Why is a dark cloud interesting to a telescope?
In a visible-light photograph, a dust cloud can look like a black hole between the stars. Yet its material has not disappeared: dust has blocked some light, while cold gas and the dust itself remain. To investigate them, we need to choose radiation they produce and a receiver that can detect it.
Millimeter and submillimeter telescopes study molecular lines, thermal emission from cold dust and the microwave background. A dry, high site reduces atmospheric interference, an accurate antenna collects the wave without excessive phase errors, and a cold receiver contributes less noise of its own. Each solves a different part of the problem.
We will follow the path from cloud to data using ALMA and the space instrument Planck/HFI. From the radio-telescope article, we need the idea that an antenna directs a wave into a receiver. From the infrared-telescope article, we need the idea that a warm instrument emits radiation too. The main thread works without the optional mathematical branches.
From excess radio noise to molecular maps
In their 1965 paper, Arno Penzias and Robert Wilson described an excess antenna temperature at 4.08 GHz, a wavelength of about 7.35 cm. They carefully accounted for atmospheric and antenna contributions; those known causes did not explain the remainder. This was a centimeter measurement, rather than a discovery at millimeter wavelengths. Its cosmological interpretation developed alongside an accompanying theoretical paper. Penzias & Wilson, 1965.
In 1970, Robert Wilson, Keith Jefferts and Penzias reported an interstellar line of CO, carbon monoxide, towards the Orion Nebula. Around 115 GHz, it lay in the millimeter range. Observers could search for it in different directions and compare its shifts: the cloud became a target for spectroscopic mapping. The NRAO 36-foot telescope’s receiver then used a diode mixer; it was not a modern ALMA SIS receiver. Wilson, Jefferts & Penzias, 1970.
Later developments provided two different capabilities. The space instrument COBE/FIRAS measured a microwave-background spectrum very close to that of a blackbody at about 2.725 K. Planck, launched in 2009, mapped the sky in detail across several bands. Ground-based ALMA, which began early science in 2011 and was inaugurated in 2013, combined accurate antennas, sensitive receivers and interferometry. These are different tools for different questions. NASA: COBE, Planck archive, ALMA Technical Handbook, introduction.
Where is the boundary between radio and infrared?
Here we adopt millimeter wavelengths of 1–10 mm and submillimeter wavelengths of 0.1–1 mm. Approximate frequencies are 30–300 GHz and 300–3000 GHz respectively. GHz means a billion oscillations per second. Shorter wavelengths correspond to higher frequencies.
These names overlap with other classifications: the previous article’s far infrared, 30–1000 μm, includes submillimeter waves. The physical nature of the radiation stays the same: this is one electromagnetic spectrum. A separate chapter is useful because of characteristic sources, atmospheric windows and receiving techniques. NASA HEASARC: a conventional spectrum division.
Branch · With equationsAn optional calculation; you can continue with the main thread.
Converting a line frequency into wavelength
For a wave in a vacuum, frequency ν and wavelength λ are linked by the speed of light, c = 299 792 458 m (s)^( − 1):
λ = (c)/(ν).
The CDMS catalog lists 115.2712018 GHz for the CO J = 1→0 transition. J denotes the molecule’s rotational quantum state: a transition between states produces a particular photon energy. Substitution gives λ≈2.600758 mm. The catalog uncertainty in this frequency is 500 Hz; those digits do not mean our telescope measured it that accurately. CDMS: CO, October 2000 version, transition catalog.
Photon energy is E = hν. Dividing it by Boltzmann’s constant k gives a characteristic scale, hν/k ≈ 5.53 K. This is not a ready-made cloud thermometer: level populations and line brightness also depend on density, collisions and radiative transfer. That number alone cannot tell us the gas temperature.
Gas, dust and background tell different stories
A molecular line is a narrow spectral feature. A molecule has allowed energy states; a transition between them emits or absorbs a photon of the corresponding frequency. H₂ dominates cold molecular gas but is difficult to see directly through its emission. CO is often a more convenient tracer. A CO map is not automatically an accurate mass map: excitation, line opacity and relative abundances matter.
The dust continuum is broader: many frequencies together describe the thermal emission of grains. Several bands help estimate dust temperature and amount within a model. Grain properties and optical depth also matter; one bright patch cannot determine all these parameters.
The microwave background fills the sky. Here both the spectrum and tiny directional differences matter, beyond individual molecular clouds. Galactic dust and gas become foregrounds that must be separated. Even CO lines can contribute noticeably to Planck’s broad bands. NASA: CO in Planck data, COBE: background spectrum and variations.
Branch · With equationsThis explains the synthetic chart's scale; it does not process a real observation.
Why a shifted line needs a reference frame
Let ν_(0) be the laboratory frequency and ν the received frequency. A useful radio definition of the velocity coordinate is:
v_(radio) = c (ν_(0) − ν)/(ν_(0)).
Under this convention, lower frequency means positive velocity. For slow recession along the line of sight, ∣v∣≪c, this coordinate approximates physical radial velocity. Large shifts require an appropriate relativistic or cosmological interpretation; the radio definition alone does not become a universal law of motion.
The chart specifies a Gaussian profile centered on 20 km/s, with width σ=5 km/s, amplitude 1 above a continuum of 0.2. These are arbitrary linear units, without noise or instrumental broadening. The center corresponds to 115.263511733 GHz. As the velocity coordinate increases, frequency decreases, so the upper scale runs in reverse.
A real observer moves with Earth. We must specify a reference frame, such as topocentric at the telescope, barycentric or LSRK, and apply the appropriate corrections. Our hypothetical stationary observer has no date and does not imitate a specific Orion spectrum. ALMA Handbook, §6.8.
Why choose a dry, high site?
Before reaching the antenna, the wave crosses the atmosphere. The atmosphere absorbs some signal, emits its own background and changes the wave’s phase through a varying optical path length. Water vapor is particularly important, but it is not the atmosphere’s only constituent. Transparency depends on frequency: observing windows remain between strong absorption bands.
A dry, high site reduces the column of water above the telescope. ALMA operates on Chile’s Chajnantor plateau, at about 5000 m. This improves conditions without guaranteeing the same sky in every band or all weather. The shortest wavelengths are especially demanding. ALMA: antennas and site, Handbook, §4.1 and fig.4.1.
Think of phase as position within an oscillation cycle: a wave crest arrives slightly earlier or later. A constant delay of an ideal wave need not reduce one receiver’s power. Different, changing delays near different antennas, however, interfere with combining their signals coherently. ALMA’s water-vapor radiometers near 183 GHz help estimate these changes and construct a phase correction. They do not return absorbed photons. For coherent phases, see the interferometer article. Handbook, §10.4.3.
- τ₂₂₅=0.04
- τ₂₂₅=0.08
- τ₂₂₅=0.12
Branch · With equationsA separate JCMT calibration example; these coefficients do not describe ALMA's atmosphere.
What atmospheric transmission means
For a plane-parallel atmosphere, let τ be zenith optical depth and X the relative airmass. Near the zenith, X ≈ 1; lower above the horizon, the path is longer. Under this approximation, transmission is:
\mathcal T=\exp(-\tau X).
For the 850 μm window of SCUBA-2 at JCMT, Mairs and colleagues obtained an empirical relation to zenith optical depth at 225 GHz:
τ_(850) = 3.71( τ_(225) − 0.040; + 0.202√(τ_(225))).
This calibrates a particular band and site from 2011–2021 data, with uncertainties in the coefficients. It is not a universal property of water. The chart uses nominal coefficients and three chosen τ₂₂₅ values: 0.04, 0.08, 0.12. At X=2, transmissions are approximately 0.741, 0.486, 0.329. This plots path length in one window, rather than an atmospheric frequency spectrum or current weather forecast.
Below about 25% transmission at 850 μm, the relation becomes considerably less certain; our examples stay above it. Dividing measured signal by transmission can correct attenuation, but it also amplifies noise. Lost photons do not reappear. Mairs et al., 2021, eq.6, Table1, §3, SCUBA-2 cookbook B.2.
Why must an antenna be accurate relative to wavelength?
A paneled metal dish reflects radio waves by the same physical process as a mirror reflects visible light. The required wavelength sets the surface requirements. An irregularity almost negligible for a centimeter wave can become a substantial fraction of a submillimeter wave.
Surface errors change the path traveled. Different aperture regions then add waves with different phases, sending some energy out of the narrow main beam into a scattered structure. This differs from atmospheric absorption. Increasing diameter alone does not fix a poor shape.
The ALMA handbook’s design target for a 12-meter antenna’s RMS surface error is 25 μm, under specified operating conditions. RMS is the root mean square deviation from the required surface. This is a design target, rather than our assessment of any antenna today. Thermal history, wind and alignment affect the actual state. ALMA Handbook, §9.2.1 and A.1.
- σ=25 μm
- σ=50 μm
Branch · DeeperA formula for a limited error model; it does not determine the antenna's whole efficiency.
What the Ruze formula actually estimates
For random Gaussian errors, relative coherent main-beam gain can be estimated as:
η_(s) = exp [ − (((4πσ)/(λ)))^(2)].
Here σ is the effective axial RMS surface error; in a simple axial approximation, the path error is about twice as large. An arbitrary height measured along the surface normal cannot always be inserted directly: geometric projection and aperture illumination matter. The correlation scale of errors must exceed the wavelength but remain smaller than antenna diameter. A regular deformation or large systematic focus error needs another assessment.
For σ=25 μm, the model gives ηₛ≈0.906 at 1 mm and 0.674 at 0.5 mm. This is a ratio to an ideal surface with the same illumination, rather than the total fraction of collected energy. Illumination, blockage and other losses are excluded; scattered radiation can matter for extended targets. The comparison curve at 50 μm is a teaching assumption, rather than an ALMA measurement. NRAO, Essential Radio Astronomy, §3.3.4, ALMA Memo456, §6.1.
How ALMA translates a fast oscillation into a convenient band
After the primary and secondary reflectors, the wave enters the receiving system. A horn couples the field into a waveguide chain; subsequent components select polarizations and bands. Consider ALMA Band3: 84–116 GHz. Its cold SIS mixer combines a weak sky signal with a stronger known signal from the local oscillator.
SIS means “superconductor–insulator–superconductor.” A thin insulating barrier between films permits quantum tunneling; its nonlinear response under local-oscillator excitation enables frequency conversion. It is a microfabricated chip with electrical bias and matching structures inside a waveguide block, rather than two wires twisted together. Cold is necessary for the superconducting device’s properties and low noise. JPL: SIS mixers, Köln development group.
In Band3, the mixer operates at a stage around 4 K, with intermediate-frequency amplification around 15 K; there is also a 110 K thermal stage and warm components. The cryostat maintains vacuum, thermal shields and separate cold platforms. The whole telescope is not at 4 K. Other ALMA bands can use different architectures, including cold radio-frequency amplification. ALMA Handbook, §4.5.3.
The output remains an electrical signal at an intermediate frequency, IF. It can be amplified, digitized and transmitted for correlation. Within the received band, a heterodyne chain preserves amplitude and phase information relative to a reference signal. This matters for interferometry and detailed spectra.
ALMA Band3 · coherent chain
- Reflectors → horn / waveguide
- Polarization and selected RF band
- SIS + local oscillator → IF · 4 K
- IF amplification · about 15 K
- Digitization → correlator
Output: amplitude and phase relative to a reference in the IF band.
Planck/HFI · bolometric chain
- Reflectors → horn / filters
- Absorber on a weak thermal link
- Heating by absorbed power · 0.1 K
- Thermistor: temperature → resistance
- Electronic readout → time samples
Output: power samples; calibration and scanning → map.
Band3 oscillator: 92–108 GHz; sky RF: 84–116 GHz; IF: 4–8 GHz.
Branch · With equationsIdeal mixing only; SIS quantum behavior and the complete frequency plan are not calculated here.
Why one difference frequency has two possible inputs
The product of two cosines contains their sum and difference frequencies. Selecting the difference component gives:
ν_(IF) = ∣ν_(RF) − ν_(LO)∣.
For our synthetic CO line, νRF≈115.263512 GHz. Choose νLO=108 GHz, the allowed upper edge of Band3’s oscillator range, 92–108 GHz. This gives IF≈7.263512 GHz, within its 4–8 GHz. Yet an input at 100.736488 GHz, on the other side of the oscillator, produces the same difference! We need to know which sideband supplied the signal.
Band3 uses a 2SB architecture to separate the two sidebands. Our calculation illustrates why separation is needed, without modeling its real accuracy. It is also not a complete observation setup: subsequent band centers and widths, and RF edges, need separate checks. We can digitize a more convenient intermediate frequency; the astronomical molecule has not changed its transition. ALMA Handbook, §4.1, §4.5.3, §6.3.3.
How Planck measured heating instead of carrier phase
A broad-band survey can use another principle. A bolometer absorbs radiation, warms slightly and converts that temperature change into an electrical signal. In Planck’s HFI, bolometers operated around 0.1 K in six bands from 100 to 857 GHz. Their output consists of absorbed-power samples over time; pointing during scanning and calibration turn these into a map.
An HFI spider-web bolometer’s absorber is a lightweight silicon-nitride structure with a metal coating. A small doped-germanium thermometer changes resistance as it warms. Thin supports provide a weak but necessary thermal link to the cold platform: without heat removal, the absorber could not return to its operating state. Low heat capacity helps reveal small absorbed energies. IPAC: HFI, Lamarre et al., 2010, §3.3, figs.6–7.
This chain does not preserve the astronomical carrier’s phase as a heterodyne receiver does. A bolometer can nevertheless be polarization-sensitive: two appropriately oriented absorbing grids compare two linear polarizations. Filters or an additional spectral instrument can separate spectral information too. It would be wrong to conclude that a bolometer “cannot measure polarization or spectra at all.”
A 0.1 K sensitive platform does not put all electronics at that temperature. HFI’s design includes separate thermal stages and warm readout. Alternating electrical bias and its demodulation belong to electronic readout; their phase is not preserved sky-wave phase. Lamarre et al., fig.13 and §§4–5.
Turning panels and microchips into an instrument
An antenna first needs a mechanically stable support, a paneled primary reflector and a secondary reflector directing the field into a receiver cabin. In one MELCO ALMA design, aluminum panels were machined together with stiffening ribs on the back. Chemical surface treatment scattered sunlight while retaining the required radio reflectivity. Other manufacturers used other techniques; there is no universal “material of every ALMA panel.” Ishiguro, Chiba & Sakamoto, 2022, §7.2.
A panel sits on adjustable supports. Holographic field measurements allow surface errors to be reconstructed; engineers adjust screws from that map, repeat the measurement and check the result. This surface alignment differs from moving the secondary mirror to compensate focus. The handbook describes up to five screws per panel, rather than continuous automatic motion of every panel during observations. ALMA Handbook, A.1.
The receiving unit needs different tools. Its microchip must connect to the waveguide block, bias circuits and oscillator; the horn, polarization components and cold stages must remain aligned and matched after cooling. Cartridges are integrated into the cryostat, checking vacuum, heat leaks and achieved temperatures. Noise, gain, frequency response and sideband separation are then measured. Common integration-center procedures allow cartridges to be compared meaningfully. ALMA: receivers and Front End Integration Centres.
For HFI, absorbing grids, thermometers and thermal suspensions were made separately and their response tested at operating temperature. The primary description reports JPL bolometer fabrication and repeated measurements at 100 mK in Cardiff. Weak thermal coupling, sensitivity and response speed must be chosen together; excessively weak coupling can slow the instrument. The 4 K cooler also underwent its own thermal-vacuum tests. Lamarre et al., §3.3, §5.2.
These are documented activities of the engineering teams. No antenna, SIS receiver or bolometer was built or physically tested for this article. Sources and teaching calculations were checked; they explain the design but do not constitute its acceptance tests.
What a dry site and a cold receiver actually provide
A cloud dark in visible light can become bright at another wavelength. To understand it, first choose the measurement: a narrow gas line, dust continuum or a broad-band background map. Then match the atmospheric window, antenna shape, receiving principle and calibration.
A dry, high site reduces atmospheric interference. An accurate surface keeps the required signal contribution in the main beam. A cold SIS chain translates oscillations into a convenient band, while a bolometer measures absorbed power through heating. No single one of these measures performs every task.
Check your explanation: can stronger receiver cooling replace a dry atmosphere? No: water vapor ahead of the antenna still absorbs and alters the signal. Does losing carrier phase prevent a bolometer from measuring polarization? Again no: the absorber can be made sensitive to polarization.
An available next step is how several antennas align phases and reconstruct images. Keep the difference between photon loss, surface error and changing path delay in mind.