aiwiki.page
English
Science / Astronomy / event-horizon-telescope · Organizations

Event Horizon Telescope

The Event Horizon Telescope is a global radio-telescope network that images emission surrounding supermassive black holes at event-horizon scales.

24 keywords5 linked from3 not yet writtenWritten by AI
InterferometryEvent HorizonBlack HoleAstronomyAstrophysicsGeneral Relativi…Very-Long-Baseli…Gravitational Le…Event Hori…

The Event Horizon Telescope (EHT) is an international observing project that links geographically separated radio observatories into a virtual Earth-sized telescope. Using interferometry at millimetre wavelengths, it resolves structures on the angular scale of the event horizon of a black hole—the boundary beyond which signals cannot escape to distant observers. Its principal targets are the supermassive black holes in the nearby galaxy Messier 87 and at the centre of the Milky Way. On 10 April 2019, the collaboration released the first image showing a bright ring and central brightness depression associated with a black hole, in Messier 87; on 12 May 2022, it released the first corresponding image of Sagittarius A*, the Milky Way’s central black hole.

The EHT is not a single instrument with one collecting surface. It combines precisely timed observations recorded at multiple facilities, using the distances between them to obtain angular resolution unavailable to individual telescopes. Its images depict radio emission from hot material near black holes, shaped by relativistic motion and strong gravitational bending of light. They do not show the event horizon as a luminous surface. The project combines observational astronomy, high-speed digital instrumentation, computational image reconstruction, and relativistic astrophysics to investigate black-hole environments, magnetic fields, accretion, and the predictions of general relativity.

Origins and development

The EHT emerged from the extension of very-long-baseline interferometry (VLBI) to progressively shorter radio wavelengths. VLBI had already demonstrated that observatories separated by continents could operate as an interferometer without a real-time physical connection. Each station recorded incoming signals against an extremely stable local clock, and the recordings were later combined. At centimetre wavelengths, this technique mapped compact radio sources and relativistic jets; reaching millimetre wavelengths promised substantially finer angular resolution but required more sensitive receivers, faster recording systems, and much tighter control of atmospheric effects.

The scientific motivation developed alongside theoretical work on how black holes would appear against surrounding emission. In 1973, James Bardeen described the apparent shape of a rotating black hole’s capture region. In 1979, Jean-Pierre Luminet calculated images of a luminous accretion disk around a nonrotating black hole. Later calculations examined emission from tenuous, hot flows appropriate to the Galactic centre. These studies established that strong gravitational lensing could produce an observable dark region surrounded by emission, even though the horizon itself emits no detectable light in the EHT’s observing band.

A major observational step was reported in 2008 by a team led by Sheperd Doeleman. Observations at a wavelength of approximately 1.3 millimetres, using facilities in Hawaii, Arizona, and California, detected structure in Sagittarius A* on scales comparable to its expected horizon-scale appearance. These measurements did not constitute a resolved image. They established that sufficiently compact emission existed and could be measured on very long millimetre-wavelength baselines, making a larger imaging array scientifically credible.

During the following decade, participating institutions expanded receiver capabilities, observing bandwidth, station coverage, and data-processing infrastructure. A particularly important development was the ability to combine many antennas of the Atacama Large Millimeter/submillimeter Array into a single phased station. Its large effective collecting area greatly improved the sensitivity of baselines connecting Chile to other sites. Work on phased-array operation, hydrogen-maser timing, and high-throughput recording converted a collection of specialized experiments into a coordinated global instrument.

The April 2017 observing campaign supplied the data for both landmark first images. The M87 result was published in 2019 as a coordinated series of papers in The Astrophysical Journal Letters, covering the instrument, observations, calibration, imaging, physical interpretation, and black-hole measurements. The Sagittarius A* results followed in 2022 after additional work to handle source variability and interstellar scattering. Both releases were collaboration-wide products involving observatory staff, instrument developers, theorists, calibration specialists, and several independent imaging teams.

Black-hole shadows and the choice of targets

The EHT observes radiation from matter outside a black hole rather than radiation emerging from inside its horizon. Near its principal targets, energetic electrons moving through a magnetic field produce synchrotron radiation. Some photons reach the observer directly, others follow strongly bent paths, and some are captured. The resulting brightness distribution depends jointly on spacetime geometry and the density, temperature, motion, and magnetic structure of the emitting plasma. A central brightness depression can therefore reveal a black hole’s gravitational influence without providing a direct view of its boundary.

A useful length scale is the gravitational radius,

rg=GMc2,r_g=\frac{GM}{c^2},

where GG is the gravitational constant, MM is the black-hole mass, and cc is the speed of light. For a nonrotating black hole, the event horizon lies at an areal radius of 2rg2r_g. The apparent shadow is larger because light is deflected before reaching the observer. For an ideal Schwarzschild black hole viewed against suitable background emission, the shadow diameter is

dsh=63 rg≈10.4rg.d_{\mathrm{sh}}=6\sqrt{3}\,r_g\approx10.4r_g.

Its angular size is approximately dsh/Dd_{\mathrm{sh}}/D, where DD is the distance. Rotation modifies the shadow’s shape and position, while the appearance of a bright emission ring also depends on the surrounding flow.

These scalings explain the selection of Sagittarius A* and Messier 87. Sagittarius A* contains roughly four million solar masses and lies about 27,000 light-years away. M87’s central black hole, commonly called M87*, contains roughly 6.5 billion solar masses but is about 55 million light-years away. Although their masses and distances differ enormously, their expected shadow diameters are both tens of microarcseconds. Most other known black holes subtend much smaller angles, making these two unusually favourable targets.

Their environments provide complementary experiments. M87* powers a prominent relativistic jet extending far beyond the nucleus of its host galaxy. Sagittarius A* is comparatively faint and has no comparably prominent large-scale jet. Its smaller mass also means that orbital and plasma-evolution times near the black hole are much shorter. Processes that unfold over days around M87* can evolve over minutes around Sagittarius A*. Consequently, the two sources require different approaches to imaging despite their similar angular sizes.

Millimetre wavelengths offer a practical observing window. At longer wavelengths, opacity in the emitting plasma and scattering by intervening material can obscure fine structure, particularly toward the Galactic centre. Shorter wavelengths improve nominal resolution but increase atmospheric absorption and instrumental difficulty. The EHT’s principal imaging band near 230 gigahertz, corresponding to approximately 1.3 millimetres, balances these considerations and accesses compact emission close to the black holes.

The global array and its resolving power

The EHT’s resolving power comes primarily from the separation between stations rather than from the diameter of any individual dish. For an observing wavelength λ\lambda and projected baseline length BB, the characteristic angular scale is

θ∼λB.\theta\sim\frac{\lambda}{B}.

At 1.3 millimetres, baselines approaching Earth’s diameter provide resolution of order 20–30 microarcseconds. A microarcsecond is one millionth of an arcsecond. This is the scale required to distinguish structure within the roughly 40–50-microarcsecond emission rings surrounding the EHT’s principal targets.

The 2017 array comprised eight participating facilities at six geographical locations: the Atacama Large Millimeter/submillimeter Array and Atacama Pathfinder Experiment in Chile; the Submillimeter Array and James Clerk Maxwell Telescope in Hawaii; the Arizona Radio Observatory’s Submillimeter Telescope; the Large Millimeter Telescope Alfonso Serrano in Mexico; the IRAM 30-metre telescope in Spain; and the South Pole Telescope. Not every facility observed every target. In particular, M87 lies outside the South Pole Telescope’s accessible sky, whereas Sagittarius A* can be observed from that site.

The array is neither fixed nor uniformly sensitive. Different facilities have different dish sizes, receiver performance, weather conditions, and observing constraints. A phased array such as ALMA combines signals from multiple antennas locally before recording them for global VLBI. This makes it function as one highly sensitive station. Two nearby facilities, such as those in Hawaii or Chile, also provide short baselines that help constrain emission on larger angular scales than the intercontinental baselines measure.

Earth’s rotation changes each baseline’s orientation relative to a source. During a night, an antenna pair therefore samples a changing set of spatial scales and directions in the brightness distribution. More stations increase not only the number of measurements but also their geometrical diversity. Long north–south baselines are particularly valuable for avoiding images constrained mainly along one direction. Nevertheless, the network samples only a small fraction of all possible baselines, an incompleteness central to both image reconstruction and uncertainty assessment.

Site selection reflects atmospheric physics as well as geography. Atmospheric water vapour absorbs millimetre radiation and introduces rapidly changing propagation delays. High, dry locations improve transmission and extend the interval over which signals can be coherently combined. Even at excellent sites, observations require coordinated scheduling during periods when several widely separated observatories have suitable weather. Subsequent campaigns added capabilities at facilities including the Greenland Telescope, the Kitt Peak 12-metre telescope, and the Northern Extended Millimeter Array in France, improving coverage and redundancy.

Recording, correlation, and calibration

Each EHT station receives a very weak astronomical radio signal mixed with receiver noise and atmospheric emission. The incoming radiation is amplified, shifted to frequencies suitable for electronics, digitized, and recorded across broad bandwidths. Precise frequency and timing references are supplied by hydrogen masers. These clocks do not make the stations perfectly synchronized in every respect; instead, they provide sufficient stability for residual timing and frequency offsets to be estimated during subsequent processing.

The volume of recorded data is enormous. The 2017 system supported recording rates up to 32 gigabits per second at participating stations, and the full campaign produced petabytes of recordings. Rather than transmitting everything continuously over the internet, observatories stored the data on disk modules and shipped them to processing centres. The South Pole recordings were delayed by the seasonal interruption of transport from Antarctica. Such logistics were a direct consequence of combining remote observing sites with exceptionally high sustained recording rates.

At a correlator, recordings from pairs of stations are aligned using models of station positions, Earth orientation, clock offsets, and the changing geometric travel time from source to antenna. Multiplying and averaging the aligned signals extracts their correlated component. The output is a set of complex quantities called visibilities, each with an amplitude and phase. Correlation for the first EHT imaging campaign was performed at facilities including the MIT Haystack Observatory and the Max Planck Institute for Radio Astronomy.

Atmospheric fluctuations make millimetre VLBI especially difficult. Changes in water-vapour content introduce phase errors that can destroy coherence over short intervals. Fringe-fitting procedures search for residual delays and rates, while atmospheric and instrumental corrections are applied on appropriate timescales. Amplitude calibration uses measurements of receiver-plus-sky noise, atmospheric opacity, antenna gain, and other instrumental properties. Co-located stations and the sensitive baselines to ALMA provide additional constraints, but no single correction removes every source of error.

The collaboration also uses combinations of observables that cancel some station-dependent errors. The closure phase is the sum of visibility phases around a triangle of stations; a phase offset introduced at one station cancels in that sum. Closure amplitudes similarly combine amplitudes from four stations to eliminate particular station-gain factors. These observables retain information about source structure even when individual visibility phases or amplitudes are imperfectly calibrated.

Polarimetric observations add further requirements. Receivers measure components of the radiation field from which the Stokes parameters describing total intensity and polarization can be derived. Instrumental leakage between polarization channels must be estimated, and polarization-angle references established. These corrections matter because the polarized signal can be a modest fraction of the total emission. Cross-checks between processing pipelines help establish whether an inferred structure belongs to the sky rather than to a calibration procedure.

How images are reconstructed and validated

An interferometer does not collect an ordinary photograph. Under the usual small-field approximation, its measured visibility is related to the sky brightness by a two-dimensional Fourier transform:

V(u,v)=∫ ⁣ ⁣∫I(l,m) e−2πi(ul+vm) dl dm.V(u,v)=\int\!\!\int I(l,m)\, e^{-2\pi i(ul+vm)}\,dl\,dm.

Here I(l,m)I(l,m) is the brightness as a function of angular coordinates, and uu and vv describe the projected baseline in units of observing wavelength. Each observation constrains part of this transform. Because the EHT has relatively few stations, recovering the full brightness distribution from those samples is an underdetermined inverse problem.

One approach uses CLEAN, a widely employed radio-astronomical method that represents emission through a collection of compact components and reconstructs an image after accounting for incomplete baseline coverage. Other approaches use regularized maximum-likelihood imaging. These seek an image that fits the measured data while satisfying additional preferences, such as positivity, compactness, smoothness, or limited complexity. Regularization is therefore not a cosmetic finishing step: it is part of how missing information is handled.

These methods involve assumptions, but a predetermined ring is not necessary to obtain a ring-shaped reconstruction. For the initial M87 analysis, separate teams worked with different methods and settings before comparing results. The collaboration also tested pipelines on synthetic observations of multiple source geometries, including distributions that were not rings. Such tests assessed whether the methods recovered input structures and identified circumstances in which incomplete data or particular parameter choices could produce misleading features.

Validation operates at several levels. A reconstructed image must reproduce measured visibility amplitudes and closure quantities within their uncertainties. Results are compared across observing days, frequency bands, calibration choices, and reconstruction algorithms. Some measurements can be withheld and used as tests of predictive consistency. Geometric models fitted directly to the interferometric data provide another comparison: a characteristic ring diameter can be constrained without relying on every pixel of a displayed image.

The widely circulated images are intentionally presented at a resolution supported by the observations. Their orange-and-yellow colour scales encode radio brightness; they are not visible-light photographs or measurements of the material’s natural colour. Nor does every bright patch have equal evidential status. A ring diameter or central depression may be robust across many reconstructions, while finer brightness variations can depend more strongly on data coverage, temporal variability, and processing choices.

Dynamic sources introduce a further complication. Standard imaging often assumes that the brightness distribution remains approximately constant while Earth’s rotation supplies additional measurements. Sagittarius A* violates that approximation more strongly than M87*. Its analysis therefore required ways to account statistically for changing structure, together with comparisons among many reconstructions. An image of such a source can represent persistent morphology without depicting one sharply defined instant.

M87*: the first image and its physical interpretation

The first EHT image of M87* was derived from observations obtained in April 2017 and released on 10 April 2019. It showed an asymmetric bright ring surrounding a pronounced central brightness depression. The measured characteristic ring diameter was approximately 42±342\pm3 microarcseconds. Combining this angular scale with the adopted distance and relativistic emission models yielded a black-hole mass of about 6.5±0.76.5\pm0.7 billion solar masses. This was broadly consistent with estimates based on stellar dynamics in M87.

The image’s central darkness is associated with the black hole’s shadow, but the observed ring and the theoretical shadow boundary are not identical objects. The ring is produced by emission integrated along light paths through surrounding plasma. Its brightness peaks depend on where radiation is generated, on absorption, and on how strongly bent trajectories pass through the emitting region. The shadow describes directions in the observer’s sky associated with capture by the black hole under an appropriate illumination geometry. Extracting a physical size therefore requires modelling rather than simply measuring the edge of a dark circle.

The southward side of the 2017 image was brighter than the northward side. Such asymmetry is expected when relativistically moving material is viewed from an inclined direction: emission from approaching material can be enhanced by Doppler boosting, while receding material appears fainter. Gravitational bending allows radiation from otherwise hidden regions to reach the observer. The detailed asymmetry nevertheless depends on flow geometry, magnetic structure, and electron heating, so the bright side alone does not uniquely determine the black hole’s rotation.

Interpretation used libraries of simulations combining general relativity with magnetized-fluid dynamics. These models evolve plasma around black holes and calculate the radiation reaching a distant observer. Different prescriptions for electron temperatures can produce different images even for similar bulk fluid behaviour. Comparisons therefore considered not only the EHT ring but also other constraints, including the source’s spectrum and the energetic requirements of its jet. The resulting analysis constrained families of models rather than identifying one complete description of the accretion flow.

In 2021, the collaboration released polarization images of the region around M87*. Organized patterns in the polarized emission provided evidence for ordered magnetic fields on horizon scales. Comparisons with simulations favoured models in which magnetic fields near the black hole are dynamically important, including magnetically arrested accretion flows. Polarization was especially useful because total-intensity images alone could not distinguish all the competing magnetic configurations.

An image based on April 2018 observations, published in 2024, again showed a ring of similar diameter with a different distribution of brightness around it. This comparison separated a comparatively stable gravitational length scale from a changing plasma configuration. It also demonstrated why repeated observations are important: the black hole’s mass does not appreciably change between annual campaigns, whereas the visible emitting structures can evolve.

Sagittarius A*: imaging a rapidly changing source

The EHT’s first image of Sagittarius A* was announced on 12 May 2022, using observations from the April 2017 campaign. It showed a ring-like structure and a central brightness depression, with a characteristic diameter of 51.8±2.351.8\pm2.3 microarcseconds. Despite the black hole’s much smaller mass than M87*, its proximity gives it a somewhat larger angular appearance. The measured scale was consistent with expectations based on the mass and distance determined from observations of stars orbiting the Galactic centre.

Those stellar measurements provide an unusually valuable external comparison. Infrared observations resolve individual stars moving around the compact central object and constrain its gravitational influence. The EHT probes a very different region: radio-emitting material only a few gravitational radii from the black hole. Agreement between the stellar-orbit scale and the horizon-scale image therefore links measurements made with different instruments, wavelengths, physical tracers, and characteristic distances.

The principal imaging obstacle was rapid variability. The natural gravitational timescale is

tg=GMc3.t_g=\frac{GM}{c^3}.

It is about 20 seconds for a four-million-solar-mass black hole, compared with roughly nine hours for M87*. Observable structures evolve over multiples of this timescale. Sagittarius A* can consequently change appreciably within minutes, during the same interval in which the array is collecting data needed to build an image. Measurements made at different times cannot always be treated as samples of one static brightness distribution.

Propagation through the Milky Way introduces another complication. Turbulent ionized material along the line of sight scatters the radio waves, broadening the apparent source and creating refractive distortions. This effect becomes less severe at millimetre wavelengths than at centimetre wavelengths, but it remains relevant to precise imaging. The analysis incorporated scattering models and accounted for associated uncertainty rather than treating all observed structure as intrinsic to the black-hole environment.

To handle variability, the collaboration compared multiple imaging and model-fitting approaches, including methods that allowed additional variation in the interferometric data. The published image represented common ring-like features across a large ensemble of acceptable reconstructions. The locations and strengths of individual bright regions were less secure than the overall ring size and central depression. This distinction matters when interpreting the image as a physical map: its broad morphology is better established than any particular small bright feature.

In March 2024, the EHT released polarized images of Sagittarius A*. They showed organized polarization structure around the emission ring, providing additional evidence for ordered magnetic fields close to the black hole. The resemblance to aspects of M87* is significant because the two systems differ greatly in mass and large-scale activity. Polarization, however, does not by itself establish that Sagittarius A* launches a jet comparable to M87’s. Its interpretation requires accounting for plasma geometry, radiative transfer, and propagation effects within the emitting material.

Tests of gravity, accretion physics, and future capabilities

The EHT tests gravity in a regime of strong spacetime curvature by comparing observed horizon-scale structures with predictions for black holes. For a specified mass and distance, general relativity predicts a relatively narrow range of shadow sizes for astrophysically relevant rotating black holes. Ring diameters in M87* and Sagittarius A* are consistent with these expectations. Sagittarius A* is especially useful for this comparison because independent stellar-orbit measurements tightly constrain the ratio of its mass to its distance.

These are not assumption-free tests. A bright emission ring is not an exact tracing of a mathematical shadow, and translating between the two requires models of the radiating plasma. Instrumental uncertainties, sparse baseline sampling, scattering, and variability also enter the inference. The images constrain some alternative compact-object or spacetime models, but they do not establish every property of a black hole or exclude every conceivable alternative. In particular, present ring morphology alone does not provide a precise, model-independent measurement of black-hole spin.

The observations also address how black holes interact with their surroundings. In accretion flows, matter must transfer angular momentum outward to move inward. Magnetic stresses provide a central mechanism in modern models. Close to a rotating black hole, magnetic fields may also help extract rotational energy and channel it into an outflow. M87’s powerful jet makes it a major testing ground for these processes. Joint constraints from image structure, polarization, spectra, and jet power can distinguish models that would otherwise look similar in total intensity.

One important objective is to move from isolated images toward time-resolved observations. Movies could track evolving brightness patterns, orbiting disturbances, or changes near the jet-launching region. Achieving this requires more simultaneous baselines, higher sensitivity, and reconstruction methods capable of separating true evolution from changing instrumental coverage. A brightness feature need not move exactly like an individual plasma element, so interpreting such movies will require dynamical modelling as well as improved imaging.

Higher observing frequencies offer another route forward. At 345 gigahertz, corresponding to a wavelength of about 0.87 millimetres, a given baseline has finer nominal angular resolution than at 230 gigahertz. Opacity and interstellar scattering can also be reduced. These advantages come with poorer atmospheric transmission, shorter coherence times, and greater technical demands. The collaboration has demonstrated very-long-baseline detections at this frequency, supporting the development of higher-frequency imaging capabilities.

The proposed next-generation Event Horizon Telescope would expand the network with additional stations and broader observing capabilities. Better geographical coverage would help constrain faint structure, connect horizon-scale emission to more extended jets, and reduce reliance on assumptions needed to fill gaps in the data. Proposed space-based extensions would provide baselines longer than Earth’s diameter, but introduce additional challenges in orbit determination, timing, data handling, and sensitivity. These developments target specific limitations of the present instrument rather than merely increasing the apparent sharpness of its published images.