Electricity-powered artificial root nodule

Root nodules are agricultural-important symbiotic plant-microbe composites in which microorganisms receive energy from plants and reduce dinitrogen (N2) into fertilizers. Mimicking root nodules using artificial devices can enable renewable energy-driven fertilizer production. This task is challenging due to the necessity of a microscopic dioxygen (O2) concentration gradient, which reconciles anaerobic N2 fixation with O2-rich atmosphere. Here we report our designed electricity-powered biological|inorganic hybrid system that possesses the function of root nodules. We construct silicon-based microwire array electrodes and replicate the O2 gradient of root nodules in the array. The wire array compatibly accommodates N2-fixing symbiotic bacteria, which receive energy and reducing equivalents from inorganic catalysts on microwires, and fix N2 in the air into biomass and free ammonia. A N2 reduction rate up to 6.5 mg N2 per gram dry biomass per hour is observed in the device, about two orders of magnitude higher than the natural counterparts.


Supplementary Note 1
The advantage of using microwire array system compared to a biofilm system.
Besides our microwire array system, another approach that utilizes biofilm to fix nitrogen has been reported recently. [1][2][3][4] The formation of biofilm is usually concurrent with the introduction of other non-N2-fixing microbial strains 1,4 , the synthesis of extracellular polymeric substances, 3,5 as well as changes in genetic regulation of N2 fixation. 6 These factors probably contribute to our observation that the rate of nitrogen fixation reported in existing literature using electrochemical method (2.6 mg•L ─1 •hr ─1 and 0.008 mg•L ─1 •hr ─1 in reference 11 and 12) are smaller than the value reported in our current work (4.8 mg•L ─1 •hr ─1 for X. autotrophicus strain, see Methods). Moreover, as fundamentally both biofilm and our approach require a material's surface for attachment, we consider the scalability of these two approaches comparable. In the future, scaling up our integrated approach can be much accelerated by taking advantages of the know-how developed in biofilm catalysis. 5 Our current approach offers a more general method of employing electricity to culture N2fixing diazotrophs, as compared to the alternative approach based on biofilms. The biofilmbased approach requires that the diazotroph maintains its N2-fixing functionality in the biofilm.
Due to the complexity of the genetic regulation of N2 fixation, 6 not all diazotrophs can remain functional in this approach and to our knowledge up to now only a few examples of biofilmbased N2-fixation in air are reported. 1, 3,4 In our approach, the introduction of inorganic wire electrodes removes the restriction of biofilm formation and allows a broader range of microbes to be incorporated. This is particularly attractive in the context that many diazotrophs as plant growth-promoting microbes 7 are desirable not only because they fix N2 but also because they secrete beneficial plant hormones. We posit that our approach can be a general method for microbes that deliver nitrogen fertilizer and/or plant hormones, which will not be genetically expressed with the formation of biofilm.

Supplementary Note 2 Design rationale of the wire array morphology
Here we aim to discuss the rationale underlying our design of the wire array morphology. Three major factors are taken into the consideration: 1) the efficacy of constructing O2 gradient without significantly hampering the mass transport of N2; 2) the practicality of device characterization; 3) the practicality of wire array synthesis. First, a wire array of a suitable wire length l is desired, as a short wire is not effective to create the hypoxic domain for biological N2 fixation while an excessively long wire will mitigate N2 transport. As the diffusion layer S7 thickness is about 20 μm in our setup, we consider that wires of l = 50 μm are suitable as a proof-of-concept. Second, the validation of O2 gradient and the characterization of microbial population relies on optical microscopes, which demands a sufficiently high transparency of the device. A wire array of l = 50 μm with periodicity p larger than 10 µm are needed in order to satisfy this requirement, as longer wires with smaller periodicity pose additional characterization challenges. Third, as a proof-of-concept, chemical etching was used to create the desirable microstructures with high fidelity. Such an etching method will difficult to yield long wires of smaller diameters. Due to this consideration, a diameter d = 4 µm was set to facilitate sample preparation while leave enough "open space" among the wire arrays for the diazotrophic microbial population. In general, the concept of electricity-driven artificial rootnodule is not limited by these practical constraints. We envision that advanced preparation techniques are capable to scale up the device with lower cost, which will be beneficial to the proposed application in the long run.

Comparison of doubling time between bacteria grown in the artificial root nodule and grown in autotrophic environment.
The literature reported doubling time for planktonic B. japonicum and X. autotrophicus are 10 and 12 hours, respectively, in liquid medium under autotrophic conditions. 8,9 Our experimental observation leads to doubling times of about 28 and 90 hr, respectively. While our observed doubling times are larger than the literature values, we would like to note that these literature values were obtained under optimal environments with well-defined conditions (e.g. relatively low cell density and strictly controlled atmosphere). In contrast, the heterogeneity of our environment in air may yield additional oxidative stress for the microbial growth.
Nonetheless, the slowed microbial growth in comparison to aqueous cultures in literature is not in conflict with the observed rate of N2 fixation, higher than the ones in symbiotic root nodules, as noted in Supplementary Table 1. As speculated in literature, 10-12 it is likely that the absence of plant tissue and their regulation on microbial metabolism in the natural symbiotic system helps to maintain the high rate of N2 fixation observed in our system.

Supplementary Note 4 Comparison between energy efficiencies of electricity-powered artificial root nodule and natural root nodule
Though the energy cost in our system is seemingly large (1.5×10 4 and 2.6×10 4 kJ per g nitrogen for X. autotrophicus and B. japonicum strains, respectively), our biological | inorganic hybrid systems are already much more efficient than the symbiotic systems in natural root nodules.

S8
The energy costs of our approach are lower than the value for the natural symbiotic N2 fixation (4.2×10 4 ~ 8.4×10 4 kJ per g nitrogen, see the estimation below).
In nature, symbiotic bacteria inside the root nodule consume carbohydrates produced by plant photosynthesis to generate energy to power nitrogen fixation process. The energy efficiency for nitrogen fixation in natural root nodules are based on the following assumptions: 1) The energy efficiency for natural photosynthesis typically do not exceed 1%, 13 and 2) in root nodules, 10 ~20 g of fixed is consumed to power the fixation of 1 g of nitrogen. 14 We take glucose (C6H12O6), as the representative molecule of carbohydrates. The energy needed to generate one mole of glucose from CO2 and H2O, the materials for photosynthesis, is ~3×10 3 kJ. 15 According to assumption 1, the energy needed to produce one mole of glucose is 3×10 5 kJ via plant photosynthesis. A mole of glucose, consumed by the root nodule by respiration, is enough for the fixation of 3.6 ~ 7.2 g nitrogen, according to assumption 2. Based on the above calculation, ignoring the energy lost in the transporting of molecules, 3×10 5 kJ energy is needed for the fixation of 3.6 ~ 7.2 g nitrogen, which corresponds to an energy efficiency of 4.2×10 4 ~ 8.4×10 4 kJ per g nitrogen for the root nodule.

Delivery of reducing equivalent to microbes by Pt
Pt is a good electrocatalyst for the reduction of both proton and dioxygen. However, we note that there are literature precedence suggesting that the presence of metals, including Pt, indeed promote the electron transfer between the microbe and inorganic electrode, when both moieties are in close proximity. [16][17][18][19] Indeed, the in situ fluorescent image in Fig. 3c suggests that after a 120-hr operation the N2-fixing microbes prefer an electrode/microbe interface. There are many possible explanations for the observation in Fig. 3c, and one of them is the promoted charge transfer due to the presence of Pt. Therefore, we are unable to experimentally exclude the possibility that Pt can facilitate a direct pathway of charge transfer between microbes and electrodes.
Although applied electrochemical potential (Eappl) does influence the O2 gradient at low overpotential ( Supplementary Fig. 5), when the value of Eappl grants large enough overpotential, to some extent the O2 gradient are no longer sensitive to Eappl. This is because that the overall O2 gradient in the liquid within the wire array is governed by the mass transport of O2, and follows the theory of porous electrode. 20 The observed O2 gradients under Eappl = 0.5 V vs.

Supplementary Note 7 Definition of "porous electrode"
In general, any electrodes that has an intercalating pores should be considered as porous electrodes. This definition includes the majority of nanomaterial-based electrochemical catalysts. [21][22][23][24][25][26][27] For example, there are many options of loading catalytic materials on a porous electrode made of graphite granules, and loading Pt nanoparticles on graphite granules remain an industrial practice for electrochemical hydrogen evolution reaction (HER) and oxygen reduction reaction (ORR); 28 . Indeed, most of nanomaterial-based catalysts benefit from the introduction of pores within the electrode, either intentionally or unintentionally, and take advantage of the increased surface area introduced by the pores.

Cheaper replacement for Pt as electrochemical catalyst.
Other earth-abundant materials can be used in lieu of Pt in our proposed approach. The design of our platform requires the surface of microwire array electrodes to be electrochemically active towards ORR and HER, while these two catalytic functionalities can be fulfilled by two different types of materials deposited onto the surface of wire electrodes. Under such a guiding principle, many earth-abundant, biocompatible ORR and HER catalysts, such as the cobaltphosphorous alloy (CoP) for HER and cobalt sulfide (CoS) for ORR, are suitable candidates. 29- 35 Incorporating earth-abundant catalysts on microwires will help to scale up the developed platform in the future.

Supplementary Note 9
Determination of exchange current density (i0) for ORR Tafel analysis 36 was implemented to determine the exchange current density i0 shown in Supplementary Table 2. Tafel analysis was conducted based on the data of linear scan voltammograms (LSV) for electrochemical O2 reduction with the fabricated wire array electrode ( Supplementary Fig. 3). Here we focused on the rising edge of the cathodic current (between 0.65 ~ 0.75 V vs. RHE, i.e. | | = 0.48 ~ 0.58 V), a practice that mitigates possible interference from the mass transport of O2. As shown in Supplementary Fig. 12 (black dots), Tafel analysis based on LSV data yields the value of i0 as 3.25 × 10 ─7 mA•cm ─2 , with a Tafel slope of 116 mV•dec ─1 .

S10
Below, we present the numerical analysis that demonstrates a constant proton concentration in our experiments, following the reported procedure: 37,38 In order to quantitatively evaluate the possible deviation of local pH near the electrode, here we define the concentration overpotential as reported: 37 Here Based on the derivation, 37 in phosphate buffer under a given electrochemical condition can be calculated as: conc = ln (− 10 (pH−pK a ) ( + 10 (pH−pK a ) − 1) + 10 (pH−pK a ) + 10 (pH−pK a ) ) (S2) in which, Here pH is the pH value of the bulk solution, pK a the pK a value for the buffer, i the current   Therefore, we conclude that our characterization method of O2 profile is not sensitive to the composition of the medium within the range of O2 concentrations measured in this work.

Supplementary Note 14 Association between microbes and wire array electrode
There are no specific driving forces such as an electrochemical one when the wire array electrode was initially exposed to microbial cultures. We also note that the gravity should not contribute to the microbial accumulation, because the whole setup, shown in Supplementary   Fig. 4, was mounted on an inverted microscope and the wire array electrodes are indeed facing down (our illustration in the main figures are schematic and meant for the readers easy to understand). We postulate that the adhesion between the microbes and electrode should contribute to the initial retention of microbes within the wire array during the inoculation.

Biofilm formation during 120-hr incubation
Significant formation of biofilms was not found after 120-hr incubation of microbes in our nitrogen fixation experiments, based on the optical fluorescent images in Fig. 3c and 3f  The components were dissolved in 1 L DI water and filter-sterilized. The solution is stored under 4 °C, and vigorously shaken and sonicated before use.         Supplementary Fig. 11: Schematic of the electricity-driven artificial root nodule that yields free-ammonia that can be potentially applied to crops. DW, dry weight of biomass in the reactor. The proposed device can be a standalone system in the farm. In such a system, renewable electricity will power the proposed platform, yielding an aqueous mixture of nitrogen fertilizer. The aqueous solution will be applied to the crop field subsequently.