The telescope has an 80 cm aperture and a focal ratio of f/8 to reduce the influence of skylight background and ambient thermal radiation. It employs a reflecting Cassegrain architecture with two Nyquist focal points, housing one indium gallium arsenide J-band camera and one mercury cadmium telluride K-band camera. The optical path switches between the J and K bands by rotating the three mirrors, and uses J and K filters from the MKO system. This filter system is optimized for ground sites at mid-altitude regions to reduce the impact of atmospheric background radiation.
1.2 Mercury cadmium telluride infrared focal plane camera
The mercury cadmium telluride(HgCdTe) infrared focal plane array camera developed by a team from the University of Science and Technology of China employs a short-wave infrared focal plane array detector(FSA) developed by the Shanghai Institute of Technical Physics, Chinese Academy of Sciences, for astronomical K-band observations. The chip size is 9.60 mm×7.68 mm, the focal plane area is 640×512, and the pixel pitch is 15 μm. Laboratory tests show that its dark current is 4.7 e-/s/pixel, readout noise is 65 e-, K-band quantum efficiency reaches 85%, full-well charge is 97.3 ke-, and blind cell rate is less than 0.6%.
The camera splices three chips together, arranged in a 3×1 configuration with short sides connected, with an imaging area spacing of 3 mm and a flatness error controlled within 15 μm. The chips and other low-temperature optical components are mounted in the dewar. When paired with the 80 cm infrared telescope of Sun Yat-sen University, the final field of view is 3 chips×5'×4', with an angular resolution of 0.484''/pixel. The lens assembly inside the camera is cooled to 150 K to suppress its own thermal radiation, while the focal plane array is cooled to 75 K to suppress dark current as much as possible [16].
In 2024, the team from the University of Science and Technology of China packaged the engineering-grade chip and conducted preliminary tests on the telescope [16]; in 2025, the chip was replaced with a scientific-grade chip and laboratory tests of the mercury cadmium telluride infrared focal plane camera were conducted again(the main parameters are shown in Table 1).
Table
1
Main parameters of the mercury cadmium telluride infrared focal plane camera (dark current is the pure dark current value before chip packaging)
2 Data processing
2.1 Data correction
The raw image output by the camera contains various systematic errors that can be separated and subtracted, as shown in Figure 3(a). Therefore, it is necessary to perform systematic error correction processing on the image to highlight the flow signal. The exposure is set to 1 ms when shooting the bias, and the main bias Biasis created by median superposition.
Bright-field images in the K-band contain thermal radiation from the telescope and its surrounding environment. We take an image with the same exposure when the lens cap is closed and approximate it as the background field, denoted as Bkg. However, the background changes with ambient temperature, and in the future we will combine shake-based shooting to create a real-time background.
The sky plane was captured by shooting a uniform sky at dawn and dusk, with the K-band image taken when the solar altitude angle was approximately 0°. To minimize the impact of thermal radiation, a slightly higher solar altitude angle was chosen, along with a shorter exposure time(100 ms), resulting in a larger proportion of skylight and a smaller proportion of thermal radiation. After subtraction Bias and corresponding exposure time adjustments Bkg', the median values were superimposed and normalized to create the master plane Flat. Future plans include combining the metering plane with corrections for large-scale inhomogeneities caused by thermal radiation and stray light.
Finally, for the original captured image light, image correction is performed according to equation(1):
（1）
The resulting images sci are used for subsequent processing. Figure 3(b) shows an example image after data correction. Future studies will consider the effects of nonlinearity and add correction steps.
Fig.
3
Images before and after data correction
2.2 Aperture metering
After completing the data correction, we performed star source detection and aperture photometry using SExtractor software [18]. We used apertures 3, 6, and 8 for photometry, and used aperture 3 when analyzing faint stars and aperture 6 when analyzing bright stars, in order to balance the ratio of star flux to background flux contained in the aperture, thereby improving the signal-to-noise ratio. We used SCAMP software [19] to calculate the WCS parameters of the image, determined the right ascension and declination of each star, and used the 2MASS database [20] as a position reference.
2.3 Flow rate calibration
We use the 2MASS database as a reference for aperture photometric flow rate calibration. Since the response curves of Ks filters used in 2MASS differ from those in our MKO system K, we perform band conversion on 2MASS using color relationships before flow rate calibration Ks. As shown in Figure 4, our conversion relationship is as follows:
（2）
The results show that our filter has a very small color difference compared to 2MASS, but is slightly bluish, contrary to expectations. This may be due to the difference in atmospheric transmittance between the two sites, or the slight difference in the filter being masked by metering errors, or the influence of the high extinction of the silver disk. After conversion, the calibration method is as follows:
（3）
In the formula, the magnitude zero point Zeropoint is obtained by cross-matching with the converted 2MASS database.
Fig.
4
The color conversion relationship with 2MASS
3 Results
3.1 Metering accuracy and stability
In astronomical applications, the photometric accuracy of an image is typically expressed using limiting magnitude. The deeper the limiting magnitude, the greater the telescope's detection capability. For measuring light curves, the focus is on the stability of time-domain photometry.
For most targets, we used high gain and 20 s exposures for observation. To evaluate the photometric performance of a single exposure, we analyzed a set of NGC 6819 images(78 valid images in total) taken on October 19, 2025. After flow calibration with 2MASS, for each set of two adjacent images, we cross-matched the photometric results and plotted the average magnitude/magnitude difference data for all the same star in the two images. Figure 5(a) shows the results for aperture 3, demonstrating a trend of dimmer magnitude and greater dispersion. On this basis, with a bin size of 0.5 mag, the standard deviation of all data points within each bin was calculated and divided by 2 to obtain the relationship between photometric precision and magnitude(see Fig.5(b)). When the photometric precision is 0.2 mag, the corresponding 5σ limiting magnitude is 15.3 mag. It can be seen from the figure that the photometric error increases instead when targets are brighter than 11.5 mag, indicating that under the observational conditions of that night, a single frame with high gain and 20 s exposure becomes saturated at 11.5 mag.
Fig.
5
Metering accuracy
To evaluate the telescope's photometric stability during time-domain observations, as shown in Figure 6, we plotted light curves for at least 50 sources observed in NGC 6819 and calculated the sigma for sources with stable light curves. For faint stars, aperture 3 provides higher accuracy, essentially consistent with the aforementioned photometric accuracy; for bright stars, aperture 6 achieves a maximum time-domain photometric accuracy of 12 mmag.
Fig.
6
Time-domain photometric stability
3.2 Image overlay
A single 20-second image is primarily intended for applications requiring high temporal resolution. If the target does not have high temporal resolution requirements, image stacking can reduce random noise intensity, thereby improving the signal-to-noise ratio and detection depth, while also significantly reducing the impact of dead pixels on detection. Assuming that faint star metering is dominated by random noise, theoretically, the relationship between the limiting magnitude derived from image stacking and the total exposure time is as follows:
（4）
In the formula, t0 is the exposure time for a single image, t1 is the total exposure time for the stacked images. That is, when the exposure time is increased by 10 times, the limiting magnitude increases by 1.25. In reality, residual systematic errors will cause the increase in limiting magnitude to be slightly less than the theoretical value.
After completing the WCS parameter calculation, we used SWarp software to perform mean stacking on a batch of images, including sigma-clipping, and tested scenarios with stacking time of 30 minutes and 2 hours. Since the stacked images lack neighboring images for subtraction, we used the magnitude error output by SExtractor software for depth calculation, and took the magnitude when MAGERR_APER was 0.2 as the limiting magnitude.
Figure 7 shows the image quality of a single image and a 30-minute stacked image, and compares it with international telescopes 2MASS and UKIRT. As the first K-band scientific imaging achievement in China, the 20-second single image depth is 15.3 mag (Vega system), approaching the level of the international 2MASS; the 30-minute stacked depth is 17.5 mag, and the 2-hour stacked depth is 18.2 mag. If the noise is dominated by random noise, a 4-hour stacked image is expected to approach the UHSDR2 depth (18.5 mag). The telescope can perform temporal monitoring at 20-second and 30-minute resolutions, and can also achieve deep-field imaging of small sky areas through long-term stacking when needed.
Fig.
7
Comparison of NGC6819 overlays (from left to right: 2MASS, 20 s single image, 30 min overlay (20 s×90 images), UKIRT).
3.3 Error analysis
The noise in astronomical image photometry mainly includes photon noise, readout noise, dark current noise, skylight background noise, and thermal radiation noise(ignoring flicker noise). In the current data processing stage, thermal radiation appears in the background of the image and can not be separated from the skylight. Therefore, the total noise of the skylight background and ambient thermal radiation is denoted as background noise. Considering the above factors, the photometry accuracy of star points can be calculated:
（5）
In the formula, N⋆ is the flux of the target star, Ndark is the dark current, Nbkg is the flux of the background, and the units are all e−. The square root of the flux is the noise caused by each factor; Nread is the readout noise; npix is the number of pixels contained in the aperture, where the aperture aper is the diameter, is used to calculate the noise contained in the aperture during photometry.
As shown in Figure 8, the relationship between photometric accuracy and magnitude under a single 20 s image was plotted according to Equation(5), and different noise sources were separated. It can be seen from the figure that the current photometric noise is dominated by the background rather than the detector, indicating that the detector's performance can meet the requirements of astronomical imaging. The background includes the sky background and additional thermal radiation stray light. Currently, there is no K-band sky background measurement at the Lenghu site, but its J and H band sky background measurement results are close to those of world-class sites. Therefore, it is speculated that its K-band sky background is comparable to that of world-class sites, approximately 13 mag/arcsec2(corresponding to our equipment approximately 200 e−/s/pix). In actual observation, the mean value of dark current is taken as the laboratory value 5 e−/s/pix and subtracted from the dark current. At an ambient temperature of 0 °C, the measured total background is approximately 1200 e-/s/pix. At −12 °C, the background decreases significantly to around 700 e-/s/pix. Both results are significantly higher than the sky background of other sites and vary with ambient temperature. Therefore, the background we measured also includes strong thermal radiation in addition to sky light. Meanwhile, the temperature inside the camera's Dewar remained stable(lens group 150 K, chip 75 K), and laboratory tests showed that its own background radiation was less than 100 e−/s/pix. Therefore, it is currently believed that the additional thermal radiation mainly comes from the telescope and its surrounding environment. This paper focuses on detector performance testing, and the testing and analysis of thermal radiation stray light will be the focus of our next work.
Fig.
8
Noise Source Decomposition
3.4 Preliminary scientific observations
After the camera was mounted on the telescope, a series of test observations were conducted from October 18 to November 7, 2025, which also included scientific targets such as comets, supernovae and variable stars.
On November 2, we observed a known ZTF optical variable star(period of 0.0689195 days). Figure 9 shows the light curve, with a photometric accuracy of about 20 mmag. Compared with the optical band, the phase of the K band is delayed and the shape of the light curve is different, with a light variation of about 0.1 mag. A series of time-domain astronomical studies can be carried out by high-precision measurement of the light curve. For example, M-type dwarf stars are important targets in the search for exoplanets. Due to their low effective temperature(1500–3500 K) and concentrated radiation energy distribution in the infrared band, photometric measurements in the K band can obtain a higher signal-to-noise ratio than those in the visible band, making them more sensitive to signals such as transit light variations of exoplanets around these stars. In addition, for standard candle targets with period-luminosity relationships such as classical Cepheid variable stars, the period-luminosity relationship in the near-infrared K band shows significantly lower dispersion: a study of Cepheid variable stars in LMC showed that the period-luminosity relationship in the K band is significantly lower than that in the optical band [4]. Lower dispersion means that time-domain monitoring of Cepheid variable stars in the K-band allows for more accurate inference of distance and galaxy structure information through period-luminosity relations.
Fig.
9
Variable star measurement results (compared with ZTF optical band, RA=77.7209903, DEC=7.2898189, period is 0.0689195 days)
Meanwhile, we have conducted multi-day monitoring of the supernova AT2025annt in both J and K bands, with the K band being used for observation on November 3. After stacking 20 s×65 images, Figure 10 shows the comparison between our obtained images and historical images from UKIRT(aperture photometry result: K= 15.9 mag). Transient sources are one of the main observation targets of the telescope, and we can conduct follow-up observations on transient source alerts issued by space telescopes such as EP and SVOM. Among them, the peak luminosity of Type Ia supernovae is relatively stable and is often used as standard candles to measure cosmological distances. The average absolute magnitude of the radiation peak of Type Ia supernovae in the K band is approximately–18.4 mag [20]. Therefore, for a 30-minute stacking observation strategy(limiting magnitude of 17.5 mag), we can detect Type Ia supernova peaks within a range of 150 Mpc. In addition, many supernovae appear in dusty regions of the host galaxy, and the K band has the advantage of being less affected by extinction. In the framework of multi-messenger astronomy, the electromagnetic counterpart of gravitational wave events(kilonova) is also one of the key scientific targets. Its light variation timescale is shorter, its radiation tends to be more red-end, and its radiation continues to shift towards the infrared over time [22]. Therefore, time-domain follow-up observations in the K-band are more advantageous than those in the optical band. For high-redshift objects, their radiation peaks may shift from the ultraviolet and optical bands to the near-infrared band, making K-band observations an effective means of studying these high-redshift transient sources.
Fig.
10
Images of the supernova AT2025aant: (a) 65 20-second stacked images from an 80 cm telescope;(b) Historical images from UKIRT (the supernova′s location is within the green circle)
4 Conclusion
To meet the needs of K-band astronomical observations, the first K-band mosaic astronomical camera, developed using a domestically produced novel mercury cadmium telluride infrared focal plane array detector, was used to conduct astronomical K-band imaging observations on the 80 cm infrared telescope at Sun Yat-sen University. Test results show that the telescope achieves a single-frame 20-s imaging depth of 15.3 mag, a stacked 30-min depth of 17.5 mag, and a stacked 2 h depth of 18.2 mag in the K-band. Compared with international standards, the single-frame depth reaches the 2 MASS level, and the stacked 2-h depth is close to UHSDR2. These results demonstrate that the telescope can conduct time-domain astronomical research and deep-field imaging of small sky areas at different depths and time resolutions, meeting the diverse scientific objectives of the Chinese astronomical community, and has already conducted several scientific observations. This achievement represents the first scientific-grade K-band imaging observation using domestically produced equipment, providing technical reference for related fields in China and offering important insights for future near-infrared astronomical research. The successful application of the mercury cadmium telluride mosaic camera and detector on the 80 cm infrared telescope at Sun Yat-sen University has sparked considerable interest within the Chinese astronomical community. Many organizations have begun developing or plan to develop infrared observation equipment.
Noise analysis shows that the performance bottleneck in K-band imaging is no longer detector noise, but mainly limited by the sky background. Therefore, applying this detector to larger-aperture telescopes at excellent domestic sites will result in a significant performance improvement. If the same camera is placed on a 2-meter aperture telescope with the same focal length, the expected effective exposure depth for 30 minutes can reach 19 mag, enabling the study of various important targets. For example, follow-up observations can be conducted on transient source alarms(gamma-ray bursts, supernovae, kilonovaes, etc.) generated by space telescopes such as EP and SVOM. In the future, with further improvements in detector performance and array size, as well as the construction of large optical/infrared telescopes in China, China's infrared astronomy is expected to enter a phase of rapid development.