**1. Introduction**

According to the World Health Organization (WHO), approximately 785 million people lack access to clean drinking water, mainly in poor countries [1], but also in some rural and even highly urbanized areas [2] in the developed world. Heavy metal contamination of the water supply is a persistent problem that dates back to antiquity [3], and some researchers speculate that lead contamination of the water supply may have played a role in the downfall of the Roman empire [3]. More recently, Fernández-Luqueño et al. [4] summarized the health effects of heavy metals, listing their contributions to disease ranging from cancer to lung failure. Despite widespread acknowledgement of these negative human health outcomes, providing uniform access to drinking water free of heavy metal contamination has proven remarkably difficult to achieve, as evidenced by recent widely reported examples [2].

Different technologies have been developed to remove heavy metals from water [5] including precipitation [5–7], sedimentation [8], flotation [8,9], membrane processes [8,10–12], electrochemical

processes [13–15], adsorption [16,17], and ion exchange [18,19]. Most of these technologies are either costly, wasteful, reliant on non-renewable resources, energy inefficient, or unable to achieve sufficiently low levels of metal concentration on their own. Of the available options, adsorption and the kindred technique of ion exchange are energy efficient and capable of achieving suitably low metal concentrations [20]. However, commercial adsorbents and ion exchange resins are derived from non-renewable resources, either petrochemicals or coal, meaning that their use has associated negative environmental impacts, creating a tradeoff between clean drinking water and mitigating climate change [21]. Accordingly, development of renewable, cost-effective, and high-capacity metal adsorbates has potential to greatly expand access to clean drinking water while minimizing other negative environmental impacts [22,23].

Recently, pyrolysis biochar has emerged as a renewable sorbent for heavy metal removal [24,25]. Unlike most activated carbons, which are produced from coal, pyrolysis biochar is produced from biomass or agricultural wastes [26]. In some cases, heavy metal capacity on biochar can reach 40 mg g<sup>−</sup>1, which is comparable with activated carbons (20–80 mg g−1) [25] or ion exchange resin capacity (20–30 mg g−1) [25]. Surface precipitation and electrostatic interactions are thought to be the key adsorption mechanisms onto biochar, and abundant oxygenated functional groups (OFG) are associated with effective cation adsorption [27,28]. Unfortunately, pyrolysis at typical conditions (>400 ◦C) tends to be ineffective for OFG formation, meaning that biochar must usually be activated to increase its adsorption capacity [29]. Various activation procedures can increase the sorption capacity of biochar, but only by adding cost, energy use, or waste generation [30]. Lastly, pyrolysis is performed in the vapor phase and requires a dry feedstock, negatively impacting the process energy balance for utilization of abundant wet wastes [31,32].

Hydrothermal carbonization (HTC) [33–35] of carbon-rich feeds, including carbohydrates, biomass, and food waste, at moderate temperature (130–250 ◦C) and autogenous pressures is a versatile, low-energy, and renewable way to produce carbon-rich materials with abundant OFGs that are known as hydrothermal chars (hydrochars) [36,37]. Possibly due to the abundance of OFGs present in hydrochars, they have greater metal sorption capacity than most pyrolysis biochars, making them especially attractive for drinking water purification [5,38,39]. For example, Regmi et al. [33] reported a hydrochar with greater Cu(II) sorption capacity than conventional activated carbon (4.8 compared with 1.8 mg g−1) [39]. Similarly, HTC compatibility with wet feeds eliminates the need for drying, benefiting process-level energy balance especially for agricultural and food waste streams [40,41].

Although HTC is a promising technology, process costs and uncertainties must be reduced to de-risk further investment [42]. Similarly, HTC can benefit by maximizing its value, which in the case of sorbent manufacture, can be accomplished by maximizing hydrochar sorption capacity [16]. Accordingly, a persistent mystery in the field of hydrochar sorption is the high capacity that the material has for metal cations despite its relatively low (<10 m2 g<sup>−</sup>1) measurable surface area. For this reason, many reports describe methods to increase hydrochar surface area [29,37,43]. Unfortunately, hydrochar activation to increase surface area again produces wastes and requires energy; moreover, the resulting capacity of the activated material often decreases on a per area basis; for example, Jain et al. [44] reported that the phenol capacity of activated carbon produced from hydrochar decreased with increasing surface area, from 0.16 to 0.13 mg g−<sup>1</sup> [16].

While pyrolytic treatment and various chemicals can increase hydrochar sorption capacity, immersion in an alkali solution at room temperature reportedly increases OFG abundance and heavy metal adsorption capacity with minimal energy requirements while generating minimal amounts of waste [28,33,37]. The mechanism of the alkali-treatment promotion effect is not clear since mild alkali treatment does not increase hydrochar surface area [45], and the conditions are not sufficient to make or break covalent bonds. Moreover, alkali treatment is sometimes reported as a necessary step for hydrochar to exhibit any heavy metal sorption capacity [29,37], while for others, alkali treatment is not reported [33]. Understanding alkali treatment is, therefore, one goal of this work.

Complicating analysis further, hydrochar OFG abundance, and sorption capacity vary depending on the properties of the feed and reaction conditions. Guo et al. [46,47] have shown that, starting with a lignocellulosic raw material, a hydrochar can be produced with greater OFG abundance than produced from carbohydrate precursors. For example, HTC from wood (initially consisting of 52.7 wt% C) produces a hydrochar with 77 wt% C [48], whereas the carbon content of hydrochar produced from glucose typically contains 53–62 wt% C [49]. Hydrochar OFG abundance tends to decrease with increasing reaction conditions such as reaction time, temperature, and water-to-biomass ratio [46–48]. For many feeds [50–54], HTC can be performed over the temperature range from 160–180 ◦C and at reaction times <12 h to promote formation of OFGs. More aggressive conditions (≥200 ◦C) are preferred for other applications [55].

The relationship between feed properties, reaction conditions, and sorption capacities provides an opportunity to synthesize a hydrochar tailor-made for a specific application, such as heavy metal adsorption. For example, Demir-Cakan et al. [56] reported that co-HTC of glucose and acrylic acid produced a hydrochar with exceptionally high OFG abundance; the resulting materials exhibited sorption capacities up to 350 mg g−<sup>1</sup> for Pb(II) and 90 mg g−<sup>1</sup> for Cd(II). In comparison, Xue et al. [16] reported peanut-hull based hydrochar sorption capacity for Cd(II) as 12.38 mg g<sup>−</sup>1.

Ideally, hydrochars could be designed for a specific application at the molecular level. Computational methods have proven effective at understanding sorption mechanisms and thereby enabling the molecular level design of metal-organic frameworks for sorption of perfluoroalkyl substances [57], nanopores for CO2 adsorption [58], and ion exchange resins and activated carbon for heavy metal sorption [59,60]. Density Functional Theory (DFT) is an especially valuable tool for studying the geometry and energetics of sorbate binding to the active site, provided that molecular structures are known that can be used as targets of rational design. Unfortunately, the structural models of hydrochar have only recently converged [61–63], making attempts at molecular level design or computational modeling of sorbate-sorbent binding difficult until now.

Recent work by Brown et al. [61] reconciled several disparate models proposed for the structure of hydrochar synthesized from glucose. Previous models inferred from infrared spectroscopy and Raman microscopy [62,64] indicated that hydrochar structures resembled activated carbon and consisted of fused aromatic cores, comprised of many aromatic rings and with OFGs present primarily as side chains. In contrast, solid-state Nuclear Magnetic Resonance (NMR) [65] and Near-Edge X-ray Absorption Fine Structure (NEXAFS) [63] indicate a structure consisting primarily of individual furan and arene groups, polymerized via short alkyl chains, and decorated with OFGs. Brown et al. [61] recognized that previously reported Raman spectra of hydrochar contained artifacts due to laser-induced pyrolysis of the hydrochar material, causing it to collapse into a condensed aromatic structure. DFT simulation of hydrochar Raman vibrations [61] then indicated that artifact-free Raman spectra were indeed consistent with the furan-arene polymer previously inferred from NMR and NEXAFS [62–64,66]. This paves the way for a molecular-level study of metal binding to hydrochar as a furan-rich polymer, thereby enabling rational design of hydrochar.

The objective of this work was molecular-level design of a hydrochar adsorbent using both experiments and simulation. To focus on generalizable mechanisms, we studied a model hydrochar synthesized from glucose for sorption of a model heavy metal, Cu(II) cations. Sorption capacity was studied before and after alkali treatment and compared with capacities measured for several activated carbon materials. Similarly, the hydrochar was characterized for OFG type and density using Fourier transforms-infrared spectroscopy (FT-IR), solid-state titration, and zeta potential measurement. Metal-hydrochar binding interactions and geometries were evaluated using DFT simulations. Chars with different types and/or densities of OFGs and other metal binding groups were custom synthesized for comparison with glucose hydrochar. The experimental and simulation results described here establish a new method for the rational design of hydrochar sorbents at the molecular level.
