Introduction

Since the identification of the simplest aldehyde—formaldehyde (H2CO, 1)—in the interstellar medium (ISM) by Snyder et al. more than half a century ago (1969)1, aldehydes (RCHO), with R being organic groups, have received extensive attention from the astronomy2,3,4, astrobiology5,6, astrochemistry7,8,9, and physical organic chemistry communities10,11,12 due to their role as crucial intermediates in metabolic pathways13 and in molecular mass growth processes to vital biomolecules necessary for the origins of life8,14,15. Although the deep ocean hydrothermal vents are considered a likely scenario for the emergence of life16,17,18, a substantial fraction of the prebiotic organic molecules on proto-Earth may have been of extraterrestrial origin19. In prebiotic chemistry, 1 can be converted to glycolaldehyde (HOCH2CHO, 2) in the formose or Butlerov reaction14,20 thus serving as a starting material for the synthesis of complex sugars21,22. Mediated via quantum tunneling, acetaldehyde (CH3CHO, 3) can react with methanol (CH3OH, 4) to form the hemiacetal 1-methoxyethanol (CH3OCH(OH)CH3), a precursor to sugars and sugar-related molecules11. Triggered by energetic radiation, 3 reacts with carbon dioxide (CO2) to form biorelevant pyruvic acid (CH3COCOOH, 5), which is a vital molecule for metabolism processes in modern biochemistry23. Besides the nine aldehydes identified in the ISM (Fig. 1)24, sixteen aldehydes including 1, 3, and propionaldehyde (CH3CH2CHO, 6) have been detected in carbonaceous chondrites (Fig. 1)9. This indicates that aldehydes can not only be synthesized in deep space, but also survive the entrance of the parent bodies of the meteorites into the atmosphere of the early Earth25; this process could have provided an exogenous source of prebiotic molecules for the early evolution of life. Although aldehydes are of particular significance to the synthesis of astrobiologically relevant molecules, the fundamental formation routes of these molecules in the interstellar environment have remained largely elusive, especially the abiotic synthesis of the biomolecule lactaldehyde (CH3CH(OH)CHO, 7).

Fig. 1: Aldehydes identified in the interstellar medium (bold) and carbonaceous chondrites (italics).
figure 1

The atoms are color-coded in white (hydrogen), gray (carbon), red (oxygen), and blue (nitrogen).

In prebiotic chemistry, 7 can form methylglyoxal (CH3COCHO, 8) through oxidation and further yield 5—a molecular building block of metabolites and amino acids (Fig. 2). Through nucleophilic addition, 7 reacts with hydrogen cyanide (HCN) and ammonia (NH3) to produce 2,3-dihydroxybutanenitrile (CH3CH(OH)CH(OH)CN, 9) and 1-amino-1,2-propanediol (CH3CH(OH)CH(OH)NH2, 10), respectively, which are molecular precursors to the proteinogenic amino acid threonine (NH2CH(CH(OH)CH3)COOH). Oxidation of 7 results in the formation of biomolecule lactic acid (CH3CH(OH)COOH, 11)26, eventually contributing to the formation of sugar acids such as glyceric acid (HOCH2CH(OH)COOH, 12)27. Molecule 7 can be reduced to produce 1,2-propanediol (CH3CH(OH)CH2OH, 13). Undergoing carbon–oxygen bond cleavage, 7 can be converted into 6, which can react with two 1 molecules to form trimethylolethane (CH3C(CH2OH)3) through condensation reactions28. Further, the cleavage of the carbon–carbon bond in 7 prepares 2, which forms the simplest sugar molecule glyceraldehyde (HOCH2CH(OH)CHO, 14)12 via aldol condensation. Consequently, 7 serves as a fundamental precursor to important biomolecules such as amino acids, sugars, and sugar acids (Fig. 2), which could have seeded the emergence of life on early Earth29. An elucidation of the interstellar formation of 7 is therefore of crucial importance to unraveling the synthesis routes of astrobiologically relevant molecules in deep space.

Fig. 2: Formation of lactaldehyde in interstellar ices and its role as a molecular building block of biorelevant molecules in the methylglyoxal cycle.
figure 2

Lactaldehyde (7) is prepared in interstellar ice analogs composed of carbon monoxide and ethanol (15) through energetic processing by galactic cosmic ray proxies. This process involves carbon–carbon bond coupling via radical–radical recombination of the formyl (HĊO, 16) with the 1-hydroxyethyl (CH3ĊHOH, 17) radical. Lactaldehyde (7) serves as a precursor to sugar-related molecules such as glycolaldehyde (2) and lactic acid (11) thus contributing to the synthesis of sugars and sugar acids, respectively. In contemporary biochemistry, seven is a key intermediate in the methylglyoxal pathway (purple arrows) and a precursor to the formation of critical biorelevant molecules including pyruvic acid (5), methylglyoxal (8), and lactic acid (11), linking to the tricarboxylic acid (TCA) cycle.

In this work, we demonstrate the bottom-up formation of 7 in interstellar ice analogs upon exposure to energetic irradiation in the form of proxies of galactic cosmic rays (GCRs). This is accomplished in low-temperature (5 K) carbon monoxide (CO)–ethanol (CH3CH2OH, 15) ice mixtures through the barrierless radical–radical recombination of formyl (HĊO, 16) with 1-hydroxyethyl (CH3ĊHOH, 17) radicals (Figs. 2 and 3). Combining VUV photoionization reflectron time-of-flight mass spectrometry (PI-ReToF-MS) and isotopic substitution studies, 7 along with its isomers 3-hydroxypropanal (HOCH2CH2CHO, 18), ethyl formate (CH3CH2OCHO, 19), and 1,3-propenediol (HOCH2CHCHOH, 20) were identified in the gas phase during the temperature-programmed desorption (TPD) of the irradiated ice mixtures. As one of the most commonly detected molecules in interstellar ices, carbon monoxide was found with a fractional abundance of up to 55% with respect to water toward IRAS 08375–410930. Ethanol (15) is abundant in the ISM and has been tentatively identified in the ices towards background stars31 and young protostars with an abundance of up to 1.8% with respect to water32. Therefore, 7 and its isomers 1820 form in interstellar ice composed of carbon monoxide and 15. Once synthesized, these organics can be incorporated into planetesimals and may have eventually been delivered to planets such as early Earth through meteoritic impacts33, providing an exogenous source for the formation of essential biorelevant molecules and thus helping to decipher the enigma of the molecular origins of life.

Fig. 3: Reaction scheme leading to five C3H6O2 (m/z = 74) isomers in irradiated carbon monoxide–ethanol ices.
figure 3

Barrierless radical–radical reactions of 16 with 17, 22, and 23 produce 7, 18, and 19, respectively; tautomerization of 7 and 18 may lead to the enols 21 and 20 (top). The computed relative energies (∆E) of radicals47 and products are shown as ranges containing all conformers. The bottom figure compiles the computed adiabatic ionization energies (IEs) of isomers (black solid line) and ranges of their conformers (gray area) after error analysis (Supplementary Tables 712). Five VUV photon energies (dashed lines) were used to photoionize sublimed molecules during TPD.

Results

Infrared spectroscopy

Fourier transform infrared (FTIR) spectroscopy was utilized to monitor the CO–CH3CH2OH ice mixture and its isotopically labeled system (CO–CD3CD2OD) before and after low dose (23 nA, 5 min) and high dose (123 nA, 10 min) irradiation at 5 K (Supplementary Figs. 14). Detailed assignments of FTIR spectra are compiled in Supplementary Tables 14. Prior to the electron irradiation, the absorptions are attributed to the fundamentals and combination modes of the reactants; the prominent absorptions in unprocessed CO–CH3CH2OH ice include the CO stretching (CO, 2136 cm−1; 13CO, 2090 cm−1) and overtone (4249 cm−1) modes for carbon monoxide34,35, and the broad O–H stretching mode (3050–3550 cm–1), asymmetric C–H stretching mode (2977 cm–1), symmetric C–H stretching modes (2935 cm–1, 2900 cm–1, and 2876 cm–1), and the C–O stretching mode (1052 cm–1) for ethanol36. After the irradiation, several absorption features emerged for the CO–CH3CH2OH ice mixtures (Supplementary Figs. 12). The absorption band at 2342 cm–1 is assigned to the C=O stretching of carbon dioxide (ν3)34. The absorptions at 1853 cm–1, 1843 cm–1, 1726 cm–1, and 1713 cm–1 are linked to 16 (HĊO, ν3), trans-hydroxycarbonyl (HOĊO, ν2), 1 (H2CO, ν2), and 3 (CH3CHO, ν4), respectively34,37,38; these absorptions are shifted to 1796 cm–1 for 16-d1 (DĊO, ν3), 1780 cm–1 for trans-hydroxycarbonyl-d1 (DOĊO, ν2), 1695 cm–1 for 1-d2 (D2CO, ν2), and 1715 cm–1 for 3-d4 (CD3CDO, ν4) in irradiated CO–CD3CD2OD ice mixtures (Supplementary Figs. 3 and 4)10,34,37. The assignments to these absorptions were further confirmed in irradiated 13CO–13CH13CH2OH ice and C18O–CHCH2OH ice (Supplementary Figs. 5 and 6 and Supplementary Tables 5 and 6)34,37,39,40,41. It is worth noting that the absorptions at 1431 cm–1 and 1352 cm–1 observed in high-dose irradiated CO–CH3CH2OH ice (Supplementary Fig. 2) can be tentatively associated with the 1-hydroxyethyl (17) radical based on calculated frequencies at 1426 cm–1 (ν7) and 1344 cm–1 (ν9) of 17 at the CCSD(T)/cc-pVTZ level of theory42. Due to the limited molecular mobility at 5 K, these radicals are preserved within the ice; radicals may not be able to react if they form without nearby radicals43. The absorptions at 1681 cm–1 and 1580 cm–1 in irradiated CO–CD3CD2OD ices may link to one or more carbonyl (C=O) containing species such as 7, 18, and 19. Due to the overlapping absorption features of the formed complex organics during radiation processing, FTIR spectra cannot uniquely detect complex compounds such as 7 and its isomers alone, highlighting that an alternative experimental technique is needed to identify individual reaction products.

Mass spectrometry

The photoionization reflectron time-of-flight mass spectrometry (PI-ReToF-MS) technique is exploited here to identify the reaction products isomer-specifically based on their desorption temperatures and adiabatic ionization energies (IEs)10,44. The PI-ReToF mass spectra of the irradiated carbon monoxide−ethanol (15) ices during TPD are compiled in Figs. 4 to 6 and Supplementary Fig. 7. Focusing on C3H6O2 isomers, four photon energies at 11.10 eV, 10.23 eV, 9.71 eV, and 9.29 eV were selected to distinguish isomers 7, 18, and 19 formed via radical–radical recombination after the low dose irradiation (Fig. 3). At 11.10 eV, the TPD profile of ions at m/z = 74 for CO–CH3CH2OH ice was deconvoluted by fitting to four split Pearson VII distributions peaking at 118 K, 144 K, 178 K, and 222 K, respectively (Fig. 4b). Given the molecular weights of the reactants in CO–CH3CH2OH ice, the ion signal at mass-to-charge (m/z) of 74 can belong to organic compounds with formulae including C6H2, C4H10O, C3H6O2, and/or C2H2O3. Hence it is imperative to confirm the molecular formula using isotopically labeled precursors. Utilizing the fully deuterated ices with CO–CD3CD2OD ice, the TPD profile of m/z = 80 in irradiated CO–CD3CD2OD ice matches well with that of m/z = 74 in irradiated CO–CH3CH2OH ice (Fig. 5a), indicating the presence of exactly six hydrogen atoms. In addition, the TPD profile of m/z = 74 in irradiated CO–CH3CH2OH ice was found to undergo an isotopic mass shift to m/z = 77 in fully carbon-13 isotopically labeled ice (13CO–13CH313CH2OH), confirming the inclusion of exactly three carbon atoms. Furthermore, the TPD profile of m/z = 74 in CO–CH3CH2OH ice shifts two atomic mass units (amu) to m/z = 76 in C18O–CH3CH2OH ice (Supplementary Fig. 8), indicating the presence of at least one oxygen atom. These findings validate the assignment of the reaction products of the molecular formula C3H6O2 for the TPD profile (Supplementary Note 1).

Fig. 4: PI-ReToF-MS data during TPD of carbon monoxide–ethanol ices with low dose (23 nA, 5 min) irradiation.
figure 4

Data were recorded for the unirradiated (blank) CO–CH3CH2OH ice at 11.10 eV, the irradiated CO–CH3CH2OH ice at 11.10 eV, 10.23 eV, 9.71 eV, and 9.29 eV, and the irradiated CO–CD3CD2OD ice at 11.10 eV (a). TPD profiles of m/z = 74 in CO–CH3CH2OH ice were measured at 11.10 eV (b), 10.23 eV (c), 9.71 eV (d), and 9.29 eV (e). The solid magenta lines indicate the total fit of the spectra. At 11.10 eV, the TPD profile can fit with four Peaks I–IV. The red and purple shaded regions indicate the peak positions corresponding to lactaldehyde (7) and 3-hydroxypropanal (18), respectively.

Fig. 5: Ion signal during TPD of isotopically labeled carbon monoxide–ethanol ices as a function of temperature.
figure 5

TPD profiles were measured at 11.10 eV with low dose irradiation (a) and at 9.29 eV with high dose irradiation (b).

Fig. 6: PI-ReToF-MS data during the TPD of carbon monoxide–ethanol ices with higher dose (123 nA, 10 min) irradiation.
figure 6

Data were recorded for the irradiated CO–CH3CH2OH ice at 11.10 eV, 9.29 eV, and 8.25 eV, and the irradiated CO–CD3CD2OD ice at 9.29 eV (a). TPD profiles of m/z = 74 in CO–CH3CH2OH ice were measured at 11.10 eV (b), 9.29 eV, and 8.25 eV (d). TPD profile of m/z = 74 in a blank experiment with 1% ethyl formate (19) was recorded at 11.10 eV (c). The solid magenta line indicates the total fit of the spectra. The green and blue shaded regions indicate the peak positions corresponding to 19 and 1,3-propenediol (20), respectively.

As previously mentioned, the TPD profile of m/z = 74 (C3H6O2+) at 11.10 eV in the low dose irradiated CO−CH3CH2OH ice (Fig. 4b) reveals peaks at 118 K (Peak I), 144 K (Peak II), 178 K (Peak III), and 222 K (Peak IV). Note that the peak sublimation temperatures of acetaldehyde (3) at m/z = 44 and ethanol (15) at m/z = 46 are at 118 K10 and 149 K, respectively (Supplementary Fig. 9), indicating that Peaks I and II are due to the co-sublimation of acetaldehyde (3) and 15, respectively. A blank experiment without electron irradiation of the ice was carried out under identical conditions (Fig. 4b); no sublimation event was detected, confirming that Peaks I–IV is caused by the electron irradiation of the ice. At 11.10 eV, all isomers 7 (IE = 9.38–9.70 eV), 18 (IE = 9.73–10.10 eV), 19 (IE = 10.46–10.69 eV), 20 (IE = 8.45–9.20 eV), and 1,2-propenediol (HOCHCH(OH)CH3, 21; IE = 7.76–8.10 eV) can be photoionized (Fig. 3 and Supplementary Tables 712). Therefore, these peaks can be associated with any isomers 7, 1821. Thereafter, the photon energy was reduced to 10.23 eV, at which isomer 19 (IE = 10.46–10.69 eV) cannot be ionized. At 10.23 eV, ion signals of Peaks I–IV remain (Fig. 4c), indicating that 19 is not present in detectable quantities and hence not formed. Upon reducing the photon energy further to 9.71 eV, at which isomers 7 (IE = 9.38–9.70 eV), 20 (IE = 8.45–9.20 eV), and 21 (IE = 7.76–8.10 eV) can be ionized but not isomer 18 (IE = 9.73–10.10 eV), Peak III at 178 K vanishes (Fig. 4d). Therefore, Peak III can be associated with 18. Further lowering the photon energy to 9.29 eV, at which enols 20 and 21 can be ionized but not isomer 7, no sublimation events were observed (Fig. 4e), suggesting that the ion signals of Peaks I, II, and IV recorded at 9.71 eV can be linked to 7; no evidence for 20 and 21 can be provided in low dose irradiation experiments. In summary, the low-dose irradiation experiments revealed the formation of isomers 7 and 18.

To further probe the synthesis of isomers 1921, high-dose experiments were carried out (Fig. 6). Compared with the results in low-dose irradiated CO–CH3CH2OH ice at 11.10 eV, the TPD profile of m/z = 74 shows a sublimation event peaking at 128 K (Peak V) (Fig. 6b). To assist in the identification of isomer 19, a calibration experiment without irradiation (blank) was performed at 11.10 eV by adding 1% of 19 (IE = 10.53–10.63 eV)45 into the reactants under identical experimental conditions. The TPD profile of 19 at m/z = 74 shows two sublimation events peaking at 131 K and 149 K (Fig. 6c); the latter peak is caused by the co-sublimation with ethanol (15). The first sublimation event (131 K) of 19 matches Peak V (128 K), indicating that Peak V can be associated with 19. Upon reducing the photon energy to 9.29 eV, the TPD profile of m/z = 74 (Fig. 6d) shows a sublimation event that starts at 160 K, peaks at 197 K, and returns to the baseline level at 220 K (Peak VI). Recall that no sublimation event was observed at 9.29 eV in low dose experiment, it is necessary to confirm the formula of this ion signal using isotopically labeled ice. The substitution of CH3CH2OH by CD3CD2OD results in products with six deuterium atoms that can be observed at m/z = 80 in the CO–CD3CD2OD ice (Fig. 5b). The first event peaking at 121 K in the TPD profile of m/z = 80 is linked to C4H8O isomers (Supplementary Fig. 10). By matching the TPD profiles for the deuterated molecules in irradiated CO–CD3CD2OD ice (C3D6O2+, m/z = 80), the assignment of Peak VI can be clearly connected with C3H6O2 isomers. Since only enols 20 (IE = 8.45–9.20 eV) and 21 (IE = 7.76–8.10 eV) can be ionized at 9.29 eV, Peak VI can be linked to 20 and/or 21. To further identify 20 and 21, we lowered the photon energy to 8.25 eV, at which 20 cannot be ionized. In comparison to the result at 9.29 eV, Peak VI vanishes at 8.25 eV (Fig. 6d), suggesting that this sublimation event peaking at 197 K must be linked to 20; no evidence for 21 could be provided. Overall, the low-dose irradiation experiments revealed the formation of isomers 7 and 18, whereas the high-dose studies further revealed the formation of 19 and enol 20.

Discussion

Having provided compelling evidence on the formation of isomers 7 and 1820, we now focus on their potential formation mechanisms. First, upon interaction with GCRs, the unimolecular decomposition of ethanol (15) can lead to the formation of atomic hydrogen (Ḣ) plus 1-hydroxyethyl (17), 2-hydroxyethyl (22), and/or ethoxy (23) radicals46,47. Reactions [1–3] are endoergic by 390 kJ mol1 to 437 kJ mol−148,49 compensated by the energy from the impinging electrons. Recall that 17 was tentatively assigned via the absorptions at 1431 cm−1 (ν7) and 1352 cm–1 (ν9) in the irradiated CO–CH3CH2OH ice. Reactions [1–3] closely resemble the decomposition of methanol (CH3OH) forming the hydroxymethyl (ĊH2OH) and methoxy (CH3Ȯ) radicals via hydrogen–carbon and –oxygen bond cleavages10,50. The hydrogen atoms can then be added to the carbon monoxide, forming the formyl (16) via reaction [4] with a reaction exoergicity of 61 kJ mol−1. Previous work by Bennett et al. suggested an entrance barrier for the reaction [4] to be 11 kJ mol−1 (0.114 eV)51. This entrance barrier can be overcome by the suprathermal hydrogen atoms having excess kinetic energies of a few eV leading to the formation of 16 as identified via FTIR spectroscopy at 1853 cm−1 (HĊO, ν3)37,38 in irradiated CO–CH3CH2OH ice and at 1796 cm−1 for 16-d1 (DĊO, ν3) in irradiated CO–CD3CD2OD ice37.

$$\begin{array}{cc}{{{{\rm{CH}}}}}_{3}{{{{\rm{CH}}}}}_{2}{{{\rm{OH}}}}({{{\bf{15}}}})\to {{{{\rm{CH}}}}}_{3}\dot{{{{\rm{C}}}}}{{{\rm{HOH}}}}({{{\bf{17}}}})+\dot{{{{\rm{H}}}}} & (+390\,{{{\rm{kJ}}}}\,{{{{\rm{mol}}}}}^{-1})\end{array}$$
(1)
$$\begin{array}{cc}{{{{\rm{CH}}}}}_{3}{{{{\rm{CH}}}}}_{2}{{{\rm{OH}}}}({{{\bf{15}}}})\to {\dot{{{{\rm{C}}}}}{{{\rm{H}}}}}_{2}{{{{\rm{CH}}}}}_{2}{{{\rm{OH}}}}({{{\bf{22}}}})+\dot{{{{\rm{H}}}}} & (+422\,{{{{\rm{kJ}}}}\;{{{\rm{mol}}}}}^{-1})\end{array}$$
(2)
$$\begin{array}{cc}{{{{\rm{CH}}}}}_{3}{{{{\rm{CH}}}}}_{2}{{{\rm{OH}}}}({{{\bf{15}}}})\to {{{{\rm{CH}}}}}_{3}{{{{\rm{CH}}}}}_{2}\dot{{{{\rm{O}}}}}({{{\bf{23}}}})+\dot{{{{\rm{H}}}}} & (+437\,{{{{\rm{kJ}}}}\;{{{\rm{mol}}}}}^{-1})\end{array}$$
(3)
$$\begin{array}{cc}\dot{{{{\rm{H}}}}}+{{{\rm{CO}}}}\to {{{\rm{H}}}}\dot{{{{\rm{C}}}}}{{{\rm{O}}}}({{{\bf{16}}}}) & (-61\,{{{\rm{kJ}}}}\,{{{{\rm{mol}}}}}^{-1})\end{array}$$
(4)

Second, 7, 18, and 19 form via barrierless radical–radical recombination via reactions [5–7]. These reactions are exoergic by 334 kJ mol1 to 420 kJ mol−148,49. Due to the limited molecular mobility caused by the low temperatures of 5 K, these radicals are preserved within the ice; radicals may not be able to react if they form without nearby radicals43. This is especially relevant for relatively large radicals such as 17, 22, and 23, indicating that the formation of lactaldehyde and its isomers is likely to proceed via non-diffusive radical recombination mechanisms52. However, at higher temperatures such as 20 K, the diffusive mechanism may compete with the non-diffusive pathways as radicals such as HĊO become more mobile52,53. Recall that the TPD profile of m/z = 74 (C3H6O2+) in CO−CH3CH2OH ice shifts 2 amu to m/z = 76 (C3H6O18O+) in C18O–CH3CH2OH ice, indicating one carbon monoxide and one 15 molecules are involved in the formation of the products. Recent studies by Zasimov et al. revealed the predominant formation of 1-hydroxyethyl (17) resulting from the X-ray irradiation of matrix-isolated 15 molecules43,44. Although indirect evidence was provided for the primary formation of ethoxy (23), the isomerization of 23 to form 17 is rapid even at 7 K due to tunneling46. This agrees with our experimental results suggesting that 19 resulting from the recombination of 16 with 23 remained undetectable at low dose studies and can only be detected in high-dose irradiation experiments.

$$\begin{array}{cc}{{{\rm{H}}}}\dot{{{{\rm{C}}}}}{{{\rm{O}}}}({{{\bf{16}}}})+{{{{\rm{CH}}}}}_{3}\dot{{{{\rm{C}}}}}{{{\rm{HOH}}}}({{{\bf{17}}}})\to {{{{\rm{CH}}}}}_{3}{{{\rm{CH}}}}({{{\rm{OH}}}}){{{\rm{CHO}}}}\,({{{\bf{7}}}}) & (-334\,{{{\rm{kJ}}}}\,{{{{\rm{mol}}}}}^{-1})\end{array}$$
(5)
$$\begin{array}{cc}{{{\rm{H}}}}\dot{{{{\rm{C}}}}}{{{\rm{O}}}}({{{\bf{16}}}})+{\dot{{{{\rm{C}}}}}{{{\rm{H}}}}}_{2}{{{{\rm{CH}}}}}_{2}{{{\rm{OH}}}}({{{\bf{22}}}})\to {{{{\rm{HOCH}}}}}_{2}{{{{\rm{CH}}}}}_{2}{{{\rm{CHO}}}}\,({{{\bf{18}}}}) & (-354\,{{{\rm{kJ}}}}\,{{{{\rm{mol}}}}}^{-1})\end{array}$$
(6)
$$\begin{array}{cc}{{{\rm{H}}}}\dot{{{{\rm{C}}}}}{{{\rm{O}}}}({{{\bf{16}}}})+{{{{\rm{CH}}}}}_{3}{{{{\rm{CH}}}}}_{2}\dot{{{{\rm{O}}}}}({{{\bf{23}}}})\to {{{{\rm{CH}}}}}_{3}{{{{\rm{CH}}}}}_{2}{{{\rm{OCHO}}}}\,({{{\bf{19}}}}) & (-420\,{{{\rm{kJ}}}}\,{{{{\rm{mol}}}}}^{-1})\end{array}$$
(7)

Third, through hydrogen migrations, 18 can tautomerize to enol 20 with a reaction endoergicity of 37 kJ mol−1 via reaction [8]48,49. The barrier of keto-enol tautomerization can be overcome by the energy contributed by GCR proxies35. Although 7 may tautomerize to 21, no evidence of 21 was observed in our experiments. Previous studies revealed that the interconversion of 7 can bypass the enol 21 and access its more stable ketone isomer hydroxyacetone (CH3C(O)CH2OH, 24)54. However, the identification of 24 is inconclusive under current experimental conditions (Supplementary Note 2).

$$\begin{array}{cc}{{{{\rm{HOCH}}}}}_{2}{{{{\rm{CH}}}}}_{2}{{{\rm{CHO}}}}({{{\bf{18}}}})\leftrightharpoons {{{{\rm{HOCH}}}}}_{2}{{{\rm{CHCHOH}}}}({{{\bf{20}}}})\to & (-37\,{{{\rm{kJ}}}}\,{{{{\rm{mol}}}}}^{-1})\end{array}$$
(8)

Altogether, we present the bottom-up formation pathways of lactaldehyde (7) and its isomers (1820) in low-temperature carbon monoxide–ethanol ice mixtures upon exposure to energetic electrons, which simulate secondary electrons produced by GCRs as they penetrate ices within a cold molecular cloud aged up to seven million years55. These molecules were identified in the gas phase during the TPD phase utilizing photoionization reflectron time-of-flight mass spectrometry (PI-ReToF-MS) along with isotopic substitution experiments. The CO–CH3CH2OH ices selected in our laboratory simulations present model ice to understand the formation pathways of lactaldehyde and its isomers in a comprehensive way. Future experiments can explore the effects of ice composition on these molecules by incorporating other simple molecules common to interstellar ice such as water (H2O), carbon dioxide (CO2), and methanol (CH3OH) into the ice mixture. Isomers 7, 18, and 19 were formed via radical–radical recombination of the formyl (HĊO, 14) with the 1-hydroxyethyl (17), 2-hydroxyethyl (ĊH2CH2OH, 22), and ethoxy (CH3CH2Ȯ, 23) radicals, respectively; enol 20 was accessed through keto-enol tautomerization of 18. These findings provide fundamental steps toward the understanding of the fundamental formation mechanisms of complex aldehydes (RCHO) and their enols (RC = C(OH)H) under astrophysical conditions.

Within cold molecular clouds, galactic cosmic rays can trigger non-equilibrium reactions in the interstellar ice that are composed of simple molecules such as water (H2O), carbon dioxide (CO2), carbon monoxide (CO), methanol (CH3OH), and ammonia (NH3)55,56. Carbon monoxide and ethanol (15) are abundant in the interstellar environment24. In interstellar ices, carbon monoxide has a fractional abundance of up to 55% with respect to water30, and 15 has been tentatively identified with an abundance of up to 1.8% with respect to water32. Therefore, through the feasible formation mechanisms as demonstrated here, 7 and its isomers 1820 can form abiotically in interstellar ices containing carbon monoxide and 15 upon interaction with ionizing radiation. Once formed, these molecules can sublime into the gas phase in the hot core stage, representing promising candidates for future astronomical searches via radio telescopes such as the Atacama Large Millimeter Array (ALMA). In fact, isomer 18 was identified toward Sagittarius B2(N)57 and Orion58—two star-forming regions. Laboratory simulation experiments by Marcellus et al. revealed the identification of 7 in the simulated pre-cometary organic residues after the VUV irradiation of an ice mixture containing water, methanol, and ammonia59. Although 7 is one of the simplest chiral molecules that could reasonably exist in deep space, it has not yet been detected24. Recent work by Alonso et al. searched for 7 in three high-mass star-forming regions and reported an upper limit of fractional abundance of 1 × 10−7 with respect to hydrogen (H2)4.

In biochemistry, lactaldehyde (7) serves as a key intermediate in the methylglyoxal pathway, in which glucose (C6H12O6) can be converted into pyruvate60. The methylglyoxal pathway is strictly linked to glycolysis61,62, a contemporary biochemical process vital to cellular metabolism. After being formed from methylglyoxal (8) via methylglyoxal reductase63, 7 can be converted into lactic acid (11) through aldehyde dehydrogenase64 and then leads to the formation of 5 via lactate dehydrogenase (Fig. 2)63, which along with its deprotonated pyruvate anion (CH3COCOO) are critical molecules linked to the tricarboxylic acid (TCA) cycle65. In addition, 7 can be reduced to form 1,2-propanediol (13) and acts as an intermediate in the 1,2-propanediol synthesis pathway66, contributing to the carbon metabolism13. Our laboratory experiments revealed that biorelevant aldehydes such as 7 can form through radical–radical recombination from carbon monoxide and alcohols on interstellar ice grains, serving as fundamental precursors to important biomolecules such as sugars, sugar acids, and amino acids. 7 could produce the sugar-related molecule 2, which can be formed via the recombination of 16 and hydroxymethyl radicals within ices of 4 and carbon monoxide—4 upon exposure to ionizing radiation67,68. Recent studies revealed that the thermal reaction of 3 with ammonia yields 1-aminoethanol (CH3CH(OH)NH2) at a low temperature of 65 K, contributing to the synthesis of prebiotic chelating agents5. Similarly, 7 could react with ammonia through a nucleophilic reaction to form 1-amino-1,2-propanediol (12), a precursor to the proteinogenic amino acid threonine. Once formed within cold molecular clouds, these molecules can be eventually incorporated into planetoids, asteroids, and comets23,69, and ultimately delivered to planets like the early Earth, providing an exogenous source of prebiotic molecules25. In fact, extraterrestrial aldehydes, sugars, and amino acids have been detected in carbonaceous chondrites9,33,70. Under prebiotic conditions, the presence of 7 can contribute to the synthesis of biomolecules such as 5 and 9 (Fig. 2), thus critically advancing our fundamental knowledge of the synthesis routes to critical biorelevant molecules in deep space and on prebiotic Earth.

Methods

Experimental

All experiments were conducted in a stainless steel ultrahigh vacuum (UHV) chamber maintained at pressures of 5 × 10−11 Torr by magnetically levitated turbomolecular pumps (Osaka, TG1300MUCWB, TG420MCAB) backed by a hydrocarbon-free dry scroll pump (XDS35i, BOC Edwards)27. A polished silver substrate (12.6 × 15.1 mm2) was interfaced to a cold head that was cooled to 5 K by a two-stage closed-cycle helium refrigerator (Sumitomo Heavy Industries, RDK-415E). The cold head can rotate freely and translate vertically through a doubly differentially pumped rotational feedthrough (Thermionics Vacuum Products, RNN-600/FA/ MCO) and an adjustable bellows (McAllister, BLT86). Ethanol (C2H5OH; Pharmco, anhydrous, ≥99.5% purity), ethanol-d6 (C2D5OD; Cambridge Isotope Laboratories, anhydrous, 99% atom D), or ethanol-13C2 (Sigma Aldrich, 99 atom% 13C) was filled into a glass vial interfaced to a UHV chamber. The samples were subjected to several freeze-thaw cycles to remove residual atmospheric gases using liquid nitrogen. Carbon monoxide (CO; Sigma Aldrich, >99%), carbon monoxide-13C (13CO; Sigma Aldrich, ≥99 atom% 13C, ≤6 atom% 18O), or carbon monoxide-18O (C18O; Sigma Aldrich, 99.9 atom% 12C, 95 atom% 18O) was premixed with ethanol, ethanol–d6, or ethanol–13C2 vapor to prepare a gas mixture with a 2:1 ratio of carbon monoxide to ethanol. To prepare the ice, the gas mixture was leaked into the main chamber at pressures of 4 × 10−8 Torr via a glass capillary array and deposited onto the silver substrate, which was cooled to 5 K. Although the temperatures of 5 K used in these experiments are slightly lower than that typically found in molecular clouds, intact reactive intermediates can be preserved to provide valuable mechanistic insights in such cold ice. Laser interferometry was used to monitor the ice thickness during the deposition via a photodiode and a helium-neon laser (632.8 nm)71. The average index of 1.26 ± 0.04 was used to derive the thickness of the mixture ice (CO–CH3CH2OH) from the refractive indexes of carbon monoxide ice (n = 1.25 ± 0.03)34 and that of ethanol ice (n = 1.26)72. This average index was used for the isotopically labeled ice mixtures. The ice thicknesses of CO–CH3CH2OH ice were determined to be 880 ± 50 nm. The densities of 0.80 ± 0.01 g cm−3 for CO ice34 and 0.584 g cm−3 for CH3CH2OH ice72 were used. For isotopically labeled ice mixtures, variations in density were considered based on the masses of reactants. A Fourier transform infrared (FTIR) spectrometer (Thermo Electron, Nicolet 6700) measured the ice after deposition in the range of 6000 − 500 cm−1 with a resolution of 4 cm−1. FTIR spectra of pure ethanol, ethanol-d6, and ethanol-13C2 ices were collected at 5 K with thicknesses of 760 ± 50 nm, 810 ± 50 nm, and 450 ± 50 nm, respectively (Supplementary Figs. 1113), which were used to determine the thicknesses and column densities of ethanol, ethanol-d6, and ethanol-13C2 in the mixture ices based on the integrated area of multiple absorption bands. Utilizing the integrated infrared absorptions of carbon monoxide at 2091 cm−1 (ν1, 13CO, 1.32 × 10−17 cm molecule−1) and 4249 cm−1 (2ν1, CO, 1.04 × 10−19 cm molecule−1)34 and the absorption bands of pure ethanol ices with known thickness (Supplementary Figs. 1113 and Supplementary Tables 1315), the ratio of carbon monoxide to ethanol in the ice mixture (CO–CH3CH2OH) was estimated to be 2.5 ± 0.4:1 (Supplementary Table 16). It is necessary to note that the absorption coefficients of carbon monoxide were obtained via transmission absorption IR spectroscopy, which may differ from those obtained using reflection absorption IR spectroscopy73,74. Other factors, such as the thickness of the ice and the angle of incidence of the IR beam, can affect the relative peak heights in reflectance IR spectra. Here, we use the absorption coefficients in reflection as a means to estimate the ratio of components in the ice mixtures. Although the ratio of carbon monoxide to ethanol used in the experiments may not be a typical abundance ratio observed in molecule clouds, this ratio ensures the highest possible yield of C3H6O2 isomers and thus facilitates their detection.

After deposition, the ice mixtures were exposed to 5 keV electrons released from an electron gun (SPECS, EQ PU-22); the electron beam was scanned over an area of 1.6 cm2 for low dose (23 nA, 5 min) and high dose (123 nA, 10 min) irradiations. Prior to irradiation, a phosphor screen was used to monitor and adjust the electron beam, ensuring uniform exposure across the sample substrate. The electron beam current was measured using a Faraday cup before and after irradiation, with fluctuations kept within 3 nA. Based on Monte Carlo simulations carried out with the CASINO software suite75, the high dose irradiation for CO–CH3CH2OH ice corresponds to doses of 1.50 ± 0.25 eV molecule−1 for CO and 3.31 ± 0.54 molecule−1 for CH3CH2OH, respectively (Supplementary Table 16). These doses simulate secondary electrons generated in the track of galactic cosmic rays (GCRs) in cold molecular clouds (Supplementary Note 3) aged up to 7 × 106 years55. The average penetration depth of electrons in CO–CH3CH2OH ice was calculated to be 420 ± 70 nm using CASINO 2.4275; 99% of the electron energy was deposited in the top 710 ± 50 nm sample layers, which is less than the ice thickness (880 ± 50 nm), preventing the interaction between the substrate and electrons. The infrared spectra of ice were measured by the FTIR spectrometer in situ before, during, and after irradiation.

After irradiation, the ices were heated from 5 K to 320 K at  1 K min−1 during the temperature-programmed desorption (TPD) process. Subliming molecules were photoionized by pulsed 30 Hz vacuum ultraviolet (VUV) lights, which were generated through resonant four-wave mixing schemes inside the noble gas jet10. The VUV photons were generated via sum frequency generation (2ω1 + ω2; 11.10 eV) and difference frequency generation (2ω1 − ω2; 10.23 eV, 9.71 eV, 9.29 eV, and 8.25 eV) utilizing two Nd:YAG lasers (Spectra-Physics, Quanta Ray PRO 250-30 and 270-30) and two tunable dye lasers (Sirah Lasertechnik, Cobra-Stretch). Detailed parameters are listed in Supplementary Table 17. The produced VUV light was spatially separated from other laser beams via a biconvex lithium fluoride lens (Korth Kristalle, R1 = R2 = 131 mm) in an off-axis geometry and passed 2 mm above the ice surface to ionize the subliming molecules. The VUV flux was monitored by a Faraday cup during TPD and was used to correct for variations of the TPD profiles throughout each experiment7. The resulting ions from VUV photoionization were mass-analyzed through reflectron time-of-flight mass spectrometry and detected by a dual microchannel plate (MCP) detector (Jordan TOF Products). The ion signals were then amplified with a preamplifier (Ortec, 9305), discriminated, and recorded by a multichannel scaler (FAST ComTec, MCS6A). For each recorded mass spectra, the ion arrival time and the accumulation time of ion signals were 3.2 ns accuracy and 2 min (3600 sweeps), respectively. An additional experiment was carried out without electron irradiation (blank) at 11.10 eV for CO–CH3CH2OH ice, and no sublimation event at m/z = 74 was observed. The gas phase mass spectra were collected for background gases in the main chamber and ethanol samples at 11.10 eV; no contamination molecules were detected (Supplementary Fig. 14). To assist in the identification of specific molecules, blank experiments were performed at 11.10 eV by adding 1% of ethyl formate (C2H5OCHO; Sigma Aldrich, >97%) or 1% of hydroxyacetone (HOCH2C(O)CH3, Thermo Scientific Chemicals, 95%) into the premixed CO–CH3CH2OH gas mixture (Supplementary Table 16).

Computational

Calculations were performed using the CBS–QB3 composite scheme to obtain accurate values for the energies of the neutral and cationic states of each species. This approach allows obtaining molecular parameters and energies with an accuracy of 0.01–0.02 Å for bond lengths, 1–2° for bond angles, and 4–8 kJ mol−1 for relative energies. All ab initio calculations of the electronic structure were performed with the GAUSSIAN 09 package76. Five backbone isomers of reaction products were identified, differing in the position of the methyl, ethyl, carbonyl, and hydroxyl groups, namely lactaldehyde (7), 3-hydroxypropanal (18), ethyl formate (19), 1,3-propenediol (20), and 1,2-propenediol (21). For each backbone isomer, all possible isomers produced by rotation around selected C–O or C–C single and double bonds (subsequently referred to as conformers) were considered to obtain reliable ionization potential data. For this purpose, a computer program was developed to automatically prepare GAUSSIAN 09 input files for all possible conformers. The most energetically favorable product was found to be ethyl formate (19). The calculated conformer energies relative to ethyl formate (19) and adiabatic ionization potentials are shown in Fig. 3 and agree well with previous results (Supplementary Table 18). The same CBS-QB3 method was used for the optimization of transition states (TS), which are involved in the potential energy surfaces leading to 1,2-propenediol (21) and hydroxyacetone (22) from lactaldehyde (7). Initial geometries were chosen to resemble the proposed TS structures, and optimization was run to obtain the TS geometry. After optimization, TS was confirmed by inspection of normal mode eigenvalues and via internal reaction coordinate calculations to yield the correct stable products on either side of the barrier. The Cartesian coordinates, harmonic frequencies, and infrared intensities of the computed structures are available in Supplementary Data 1.