Terahertz Intensity Mapper (TIM)

• TIM is a balloon-borne far-infrared instrument that uses long-slit spectrometry and horn-fed KID arrays to perform [C II] line-intensity mapping.
• It maintains a precise 50 μm detector-horn gap and employs μT-level magnetic shielding to ensure high detector yield and minimal cosmic-ray interference.
• TIM cross-correlates its three-dimensional spectral-imaging cubes with galaxy surveys to robustly constrain cosmic star formation history and large-scale structure over z ≈ 0.5–1.7.

Searching arXiv for relevant TIM papers to ground the article in current literature. to=arxiv_search.query ฝ่ายขายละคร 微信天天中彩票 ฝ่ายขายออนไลน์ code 中国福利彩票天天 "query": "Terahertz Intensity Mapper TIM [CII] KID focal plane cosmic ray susceptibility Euclid", "max_results": 10} The Terahertz Intensity Mapper (TIM) is a balloon-borne far-infrared line-intensity mapping experiment designed to measure the aggregate [C II] 157.7 μm fine-structure emission from dust-obscured star-forming galaxies during the epoch around the peak and decline of cosmic star formation. It does so with two long-slit grating spectrometers of resolving power $R \approx 250$ spanning 240–420 μm, coupled to horn-fed aluminum kinetic inductance detector (KID) focal planes operated at about 250 mK. In TIM’s intended science mode, the slit and scan strategy produce three-dimensional spectral-imaging data cubes, enabling measurements of both mean [C II] intensity and its clustering, and thereby constraining the star-formation history and large-scale structure over approximately $0.5 \lesssim z \lesssim 1.7$ (Vieira et al., 2020, Liu et al., 2022).

1. Scientific target and line-intensity-mapping role

TIM is optimized for the redshifted [C II] line, whose rest wavelength is $\lambda_{\rm rest}=157.7\,\mu{\rm m}$. Using

$z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$

TIM’s 240–420 μm band maps [C II] over approximately $z \approx 0.52$ to $1.66$, often summarized as $z \approx 0.5$–$1.7$. This interval is central to the rise and fall of the cosmic star formation rate density. The instrument is motivated by the fact that dust absorbs starlight and re-emits in the far-infrared, while [C II] is a bright, dust-immune tracer of star formation and interstellar-medium conditions, so [C II] intensity mapping can recover the three-dimensional structure of star formation without the dust-selection biases of rest-frame UV measurements (Liu et al., 2022).

The experiment was framed from the outset as an integral-field far-infrared spectrometer intended to open a new observational window on galaxy evolution and to demonstrate the technical basis for future space far-infrared spectroscopy. In addition to [C II], TIM’s bandpass also admits [N II] 205 μm, [O III] 88 μm, [O I] 63 μm, [C I] 370 μm, and multiple mid-$J$ CO lines over different redshift intervals, supporting metallicity, ionized-gas, and feedback diagnostics beyond the core [C II] program (Vieira et al., 2020).

A recurrent misunderstanding is to treat TIM primarily as a source-finding telescope for individually resolved galaxies. The published science case instead emphasizes line-intensity mapping: measuring aggregate line emission, the [C II] power spectrum, [C II]$\times$[N II] cross-power, and stacked or cross-correlated observables, while still expecting roughly 100 blind [C II]-bright galaxy detections in survey mode (Vieira et al., 2020). This distinction is central to the instrument concept.

2. Platform, optics, and spectrometer architecture

TIM is a high-altitude balloon instrument intended for Antarctic long-duration balloon flight at about 37 km altitude. The 2020 mission overview describes a 2.0-meter on-axis segmented carbon-fiber-reinforced polymer primary mirror with gold-metallized reflecting surfaces, a carbon-fiber secondary, and a low-emissivity optical train designed for near-atmosphere-limited performance. The telescope is reimaged from $0.5 \lesssim z \lesssim 1.7$0 to $0.5 \lesssim z \lesssim 1.7$1 by a cold Offner-style relay including a cold pupil/Lyot stop, and two azimuth-aligned entrance slits feed two separate spectrometer modules (Vieira et al., 2020).

TIM uses two independent long-slit grating spectrometers. The short-wavelength channel covers 240–317 μm and the long-wavelength channel covers 317–420 μm, each at resolving power $0.5 \lesssim z \lesssim 1.7$2. Long slits provide wide instantaneous spatial coverage in one dimension, while detector arrays sample the dispersed spectrum across the slit; scanning the payload sweeps the slit across the sky to build a spectral-spatial cube. In this architecture, $0.5 \lesssim z \lesssim 1.7$3 is chosen as a compromise between spectral resolution and mapping speed, with the 2020 overview quoting a velocity resolution of about 830 km s$0.5 \lesssim z \lesssim 1.7$4 and an instantaneous fractional bandwidth of about 32% per module (Vieira et al., 2020, Liu et al., 2022).

The mapping program originally emphasized two deep fields with substantial ancillary data, GOODS-S and the South Pole Telescope Deep Field, with net science time of about 280 hours in a nominal two-week flight and approximately 10% of time allocated to calibration and pointing. The slit-scan strategy, combined with the balloon platform’s access to an atmospheric window inaccessible from the ground, is what allows large-area, wide-bandwidth far-infrared spectral mapping at these wavelengths (Vieira et al., 2020).

Later subsystem papers present a more detailed focal-plane segmentation than the earlier mission overview. The 2020 overview describes 3600 KIDs in total (Vieira et al., 2020), whereas later focal-plane papers describe each spectrometer’s focal plane as four quadrant arrays with approximately 900 horn-coupled aluminum KIDs per quadrant, corresponding to about 3600 pixels per focal plane and about 7200 across both bands (Liu et al., 2024). This suggests an instrument-development evolution in the detector implementation rather than a change in the core spectroscopic concept.

3. Focal-plane layout, KID technology, and package design

Each spectrometer’s focal plane is divided into four wafer-sized subarrays, or quadrants. In the focal-plane assembly design paper, the full focal plane covers 64 spectral by 51 spatial pixels in each band, with pixels hex-packed on 2.3 mm pitch and optically coupled to direct-machined aluminum feedhorns. The KIDs are lumped-element resonators fabricated from 30 nm aluminum on intrinsic silicon wafers of thickness 675 μm, with each resonator consisting of an inductive absorber plus an interdigitated capacitor and coupled capacitively to a microstrip feedline (Liu et al., 2022).

A central engineering requirement is the strict maintenance of a 50 μm air gap between the KID wafer and the horn block. Precision-machined 50 μm bosses on the horn block set this gap, while spring-loaded pins press the silicon wafer against the bosses and provide compliance against differential thermal contraction. This geometry is not a minor packaging detail: HFSS modeling shows that increasing the gap from 40 to 80 μm increases radiation leakage by approximately an order of magnitude, significantly elevating optical crosstalk. The horn block surface facing the silicon also serves as the dominant ground plane over the optically active region, so the same 50 μm gap sets the readout microstrip geometry and helps control the electromagnetic environment (Liu et al., 2022).

The absorber uses a “chain-link” inductive pattern. In the published design studies, 52 chain-link cells on a single meander yield simulated $0.5 \lesssim z \lesssim 1.7$5 optical absorption efficiency in both linear polarizations, and the focal-plane paper summarizes the design as exceeding $0.5 \lesssim z \lesssim 1.7$6 in-band absorption in both polarizations. The horn-grounded microstrip architecture is correspondingly tuned to a matched impedance of $0.5 \lesssim z \lesssim 1.7$7: under the horn block and 50 μm gap, the matched line width is 91 μm; in the bonding region, where the horn block is absent and the aluminum backside becomes the dominant ground, the matched line width is 444 μm; and a 2 mm linear taper connects the two regimes with modeled added loss below 0.1 dB over a 23 mm reference path (Liu et al., 2022).

The prototype quadrant package preserves the same electromagnetic strategy in a mechanically compact, flight-like aluminum enclosure. In a dark test at 213 mK, a vector network analyzer sweep from 360 to 800 MHz identified 823 resonances out of 864 designed resonators, corresponding to a 95.25% electrical yield, with resonant frequencies within the design range and coupling-limited behavior near the design value $0.5 \lesssim z \lesssim 1.7$8 (Liu et al., 2022). A later prototype test at about 250 mK again reported 823 resonance features for an 864-pixel long-wavelength quadrant, or greater than 90% yield, before magnetic effects were isolated as the main cause of degraded internal quality factors in some cooldowns (Liu et al., 2022).

4. Resonator physics and magnetic-field control

TIM’s KIDs are superconducting microresonators, and their resonant transmission is modeled as

$0.5 \lesssim z \lesssim 1.7$9

with

$\lambda_{\rm rest}=157.7\,\mu{\rm m}$0

while the resonance frequency is

$\lambda_{\rm rest}=157.7\,\mu{\rm m}$1

Here $\lambda_{\rm rest}=157.7\,\mu{\rm m}$2 is the kinetic inductance, so magnetic fields can shift $\lambda_{\rm rest}=157.7\,\mu{\rm m}$3 by altering $\lambda_{\rm rest}=157.7\,\mu{\rm m}$4 and can degrade the internal quality factor $\lambda_{\rm rest}=157.7\,\mu{\rm m}$5 through trapped vortices and related dissipation (Liu et al., 2022).

This susceptibility proved quantitatively important in prototype tests. Smartphone magnetometer measurements indicated a local Earth field of about $\lambda_{\rm rest}=157.7\,\mu{\rm m}$6 at the test site. In an initial cooldown without magnetic shielding, many resonances showed collisions and shallow notches indicative of degraded $\lambda_{\rm rest}=157.7\,\mu{\rm m}$7. With a cylindrical mu-metal shield present, the $\lambda_{\rm rest}=157.7\,\mu{\rm m}$8 distribution centered at $\lambda_{\rm rest}=157.7\,\mu{\rm m}$9 while $z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$0 remained near the design value of approximately $z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$1. Without shielding, ambient Earth field decreased $z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$2 by a factor of about 2–5, producing bimodal peaks near approximately $z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$3 and $z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$4, while $z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$5 was largely unchanged apart from a small shift from about $z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$6 to $z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$7 (Liu et al., 2022).

The published mitigation strategy uses both passive shielding and active nulling. A Helmholtz coil was deployed around the cryostat, with field

$z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$8

and the coil current was tuned empirically to cancel the net field seen by the wafer during cooldown. Because the thin-film aluminum KIDs transition before the surrounding aluminum package, the devices are especially susceptible to flux trapping if residual field remains during the superconducting transition. When the net field $z=\lambda_{\rm obs}/\lambda_{\rm rest}-1,$9 was tuned to within a few μT of zero, the median $z \approx 0.52$0 rose toward about $z \approx 0.52$1, and resonance line profiles approached the shielded case; performance was best for $z \approx 0.52$2 (Liu et al., 2022).

The resulting instrument-level requirement is explicit: TIM adopts a magnetic-field requirement at the detectors of $z \approx 0.52$3 during the superconducting transition and operation. The focal-plane studies therefore recommend high-$z \approx 0.52$4 shielding near the focal plane unit, optional complementary superconducting shielding at cold stages if cooled in already-low fields, degaussing of high-$z \approx 0.52$5 shields, avoidance of ferromagnetic components near the detectors, and cooldown procedures that minimize the component of Earth’s field normal to the wafer (Liu et al., 2022). In TIM’s dense multiplexing environment, this is not merely a materials issue; higher $z \approx 0.52$6 directly reduces resonance collisions and preserves usable readout bandwidth.

5. Dense multiplexing, detector identification, and cosmic-ray susceptibility

TIM’s detector arrays operate in a densely multiplexed regime. One 864-pixel array was measured with 791 resonators observed in a 0.5 GHz band, underscoring the calibration difficulty created by near-neighbor resonances and collisions. To manage this, TIM uses a unit-cell frequency plan in which each 16-pixel unit cell contains resonances that are widely separated within the cell by varying the interdigitated-capacitor tine count, while successive unit cells are shifted incrementally by tine-length adjustments of $z \approx 0.52$7, filling the 500–1000 MHz readout band with about 864 designed resonances (Liu et al., 2024).

The calibration model near a resonance is written as

$z \approx 0.52$8

and TIM’s improved routine narrows the fit window around each nominal resonance and imposes a collision threshold

$z \approx 0.52$9

approximately 50 kHz near mid-band, matched to typical $1.66$0. Using this approach, 629 of the 791 observed resonances were properly calibrated, up from 248 with a conventional widely separated-resonator approach, a gain of about 150%. For the recovered detectors, the average white fractional-frequency-noise power spectral density at 1 Hz was approximately $1.66$1, consistent with single-pixel results (Liu et al., 2024).

A cryogenic LED mapping scheme is used to connect resonant frequencies to physical pixel positions. An LED board mounted above the horn array illuminates one 16-pixel unit cell at a time, exploiting the designed frequency ordering within each cell. Sequential LED biasing identified 788 of 791 resonators, or 99.6%, and even revealed seven resonances that were fully merged with others when unilluminated (Liu et al., 2024). This identification step is essential for array-level diagnostics, including the study of high-energy particle events.

Cosmic-ray susceptibility was quantified directly. In 5 minutes of time-ordered data from the properly calibrated 629-KID subset, TIM recorded 434 unique events over an effective KID-populated area of about 41.23 cm$1.66$2, corresponding to a lower-limit event flux

$1.66$3

About 66% of events affected a single pixel, and about 33% affected fewer than 5 KIDs; the typical affected area was about 0.66 cm$1.66$4, corresponding to a radius of roughly two pixel pitches. The transient signature was a sharp excursion followed by an exponential decay with time constant about 100 μs at 250 mK, consistent with quasiparticle recombination after ballistic-phonon production in the substrate (Liu et al., 2024).

TIM’s measured laboratory data-loss fraction from these events was

$1.66$5

with projected in-flight values of

$1.66$6

The design features credited with this robustness are a 100 nm aluminum backside phonon-thermalization layer, a 25 μm thick and 1.3 mm diameter backshort membrane that constrains phonon transport, and front/back annular phonon-absorption structures including choke rings. These results imply that, even in the stratospheric environment, cosmic-ray-induced dead time should remain well below 1% and therefore should not materially compromise TIM’s observing efficiency (Liu et al., 2024).

6. Forecast observables, Euclid cross-correlation, and systematic limitations

TIM’s science case has increasingly emphasized cross-correlation as the most robust route to extracting [C II] information in the presence of bright continuum foregrounds and line interlopers. A 2026 forecast studies cross-correlation between TIM [C II] maps and Euclid’s spectroscopic galaxy survey in the Fornax Euclid Deep Field. The redshift range is split into four bins centered at approximately $1.66$7, with the short-wavelength array covering 0.52–1.0 and the long-wavelength array covering 1.0–1.67 (Bracks et al., 27 Feb 2026).

The fiducial cross-power model is

$1.66$8

with the first term representing large-scale clustering and the second the cross-shot component. In the forecast, the [C II] bias inferred from SimIM is approximately $1.66$9 across the four bins, and the model sets $z \approx 0.5$0 for simplicity. The forecast also includes beam and spectral attenuation through

$z \approx 0.5$1

and cross-spectrum variance using the Villaescusa-Navarro et al. formalism (Bracks et al., 27 Feb 2026).

For the instrument model used in that forecast, TIM has a 2.0 m telescope, $z \approx 0.5$2, beam FWHM of approximately 37 arcsec in the short-wavelength array and 48 arcsec in the long-wavelength array, detector array format 51 × 64 per module, detector yield of about 85%, total transmission of about 8%, detector focal-plane temperature of about 250 mK, and nominal integration time of about 200 hours. The baseline survey area used in the sensitivity curves is $z \approx 0.5$3, and the forecast quotes median noise-equivalent intensity values per module of $z \approx 0.5$4 and $z \approx 0.5$5 Jy sr$z \approx 0.5$6 for the short- and long-wavelength arrays, respectively (Bracks et al., 27 Feb 2026).

The principal forecasted result is that TIM cross-correlated with the Euclid Fornax deep field should detect the [C II]–galaxy cross-power spectrum at a median redshift of 1.1 with at least $z \approx 0.5$7 confidence, and should detect the Poisson component of the cross-power spectrum, or cross-shot noise, at at least $z \approx 0.5$8 in four bins spanning $z \approx 0.5$9 over $1.7$0 (Bracks et al., 27 Feb 2026). The mean [C II] specific intensity can then be estimated via

$1.7$1

In the published interpretation, this would constrain the mean [C II] intensity over roughly half of cosmic history and quantify the fraction of TIM’s [C II] background traceable to Euclid-selected galaxies (Bracks et al., 27 Feb 2026).

The same forecast also makes explicit the dominant systematic challenge: far-infrared continuum is expected to be at least two orders of magnitude brighter than the [C II] line, and line interlopers may contaminate the maps. The argument for cross-correlation is therefore methodological rather than incidental. Because continuum and unrelated interlopers are not shared between TIM’s [C II] cube and Euclid’s spectroscopic galaxy overdensity field, the cross-spectrum is comparatively robust to those contaminants, even when the TIM auto-spectrum is foreground-limited (Bracks et al., 27 Feb 2026). This is a central clarification of TIM’s mature science strategy.

7. Development trajectory and broader significance

TIM occupies a specific niche in far-infrared instrumentation. It is designed to demonstrate background-limited, integral-field far-infrared spectroscopy at THz frequencies from a suborbital platform, combining a low-emissivity 2 m telescope, large-format horn-coupled KID arrays, wide-band spectrometers, and multiplexed readout. The 2020 overview places this explicitly in the context of de-risking future observatories such as Origins Space Telescope or a Probe-class mission, while the focal-plane papers show how apparently localized hardware details—air-gap control, ground-plane geometry, magnetic shielding, and phonon management—are decisive for system performance (Vieira et al., 2020, Liu et al., 2022).

The published subsystem results indicate that TIM’s engineering risks are concentrated in precisely those areas most relevant to a balloon-borne, large-format, cryogenic spectrometer: maintaining a 50 μm array–horn gap through thermal cycles, preserving high $1.7$2 under geomagnetic conditions, calibrating resonators in a densely packed 0.5 GHz band, and limiting data loss from high-energy events. The reported prototype results—greater than 90% or 95.25% resonance yield depending on configuration, shielded $1.7$3 at or above $1.7$4, recovery of magnetic degradation through μT-level field control, 99.6% LED-based pixel identification, and projected in-flight cosmic-ray dead fraction of about 0.165% at worst—show that TIM’s focal-plane implementation has been developed with quantitative margins rather than qualitative plausibility alone (Liu et al., 2022, Liu et al., 2024).

In that sense, TIM is best understood not as a generic far-infrared spectrometer, but as a line-intensity-mapping system whose science return depends directly on detector-package electromagnetics, multiplexing fidelity, and cross-correlation methodology. Its instrument papers and forecast studies consistently connect those engineering choices to a specific measurement program: [C II] intensity mapping over $1.7$5, recovery of the mean [C II] history and clustering signal, and robust joint analyses with external galaxy surveys such as Euclid (Vieira et al., 2020, Bracks et al., 27 Feb 2026).