UV Transmission in Natural Waters on Prebiotic Earth

Abstract

Ultraviolet (UV) light plays a key role in surficial theories of the origin of life, and numerous studies have focused on constraining the atmospheric transmission of UV radiation on early Earth. However, the UV transmission of the natural waters in which origins-of-life chemistry (prebiotic chemistry) is postulated to have occurred is poorly constrained. In this work, we combine laboratory and literature-derived absorption spectra of potential aqueous-phase prebiotic UV absorbers with literature estimates of their concentrations on early Earth to constrain the prebiotic UV environment in marine and terrestrial natural waters, and we consider the implications for prebiotic chemistry. We find that prebiotic freshwaters were largely transparent in the UV, contrary to assumptions in some models of prebiotic chemistry. Some waters, such as high-salinity waters like carbonate lakes, may be deficient in shortwave (≤220 nm) UV flux. More dramatically, ferrous waters can be strongly UV-shielded, particularly if the Fe²⁺ forms highly UV-absorbent species such as $Fe{\left(CN\right)}_6^{4-}$. Such waters may be compelling venues for UV-averse origin-of-life scenarios but are unfavorable for some UV-dependent prebiotic chemistries. UV light can trigger photochemistry even if attenuated through photochemical transformations of the absorber (e.g., $e_{aq}^{-}$ production from halide irradiation), which may have both constructive and destructive effects for prebiotic syntheses. Prebiotic chemistries that invoke waters that contain such absorbers must self-consistently account for the chemical effects of these transformations. The speciation and abundance of Fe²⁺ in natural waters on early Earth is a major uncertainty and should be prioritized for further investigation, as it played a major role in UV transmission in prebiotic natural waters.

1. Introduction

Akey challenge for origin-of-life studies is to constrain the range of environmental conditions on early Earth under which life arose. Knowledge of these environmental conditions informs development of theories of the origin of life and enables assessment of the plausibility and probability of postulated prebiotic chemistries (Pace, 1991; Cleaves, 2013; Barge, 2018; Lyons et al., 2020). Consequently, considerable work has been done to constrain the range of physicochemical conditions available in prebiotic environments (Sleep, 2018; Sasselov et al., 2020).

An important prebiotic environmental factor, particularly for surficial prebiotic chemistry, is the ultraviolet (UV) irradiation environment. UV photons can destroy nascent biomolecules, motivating a search for mechanisms to protect prebiotically important molecules from UV irradiation (Sagan, 1973; Holm, 1992; Cleaves and Miller, 1998), but UV light has also been suggested useful to the origin of life as a source of chemical free energy and selectivity (Deamer and Weber, 2010; Pascal, 2012; Beckstead et al., 2016). Indeed, UV light has been experimentally shown to drive a range of prebiotic chemistries (Ferris and Orgel, 1966; Sagan and Khare, 1971; Flores et al., 1977; Pestunova et al., 2005; Bonfio et al., 2017; Mariani et al., 2018; Xu et al., 2020). Surficial prebiotic chemistries, therefore, broadly fall into two classes: those for which UV light is strictly destructive and must be mitigated or avoided, and those for which UV light is essential and must be sought (Ranjan and Sasselov, 2016; Todd et al., 2020). The self-consistent availability of UV radiation is consequently a crucial part of the assessment of the plausibility of proposed prebiotic chemistries.

The pivotal roles that UV radiation plays in diverse prebiotic chemistries have motivated increasingly sophisticated estimates of the prebiotic UV environment, with particular emphasis on atmospheric transmission. Models generally predict abundant steady-state UV at wavelengths $\gtrsim$204 nm and a shortwave cutoff of ≈200 nm driven by atmospheric CO₂ and H₂O, though transient eras of low UV are possible after large volcanic eruptions and an occurrence of impacts (Cockell, 2000; Rugheimer et al., 2015; Ranjan and Sasselov, 2017; Ranjan et al., 2018; Zahnle et al., 2020).

However, prebiotic chemistry is generally proposed to occur in aqueous reservoirs such as ponds or oceans, which may host UV-blocking compounds of geological origin (Martin et al., 2008; McCollom, 2013; Patel et al., 2015; Benner et al., 2019; Becker et al., 2019). To date, estimates of UV transmission in prebiotic natural waters are generally based on pure water or modern pond water (Cockell, 2000; Ranjan and Sasselov, 2016; Pearce et al., 2017).

Neither modern pond water nor pure water is likely representative of prebiotic waters. The UV-opacity of modern pond waters is largely driven by biogenic dissolved carbon species (Morris et al., 1995; Markager and Vincent, 2000; Laurion et al., 2000). In the limit of low biological productivity, as expected in a prebiotic world, natural waters can be clear down to 300 nm; their transmission from 200 to 300 nm is unconstrained but has been extrapolated to be similarly low (Smith and Baker, 1981; Morel et al., 2007).

At the other extreme, although pure water is clear in the near-UV (Quickenden and Irvin, 1980), the waters in which prebiotic chemistry occurred must by definition have been impure, since they must have contained, at minimum, the feedstocks for that chemistry, and likely other geogenic constituents as well (e.g., Toner and Catling, 2020). Absorption due to these constituents may influence UV transmission. The UV transmission of prebiotic waters likely lies, therefore, between the extremes of high transparency and high opacity represented by the previously utilized proxies of pure water and modern pond water.

In the present study, we constrain wavelength-dependent UV transmission in prebiotic waters. We focus on absorption due to a subset of UV-active species that have been proposed to be present in prebiotic waters. We derive molar absorptivities of these compounds from a combination of literature reports and our own measurements. We draw on literature proposals for the concentrations of these species in representative prebiotic waters. We do not attempt self-consistent geochemical modeling of these waters; such work is important but requires improved measurements of the relevant reaction kinetics under prebiotically relevant conditions, which is beyond the scope of the present study. Our work is similar in spirit to the work of Cockell (2000, 2002), who drew on literature proposals for the composition of the early atmosphere to constrain its UV transmission, without attempting to model its self-consistent photochemistry. Their simple initial analysis guided early work (e.g., Pierce et al., 2005; Gómez et al., 2007) and motivated more atmospherically sophisticated follow-up (e.g., Rugheimer et al., 2015), which came to largely similar conclusions. Similarly, we hope our simple study will guide prebiotic chemistry, while motivating more geochemically sophisticated follow-up work.

2. Background

Relatively few studies have constrained the UV transmission of prebiotic natural waters. A notable exception is the work of Cleaves and Miller (1998), who considered potential “sunscreens” for the prebiotic ocean. Cleaves and Miller (1998) measured the UV absorptivities of salts, hydrogen cyanide (HCN) and spark discharge-derived polymers, HS⁻, and marine Fe²⁺. Based on a combination of calculated and literature estimates of the concentrations of these species, Cleaves and Miller (1998) concluded that, in favorable circumstances, HS⁻, marine Fe²⁺, or spark discharge polymer could have extinguished UV in the surface layer of the prebiotic ocean. The conditions required for accumulation of a spark discharge polymer to high concentrations in the prebiotic ocean would have included the emission of prebiotic volcanic carbon as CH₄, its conversion with unity efficiency to a spark discharge polymer, and its efficient deposition to the ocean. However, more recent modeling studies have indicated that volcanogenic carbon is emitted even on early Earth-like planets as CO₂ and not CH₄ and the primary fate of abiotic CH₄ should be oxidation to CO₂ (Kasting, 2014). Such models suggest that the conditions required for accumulation of optically relevant concentrations of spark discharge polymers are unlikely to be met in the steady state, although high hydrocarbon concentrations may be transiently possible after large impacts (Genda et al., 2017; Benner et al., 2019; Zahnle et al., 2020). Similarly, recent studies suggest that the early ocean was ferrugious (Fe²⁺-rich) and would have titrated out HS⁻ as pyrite. Along with the low solubility of HS⁻, [HS⁻] was most likely very low in most natural waters on early Earth (Walker and Brimblecombe, 1985; Poulton and Canfield, 2011; Ranjan et al., 2018). However, elevated [Fe²⁺] was a possibility for the early ocean (Konhauser et al., 2017).

3. Methods

3.1. Calculating aqueous UV attenuation

We approximate the transmission of UV radiation in homogenous (well-mixed), nonscattering aqueous solutions at low concentrations and low light intensities by the Beer-Lambert law (IUPAC, 1997):

$$\begin{array}{c}lo{g}_{10}\left(I\left(\lambda \right)∕I_0\left(\lambda \right)\right)=lo{g}_{10}\left(T\left(\lambda \right)\right)=\\ -\left({\Sigma }_i{\epsilon }_i\left(\lambda \right)c_i\right)d=-\left({\Sigma }_i{a}_i\left(\lambda \right)\right)d,\end{array}$$ (1)

where I₀ is the incident irradiance, I is the transmitted irradiance, T is the fraction of transmitted radiation, ɛ_i is the molar decadic absorption coefficient for the ith component of the solution, c_i is the concentration of the ith component of the solution, a = ɛc is the linear decadic absorption coefficient, and d is the path length. This approach considers only absorption and neglects scattering. This is a reasonable approximation because in the 200–300 nm wavelength range on which we focus, the single-scattering albedo of liquid water ω₀ << 1, that is, absorption dominates scattering (Appendix Fig. A1; Quickenden and Irvin, 1980; Kröckel and Schmidt, 2014). We follow previous workers in assuming that ɛ_i does not vary as a function of pH for the simple inorganic molecules we consider, that is, that changes in the absorbance of solutions of these molecules with pH are due to changes in speciation or complexation and not intrinsic changes to ɛ_i (Braterman et al., 1983; Anbar and Holland, 1992; Nie et al., 2017; Tabata et al., 2021). We propagate the uncertainties on ɛ_i under the assumption that they are independent and normally distributed (Bevington and Robinson, 2003).

3.2. Molar decadic absorption coefficients for potential prebiotic absorbers

We consider absorption due to halide anions (Cl⁻, Br⁻, I⁻), ferrous iron species (Fe²⁺, $Fe{\left(CN\right)}_6^{4-}$), sulfur species (HS⁻, ${HSO}_3^{-}$, ${SO}_3^{2-}$), dissolved inorganic carbon (${HCO}_3^{-}$, ${CO}_3^{2-}$), and nitrate (${NO}_3^{-}$). Halides are ubiquitous in natural waters on modern Earth due to their high solubility and robust geological source, and hence they are thought to have been present in natural waters on early Earth as well (Knauth, 2005; Marty et al., 2018; Hanley and Koga, 2018; Toner and Catling, 2020). Ferrous iron is inferred to have been on early Earth on the basis of banded iron formations (Walker and Brimblecombe, 1985; Li et al., 2013; Konhauser et al., 2017; Toner and Catling, 2019). Dissolved inorganic carbon was inevitable in natural waters on early Earth due to dissolution of atmospheric CO₂ (Krissansen-Totton et al., 2018; Kadoya et al., 2020). Sulfur species may have been present on early Earth due to dissolution of volcanically outgassed sulfur species (Walker and Brimblecombe, 1985; Ranjan et al., 2018). Nitrate is predicted to have accumulated in natural waters on early Earth as a product of lightning in an N₂–CO₂ atmosphere (Mancinelli and McKay, 1988; Wong et al., 2017; Laneuville et al., 2018; Ranjan et al., 2019).

We draw on both the literature and our own measurements for the molar decadic absorption coefficients of potential prebiotic aqueous-phase absorbers. We use our own measurements when available because our measurements typically feature broader wavelength coverage than those in the literature, and they include estimates of uncertainty and are collected by using uniform techniques (Appendix A1.1). For the ferrous iron species, we check that the dominant ion speciation does not vary due to pH drift over the course of our dilutions (Appendix A1.1.4). Our measurement techniques are potentially inaccurate for weak acids and bases because we do not control pH, which may drift during dilution steps and affect speciation. For such species, we rely instead on literature data (Appendix A1.3). Table 1 summarizes the sources of the molar decadic absorption coefficients used in this work.

Table 1.

Molar Decadic Absorption Coefficients Used in this Work

Species	Source
Br−	As NaBr, this work
Cl−	As NaCl, this work
I −	As NaI and KI, this work
${NO}_3^{-}$	As NaNO3, this work
Fe2+	As Fe(BF4)2, this work
$Fe{\left(CN\right)}_6^{4-}$	As K4Fe(CN)6, this work
HS−	Guenther et al. (2001)
${HSO}_3^{-}$	Fischer and Warneck (1996); Beyad et al. (2014)
${SO}_3^{2-}$	Fischer and Warneck (1996); Beyad et al. (2014)
${HCO}_3^{-}$	Birkmann et al. (2018)
${CO}_3^{2-}$	Birkmann et al. (2018)
H2O	Quickenden and Irvin (1980)

3.3. Abundances of potential prebiotic absorbers in fiducial prebiotic waters

Constraining the impact of potential absorbers on prebiotic aqueous transmission requires estimates of their abundances in natural waters on early Earth. Natural waters are diverse, and we cannot hope to explore this full diversity. Instead, we focus on fiducial waters motivated by proposed origin-of-life scenarios. We focus primarily on shallow terrestrial waters, for example, ponds and lakes. Such waters are of interest for prebiotic chemistry because of their propensity for wet–dry cycles, their capability to accumulate atmospherically delivered feedstocks more efficiently than the oceans, and their potentially diverse palette of environmental conditions (Patel et al., 2015; Deamer and Damer, 2017; Pearce et al., 2017; Becker et al., 2018; Rimmer and Shorttle, 2019; Ranjan et al., 2019; Toner and Catling, 2020). We specifically consider freshwater lakes, carbonate lakes, and ferrocyanide lakes. We also consider the early ocean, to offer a basis of comparison for the terrestrial waters. Our oceanic calculations may also be relevant to origin-of-life scenarios that invoke shallow waters at the land–ocean interface (Commeyras et al., 2002; Lathe, 2005; Bywater and Conde-Frieboes, 2005), but they are not relevant to deep-sea origin-of-life scenarios (e.g., Corliss et al., 1981; Sojo et al., 2016) for which water alone is enough to extinguish UV. The composition of these waters is uncertain, and so we draw on the literature to construct high and low transmission endmember cases to bound their potential UV transmission. Our construction of these endmember cases is summarized in Table 2 and detailed in Sections 3.3.1–3.3.4.

Table 2.

Estimated Range of Concentrations of Potential Prebiotic Absorbers in Prebiotic Waters

Species	Ocean		Freshwater pond		Carbonate lake		Ferrous lake
	Low	High	Low	High	Low	High	Low	High
Br−	0.45 mM	1.8 mM	0.15 μM	0.15 μM	1 mM	10 mM	0.15 μM	0.15 μM
Cl−	0.3 M	1.2 M	0.2 mM	0.2 mM	0.1 M	6 M	0.2 mM	0.2 mM
I −	0.25 μM	1 μM	40 nM	600 nM	40 nM	600 nM	40 nM	600 nM
Fe2+	1 nM	0.1 mM	0.1 μM	0.1 mM	0	0	0	0
$Fe{\left(CN\right)}_6^{4-}$	0	0	0	0	0	0	0.1 μM	0.1 mM
NO3−	2 × 10−15 M	0.5 μM	0.05 nM	10 μM	5 nM	1 mM	0.05 nM	10 μM
HS−	0	0	0	8 × 10−11 M	0	8 nM	0	8 × 10−11 M
${HSO}_3^{-}$	0	0	0	200 μM	0	200 μM	0	200 μM
${SO}_3^{2-}$	0	0	0	100 μM	0	400 μM	0	100 μM
${HCO}_3^{-}$	2 mM	0.2 M	1 mM	1 mM	50 mM	0.1 M	1 mM	1 mM
${CO}_3^{2-}$	0.2 μM	1 mM	100 nM	100 nM	7 μM	7 mM	100 nM	100 nM

3.3.1. Ocean

For halide species, we scale the composition of the modern oceans. We consider a salinity range of 0.5–2 × modern, motivated by theoretical arguments and isotopic evidence (Knauth, 2005, 1998; Marty et al., 2018). On modern Earth, seawater halide concentrations are [Cl⁻] = 0.6 M, [Br⁻] = 0.9 mM, and [I⁻] = 0.5 μM (Channer et al., 1997; ASTM, 2013). Cl⁻ and Br⁻ covary in natural waters, leading us to fix their prebiotic ratios to the modern value (Hanley and Koga, 2018). De Ronde et al. (1997) used fluid inclusions to infer high [I⁻] at 3.2 Ga, but the interpretation of these samples is strongly contested (Lowe and Byerly, 2003; Knauth, 2005; Farber et al., 2015). We, therefore, fix our prebiotic [I⁻]/[Cl⁻] to the modern value as well and assume these ions to covary in the prebiotic ocean.

Estimates of ferrous iron concentrations in the surficial prebiotic ocean vary significantly. Halevy et al. (2017) estimated [Fe²⁺] < 10⁻⁹ M at the ocean surface, largely driven by the assumption of efficient photooxidation. More recent work has reported this process to be less efficient and estimated [Fe²⁺] = 10⁻⁴ M at the ocean surface (Konhauser et al., 2007; Halevy and Bachan, 2017; Konhauser et al., 2017). Fe²⁺ is predicted to be the main ferrous species at circumneutral pH for solutions with oceanic Cl⁻ and ${SO}_4^{2-}$ concentrations in equilibrium with 0.03 atm CO₂, which approximates conditions for the Archean ocean (King, 1998; Halevy and Bachan, 2017; Krissansen-Totton et al., 2018). We, therefore, assume the ferrous iron to be present as Fe²⁺, and we adopt 10⁻⁹ and 10⁻⁴ M as our bracketing estimates for [Fe²⁺] in the photic zone.

We take the bracketing range of oceanic [${NO}_3^{-}$] from Ranjan et al. (2019). We take oceanic sulfide concentrations to be negligible due to titration with Fe²⁺ (Walker and Brimblecombe, 1985; Poulton and Canfield, 2011). We take oceanic sulfite and bisulfite concentrations to be negligible (Halevy, 2013). We take bracketing oceanic carbonate and bicarbonate concentrations spanning the range at 3.9 Ga defined by “standard” and “control” cases of the early ocean model from the work of Kadoya et al. (2020).

3.3.2. Freshwater lakes

Terrestrial waters are typically dilute, with an ionic strength of order 10⁻³ M (Lerman et al., 1995). To represent a dilute endmember scenario for natural waters, we consider freshwater lakes with riverine composition (i.e., not evaporatively concentrated). Modern surface freshwater systems average [Cl⁻] = 0.2 mM, and the same is proposed for Archean and prebiotic river waters (Graedel and Keene, 1996; Hao et al., 2017). Mean [Br⁻]/[Cl⁻]≈1 − 2 × 10⁻³ in modern terrestrial waters (Edmunds, 1996; Magazinovic et al., 2004), which suggests mean [Br⁻ = 0.1 − 0.2 μM].

Mean iodide concentrations in modern terrestrial waters are reported as 40 nM (range: 0.1 nM − 0.6 μM) (Fuge and Johnson, 1986), and it is not clear that I⁻ is generally correlated with Cl⁻ as Br⁻ seems to be (e.g., Worden, 1996). We, therefore, consider both mean-I (40 nM) and high-I (0.6 μM) compositions. Estimates of [Fe²⁺] in riverine waters on early Earth span 0.1 μM–0.1 mM (Halevy et al., 2017; Hao et al., 2017), and we consider this range. Hao et al. (2017) predicted Archean river water to have a pH ≤6.34, for which Fe²⁺ should mainly present as ${Fe}_{aq}^{2+}$ (King, 1998; Tabata et al., 2021). We take [${HCO}_3^{-}$] from Hao et al. (2017), and calculate [${CO}_3^{2-}$]from it.

We take the bracketing range of [${NO}_3^{-}$] from Ranjan et al. (2019), scaled by 0.01 × to remove the assumption of a high drainage ratio. We take upper bounds on [${SO}_3^{2-}$], [${HSO}_3^{-}$] and [HS⁻] from Ranjan et al. (2018) for circumneutral pH and steady-state out-gassing. We consider a lower bound of 0 for all sulfur species following the assumption of Halevy (2013) for rivers.

3.3.3. Closed-basin carbonate lakes

Closed-basin carbonate lakes have been proposed as venues for prebiotic chemistry, because the elevated carbonate concentrations in these lakes suppress [Ca²⁺], permitting the accumulation of phosphate to prebiotically relevant concentrations (Toner and Catling, 2020). For an endmember, we consider a closed basin carbonate lake with 10⁻² mol/kg phosphorus, which corresponds to the upper edge of phosphorus concentrations in the sample of closed-basin carbonate lakes reported by Toner and Catling (2020).

Cl⁻ and Br⁻ behave conservatively with P in carbonate lakes, with [Cl⁻] = 0.1–6 mol/kg and [Br⁻] = 10⁻³–10⁻² mol/kg for [P] = 10⁻² mol/kg (Toner and Catling, 2020). Projecting from Toner and Catling (2020), we simplify the calculation by approximating molarity as molality; we contend this approximation to suffice for the order-of-magnitude estimates we seek. Iodine is proposed to be delivered to surface waters via rainfall, and so we might expect it to be evaporatively concentrated in closed-basin lakes; however, we located no reports of evaporative iodide concentrations in composition studies of such lakes (Friedman et al., 1976; Eugster and Jones, 1979 ; Fuge and Johnson, 1986; Mochizuki et al., 2018; Toner and Catling, 2020; Hirst, 2013). We, therefore, consider a bracketing range of [I⁻] equal to the freshwater lake scenario. We take ${Fe}_{aq}^{2+}=0$, given that high carbonate concentrations should suppress ${Fe}_{aq}^{2+}$ due to siderite precipitation (Toner and Catling, 2020). Toner and Catling (2020) did not report [${HCO}_3^{-}$] or [${CO}_3^{2-}$] for closed-basin carbonate lakes on early Earth, but they calculated a bracketing range of pCO₂ = (0.01 bar, 1 bar) to correspond to pH = (9, 6.5) in such lakes. We convert these estimates to approximate bracketing ranges on [${HCO}_3^{-}$] and [${CO}_3^{2-}$] under the assumption of equilibrium (Sander, 2015; Rumble, 2017). We take [${NO}_3^{-}$] from Ranjan et al. (2019) and consider a lower bound of 0 for all sulfur species following the assumption of Halevy (2013) for riverine waters. We take upper bounds on [HS⁻], [${SO}_3^{2-}$], and [${HSO}_3^{-}$] from Ranjan et al. (2018) for steady-state outgassing, for pH = 7 and pH = 9. The upper limits listed on [${SO}_3^{2-}$] and [${HSO}_3^{-}$] cannot simultaneously be achieved because they correspond to different pHs. Our conclusions are robust to this inconsistency, because even at their respective upper limits ${SO}_3^{2-}$ and ${HSO}_3^{-}$ are not the dominant absorbers in the carbonate lake scenario.

3.3.4. Ferrous lakes

Ferrocyanide lakes have been proposed to form when ferrocyanide salt deposits are irrigated by neutral water (Toner and Catling, 2019; Sasselov et al., 2020). Ferrocyanide is an extremely potent UV absorber, and so we consider transmission in such ferrocyanide lakes. Notably, ferrocyanide has been invoked in UV-dependent prebiotic chemistries and thus provides an opportunity to examine their geochemical self-consistency (Xu et al., 2018; Mariani et al., 2018; Rimmer et al., 2018). We approximated the composition of a ferrocyanide lake to be that of a freshwater lake, but with the Fe²⁺ present as $Fe{\left(CN\right)}_6^{4-}$ instead.

4. Results

4.1. Prebiotic ocean

The prebiotic ocean efficiently attenuated shortwave UV radiation but may have admitted longer-wavelength UV radiation to depths of meters (Fig. 1). In the low-absorption endmember, the absorption is dominated by the halides, especially Br⁻ at shorter wavelengths and I⁻ at longer wavelengths. In the high-absorption endmember scenario, absorption is dominated by Br⁻ at shorter wavelengths and ${Fe}_{aq}^{2+}$ at longer wavelengths. Halides confine the shortest-wavelength UV photons to the surface of the ocean: In even the low-absorption endmember scenario, the ocean is optically thick¹ at ≤220 nm for depths d ≥ 7 ± 1 cm, driven primarily by Br⁻. Longer-wavelength radiation may penetrate to much greater depths, with the absorbers considered here permitting penetration of the ∼260 nm radiation that would be responsible for nucleotide degradation to a depth of meters in even the high-absorption endmember scenario.

FIG. 1.

Simulated linear decadic absorption coefficients of the prebiotic ocean and its component solutes, for low-absorption and high-absorption endmember cases. Not all solutes are visible in each case, because the linear decadic absorption coefficients of some solutes fall below the lower limit of the y-axis across the wavelength space plotted here. The shortest-wavelength photons are extinct in the surface layers of the ocean, but the ∼260 nm radiation responsible for nucleotide degradation can penetrate to a depth of meters. Color images are available online.

4.2. Prebiotic terrestrial waters

4.2.1. Freshwater lakes

The transparency of shallow freshwater lakes depends strongly on the abundances of Fe²⁺ and S(IV) species (sulfite, bisulfite). In the low-absorption endmember scenario, freshwater lakes are essentially transparent in the UV, with depths of meters required for non-negligible attenuation of UV across most of the UV (Fig. 2). However, in the high-absorption endmember scenario, shortwave UV is shielded due to sulfite, with secondary shielding from nitrate and ${Fe}_{aq}^{2+}$. For radiation with wavelengths ≤235 nm, the high-absorption endmember freshwater lake is optically thick for depths d ≥ 9.6 ± 0.4 cm. However, even the highly absorptive endmember remains optically thin to the $\gtrsim$260 nm radiation, which dominates nucleotide photolysis down to a depth of 1.4 ± 0.2 m.

FIG. 2.

Simulated linear decadic absorption coefficients of the prebiotic freshwater lake and its component solutes, for low-absorption and high-absorption endmember cases. Not all solutes are visible in each case, because the linear decadic absorption coefficients of some solutes fall below the lower limit of the y-axis across the wavelength space plotted here. The freshwater lake may have been largely transparent to UV, and even in the high-absorption endmember would have been largely transparent to UV at the longer wavelengths (∼260 nm) relevant to nucleotide photolysis. UV, ultraviolet. Color images are available online.

4.2.2. Closed-basin carbonate lakes

Closed-basin carbonate lakes can be more UV-opaque compared with the prebiotic ocean (Fig. 3). Elevated [Br⁻] robustly limits absorption at short wavelengths; in even the low-absorption endmember scenario, ≤220 nm radiation is efficiently extinquished (i.e., is in the optically thick regime) for depths of >3.2 ± 0.6 cm. Radiation at 260 nm is available for depths of ≤1.04 ± 0.07 m even in the high-absorption endmember scenario, but it is depleted at depths of a few meters, due to nitrate absorption with contributions from sulfite.

FIG. 3.

Simulated linear decadic absorption coefficients of the prebiotic carbonate lake and its component solutes, for low-absorption and high-absorption endmember scenario. Not all solutes are visible in each case, because the linear decadic absorption coefficients of some solutes fall below the lower limit of the y-axis across the wavelength space plotted here. The carbonate lake robustly extincts shortwave UV in even the low-absorption endmember scenario, driven by Br⁻. In the high-absorption endmember case, longer wavelength UV would have been available throughout shallow lakes (d < 1 m), but it would have been extinct at depths of a few meters. Color images are available online.

4.3. Ferrocyanide lakes

Ferrocyanide lakes are may have been low-UV environments (Fig. 4). Ferrocyanide is a stronger and crucially broader UV absorber than Fe²⁺ and can attenuate UV across the near-UV range relevant to prebiotic chemistry. Attenuation due to ferrocyanide is minimal in the low-absorption endmember scenario. However, for the high-absorption endmember, the ferrocyanide lake becomes optically thick by d = 11 ± 3 cm across 200–300 nm.

FIG. 4.

Simulated linear decadic absorption coefficients of the prebiotic ferrocyanide lake and its component solutes, for low-absorption and high-absorption endmember scenarios. Not all solutes are visible in each case, because the linear decadic absorption coefficients of some solutes fall below the lower limit of the y-axis across the wavelength space plotted here. Ferrocyanide is an effective sunscreen, and lakes hosting more than dilute ferrocyanide ($\gtrsim$100 μM) would have been low-UV environments. Color images are available online.

Photochemical derivatives of ferrocyanide, such as nitroprusside and ferricyanide, are similarly potent UV absorbers, which suggests that ferrocyanide-rich lakes will remain UV-poor even if some of the ferrocyanide is photochemically processed into these forms (Strizhakov et al., 2014; Xu et al., 2018; Mariani et al., 2018; Ross et al., 2018). Ferrocyanide and derived compounds may, therefore, have been strong “sunscreens” in select lacustrine environments on early Earth, if at elevated concentrations as postulated by some authors and invoked by others (Xu et al., 2018; Toner and Catling, 2019; Sasselov et al., 2020).

5. Discussion

Ranjan and Sasselov (2016) argued that UV radiation down to wavelengths of ∼204 nm would have been available for aqueous prebiotic chemistry on early Earth, motivated by the transmission of the atmosphere and pure water. However, prebiotic waters were likely not pure. In this article, we re-examine the conclusion of Ranjan and Sasselov (2016) by considering potential UV absorbers that might have been present in prebiotic waters.

5.1. Terrestrial freshwater systems may have been UV-transparent

Prebiotic terrestrial freshwaters may have been largely transparent in the UV. In the low-absorption endmember scenario for terrestrial freshwater lakes, shallow lakes could have been transparent down to depths of meters across most of the >200 nm wavelength range admitted by the prebiotic atmosphere, meaning that UV radiation may have been a pervasive aspect of the surficial prebiotic milieu.

Some investigators have assumed prebiotic pond waters to be UV-opaque based on extrapolation of modern natural waters; however, the opacity of modern terrestrial waters in the UV is generally due to biogenic dissolved organic compounds (e.g., Morris et al., 1995; Markager and Vincent, 2000; Laurion et al., 2000; Pearce et al., 2017). Indeed, even on modern Earth, waters with low biological productivity are transparent to a depth of ≥3 m down to the 300 nm threshold to which transmission has been characterized (Smith and Baker, 1981; Morel et al., 2007). Prebiotic fresh waters may have been similarly transparent, unless abiotic processes were comparably efficient as modern productive ecosystems in generating organics (e.g., potentially in the immediate aftermath of a large impact; Zahnle et al., 2020).

5.2. Shortwave UV was likely attenuated in diverse prebiotic waters

Although prebiotic natural waters in general were not necessarily opaque across the UV (200–300 nm), shortwave UV (≤220 nm) would have been attenuated in diverse prebiotic waters. Numerous prebiotic absorbers strongly attenuate shortwave UV, even if they absorb longwave UV only weakly. In particular, in high-salinity waters such as the prebiotic ocean and carbonate lakes, the halide anions, especially Br⁻, efficiently attenuate shortwave UV. In even the low-absorption endmembers, λ ≤ 220 nm radiation is restricted to depths d ≤ 7 ± 1 cm in the prebiotic ocean, and d ≤ 3.2 ± 0.6 cm in the carbonate lake scenario (Fig. 5).

FIG. 5.

Estimate of actinic UV flux as a function of depth in the low-absorption endmember of the carbonate lake scenario, calculated by attenuating the surface actinic flux of Ranjan and Sasselov (2017) (their “surface radiance”) by Beer-Lambert law, assuming a slant angle of 60°, atmospheric composition from Rugheimer et al. (2015), and stellar irradiation of 3.9 Ga Sun from Claire et al. (2012). In even the low-absorption endmember, attenuation of <220 nm irradiation is significant. Color images are available online.

Shortwave UV may have been even more strongly attenuated in some waters. For example, in the high-absorption endmember for the carbonate lake scenario, nitrate absorption restricts λ ≤ 230 nm UV to d ≤ 0.43 ± 0.06 cm. Such nitrate concentrations would only have been available in shallow closed-basin lakes with large drainage ratios, and only if atmospheric ${NO}_X^{-}$ production rates were at the upper edge of the predicted range (Ranjan et al., 2019). Shortwave UV may have been restricted in prebiotic terrestrial waters, in general, if prebiotic terrestrial sulfite concentrations were at the upper limit of what has been proposed in the literature (Fig. 2). However, these literature estimates of prebiotic terrestrial sulfite concentrations consider only thermal processes and neglect photolytic loss mechanisms, which means that they are overestimates; further modeling that includes sulfite photolysis will be required to determine this possibility (Deister and Warneck, 1990; Fischer and Warneck, 1996; Halevy, 2013; Ranjan et al., 2018). In summary, shortwave UV may have been strongly attenuated in diverse prebiotic waters, including saline lakes and closed-basin lakes, due to the action of species such as Br⁻ and nitrate.

5.3. Broadband UV was attenuated in some prebiotic waters

While diverse prebiotically plausible absorbers are capable of attenuating shortwave UV, fewer prebiotic absorbers are capable of broadband UV attenuation, including the longer-wavelength UV photons that dominated the early Sun's UV output and hence prebiotic photoprocesses such as nucleobase photolysis. One prebiotic proposed family of broadband UV absorbers are compounds derived from ferrous iron. In particular, ferrocyanide is a strong, broad UV absorber. In the high-absorption endmember scenario for ferrocyanide lakes, ferrocyanide would have suppressed (optical depth >1) ≤260 nm radiation for d ≥ 1.8 ± 0.5 cm, and ≤300 nm radiation for d ≥ 11 ± 3 cm (Fig. 6). Photochemical derivatives of ferrocyanide, such as ferricyanide and nitroprusside, are similarly effective broadband UV screens (Appendix Fig. A12). Ferrocyanide lakes and their derivative waters, if extant, would have been low-UV environments at depth on early Earth.

FIG. 6.

Estimate of actinic UV flux as a function of depth in the high-absorption endmember of the ferrocyanide lake scenario, calculated by attenuating the surface actinic flux of Ranjan and Sasselov (2017) (their “surface radiance”) by Beer-Lambert law, assuming a slant angle of 60°, atmospheric composition from Rugheimer et al. (2015), and stellar irradiation of 3.9 Ga Sun from Claire et al. (2012). In the high-absorption endmember, broadband UV is extincted at shallow depths. Color images are available online.

The existence and prevalence of ferrocyanide lakes on early Earth is uncertain. In particular, even if ferrocyanide lakes form as proposed in the work of Toner and Catling (2019), ferrocyanide undergoes photoaquation under irradiation by ≤400 nm radiation (Ašpergěr, 1952). Toner and Catling (2019) argued that rapid back-reaction stabilizes ferrocyanide against this photodecomposition, based on the experimental study of Ašpergěr (1952). However, the measurements of Ašpergěr (1952) were not conducted in prebiotically representative conditions. In particular, their UV irradiation was not representative of prebiotic UV irradiation, and the ferrocyanide concentrations used in their study were extremely high (≥50 mM). It is not clear whether ferrocyanide lakes would remain stable under more prebiotically representative conditions; detailed measurements and modeling of ferrocyanide photochemistry in prebiotically relevant conditions would be required. Alternately, the detection of remnants of ferrocyanide-rich environments on Mars (e.g., ferrocyanide salt deposits) might inform our understanding of the prevalence of such systems on early Earth (Sasselov et al., 2020; Mojarro et al., 2021).

In addition to ferrocyanide, other ferrous iron compounds such as FeSO₄ and FeCl₂ may also have acted as prebiotic “sunscreens,” due to their broadband UV absorption; however, much higher concentrations of these compounds would be required compared with ferrocyanide due to their much lower molar absorption in the UV (Appendix Fig. A12). In addition, their formation is not thermodynamically favored relative to Fe²⁺ under conditions relevant to natural waters on modern Earth (King, 1998), and they may not have been favored on early Earth either; modeling under prebiotic Earth conditions will be required to definitively rule on this question. Finally, the abundance of Fe²⁺ itself in natural waters on early Earth is still under debate (Konhauser et al., 2017; Halevy et al., 2017; Hao et al., 2017). Detailed, focused modeling of Fe²⁺ abundances and speciation in prebiotic waters on early Earth will also be required to estimate the prevalence of Fe²⁺-derived “sunscreens” such as ferrocyanide in natural waters on early Earth.

We do not recover the finding of Cleaves and Miller (1998) that Fe²⁺ in isolation would be an effective broadband sunscreen at concentrations that correspond to the prebiotic ocean. We report a molar absorbance for the Fe²⁺ ion (as Fe(BF₄)₂; Section A1.2.4) at 260 nm (15 ± 4 M⁻¹ cm⁻¹) that is approximately consistent with the findings of Fontana et al. (2007) and Heinrich and Seward (1990) but two orders of magnitude lower than Cleaves and Miller (1998) (1630 M⁻¹ cm⁻¹). The molar absorbances of FeCl₂, FeSO₄, and FeOH⁺ at 260 nm are also significantly below that reported by Cleaves and Miller (1998) (Anbar and Holland, 1992; Fontana et al., 2007). We speculate that one possible explanation might be the formation of Fe²⁺ complexes in the study of Cleaves and Miller (1998). At a slightly basic pH, characteristic of the modern ocean conditions simulated by Cleaves and Miller (1998), Fe²⁺ may have complexed with OH⁻ and/or ${CO}_3^{2-}$ (King, 1998), and the absorption properties of such complexes may differ from ${Fe}_{aq}^{2+}$. Our findings for the prebiotic ocean are more consistent with those of Anbar and Holland (1992) for the Archean ocean, who reported efficient attenuation of shortwave UV in the early ocean but the possibility of transmittance of longer-wavelength UV to a depth of meters.

5.4. Implications for prebiotic chemistry

Of the species considered in this study, we identify no ubiquitous absorber capable of attenuating broadband UV (200–300 nm) in prebiotic natural waters. Indeed, prebiotic freshwaters may have been essentially transparent in the UV. Prebiotic chemistries that are adversely affected by UV irradiation (UV-avoidant) must either be demonstrated to be so productive as to outpace UV degradation under prebiotic conditions or to invoke a UV-shielded prebiotic milieu (e.g., the organic-haze shielded aftermath of a large impact; Benner et al., 2019). UV light in the ∼240–300 nm wavelength range was not significantly attenuated by the geologically derived solutes considered in this study in most terrestrial prebiotic waters, meaning that low-pressure mercury lamps with primary emission at 254 nm remain reasonable proxies for UV irradiation in initial simulations of prebiotic chemistry. However, follow-up studies (e.g., characterization of action spectra) are required to verify whether pathways discovered with such sources could have functioned in realistic natural environments on prebiotic Earth (Ranjan and Sasselov, 2016; Todd et al., 2018; Rimmer et al., 2021).

Diverse prebiotic absorbers are capable of attenuating shortwave UV (≤220 nm). The general prebiotic importance of removal of shortwave UV (≤220 nm) should not be overstated, because most of the young Sun's UV flux was delivered at wavelengths >220 nm (Claire et al., 2012). In particular, shallow saline waters do not shield the canonical nucleobases, nucleosides, and nucleotides whose photolysis is expected to be dominated by the longwave (∼260 nm) bands (Voet et al., 1963; Todd et al., 2020). Similarly, the photolysis of 2-aminothiazole and the photoconversion of cytidine to uridine are not inhibited in saline waters (Todd et al., 2019, 2020). However, the photolyses of 2-aminooxazole and 2-aminoimidazole, intermediates with proposed roles in prebiotic nucleotide synthesis and activation (Powner et al., 2009; Patel et al., 2015; Li et al., 2017), are dominated by <230 nm radiation under early Earth conditions, and their photodestruction lifetimes are modestly enhanced in highly saline waters. Todd et al. (2019) estimated half-lives of these molecules to photodestruction to be ≈7, 26, and 99 h on early Earth; the half-lives of these molecules to direct photodestruction may be enhanced by up to a factor of a few in carbonate lakes by halide photoabsorption (Appendix A2). Saline systems have been disfavored as venues for prebiotic chemistry, on the argument that high salt concentrations generally inhibit lipid membrane formation (e.g., Deamer and Damer, 2017). However, experimental work suggests that some amphiphile mixtures can form stable vesicles at oceanic salinity or higher, and indeed that some families of amphiphiles require salinity for vesicle formation (Namani and Deamer, 2008; Maurer and Nguyen, 2016; Xu et al., 2017; Maurer, 2017). We therefore argue that it is premature to dismiss saline waters as venues for prebiotic chemistry. Indeed, they may be favorable environments for the accumulation of molecules whose concentrations on early Earth are limited by ≤220 nm photolysis.

Waters rich in ferrous iron compounds such as ferrocyanide were low-UV (UV-shielded) environments. Such waters may have been favorable environments for UV-avoidant prebiotic chemistries. For example, Pearce et al. (2017) proposed meteoritically delivered nucleobases in prebiotic ponds under the assumption that prebiotic pond water was UV-opaque and could shield meteoritic nucleobases from photolysis. In a 1 m-deep lake with a composition that corresponds to the high-absorption endmember scenario for ferrocyanide lakes, the lifetimes of the nucleobases to direct photolysis would be enhanced by approximately two orders of magnitude, which would make such waters potential candidates for the meteoritic delivery scenario (Appendix A2). By the same token, ferrous-rich lakes may be poor environments for UV-dependent prebiotic pathways. For example, the photoisomerization of diaminomaleonitrile (DAMN) to 5-aminoimidazole-4-carbonitrile (AICN), an intermediate for purine nucleobase synthesis, requires irradiation by ≲300 nm photons, which could be strongly attenuated in ferrocyanide-rich lakes (Ferris and Orgel, 1966; Cleaves, 2012; Boulanger et al., 2013; Yadav et al., 2020). Similarly, the UV-driven deamination of cytidine to form uridine, which is driven by 200–300 nm photons, could be inhibited in ferrocyanide-rich waters (Powner et al., 2009; Todd et al., 2020). Solvated electron production by UV photoirradiation of the ferrocyanide–sulfite system has been proposed to efficiently drive HCN homologation, but it is unknown whether ferrocyanide or sulfite (or potentially both) is the photoactive agent. If it is solely sulfite, it is unknown whether the chemistry can function with the dilute, optically thin ferrocyanide required for sulfite photolysis (Xu et al., 2018; Ritson et al., 2018; Green et al., 2021). We encourage further investigation into these questions to better understand the prebiotic potential for this chemistry.

In our discussion so far, we have focused on the implications of absorption of UV radiation by prebiotic solutes for direct photochemical processes such as excitation or photolysis of biomolecules. However, the UV energy absorbed by prebiotic absorbers must be dissipated, and although it may be dissipated thermally, it may also be dissipated through the formation of high-energy species that can participate in further chemistry. For example, the halides, sulfite/sulfide, and ferrocyanide generate solvated electrons ($e_{aq}^{-}$) and oxidized radicals on UV irradiation, which are reactive and may trigger further chemistry (Jortner et al., 1964; Sauer et al., 2004). Such further chemistry may be destructive of biomolecules, meaning that even in waters with high concentrations of UV-absorbing compounds, UV light may indirectly suppress biomolecule accumulation. This condition must be checked by prebiotic chemistries invoking an aqueous UV sunscreen. On the other hand, such further chemistry may also be productive. For example, Liu et al. (2021) reported irradiation of ${SO}_3^{2-}$ and ${HCO}_3^{-}$ to lead to abiotic carbon fixation driven by $e_{aq}^{-}$ production from ${SO}_3^{2-}$, whereas the simultaneous irradiation of ${SO}_3^{2-}$ and ferrocyanide leads to HCN homologation (J. Xu et al., 2018). In these chemistries, the absorber effectively converts solar UV energy into chemical free energy in the form of $e_{aq}^{-}$. In summary, photochemical transformations may occur even in UV-shielded waters, triggered by photoabsorption of the very absorber that shields the UV; the proposed prebiotic chemistry that invokes an aqueous-phase UV absorber must incorporate this possibility in verifying self-consistency.

5.5. Caveats and limitations

We have not considered FeOH⁺ as a prebiotic sunscreen. FeOH⁺ is a strong and broad UV absorber. Crucially, its absorption at the longer UV wavelengths where the Sun emits substantially more photons means that FeOH⁺ is projected to have played a key role in Fe²⁺ photooxidation and subsequent deposition (Braterman et al., 1983; Anbar and Holland, 1992; Tabata et al., 2021). Because of its ability to absorb longer-wavelength UV (300–450 nm), the total photoabsorption of FeOH⁺ is comparable to that of Fe²⁺ at circumneutral pH, despite much lower concentrations (Nie et al., 2017). However, its absorption at longer wavelengths does not aid its potential as a sunscreen at shorter wavelengths (200–300 nm). At these wavelengths, the potential of FeOH⁺ as a sunscreen is limited due to its low solubility at basic pH due to Fe(OH)₂ precipitation. This low solubility means that [FeOH⁺] is low, implying that large depths are required for it to significantly attenuate UV (i.e., bring the optical depth to unity). The solubility product of Fe(OH)₂ is K_sp = 4.9 × 10⁻¹⁷, and K_eq = 3.02 × 10⁻¹⁰ for the reaction $F{e}^{2+}+H_{2}O\to FeO{H}^{+}+H^{+}$ (Braterman et al., 1983; Rumble, 2017). For [Fe²⁺] = 100 μM, this implies [FeOH⁺] ≤ 2 μM, restricting the relevance of FeOH⁺ as an aqueous sunscreen for 200–300 nm radiation to deeper waters with d > 6 m.

In the present study, we considered only a few of the vast diversity of natural waters. Other natural waters may have different absorption properties. For example, hot springs host high concentrations of HS⁻ and are driven by continuous supply of H₂S from below (Kaasalainen and Stef'ansson, 2011). HS⁻ is a strong and broad absorber and, similar to the ferrous iron compounds, attenuates the longer-wavelength UV radiation that dominates solar UV output; some hot springs could, therefore, have been UV-opaque. Similarly, I⁻ is a strong and broad UV absorber; in waters rich in iodine, for example, mineral springs, I⁻ may provide significant UV attenuation (Fuge and Johnson, 1986).

We reiterate that the estimates of UV transmittance presented here are upper bounds. The longwave absorption of ${HCO}_3^{-}$ and ${CO}_3^{-}$ is unconstrained and may be important in very alkaline waters. Prebiotic natural waters may have contained absorbers other than those we have considered here, and if absorptive/abundant enough, these compounds may have served as sunscreens. Examples of such compounds include silicate or basaltic dust and meteoritically delivered or atmospherically derived organic compounds.² Further work will be required to determine the wavelength-dependent molar absorptivities of these alternate absorbers and to estimate whether they could have been present in prebiotic waters at concentrations sufficient to significantly attenuate solar UV. Finally, the products of prebiotic chemistry may themselves attenuate UV (Cleaves and Miller, 1998; Todd et al., 2021). More detailed modeling of these products and their accumulation will be required to rule on this possibility.

In this work, we have constructed the bracketing low- and high-absorptivity compositional endmembers for various prebiotic waters based on literature estimates. We have not attempted self-consistent geochemical modeling of these waters to estimate their composition, in part because the relevant kinetics have not been characterized under prebiotically relevant conditions (e.g., Ranjan et al., 2018, 2019). Similarly, our work treats UV absorbers as static and neglects further photochemical transformations triggered by their absorption of solar UV. Our work is, therefore, a first approximation and subject to revision by future models that are capable of investigating the self-consistent photochemistry and geochemistry of prebiotic natural waters.

6. Conclusions

Prebiotic freshwaters may have been essentially transparent in the UV; UV-avoidant surficial prebiotic chemistries must invoke a UV “sunscreen” agent. Shortwave (≤220) UV may have been attenuated in diverse prebiotic waters, with the most important absorbers being Br⁻ in saline waters and potentially sulfite in shallow lakes and nitrate in shallow closed-basin lakes. Better constraints on prebiotic ${NO}_X^{-}$ production rates and on prebiotic sulfite loss rates will be required to rule on the latter possibility. Regardless, the impact of shortwave UV attenuation is likely modest, because most of the UV flux from the early Sun was delivered at longer, unshielded wavelengths, with the largest impact among the photoprocesses we considered being a modest enhancement in the lifetimes of the prebiotic intermediates 2-aminoxazole and 2-aminoimidazole to photolysis. Measurements of the wavelength-dependent rates of prebiotic processes are required to identify other processes that may be particularly promoted or inhibited in shortwave-shielded waters.

The generally widespread availability of ∼240–300 nm radiation means that low-pressure mercury lamps remain suitable for initial studies of prebiotic chemistry, though more realistic irradiation is required to verify the plausibility of pathways discovered under mercury lamp irradiation.

Some natural waters may have been largely opaque in the UV. In particular, Fe²⁺-derived compounds are effective broadband “sunscreens” if present at high concentrations. Ferrocyanide is an especially potent UV absorber, and ferrocyanide lake waters would have been low-UV environments that are candidate waters for UV-avoidant origin-of-life scenarios such as the meteoritic delivery hypothesis. On the other hand, such waters may be poor environments for UV-dependent reactions such as the photodeamination of cytidine to uridine. In addition to ferrocyanide, other ferrous iron compounds such as FeSO₄ may serve as sunscreens in shallow lakes if present at proportionately higher concentrations to compensate for their lower molar absorptivities relative to ferrocyanide and to thermodynamically favor their complexation. Fe²⁺ has been suggested to be widespread on early Earth, but the abundance and speciation of Fe²⁺ compounds on early Earth remain uncertain; we highlight detailed modeling and/or geochemical constraints on ferrous iron concentrations and speciation in prebiotic natural (especially terrestrial) waters as a priority for UV-sensitive prebiotic chemistry.

UV light can trigger photochemistry even if attenuated by an absorber, through photochemical transformations of that absorber. In particular, irradiation of a wide range of prebiotic absorbers, such as the halides, sulfite/sulfide, and ferrocyanide, is predicted to generate $e_{aq}^{-}$ and oxidized radicals. The implications of this production vary according to prebiotic chemistry. To be plausible, prebiotic chemistries that invoke these absorbers must self-consistently account for the chemical effects of these by-products. The rates, timescales, and concentrations relevant for various prebiotic chemical reactions may also play a role in determining the overall self-consistency of different scenarios. We encourage further investigation in these areas to better understand the necessary environments and prebiotic plausibility of various origins-of-life chemistries.

Abbreviations Used

HCN

hydrogen cyanide

OD

optical depth

UV

ultraviolet

Appendix A1. Molar Decadic Absorption Coefficients

A1.1. Measured Molar Decadic Absorption Coefficients

A1.1.1. Samples

All salts were purchased from Sigma-Aldrich at the highest available purity grade (≥95%) and used without further purification. All samples were dissolved in Liquid Chromatography-Mass Spectroscopy (LC-MS) grade freshwater (LiChrosolv; Millipore Sigma).

A1.1.2. Setup

The pH was measured by a pH electrode (Hach SensION + PH3). The dissolved samples were kept in sealable spectrosil quartz cuvettes (9B-Q-10-GL14-C; Starna Cells) with a sample depth of 10 mm. The concentrations of the samples were kept ≤0.1 × the saturation concentration. In addition, the absorbance was kept between optical depth (OD) 0.005 and OD 1.1 throughout the spectral range of 200 to 360 nm by several dilution steps of the samples. All absorbance spectra were recorded in triplicates at 23°C by a Shimadzu UV-1900 UV-VIS spectrophotometer relative to a blank water sample, and measurements were conducted at ambient conditions unless otherwise specified.

A1.1.3. Data evaluation

After an initial solvent correction, the triplicate spectra were averaged and cropped to an absorbance range between OD 0.005 and OD 1.1. The molar decadic absorption coefficients were calculated from the absorption spectra by division by the respective sample concentrations. Spectral ranges with systematic deviations, for example, due to cuvette contamination, were excluded from the dataset. The systematic error of the spectrophotometric measurements was estimated conservatively to be about 15%. In addition, the following errors were included in the estimate: 6% for the pipetting process, 4% for the sample masses (14% for NaS), and 5% for sample impurities. In case of pH-sensitive samples, an error of up to 4% was included to account for the chemical equilibrium at the respective pH according to the Henderson-Hasselbalch equation. The statistical error for each sample was estimated from the averaging process. Appendix Table A1 gives the dilution sequence used to compile the molar absorptivity measurements.

APPENDIX FIG. A1.

Optical properties of pure water. Decadic extinction and scattering coefficients, single scattering albedo, and the depth at which the scattering optical depth ≥1 as a function of wavelength for pure liquid water under standard conditions. The decadic extinction is taken from Quickenden and Irvin (1980), which remains state-of-the-art in this wavelength range (Kröckel and Schmidt, 2014). The Rayleigh scattering is calculated following Kröckel and Schmidt (2014), with the modifications that the index of refraction, n, and the pressure derivative of n at constant temperature T, ${\left(\frac{\partial n}{\partial p}\right)}_T$, are taken to be constant at n = 1.39 and ${\left(\frac{\partial n}{\partial p}\right)}_T$ = 0.095 μm² kN⁻¹. Color images are available online.

Appendix Table A1.

Measurement Sequences Used to Determine Molar Decadic Absorption Coefficients

Sample	Measurement sequence
NaBr	80 μM, 400 μM, 860 μM, 2 mM, 20 mM, 100 mM
KBr	70 μM, 350 μM, 1 mM, 20 mM, 100 mM
NaCl	50 mM, 559 mM
NaI	50 μM, 100 μM, 2 mM, 50 mM, 100 mM
KI	100 μM, 2 mM, 100 mM
Fe(BF4)2	1 mM (pH = 3.4), 2 mM (pH = 3.2), 20 mM (pH = 2.6), 50 mM (pH = 2.3), 200 mM (pH = 1.7)
K4[Fe(CN)6]	44 μM (pH = 6.5), 87 μM (pH = 6.4), 174 μM (pH = 6.4), 436 μM (pH = 6.0)

We report pH for the Fe(BF₄)₂ and K₄[Fe(CN)₆] measurement sequences to verify their speciation (Section A1.1.4). FeSO₄, FeCl₂, and Na₂Fe(CN)₅NO are not listed here because molar decadic absorption coefficients for those species were drawn from the literature.

A1.1.4. Experimental concerns and mitigations

We conducted our measurements at ambient conditions, that is, in oxic air. Under oxic conditions, ferrous iron slowly oxidizes. To mitigate this effect, each measurement sequences involving ferrous iron species were completed in ≤1 h. We performed control experiments for the ferrous iron species under reduced oxygen conditions. We did not observe differences in the absorption spectra within the duration of our experiments, and in both cases found good agreement with the literature, suggesting that this procedure did not impact the accuracy of our approach.

Special care was taken during the cuvette cleaning process to avoid contamination. In particular, the cuvettes were cleaned solely with water; we did not clean with acetone as we discovered in preliminary experiments that absorptivity due to residual acetone contaminated the measurements.

Fe²⁺ can complex with OH⁻ in basic solution (King, 1998). We, therefore, tracked the pH of the Fe(BF₄)₂ solutions to ensure they remained at the acid pH at which ${Fe}_{aq}^{2+}$ is the dominant form. Conversely, in acidic solution $Fe{\left(CN\right)}_6^{4-}$ can protonate. We tracked the pH of the ferrocyanide solutions to ensure they well exceeded pH = 4.2, corresponding to the pKa of the first protonation of $Fe{\left(CN\right)}_6^{4-}$ (Shirom and Stein, 1969).

A1.2. Measured Spectra of Prebiotically Relevant Compounds and Comparison to Literature Data

In this section, we compare our measured spectra with those extracted from the literature. The literature data typically do not include uncertainty estimates. However, Birkmann et al. (2018) noted that, in previous studies, the uncertainty on their measurements of molar absorption coefficients was 5%, as determined by reproduction. We assume this uncertainty estimate of 5% to apply to the measurements reported in the work of Birkmann et al. (2018) as well.

A1.2.1. Bromide (Br⁻)

Appendix Fig. A2 presents the molar absorption coefficients of NaBr and KBr derived from our measurements. The absorption of these species is identical to the precision of our measurements, confirming the findings of past work, which indicate that the anion dominates the absorption (Birkmann et al., 2018 and sources therein). Our data agree with the measurements of Birkmann et al. (2018) from 200 to 225 nm, and extend the spectral coverage to 240 nm. We follow Birkmann et al. (2018) in disagreeing with Johnson and Coletti (2002) at short (<204 nm) wavelengths.

APPENDIX FIG. A2.

Molar absorption coefficients of Br⁻ derived from our measurements, compared with literature reports from Birkmann et al. (2018) and Johnson and Coletti (2002). Our data extend spectral coverage of Br⁻ absorption from 225 to 240 nm. We do not detect Br⁻ absorption at >240 nm. Color images are available online.

A1.2.2. Chloride (Cl⁻)

Appendix Fig. A3 presents the molar absorption coefficients for NaCl derived from our measurements. We take the absorption of NaCl to be representative of Cl⁻ generally, based on past work (Birkmann et al., 2018), and on our own measurements of KCl absorption. Our measurements agree with Birkmann et al. (2018) from 200 to 215 nm, and they indicate that absorption of Cl⁻ is much lower than reported by Perkampus (1992).

APPENDIX FIG. A3.

Molar absorption coefficients of Cl⁻ derived from our measurements, compared with literature reports from Birkmann et al. (2018) and Perkampus (1992). We do not detect Cl⁻ absorption at >215 nm. Color images are available online.

A1.2.3. Iodide (I⁻)

Appendix Fig. A4 presents the molar absorption coefficients of NaI and KI derived from our measurements. Where our data for these molecules overlap in spectral coverage, they are identical to within experimental uncertainty, consistent with past reports that the anion dominates ultraviolet (UV) absorption (Birkmann et al., 2018). We, therefore, combine the NaI and KI molar absorptivity measurements to synthesize a generalized I⁻ absorption spectrum, which extends the longwave limit of spectral coverage of I⁻ absorption from 252 to 280 nm. Our measurements agree with Birkmann et al. (2018) from 200 to 252 nm.

APPENDIX FIG. A4.

Measured molar absorption coefficients of I⁻, compared with literature reports from Birkmann et al. (2018) and Guenther et al. (2001). Our data extend spectral coverage of I⁻ absorption to 280 nm.We do not detect I⁻ absorption at >280 nm. Color images are available online.

Our measurements generally agree with Guenther et al. (2001), but they disagree slightly at their longest wavelengths of measurement.

A1.2.4. Fe²⁺, as Fe(BF₄)₂

We represent the absorption of ${Fe}_{aq}^{2+}$ by the absorption spectrum of Fe(BF₄)₂. Other investigators (e.g., Tabata et al., 2021), however, take the absorption spectrum of ${Fe}_{aq}^{2+}$ from that reported by Heinrich and Seward (1990). Heinrich and Seward (1990) derived their Fe²⁺ by reducing iron metal (Fe⁰) with HCl. This means that Cl⁻ is present as a counter-ion in this solution, which influences the complexation and hence the UV spectrum of Fe²⁺. Most significantly, the use of Cl⁻ as a counter-ion leads to higher longwave UV absorption, and lower shortwave absorption (Fontana et al., 2007). Comparing the UV molar absorption coefficients of solutions of FeCl₂, FeSO₄, Fe(NH₄SO₄)₂, and Fe(BF₄)₂, Fontana et al. (2007) reported solutions of Fe(BF₄)₂ to have the fewest bands attributable to complexes other than the $Fe{\left(H_{2}O\right)}_6^{2+}$ expected to form from ${Fe}_{aq}^{2+}$ in isolation. We, therefore, follow Fontana et al. (2007) in not using the absorption spectrum of FeCl₂ solution to represent the absorption spectrum of ${Fe}_{aq}^{2+}$ in isolation, and instead use Fe(BF₄)₂. In natural waters, anions may, be present, which could enhance the absorptivity of the ferrous iron by complexation; the absorption of Fe²⁺ therefore constitutes a lower bound on the UV absorption of ferrous compounds.

Our measurement of the Fe(BF₄)₂ molar absorption coefficients in the UV is given in Appendix Fig. A5. Our measurements generally agree with those extracted from Fontana et al. (2007), but are slightly lower than theirs from 240 to 280 nm. Our measurements may therefore slightly underestimate UV absorption in this range. Our conclusions, which are at the order-of-magnitude level, are robust to this possibility.

APPENDIX FIG. A5.

Measured molar absorption coefficients of Fe(BF4)2, taken as representative of ${Fe}_{aq}^{2+}$, compared with literature reports from Fontana et al. (2007). Color images are available online.

A1.2.5. Ferrocyanide

Appendix Fig. A6 presents the molar absorption coefficents for K₄Fe(CN)₆ (ferrocyanide) derived from our measurements. The molar absorption coefficients we report agree with those extracted from Ross et al. (2018), but we extend the spectral coverage blueward to 200 nm.

APPENDIX FIG. A6.

Measured molar absorption coefficients of K₄Fe(CN)₆, taken as representative of ferrocyanide, compared with literature reports from Ross et al. (2018). Our data extend spectral coverage of K₄Fe(CN)₆ absorption down to 200 nm. Color images are available online.

A1.2.6. Nitrate

Appendix Fig. A7 presents the molar absorption coefficents for NaNO₃ (nitrate) derived from our measurements. The molar absorption coefficients we report are in reasonable agreement to those we extract from Mack and Bolton (1999) and Birkmann et al. (2018).

APPENDIX FIG. A7.

Measured molar absorption coefficients of NaNO₃, taken as representative of the ${NO}_3^{-}$ ion, compared with the literature absorption spectrum of ${NO}_3^{-}$ of Mack and Bolton (1999) and Birkmann et al. (2018). Our measurements agree well with the literature sources. Color images are available online.

A1.3. Literature-Derived Molar Decadic Absorption Coefficients

In this section, we show the molar decadic absorption coefficients we extracted from the literature. The literature data typically do not include uncertainty estimates. As in Section A1.2, we assume an error of 5% on the measurements reported by Birkmann et al. (2018). We propagate this uncertainty under the assumption of uncorrelated, normally distributed error, as with our own uncertainty estimates.

A1.3.1. Sulfite

We combined the measurements of Beyad et al. (2014) and Fischer and Warneck (1996) to obtain an absorption spectrum for ${SO}_3^{2-}$. Specifically, we used the entire dataset of Beyad et al. (2014), and we concatenated the molar absorptivities of Fischer and Warneck (1996) at shorter wavelengths where data from Beyad et al. (2014) were unavailable (Appendix Fig. A8).

APPENDIX FIG. A8.

Molar absorption coefficients of ${SO}_3^{2-}$, derived from Beyad et al. (2014) and Fischer and Warneck (1996). Color images are available online.

A1.3.2. Bisulfite

We combined the measurements of Beyad et al. (2014) and Fischer and Warneck (1996) to obtain an absorption spectrum for ${SO}_3^{2-}$. Specifically, we used the entire dataset of Beyad et al. (2014), and we concatenated the molar absorptivities of Fischer and Warneck (1996) at shorter wavelengths where data from Beyad et al. (2014) were unavailable. We used a log-linear extrapolation to connect the two datasets (Appendix Fig. A9).

APPENDIX FIG. A9.

Molar absorption coefficients of ${HSO}_3^{-}$, derived from Beyad et al. (2014) and Fischer and Warneck (1996). Color images are available online.

A1.4. Bisulfide

We derived molar absorption coefficients for HS⁻ in the UV from the absorbance spectra of 50 μm HS⁻ reported by Guenther et al. (2001) (Appendix Fig. A10). These data terminate at 260 nm, and our procedure was inadequate to reliably constrain molar absorptions at longer wavelengths. We, therefore, caution that HS⁻ may have additional absorption at longer wavelengths than we consider.

APPENDIX FIG. A10.

Molar absorption coefficients of ${SO}_3^{2-}$, extracted from Guenther et al. (2001). Color images are available online.

A1.4.1. Carbonate and bicarbonate

We draw molar absorption coefficients for carbonate (${CO}_3^{-}$) and bicarbonate (${HCO}_3^{-}$) from the work of Birkmann et al. (2018) (Appendix Fig. A11). These measured spectra terminate abruptly at relatively short wavelengths. We are not aware of constraints on the absorption of bicarbonate and carbonate at longer wavelengths, and our efforts to reliably constrain absorption at these longer wavelengths were unsuccessful. We, therefore, caution that carbonate and bicarbonate may have additional absorption at longer wavelengths than we consider.

APPENDIX FIG. A11.

Molar absorption coefficients of ${CO}_3^{2-}$ and ${HCO}_3^{-}$, from Birkmann et al. (2018). Color images are available online.

A1.4.2. Other Fe²⁺-derived compounds

FeCl₂ and FeSO₄ are potential complexes of Fe²⁺ in natural waters, and nitroprusside and ferricyanide are potential photochemical derivates of ferrocyanide. We extract molar absorption coefficients of FeCl₂, FeSO₄, Na₂Fe(CN)₅NO (nitroprusside), and $Fe{\left(CN\right)}_6^{3-}\left(ferricyanide\right)$ from the literature (Fontana et al., 2007; Strizhakov et al., 2014; Ross et al., 2018). Iron and cyanide are extremely absorptive in the UV (Appendix Fig. A12).

APPENDIX FIG. A12.

Literature molar absorption coefficients of Na₂Fe(CN)₅NO (sodium nitroprusside, SNP; Strizhakov et al., 2014), $Fe{\left(CN\right)}_6^{3-}$ (ferricyanide; Ross et al., 2018), FeCl₂ (Fontana et al., 2007), and FeSO₄ (Fontana et al., 2007). Also shown are our derived molar absorption coefficients of K₄Fe(CN)₆ (ferrocyanide) and Fe(BF₄)₂, for context. The molar absorption of Fe²⁺ varies dramatically depending on the anions with which it is complexed. Color images are available online.

Appendix A2. Estimating Photochemical Timescale Enhancement

Aqueous absorbers can decrease the rates of photochemical processes (e.g., photolysis, photoproduction) and hence increase their timescales by attenuating the UV radiation that drives them. To gain an approximate sense of the magnitude of this effect for the prebiotic natural water endmember scenarios we considered in the present study, we developed a simple formalism to crudely estimate the enhancement of timescales of photoprocesses in natural waters of nonzero absorbance. Specifically, we crudely estimated the enhancement in timescales of photochemical processes in well-mixed aqueous reservoirs by observing that the photochemical rate is linearly proportional to the UV flux (Bolton et al., 2015). Then, integrating the rate over the wavelengths of irradiation and the depth of the reservoir, and defining the timescale T corresponding to a reaction rate r to be $T=\frac{1}{r}$:

$$\frac{T}{T_0}=\frac{{\int }_{{\lambda }_1}^{{\lambda }_2}d\lambda k\left(\lambda \right){\int }_0^{d}dx}{{\int }_{{\lambda }_1}^{{\lambda }_2}d\lambda k\left(\lambda \right){\int }_0^{d}dx{10}^{-a\left(\lambda \right)x∕cos\theta }}$$ (A1)

$$=\frac{d{\int }_{{\lambda }_1}^{{\lambda }_2}d\lambda k\left(\lambda \right)}{{\int }_{{\lambda }_1}^{{\lambda }_2}d\lambda k\left(\lambda \right){\int }_0^{d}dx{10}^{-a\left(\lambda \right)x∕cos\theta }}$$ (A2)

where T₀ is the photochemical timescale at the surface, T is the photochemical timescale in the reservoir, d is the depth of the reservoir, $a\left(\lambda \right)={\Sigma }_i{\epsilon }_i\left(\lambda \right)c_i$ is the decadic absorption coefficient of the reservoir as a function of wavelength, θ is the slant angle of the incident radiation, and k(λ) is the wavelength-dependent photoreaction rate under early Earth surface conditions. We take θ = 60° (cosθ = 0.5), which is consistent with both the angle of the diffuse stream in the two-stream radiative transfer formalism used to calculate the surface radiation field, and the assumed solar zenith angle in the atmospheric photochemical modeling on which our atmospheric transmission calculation is based (Rugheimer et al., 2015; Ranjan and Sasselov, 2017). k(λ) may be (and generally is) a relative quantity, determined relative to a reference wavelength. k(λ) have estimated for the photolysis of 2-aminooxazole, 2-aminoimidazole, and 2-aminothiazole, and for the photo-deamination of cytidine (Todd et al., 2019, 2020).

We are not aware of a measurement of k(λ) for the photolysis of the nucleobases; for this prebiotically important process, we crudely estimate the timescale following the simplifying assumption of Cleaves and Miller (1998), that is, assuming that the photolysis rate is proportional to the irradiation at 260 nm: . We emphasize this approximation to be very crude; the peak absorbance of the biogenic nucleobases, while in the broad vicinity of 260 nm, often differs from 260 nm, and varies by nucleobase, pH, and structure of the incorporating molecule (Voet et al., 1963), and the wavelength-dependent quantum yield of photolysis of the nucleobases to our knowledge has not been strongly constrained in this wavelength region. This approximation should, therefore, be considered indicative only, and a call to detailed characterization of the wavelength-dependent nucleobase photolysis rates under early Earth conditions as conducted by Todd et al. (2019) for the 2-aminoazoles.

Applying the calculation above, we estimate enhancements in the photolysis timescales of 2-aminooxazole and 2-aminoimidazole by a factor of a few (2–5 × ) in closed-basin carbonate lakes (1 m-deep lake for the low-absorption endmember, 10 cm-deep lake for the high-absorption endmember). We estimate enhancements in the photochemical timescales of all five processes considered here by two orders of magnitude in the high-absorption end-member for the ferrocyanide lake scenario (1 m-deep lake). We emphasize the need for caution in extrapolating photochemical timescales to overall chemical timescales. Just because a process is photochemically inhibited does not mean that it does not proceed. For example, photoabsorption by Br⁻ can generate Br radical (Zafiriou, 1974); if Br efficiently destroys 2-aminooxazole, then the overall lifetime of 2-aminooxazole is not necessarily enhanced in Br⁻-rich waters despite inhibition of the direct photolysis pathway. Our calculations refer to specific photochemical processes; experiments are required to confirm whether the photochemical timescales are represenative of the overall timescales in realistic prebiotic mixtures.

Author Contributions

S.R. conceived and led the study; G.G.L and C.L.K. performed the measurements; C.L.K. synthesized data into molar decadic absorptivities and estimated errors; S.R. and A.H. performed modeling and literature data extraction for comparison; and S.R., D.D.S., and Z.R.T. explored prebiotic implications.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported in part by grants from the Simons Foundation (495062, Ranjan) (3290360, Sasselov) via SCOL. Z.R.T. acknowledges support from the NASA Hubble Fellowship Program, award# HST:HF2-51471.

Associate Editor: Christopher McKay