GOTHAM Survey: TMC-1 Aromatic Chemistry

• The survey investigates aromatic chemistry in TMC-1 using broadband, high-resolution observations with the Green Bank Telescope to uncover new interstellar molecules.
• It employs advanced techniques like velocity stacking, matched filtering, and Bayesian forward modeling to precisely extract molecular signatures in a quiescent dark cloud.
• GOTHAM reports first detections such as propargyl cyanide and indene while setting stringent upper limits, thereby challenging and refining existing astrochemical models.

Searching arXiv for the cited GOTHAM survey papers and related overview work. The GOTHAM Large Program Survey—GBT Observations of TMC-1: Hunting for Aromatic Molecules—is a deep, broadband centimeter-wavelength spectral line survey of the cyanopolyyne peak in the Taurus Molecular Cloud 1 (TMC-1) conducted with the 100 m Green Bank Telescope. It was launched to investigate the depth and breadth of aromatic chemistry in the interstellar medium at the earliest stages of star formation, following the detection of benzonitrile in TMC-1, and has since developed into a broader program for constraining molecular inventories, line identifications, formation pathways, and astrochemical model deficiencies in a cold, chemically rich dark cloud (McGuire et al., 2020). Across the cited works, GOTHAM is characterized by large spectral bandwidth, high sensitivity, high spectral resolution, laboratory-supported spectroscopy, Bayesian line analysis, velocity stacking, and matched filtering; these together have enabled both secure detections of new interstellar molecules and stringent upper limits on chemically informative non-detections (McGuire et al., 2020).

1. Survey design, target selection, and observational regime

GOTHAM targets the cyanopolyyne peak of TMC-1, a cold, quiescent, chemically rich dark cloud that is nearby, well studied, and frequently treated as a benchmark source for astrochemical models (McGuire et al., 2020). The program rationale is tied to four scientific questions stated in the overview paper: whether aromatic species beyond benzonitrile exist in TMC-1, what precursors may build aromatic species through bottom-up chemistry, whether this chemistry is unique to TMC-1 or widespread in dark clouds, and whether such species survive subsequent star and planet formation (McGuire et al., 2020).

The observational architecture is based on the Green Bank Telescope with X-, K-, and Ka-band coverage. In the early overview, the survey spans 8.0–11.6 GHz, 18.0–27.5 GHz, and 26.0–39.5 GHz, using the VEGAS backend and ON-OFF position-switched observations with 1.4 kHz per channel spectral resolution (McGuire et al., 2020). Later GOTHAM analyses describe nearly continuous coverage across 7.9–36.4 GHz or comparable ranges, with uniform frequency resolution of 1.4 kHz, corresponding to roughly 0.05–0.01 km s$^{-1}$ in velocity, and high sensitivity reaching RMS noise of a few mK in parts of the band (Cooke et al., 2023, Remijan et al., 2024). A later inventory paper extends the survey description to 3.9–36.4 GHz, with $\sim 29$ GHz of spectral bandwidth and 1.43 kHz resolution (Xue et al., 8 Sep 2025).

This instrumental regime is central to the program’s scientific character. TMC-1 exhibits narrow lines and low excitation temperatures, so high frequency precision and fine channelization are necessary not merely for sensitivity but for reliable rest-frequency matching, multi-component decomposition, and statistical recovery of molecules whose individual transitions remain below the single-line detection threshold (McGuire et al., 2020, Remijan et al., 2024). A plausible implication is that GOTHAM is not only a discovery survey but also a precision spectroscopy platform for centimeter-wave astrochemistry.

2. Data reduction, calibration, and statistical analysis framework

The overview paper emphasizes detailed observing and reduction procedures, including archival data reuse, gbtidl-based processing, Doppler corrections, conversion to atmosphere-corrected $T_A^*$, polarization averaging where applicable, manual RFI cleaning, baseline fitting with polynomial orders between 1 and 20, and noise-weighted averaging (McGuire et al., 2020). Later work on the TMC-1 inventory reports a Python-based, fully automated data reduction pipeline—the GOTHAM Spectral Pipeline—for single-beam, position-switched observations, with automated RFI flagging, per-channel Doppler tracking correction, dynamic zenith opacity correction using real-time weather data, and rejection of integrations with abnormal system temperatures (Xue et al., 8 Sep 2025). This indicates an evolution from early heterogeneous reduction toward a scalable, reproducible calibration framework suited to a survey of very large bandwidth and integration time.

A consistent analysis layer across the program is Bayesian forward modeling with Markov Chain Monte Carlo (MCMC). In different papers, the fit parameters include column density, excitation temperature, linewidth, source size, and one to four velocity components depending on the species and model choice (McGuire et al., 2020, Cooke et al., 2023, Remijan et al., 2024, Xue et al., 8 Sep 2025). The later inventory work gives the likelihood explicitly as

$$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$

with $x_i$ the observed intensity, $m_i$ the model intensity, and $\sigma_i$ including noise and systematic uncertainties (Xue et al., 8 Sep 2025).

Radiative-transfer-based column density modeling is likewise explicit in the program literature. For example, the HCCCHO study gives the optically thin line-analysis expression

$$N = \frac{3k}{8\pi^3 \nu S \mu^2} \cdot \frac{Q(T_{ex})}{g_u} \exp\left(\frac{E_u}{kT_{ex}}\right) \cdot \int T_{mb}\,dv$$

for emission-line column density estimation (Remijan et al., 2024). The inventory study further provides expressions for optical depth and observed brightness temperature, together with beam dilution and partition function corrections (Xue et al., 8 Sep 2025).

For weak molecules, GOTHAM systematically combines velocity stacking and matched filtering. Spectra centered on predicted transitions are aligned in velocity, stacked with weights based on predicted intensity and RMS noise, and then cross-correlated against a simulated stacked spectrum to yield an impulse response used as a detection statistic (McGuire et al., 2020, Lee et al., 2021, Burkhardt et al., 2021, Sita et al., 2022, Remijan et al., 2023, Cooke et al., 2023). The repeated use of this framework across chemically diverse targets suggests that statistical signal extraction is a defining methodological contribution of the survey rather than an auxiliary technique.

3. Molecular detections and the expansion of the TMC-1 inventory

GOTHAM has been associated with a sequence of first detections and targeted identifications spanning unsaturated chains, substituted nitriles, pure hydrocarbons, sulfur-bearing species, and molecular anions.

The survey overview reported the first interstellar detection of propargyl cyanide $\mathrm{HCCCH_2CN}$, supported by new laboratory spectroscopy with over 110 hyperfine-split features measured between 5–40 GHz at $\sim 2$ kHz uncertainty, and an 18$\sim 29$0 matched-filter detection in TMC-1 with a total column density of $\sim 29$1 (McGuire et al., 2020).

Early GOTHAM analyses also reported the discovery of trans-(E)-cyanovinylacetylene and vinylcyanoacetylene, with matched-filter significances of 8.0$\sim 29$2 and 5.5$\sim 29$3, respectively. Their MCMC-derived column densities are $\sim 29$4 for trans-(E)-cyanovinylacetylene and $\sim 29$5 for vinylcyanoacetylene; the trans-(Z) isomer remained undetected with an upper limit of $\sim 29$6 (Lee et al., 2021).

The survey subsequently yielded the first interstellar detection of indene $\sim 29$7, described as the first interstellar detection of a pure hydrocarbon PAH, with a 5.7$\sim 29$8 matched-filter detection and total column density $\sim 29$9 (Burkhardt et al., 2021). A later study detected 2-cyanoindene at 6.3$T_A^*$0, with $T_A^*$1, while refining the indene column density to $T_A^*$2 (Sita et al., 2022). These measurements enabled the first direct hydrocarbon:CN-substituted comparison for a PAH-related system in the same source.

GOTHAM also reported the first astronomical detection of the interstellar anion $T_A^*$3. Using velocity-stacked data and matched filtering, the anion was detected at $T_A^*$4 confidence with column density $T_A^*$5, while the neutral radical $T_A^*$6 was only tentatively detected at $T_A^*$7 with $T_A^*$8 (Remijan et al., 2023).

Further detections include $T_A^*$9-1-cyano-1,3-butadiene, identified at 5.1$$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$0 with total column density $$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$1 (Cooke et al., 2023), and cyclopropenethione $$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$2, detected with $$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$3 and $$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$4, completing the three low-energy isomers of the $$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$5 family in TMC-1 (Remijan et al., 10 Jan 2025).

The later inventory analysis synthesizes these and other results into a survey-scale census, reporting constrained column densities for 102 molecular species in TMC-1, comprising 75 main isotopic species, 20 $$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$6C-substituted species, and 7 deuterium-substituted species, with the detected gas-phase inventory populated primarily by unsaturated hydrocarbons (Xue et al., 8 Sep 2025).

4. Aromatics, cyano proxies, and bottom-up chemical inference

A major theme of GOTHAM is the use of nitrile-functionalized molecules as observational proxies for otherwise radio-inactive hydrocarbons and PAHs. This strategy follows from the difficulty that fully symmetric pure hydrocarbons may lack a permanent dipole moment and therefore a detectable rotational spectrum, whereas CN substitution imparts strong dipole moments and rotational observability (Sita et al., 2022).

The indene and 2-cyanoindene pair provides a central case. The measured abundance ratio

$$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$7

with uncertainty range 27–61 is described as the first direct observation of the ratio of a cyano-substituted PAH to its pure hydrocarbon counterpart in the same source (Sita et al., 2022). That study concludes that while astrochemical models are not yet sufficiently accurate to reproduce the absolute abundances of these species, they do a good job at predicting hydrocarbon:CN-substituted ratios, thereby supporting -CN tagged species as excellent proxies for fully symmetric counterparts (Sita et al., 2022).

The survey has also used nitriles as proxies for non-polar chain precursors. The detection of $$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$8-1-cyano-1,3-butadiene constrains the abundance of 1,3-butadiene, a non-polar hydrocarbon that cannot be directly detected by radio observations. Using a modeled CN-functionalization ratio, the work infers

$$l = \sum_{i} \left[ -\ln \left(\sigma_i \sqrt{2\pi}\right) - \frac{1}{2}\frac{(x_i-m_i)^2}{\sigma_i^2} \right],$$9

and discusses 1,3-butadiene as a key precursor in laboratory, theoretical, and astrophysical models for bottom-up aromatic-ring formation (Cooke et al., 2023). The same paper gives the reaction

$x_i$0

as a route to benzene and related products (Cooke et al., 2023).

The indene paper frames pure hydrocarbon PAHs as chemically decisive because they provide a direct handle on bottom-up and top-down aromatic formation scenarios in cold clouds (Burkhardt et al., 2021). It states that indene is the most abundant organic ring detected in TMC-1 to date and that its abundance exceeds NAUTILUS predictions by several orders of magnitude (Burkhardt et al., 2021). This suggests that existing aromatic formation networks remain incomplete even when precursor chains and cyano derivatives are better constrained.

A later statistical inventory strengthens the aromatic context by reporting ten individual aromatic molecules in the GOTHAM observations and quantifying their contribution to the gas-phase carbon budget: 0.011% when including CO and 6% when excluding CO (Xue et al., 8 Sep 2025). Nine of the ten are CN-functionalized aromatics, which is presented as evidence of the practical role of nitriles as radio proxies (Xue et al., 8 Sep 2025).

5. Spectroscopic precision, laboratory support, and catalog revision

A defining feature of GOTHAM is its dependence on laboratory spectroscopy and on the revision of public spectral catalogs when survey data reveal inconsistencies. The HCCCHO study is exemplary. It combined astronomical frequencies measured at sufficiently high spectral resolution with new laboratory measurements at $x_i$1 kHz resolution to produce a new global fit to the rotational spectrum of propynal (Remijan et al., 2024). The fit comprised 2674 unique transitions—or 3230 with blended lines—covering $x_i$2 to $x_i$3 and $x_i$4–15, and achieved astronomical frequency matches within $x_i$5 kHz for the key $x_i$6 and $x_i$7 transitions (Remijan et al., 2024).

Using this revised catalog, MCMC analysis yielded a best-fit total HCCCHO column density of

$x_i$8

with an excitation temperature of approximately 3 K, around a factor of 5 times larger than reported in previous studies (Remijan et al., 2024). The paper explicitly states that care is needed when using publicly available spectral catalogs to characterize astronomical spectra and emphasizes the growing importance of such catalogs as next-generation facilities come online (Remijan et al., 2024).

Comparable laboratory support appears throughout the survey. The propargyl cyanide detection required improved hyperfine-resolved rotational measurements to predict rest frequencies within 0.03 km s$x_i$9 (McGuire et al., 2020). The indene work relied on 189 transitions measured with 2 kHz accuracy over roughly 2–40 GHz (Burkhardt et al., 2021). The cyanoindene study used cavity Fourier-transform microwave spectroscopy between 6–40 GHz for five isomers (Sita et al., 2022). The c-$m_i$0 detection was enabled by laboratory characterization of a molecule for which no previous experimental spectra existed (Remijan et al., 10 Jan 2025).

This recurring pattern shows that GOTHAM is structurally coupled to laboratory molecular spectroscopy. A plausible implication is that the survey’s discovery space is determined as much by catalog fidelity and laboratory rest-frequency accuracy as by telescope sensitivity alone.

6. Non-detections, upper limits, and constraints on chemical selectivity

GOTHAM’s scientific value is not confined to positive detections. The survey’s sensitivity and Bayesian multi-line analysis have also produced stringent upper limits that constrain chemical formation and destruction pathways.

The most systematic non-detection study in the cited set is the search for nitrogen-, oxygen-, and sulfur-bearing heterocycles in TMC-1 (Barnum et al., 2022). Using deep GOTHAM observations over 7.906–35.434 GHz and Bayesian forward modeling in molsim, the search targeted 19 heterocycles and found no evidence for the presence of any heterocyclic species (Barnum et al., 2022). The MCMC approach used all lines covered by the survey for each species and reported the 97.5th percentile of the posterior distribution as a 2$m_i$1 upper limit (Barnum et al., 2022). The paper states that the MCMC multi-line analysis systematically yielded more stringent upper limits than traditional single-line frequentist methods (Barnum et al., 2022).

The same work interprets the absence of detectable heterocycles despite the confirmed presence of cyclic and polycyclic hydrocarbons as evidence that heterocycles are significantly less abundant than pure carbocyclic counterparts in TMC-1 (Barnum et al., 2022). Specific examples given are that pyridine is at most half as abundant as benzonitrile, while pyrrole, thiophene, and furan are depleted by one to two orders of magnitude relative to cyclopentadiene (Barnum et al., 2022). The paper further hypothesizes inefficient low-temperature formation pathways, lack of suitable precursors, and possible destruction effects, though it notes that dense clouds such as TMC-1 are UV-shielded (Barnum et al., 2022).

Upper limits are also chemically informative in chain and nitrile chemistry. The methyl-chain study found no significant emission from methyltetraacetylene $m_i$2 and set a 95% confidence upper limit of $m_i$3, interpreting this as a sharp abundance drop-off after $m_i$4 (Siebert et al., 2021). The trans-(Z)-cyanovinylacetylene isomer likewise remained undetected with $m_i$5 (Lee et al., 2021). Ethyl cyanide, considered as a hydrogenation product of vinyl cyanide, showed only 4.17$m_i$6 significance and was treated as an upper limit of $m_i$7, an order of magnitude lower than previous limits toward TMC-1 (Lee et al., 2021).

These non-detections collectively indicate that GOTHAM is sensitive not just to molecular complexity per se, but to chemical selectivity: why some unsaturated carbonaceous species are abundant in TMC-1 while others, including heterocyclic analogues or more saturated products, are not.

7. Astrochemical interpretation, model tensions, and benchmark status

A recurring result across GOTHAM papers is that observations often outpace existing astrochemical models in absolute abundance, while some relative trends are reproduced more successfully.

For propargyl cyanide, NAUTILUS-v1.1 with the KIDA 2014 network and a dominant formation route

$m_i$8

reproduces the observed abundance within a factor of a few, suggesting that the main formation picture for this unsaturated precursor is approximately correct (McGuire et al., 2020).

By contrast, indene presents a much stronger challenge. The indene paper states that NAUTILUS models underpredict the observed abundance by several orders of magnitude, with modeled column densities of order $m_i$9–$\sigma_i$0 against an observed value near $\sigma_i$1 (Burkhardt et al., 2021). Likewise, the $\sigma_i$2-1-cyano-1,3-butadiene study reports that even after expanding the reaction network and increasing modeled 1,3-butadiene abundances by about two orders of magnitude relative to previous GOTHAM models, modeled abundances of aromatic species remain underestimated by 1–4 orders of magnitude (Cooke et al., 2023).

The methyl-chain study reports that methylcyanopolyynes decrease log-linearly with chain length and are broadly matched by a NAUTILUS v1.1 three-phase gas-grain model, whereas the sharp observed drop in methylpolyynes at $\sigma_i$3 is not reproduced (Siebert et al., 2021). That same paper interprets differing $\sigma_i$4 nuclear-spin isomer ratios in methylcyanopolyynes and methylpolyynes as evidence for separate interstellar formation pathways (Siebert et al., 2021).

The $\sigma_i$5 analysis exposes a different class of model discrepancy. The measured ratio

$\sigma_i$6

is described as the highest measured between an anion and neutral species to date and as being at odds with current theories of interstellar anion chemistry (Remijan et al., 2023). That work further compares the observed abundances to a gas/grain chemical model and to a machine learning analysis, finding that the machine learning approach matches the detected anion abundances much better than the gas/grain model, which is presented as evidence that current understanding of molecular anion formation chemistry is still highly uncertain (Remijan et al., 2023).

The c-$\sigma_i$7 detection adds an isomeric-chemistry perspective. Its abundance relative to the previously detected CH$\sigma_i$8CCS and HCCCHS follows the relative dipole principle (RDP), a kinetic rule-of-thumb according to which, all other chemistry among a family of isomers being the same, the member with the smallest dipole should be the most abundant (Remijan et al., 10 Jan 2025). The paper states that the RDP now holds for both S-bearing and O-bearing counterparts observed in TMC-1, although CH$\sigma_i$9CCO continues to elude detection (Remijan et al., 10 Jan 2025). This suggests that kinetic control, including ion-molecule destruction scaling with dipole moment, can dominate over thermodynamic ordering in at least some isomer families.

The 2025 inventory paper formalizes GOTHAM’s benchmark role by presenting a reference set of gas-phase molecular abundances for interstellar clouds (Xue et al., 8 Sep 2025). It states that the inventory accounts for approximately 0.12% of gas-phase carbon, 0.0015% of oxygen, 5% of sulfur, and ~80% of detected gas-phase nitrogen, while confirming up to four velocity components and identifying some molecules—such as CCS, $$N = \frac{3k}{8\pi^3 \nu S \mu^2} \cdot \frac{Q(T_{ex})}{g_u} \exp\left(\frac{E_u}{kT_{ex}}\right) \cdot \int T_{mb}\,dv$$0, and $$N = \frac{3k}{8\pi^3 \nu S \mu^2} \cdot \frac{Q(T_{ex})}{g_u} \exp\left(\frac{E_u}{kT_{ex}}\right) \cdot \int T_{mb}\,dv$$1-$$N = \frac{3k}{8\pi^3 \nu S \mu^2} \cdot \frac{Q(T_{ex})}{g_u} \exp\left(\frac{E_u}{kT_{ex}}\right) \cdot \int T_{mb}\,dv$$2—as requiring non-LTE or more complex models (Xue et al., 8 Sep 2025). In that sense, GOTHAM functions not only as a discovery program but as an empirical reference against which chemical networks, excitation treatments, isotopic fractionation schemes, and catalog accuracy can be tested.

Taken together, the survey establishes TMC-1 as a chemically rich environment dominated in the detected gas phase by unsaturated hydrocarbons, nitriles, and related chain species, in marked contrast to the oxygen-rich organics characteristic of sublimated ices around protostars (Xue et al., 8 Sep 2025). The broader implication is that cold dark-cloud chemistry supports an unexpectedly extensive reservoir of aromatic and pre-aromatic molecules, yet that the mechanistic basis of this reservoir remains only partially captured by current gas-grain models.