On the Origin of Sodium Atoms in the Comae and Trails of Comets

Abstract

Background: The observed abundance ratios of alkali species in ground-based spectra of comets deviate from solar composition, suggesting alkali ejection from phenoxides reacting with carbon dioxide at the nucleus surface (alkali-phenoxide carbonylation). Methods: Here, we search for the alkali emissions in spectra of the coma and of the trail of Comet C/2023 A3 (Tsuchinshan-ATLAS) exploiting the double-fiber entrance of the high-resolution PEPSI spectrograph at the 8.4 m Large Binocular Telescope. Results: Spectra sampling the nucleus yield $Na/K$ ratios 3.6 times higher than the chondritic value, and even higher ratios sampling the trail. This fact excludes photodesorption as the main sodium source, leaving sodium-phenoxide carbonylation at the surface of the main nucleus and the trail mininuclei as the primary sodium source. Conclusions: The nucleus temperature and the faint KI line exclude potassium-phenoxide carbonylation. For the first time, KI is detected in the trail of an Oort cloud comet, suggesting potassium photodesorbed from the trail mininuclei. Sodium-phenoxide carbonylation is at least six times more efficient than sodium photodesorption if the $Na/K$ ratio in the C/2023 A3 nuclei is chondritic. Trails composed of sub-km-sized mininuclei may be common features of Oort cloud comets.

1. Introduction

This study examines the sources and abundances of the alkali elements sodium (Na), potassium (K) and lithium (Li) in comets. Sodium emission has been widely reported in cometary comae and tails [1,2,3,4,5], whereas potassium detections remain comparatively rare, having been identified to our knowledge only in comets C/1965 S1 [6], C/2011 L4 [7], C/2012 S1 [5], C/2020 F3 and C/2024 G3 [8]. In all cases, the $Na/K$ atomic ratio significantly exceeds the chondritic value of ${(Na/K)}_c=16$ [9]. The detection of the KI line in cometary spectra allowed Fulle et al. [8] to predict that a potassium tail should be observable in future comets.

To date, lithium has not been directly detected in cometary spectra. In contrast, laboratory mass spectrometry of the components of residues of dust particles from comet 81P/Wild 2 collected by the Stardust mission revealed Li, Na, and K abundances exceeding chondritic levels all within a factor of two [10]. The Li upper limit for comet C/2020 F3 is very stringent with $Na/Li>3.4\times {10}^4$, a factor of >33 greater than the chondritic value of ${(Na/Li)}_c=1036$ [9]. Li is synthesized during standard Big Bang nucleosynthesis and further enriched by novae to present values [11]. The value found in C/2020 F3 implies a Li abundance below the primordial value, a fact that leads to the formulation of a new mechanism for alkali ejection in comets due to alkali-phenoxide carbonylation, namely, the chemical reaction between alkali phenoxides and carbon dioxide inside the dust particles located at the nucleus surface followed by the protonation (e.g., by protons photodesorbed from dust hydrocarbons) of the produced alkali salicylates actually ejecting alkali atoms [8,12]. Laboratory data of the protonation of alkali phenoxides are not available. If this process was ejecting alkali atoms as efficiently as the protonation of salicylates at temperatures different than alkali-phenoxide carbonylation, then the strong production increase in sodium at $r\approx 0.8$ au and of potassium at $r\approx 0.5$ au [8] would remain unexplained.

Alkali-phenoxide carbonylation is activated when the cometary nucleus exceeds a limit temperature, which is 363 K for sodium and 473 K for potassium [12]. Available measurements of the surface temperature of the nucleus of Comet 67P/Churyumov-Gerasimenko (67P hereafter) by means of the Visible and Infrared Thermal Imaging Spectrometer (VIRTIS) instrument onboard the Rosetta mission showed that Hapi, the most active region during the inbound orbit, reached the gray-body temperatures, with the most probable value of 214 K at $r=3.54$ au [13]. Here, we focus on spectra of Comet C/2023 A3 (Tsuchinshan-ATLAS), which is an Oort cloud comet first identified in Purple Mountain Observatory images and later independently recovered by the Asteroid Terrestrial impact Last Alert System (ATLAS) in February 2023. The spectra were taken at the heliocentric distances $0.81\le r\le 0.83$ au (

Table 1. Log of observations, with one fiber located on the nucleus and another on the trail.

UT	Exposure	Air	r	$\mathit{\Delta }$	$\dot{\mathit{r}}$	$\dot{\mathit{\Delta }}$	$\mathit{\alpha }$	${\mathbf{PA}}_{\mathit{r}}$	${\mathbf{PA}}_{\mathit{v}}$	${\mathbf{PA}}_{\mathit{o}}$	${\mathit{\rho }}_{\mathit{o}}$	${\mathit{a}}_{\mathbf{Na}}$	${\mathit{a}}_{\mathit{K}}$
October 2024	s	Mass	au	au	km/s	km/s	deg	deg	deg	deg	km	m/s2	m/s2
26.09–26.11	$3\times 720$	1.9–2.4	0.812	0.774	+33.64	+58.02	77.6	71.6	239.8	225.1	5600	0.65	0.48
27.06–27.12	$3\times 1800$	1.6–2.9	0.831	0.808	+33.61	+58.80	74.6	71.4	238.3	225.2	5800	0.62	0.46

r and $\Delta$: heliocentric and geocentric distances. $\alpha$: phase angle. $P{A}_r$ and $P{A}_v$: position angles of the antisolar and trailing orbit directions. $P{A}_o$ and ${\rho }_o$: position angle and distance from the nucleus of the trail observation. $a_{Na}$ and $a_K$: alkali antisunward acceleration due to solar radiation pressure (see Fulle et al. [7] for its computation).

), where an active nucleus reaches gray-body temperatures $442\le T\le 447$ K, consistent with sodium-phenoxide carbonylation and inconsistent with potassium-phenoxide carbonylation. The water-ice-enriched block (WEB) model [8,14] suggests lower equilibrium temperatures of 340 K at $r=0.8$ au (in Table 5 of [8]) for a water-driven active nucleus. However, the Rosetta/VIRTIS data show that the surface dust particles where carbon dioxide may react with alkali phenoxides maintain higher non-equilibrium temperatures much closer to the gray-body ones.

2. Observations and Data Analysis

Comet C/2023 A3 was observed on 26 and 27 October 2024 with the Potsdam Echelle Polarimetric and Spectroscopic Instrument (PEPSI) at the Large Binocular Telescope (LBT, Mount Graham, Arizona) [15] in monocular mode (8.4 m), with a spectral coverage from 380 to 900 nm at a resolving power of $\lambda /\Delta \lambda$ = 130,000. A fiber was located on the nucleus and another at the distance ${\rho }_o$ of 5600 and 5800 km and position angle $P{A}_o$ of 239.8 and 238.3 deg, respectively (Table 1).

The comet was receding from Earth with radial velocities of 58.02 and 58.80 km s⁻¹ at geocentric distances of 0.77 and 0.81 au in the nights of the 26 and 27 October 2024, respectively. The comet was also receding from the Sun with heliocentric radial velocities of 33.64 and 33.61 km s⁻¹ at heliocentric distances of 0.81 and 0.83 au, respectively (Table 1). The heliocentric and geocentric radial velocities were calculated by using JPL’s (Jet Propulsion Laboratory, Pasadena, California) SPK solar system and C/2023 A3 BSP ephemeris files. On 26 October, the comet was observed from 2h05m UT to 2h37m UT in three couples (each per fiber) of different exposures at different Cross Dispersers (CDs), with sequence CD6 CD5 CD4 at an air mass increasing from 1.9 to 2.4; on 27 October, the comet was observed from 1h28m to 3h00m UT in three couples of different exposures, with sequence CD5 CD4 CD6 at an air mass increasing from 1.6 to 2.9: this observing procedure introduced different night-dependent atmospheric reddening at different spectrum orders.

The spectra were reduced as described in Strassmeier et al. [16] and calibrated by means of the spectra of the standard star HR7950, observed just after each sequence at an air mass of 1.4, fifty degrees east of the comet. The spectra were divided by the telluric synthetic Transmission of the AtmosPhere for AStronomical data (TAPAS) spectrum calculated for the time of observations and local atmospheric profile [17] and then offset to the solar-reflected rest frame. A Phoenix solar spectrum [18] was used to correct the comet ones, thus providing the calibrated average values of the dust continuum computed over the whole spectral range covered by CD4, CD5 and CD6, after the correction of the different reddening at different spectrum orders. This procedure increased the local continuum noise and introduced emission artifacts, due to the high air masses, so we measured the line intensities on flux-calibrated raw spectra. The dust continuum of the raw spectra was expected to be constant (as observed in comet C/2011 L4 [7]) because the spectral slope of the solar Planckian of −18%/100 nm at 600 nm was almost balanced by the spectral reddening of cometary dust of $+(12.5\pm 1.5)$%/100 nm [19], with a systematic error <15% from the NaI to the KI wavelengths. Therefore, the raw spectra were flux-calibrated by normalizing them to the computed dust continuum, and offset to the rest frame of the comet, so that the standard air wavelength scale corresponded to atomic/molecular species (Figure 1, Figure 2, Figure 3, Figure 4 and Figure 5). The dust continuum increased by 30% from 26 to 27 October (an increase higher than the slope error and than the noise of the nucleus-centered dust continuum), consistent with possible one-day changes in the total dust cross section both on the nucleus-centered coma and at the distance ${\rho }_o$, which is covered in less than one day if the dust velocity is ≥68 m s⁻¹.

After the subtraction of the continuum, each emission line was fitted by a Gaussian, the FWHM of which varied from 60 to 80 mÅ, and converted to the intensities I (

Table 2. Line intensities and elemental abundances in C/2023 A3.

Atom	$\mathit{\lambda }$ (Å)	${\mathit{I}}_{26\mathit{N}}$ ($3\mathit{\sigma }$)	${\mathit{I}}_{27\mathit{N}}$ ($3\mathit{\sigma }$)	${\mathit{I}}_{26\mathit{T}}$ ($3\mathit{\sigma }$)	${\mathit{I}}_{27\mathit{T}}$ ($3\mathit{\sigma }$)	${\left(\frac{\mathit{Na}}{\mathit{x}}\right)}_{26\mathit{N}}$	${\left(\frac{\mathit{Na}}{\mathit{x}}\right)}_{27\mathit{N}}$	${\left(\frac{\mathit{Na}}{\mathit{x}}\right)}_{26\mathit{T}}$	${\left(\frac{\mathit{Na}}{\mathit{x}}\right)}_{27\mathit{T}}$
Li	6707.78	<2.3	<1.9				>4500
Na	5889.95	$1315\pm 15$	$2415\pm 20$	$109.5\pm 1.5$	$379.3\pm 1.5$	1	1	1	1
Na	5895.92	$720\pm 15$	$1370\pm 20$	$59.1\pm 1.5$	$205.5\pm 1.5$	1	1	1	1
K	7664.90	$32.9\pm 2.1$	$59.7\pm 1.4$	$3.3\pm 1.6$	$7.0\pm 1.6$	$53\pm 12$	$53\pm 10$	$61\pm 37$	$77\pm 30$
K	7698.96	$15.0\pm 3.1$	$32.6\pm 1.4$	<1.8	$4.6\pm 1.8$	$70\pm 23$	$57\pm 10$	>38	$75\pm 35$

Intensities I are in ${10}^{-15}$ erg cm⁻² s⁻¹ Å⁻¹ units. $I_{ij}$ specifies the intensities measured on the October day ($=i$) and on the nucleus ($j=N$) or on the trail ($j=T$). $I_{Kij}$ specifies the average of the potassium intensities. ${(Na/K)}_N$ and ${(Na/K)}_T$ specify the averages of the nucleus and trail $(Na/K)$ values, respectively. The two $Na/K$ ratios refer to the couples of lines at $\lambda =5889.95$ Å and $\lambda =7664.90$ Å, and at $\lambda =5895.92$ Å and $\lambda =7698.96$ Å, respectively.

) by dividing each Gaussian area by 70 mÅ, thus facilitating the comparison with the lines in Figure 1, Figure 2, Figure 3, Figure 4 and Figure 5. If the main sodium source is sodium-phenoxide carbonylation, then the $I_{27N}/I_{26N}$ ratios provide the increase in the sunlit nucleus cross section due to its rotation, which is higher than the continuum increase during the same period because the wide dust velocity dispersion damps the brightness variations in the nucleus-centered dust coma. The abundances of sodium related to another atom x, $Na/x$ (Table 2), are independent of the absolute value of a flat continuum, depend on the g-factor $g_x$ at 1 au computed by Fulle et al. [7] and are obtained from the line intensity I by means of the relationship

$$Na/x=(g_x/g_{Na})(I_{Na}/I_x)$$

(1)

.

3. Discussion

The detection of faint KI lines in high-resolution spectra of C/2023 A3 implies that potassium is ejected by photodesorption only, thermal desorption being much less probable [4] and potassium-phenoxide carbonylation being impossible at $T\le 447$ K. On the other hand, the detection of NaI lines implies two possible sources for sodium: photodesorption and sodium-phenoxide carbonylation. The fact that the $Na/K$ ratio of cometary nuclei is likely chondritic [10,20] allows us to determine the efficiency of sodium-phenoxide carbonylation with respect to sodium photodesorption.

As in all previous comets with a measured $Na/K$ ratio (

Table 3. Alkali abundances with respect to the chondritic ratios ${(Na/K)}_c$ and ${(Na/Li)}_c$ [6,7,8].

Comet	r	${(\mathit{Na}/\mathit{K})}_{\mathit{N}}/{(\mathit{Na}/\mathit{K})}_{\mathit{c}}$	${(\mathit{Na}/\mathit{K})}_{\mathit{T}}/{(\mathit{Na}/\mathit{K})}_{\mathit{c}}$	${(\mathit{Na}/\mathit{Li})}_{\mathit{N}}/{(\mathit{Na}/\mathit{Li})}_{\mathit{c}}$
	au
C/1965 S1	$0.14$	$3.1\pm 0.7$		>32
C/2024 G3	$0.15$	$1.6\pm 0.5$		>1
C/2020 F3	$0.36$	$1.9\pm 0.3$		>33
C/2011 L4	$0.46$	$3.4\pm 0.9$		>7.7
C/2023 A3	$0.83$	$3.6\pm 0.9$	$4.5\pm 2$	>4.5

), $Na/K$ of C/2023 A3 exceeds the chondritic value (Table 2), with ${(Na/K)}_N=(3.6\pm 0.9){(Na/K)}_c$. The uncertainties affecting the $Na/K$ ratios are higher than that of the continuum spectral slope. Munaretto et al. [21] analyzed the same spectra finding ${(Na/K)}_{27N}=145\pm 6$, which is 2–3 times higher than those shown in Table 2, with a similar uncorrected factor between the Na and K continuum, due to the color bias of a flux calibration not correcting the different atmospheric reddening of the comet and of the photometric standard star HR7950.

The sunward distance ${\rho }_o$ excludes that the spectra outside the nucleus are polluted by alkali atoms ejected from the main nucleus towards the Sun, reaching the maximum sunward distance $\rho =v^2/\left(2a\right)$, where a is the antisunward acceleration due to the solar radiation pressure (Table 1) and $v=\sqrt{7.6085kT/m}=1.25$ km s⁻¹ at $T=447$ K is the atom ejection velocity outside ten nucleus radii [22]; here, k is the Boltzmann constant and m is the molecular mass of water, because all alkali atoms are dragged at the velocity of the most abundant gas in the coma. We get $\rho \le 1700$ km $\ll {\rho }_o$ for both sodium and potassium (Table 1), thus ruling out the alkali tail model [7] as a possible explanation of $Na/K\gg {(Na/K)}_c$, because it predicts an increasing $Na/K$ ratio from the assumed alkali source in the sunward coma to the nucleus, opposite to the actual data (Table 2).

All these facts exclude that sodium is ejected by thermal or photodesorption only, which, according to laboratory experiments [23], are expected to maintain a chondritic value both on the nucleus and on the coma dust, i.e., the same refractories mainly composing the nucleus itself. Mass fractionation at ejection is also inconsistent with the $(Na/Li)/{(Na/Li)}_c$ ratios higher than the $(Na/K)/{(Na/K)}_c$ ones (Table 3), opposite to what is expected according to the respective atomic masses. Significant sticking of alkali on coma dust would imply a strong increase in the $Na/K$ ratio as r decreases [23], opposite to observations (Table 3). After ejection, the main fractionation is photoionization by the solar UV radiation [23], with lifetimes ${\tau }_{Na}=1.3\times {10}^5$ s and ${\tau }_K=3\times {10}^4$ s at $r=0.83$ au [7], thus affecting the $Na/K$ ratio by $exp-{\displaystyle \frac{\varphi }{{\tau }_{Na}v}}/exp-{\displaystyle \frac{\varphi }{{\tau }_{K}v}}=1.018$, where $\varphi =870$ km is the fiber diameter at the comet. Therefore, the observed high $Na/K$ ratios suggest that sodium is mainly produced by sodium-phenoxide carbonylation, thus more efficient than photodesorption. In this case, the alkali source should have a volume big enough to produce carbon dioxide, corresponding to sizes of some tens of meters [8], i.e., mininuclei populating a trail. A sodium trail was actually observed in C/2024 G3 [8]. Its dynamical analysis constrained its origin in the Oort cloud, where the galactic tides injecting C/2024 G3 towards the Sun possibly fragmented it into bodies larger than 30 m [8]. A sodium trail is also consistent with the very asymmetric C/2006 P1 NaI line along the comet-orbit-oriented spectrograph slit [4,8] and with the location of the fiber outside the nucleus of C/2023 A3 (Table 1). Here, we check if such a scenario is consistent with all the available data.

The very faint KI lines (a factor of >300 fainter than all other previous KI detections, which occurred at $r<0.5$ au) suggest potassium photodesorption. In all previous KI observations, potassium-phenoxide carbonylation was the main source of potassium atoms, because photodesorption could have been more efficient than in C/2023 A3 by a factor of ≈3 (given by the squared ratio of the observation heliocentric distances) times the ratio of the respective nucleus cross sections, which is ≪100. The KI lines are as intense as the continuum on the trail (Figure 5), but much less intense than the continuum on the nucleus (Figure 3), where some contribution from the nucleus added to that from the dust would imply the opposite if potassium were photodesorbed from dust. A reliable continuum-independent source of potassium is photodesorption from the trail mininuclei, the volume of which can be estimated because photodesorption depends on the sunlit cross section of the source. We computed that the trail mininuclei had a total sunlit cross section $I_{K26T}/I_{K26N}=(23\pm 18)$% of the main nucleus on 26 October, and $I_{K27T}/I_{K27N}=(16\pm 12)$% on 27 October, both consistent with the storage of CO₂ ice inside them and with the expected size of trail mininuclei [8] at the distance ${\rho }_o$ from the main nucleus (Table 1).

The fact that the KI line intensities are inconsistent with dust as the source of potassium further supports sodium-phenoxide carbonylation as the main sodium source. Sodium-phenoxide carbonylation depends on the nucleus surface area with $T>363$ K, rather than on the sunlit nucleus cross section. Let us assume that sodium-phenoxide carbonylation is a factor of $f=10$ more efficient than photodesorption in ejecting sodium atoms. In order to explain the highest $Na/K$ ratio in Table 2, namely, ${(Na/K)}_{27Tmax}=110$, it is sufficient that ${(Na/K)}_{27Tmax}/\left[{(Na/K)}_c(f+1)\right]\approx 70$% of the average trail mininuclei’s surface had $T>363$ K. If the whole mininuclei’s surface had $T>363$ K, then ${(Na/K)}_T={(Na/K)}_c(f+1)$, namely, ${(Na/K)}_{c}f$ from sodium-phenoxide carbonylation and ${(Na/K)}_c$ from photodesorption. It follows that the lower limit of the efficiency ratio between sodium-phenoxide carbonylation and photodesorption is $f_{min}={(Na/K)}_{27Tmax}/{(Na/K)}_c-1\approx 6$, which provides the highest possible area fraction of trail mininuclei at $T>363$ K: ${(Na/K)}_{26T}/\left[{(Na/K)}_c(f_{min}+1)\right]=(56\pm 34)$% on 26 October and ${(Na/K)}_{27T}/\left[{(Na/K)}_c(f_{min}+1)\right]=(68\pm 32)$% on 27 October. For the main nucleus, the highest possible area fraction at $T>363$ K is ${(Na/K)}_N/\left[{(Na/K)}_c(f_{min}+1)\right]=(61\pm 24)$%, consistent with the expected source of carbon dioxide, i.e., the water-poor nucleus matrix covering about 90% of 67P’s nucleus surface [13,14]. The trail mininuclei may have very different shape and temperature distribution, i.e., more variable $Na/K$ ratios.

From 8 to 11 October 2024, the Solar and Heliospheric Observatory (SOHO) spacecraft observed C/2023 A3 at an angle of 3–8 deg between the comet orbital plane and the line of sight (versus 63 deg of C/2024 G3 in mid January 2025), preventing the detection of any sodium trail because projected on a much brighter dust tail (Figure 6). Sodium emission may be the only way to detect trails of Oort cloud comets. Alkali-phenoxide carbonylation requires production of carbon dioxide, which occurs in the water-poor matrix of nuclei, according to the WEB model [8,14]. At $r<4$ au, such a matrix does not eject dust, consistent with the SOHO observations of the C/2024 G3 trail [8]. These SOHO observations are inconsistent with usual water-driven dust ejection: the dust continuum of, e.g., C/2023 A3 overcomes the sodium brightness at bandwidths larger than 1 nm (Figure 4), so that the dust trail should be visible at all wavelengths. It follows that the brightness slope of the dust coma of C/2023 A3 and C/2024 G3 is not affected by the presence of trail mininuclei. Degassing without any dust ejection has been inferred from non-gravitational perturbations of the orbits of 1I/’Oumuamua and of other dark comets [24,25]. Also, a direct detection of trail mininuclei of Oort cloud comets is challenging. Their distance from the main nucleus is inversely proportional to their size [8], so that the trail mininuclei’s brightness is inversely proportional to the squared distance from the main nucleus, i.e., much steeper than the brightness slope of a steady dust coma. Since the main nucleus is hidden by the much brighter inner dust coma, this occurs even more for the trail mininuclei. The biggest mininuclei, closest to the main nucleus, might be affected by significant rocket effects due to anisotropic ejection of carbon dioxide spreading them in a cloud surrounding the main nucleus, i.e., a possible additional risk for the ESA Comet Interceptor Mission [26].

As in all previous spectra of comets, lithium was not detected, because the expected ratio was $Na/Li>{(Na/Li)}_c{f}_{min}=6\times {10}^3$ in case of lithium photodesorption, and $Na/Li>{10}^7$ in case of lithium-phenoxide carbonylation [8], both higher than the measured lower limit (Table 2), which was improved by averaging the values measured on 26 and 27 October. The lower limits measured on the trail were lower than ${(Na/Li)}_c$. The measured lower limit $Na/Li>3.4\times {10}^4$ of comet C/2020 F3 Neowise [8] is still consistent with lithium photodesorption if $f_{min}>33$ during those observations. In Table 3, the values of C/1965 S1 may be affected by the saturation of the NaI lines in the photographic spectra, thus of very uncertain intensity because extrapolated. The decrease in $Na/K$ of the other comets towards chondritic values as r decreases is consistent with an increasing efficiency of potassium-phenoxide carbonylation as the nucleus temperature increases. The release of potassium should also build up an atomic tail, providing information on the distribution of this cometary component.

4. Conclusions

• We detected NaI and KI emission lines in very high resolution LBT/PEPSI spectra measured on the nucleus and on the trail of C/2023 A3 (Tsuchinshan-ATLAS).
• The C/2023 A3 nucleus was too cold to allow potassium-phenoxide carbonylation but warm enough to allow sodium-phenoxide carbonylation.
• The KI line was a factor of >300 fainter than all previous detections, consistent with potassium ejection by photodesorption.
• We suggested trail mininuclei as the most probable source of alkali observed outside the nucleus. In this case, the ratios of potassium KI lines intensities allowed us to estimate the sizes of the trail mininuclei ejecting alkali atoms. These sizes were consistent with the production of carbon dioxide reacting with sodium phenoxides.
• The $Na/K$ ratios measured on the nucleus were a factor of 3.6 higher than the chondritic value, and even higher on the trail, thus excluding photodesorption and thermal desorption as the main sodium source.
• A chondritic $Na/K$ ratio in the nuclei of C/2023 A3 is consistent with all the performed measurements if the sodium-phenoxide carbonylation is at least a factor of six more efficient than sodium photodesorption in ejecting neutral sodium atoms.
• As predicted by the alkali-phenoxide carbonylation model, lithium emission LiI was not detected in the spectra of C/2023 A3.
• A trail of mininuclei big enough to allow alkali-phenoxide carbonylation may be a common feature of Oort cloud comets.