Sequential Ammonia and Carbon Dioxide Adsorption on Pyrolyzed Biomass to Recover Waste Stream Nutrients

The amine-rich surfaces of pyrolyzed human solid waste (py-HSW) can be “primed” or “regenerated” with carbon dioxide (CO2) to enhance their adsorption of ammonia (NH3) for use as a soil amendment. To better understand the mechanism by which CO2 exposure facilitates NH3 adsorption to py-HSW, we artificially enriched a model sorbent, pyrolyzed, oxidized wood (py-ox wood) with amine functional groups through exposure to NH3. We then exposed these N-enriched materials to CO2 and then resorbed NH3. The high heat of CO2 adsorption (Qst) on py-HSW, 49 kJ mol–1, at low surface coverage, 0.4 mmol CO2 g–1, showed that the naturally occurring N compounds in py-HSW have a high affinity for CO2. The Qst of CO2 on py-ox wood also increased after exposure to NH3, reaching 50 kJ mol–1 at 0.7 mmol CO2 g–1, demonstrating that the incorporation of N-rich functional groups by NH3 adsorption is favorable for CO2 uptake. Adsorption kinetics of py-ox wood revealed continued, albeit diminishing NH3 uptake after each CO2 treatment, averaging 5.9 mmol NH3 g–1 for the first NH3 exposure event and 3.5 and 2.9 mmol NH3 g–1 for the second and third; the electrophilic character of CO2 serves as a Lewis acid, enhancing surface affinity for NH3 uptake. Furthermore, penetration of 15NH3 and 13CO2 measured by NanoSIMS reached over 7 μm deep into both materials, explaining the large NH3 capture. We expected similar NH3 uptake in py-HSW sorbed with CO2 and py-ox wood because both materials, py-HSW and py-ox wood sorbed with NH3, had similar N contents and similarly high CO2 uptake. Yet NH3 sorption in py-HSW was unexpectedly low, apparently from potassium (K) bicarbonate precipitation, reducing interactions between NH3 and sorbed CO2; 2-fold greater surface K in py-HSW was detected after exposure to CO2 and NH3 than before gas exposure. We show that amine-rich pyrolyzed waste materials have high CO2 affinity, which facilitates NH3 uptake. However, high ash contents as found in py-HSW hinder this mechanism.


■ INTRODUCTION
In communities lacking sewerage, indiscriminate disposal of nutrient-enriched liquids and pathogen-laden solid wastes poses environmental and sanitation hazards. 1,2 At the same time, between 70 and 90% of waste nitrogen (N) is excreted in the urine in the form of urea (CH 4 N 2 O), a commonly applied N fertilizer worldwide. 3−5 Technologies for N removal from wastewater can help reduce environmental contamination while promoting recovery of fertilizer nutrients as soil amendments. 6,7 Zeolites, ion-exchange resins, and activated carbon have been investigated to trap plant-available N species such as ammonium (NH 4 + ) and nitrate (NO 3 − ) from liquid streams. 8−15 Ammonia gas (NH 3 ) is another concern due to its high toxicity, with point-source pollution arising from fertilizer, 16,17 animal production, 18 and composting. 19 Ammonia capture on solid sorbents occurs through hydrogen bonding 20 and acid−base reactions, as NH 3 functions both as a proton-accepting Bronsted base and as an electrondonating Lewis base. 17,21,22 In the presence of water, NH 3 adsorbs to solid sorbents electrostatically as NH 4 + . 8,178,17 In dry systems, NH 3 adsorption occurs through nucleophilic addition via interaction of the lone electron pair on NH 3 −N and an electrophilic C on the sorbent scaffold. 21 Hydrogen bonding between NH 3 and oxygenated functional groups is another mechanism for NH 3 adsorption. 20 For this reason, oxidized carbonaceous substrates such as graphene oxide are excellent NH 3 sorbents. 17,23,34 Dry scrubbing of NH 3 has proven effective with oxidized graphene and activated carbon. 20−24 Another option for sorbents, which combines the high adsorption potential of zeolites or activated carbon, is pyrolyzed biomass or biochar. 25 Pyrolysis may convert straw, woody shrubs, and manure solids into porous, surface-functionalized adsorbents. 8 As most biomass feedstocks are waste materials, they can be locally available and do not require regeneration such as zeolites or activated carbon but can be applied as soil amendments. 25 The NH 3 retention capacity of pyrolyzed materials may vary with the physical and chemical characteristics of the original material. Pyrolyzed plant biomass or biochar contains carboxylic functional groups, which can adsorb NH 3 . 8,25 In one study, the total N content of pyrolyzed wood oxidized with hydrogen peroxide (H 2 O 2 ) increased by 9% (w w −1 ) after exposure to NH 3 gas through both physisorption and chemisorption. 26 In another study, the total N content of a low-temperature woody biochar mixed in soil increased by 0.6% (w w −1 ) through sorption of 15 NH 3 volatilizing from cow urine. Sorbed 15 NH 3 was both KCl extractable and plant available to ryegrass. 25 Unlike most plant feedstocks used for biochar production, manures including human solid waste are high in organic N and mineral ash, which may affect the sorption dynamics of NH 3 . Fecal sludge biochar was shown to be an effective NH 4 + sorbent, removing over 18 g N g −1 from NH 4 + Cl − solution. 27 Moreover, poultry biochar sorbed slightly less NH 4 + compared to wood biochar from NH 4 + NO 3 − solution, 28 19.8 vs 26.3 mg N g −1 . However, the effectiveness of ash-rich manure biochars as NH 3 sorbents is unknown. As pyrolysis gains interest for waste management, biochar derived from human waste 27 may be useful for stripping volatile NH 3 from wastewater and urine. 7,25 However, no study has investigated the creation of a N-rich fertilizer using human solid waste biochar as a sorbent for NH 3 .
The structural changes in N compounds following pyrolysis 29,30 may not enhance a material's ability to sorb NH 3 but may facilitate another type of interaction, CO 2 adsorption. Significantly greater N in 6-membered rings was detected in woody biomass with increasing pyrolysis temperature, from 300 to 700°C. A clear correlation was also observed between N-heterocycles in pyrolyzed biomass and the initial feedstock N content. 29 The acid-hydrolyzable fraction of N in biosolids was observed to decrease from 83 to 5% after pyrolysis at 550°C, with a shift toward amino− sugar compounds instead of amino acids. 30 These Ncontaining surface functional groups, primary, secondary, and tertiary amines, have properties of Lewis and Bronsted bases that are not effective in trapping NH 3 , but which are useful in chemisorbing CO 2 . Chemisorption of CO 2 onto primary amines under dry conditions generates carbamates through cooperative binding of two adjacent primary amines. Secondary amine uptake of CO 2 forms carbamic acid and does not involve cooperative binding. 31−35 Tertiary amines can only interact with CO 2 in the presence of water vapor to form bicarbonate. 35−37 What has not been investigated is whether these products resulting from the CO 2 reaction with amine groups, such as carbamate, carbamic acid, and bicarbonate, are able to bond with NH 3 .
For this study, we examined whether exposure to CO 2 enhances the affinity for NH 3 uptake in a sorbent inherently enriched with amine groups. We were interested in whether pyrolyzed human waste (py-HSW) characterized by surface basicity due to amine functional groups could sorb NH 3 following a single pre-exposure to CO 2 . To determine the robustness of surface repriming with CO 2 in facilitating NH 3 uptake without potentially confounding effects of ash minerals, we artificially enriched an ash-poor sorbent with amine functional groups through exposure to NH 3 and then evaluated whether repeated exposure to CO 2 facilitates additional NH 3 uptake. We chose a biomass-derived sorbent previously reported to have high NH 3 sorption capacity and low ash content, pyrolyzed oxidized wood (py-ox wood). 26 In using isotopically enriched 15 NH 3 and 13 CO 2 coupled with gravimetric measurements, we were able to determine total N and C uptake with repeated gas exposure. The hypotheses were: (i) a Lewis acid such as the C in CO 2 is strongly adsorbed to basic surfaces such as those of py-HSW, (ii) a Lewis base such as NH 3 is strongly retained by surface acidity created by this CO 2 adsorbed to py-HSW as well as by oxidizing pyrolyzed wood (py-ox wood), and (iii) alternating exposure to CO 2 and NH 3 increases the N retention capacity by refunctionalizing biochar surfaces to more acidic and more basic, respectively.

■ MATERIALS AND METHODS
Preparation of Ammonia Sorbents. Two types of feedstock, latrine waste or human solid waste (HSW) and maple wood chips ( Acer rubrum), were converted into NH 3 sorbents through pyrolysis at 500°C. The HSW collection process and the feedstock conversion process into biochars are outlined in the Supporting Information. Pyrolyzed maple wood was oxidized in H 2 O 2 at a ratio of 1.6:10 g mL −1 for 25 days, rinsed in deionized water (DIH 2 O), and dried at 60°C . Pyrolyzed HSW (py-HSW) was not oxidized. Prior to NH 3 or CO 2 exposure, both materials, py-HSW and pyrolyzed oxidized wood (py-ox wood), were degassed at 150°C for 12 h under vacuum on a porosimeter (ASAP 2020, Micromeritics, Norcross, GA).
Experimental Procedure. Py-HSW and py-ox wood were exposed to pure NH 3 (10 atom % (AT %) 15 N/ 14 N) and pure CO 2 (10 AT % 13 C/ 12 C) within a thermogravimetric analyzer (TGA; Q50 EGA furnace, TA instruments) in different sequences described below and outlined in the Supporting Information. Our aim was to determine whether exposure to CO 2 can enhance NH 3 uptake in a material already enriched in amine functional groups. For this, we evaluated total N uptake from NH 3 and N bonding structures in two materials exposed to CO 2 (1) py-HSW, which is characterized by inherent surface basicity due to amine functional groups, and (2) pyox wood artificially enriched with amine functional groups through NH 3 adsorption. We compared the effect of CO 2 conditioning of py-HSW on its N uptake with that of py-ox wood exposed to NH 3 , which generated a material with similar N contents yet low ash content. 26 The effectiveness of CO 2 in renewing the surface affinity for NH 3 was evaluated by repeatedly exposing our model material, py-ox wood sorbed with NH 3 , to CO 2 followed by NH 3 . Five types of gas exposure regimes were employed in triplicate: (1) py-HSW not exposed to gas (control), (2) py-HSW exposed to 13 CO 2 for 1 h followed by 15 NH 3 for 1 h (py-HSW CO 2 + NH 3 ), (3) py-ox wood not exposed to gas (control), (4) py-ox wood exposed to 15 NH 3 for 1 h (py-ox wood NH 3 ), (3) py-ox wood exposed to 15 NH 3 for 1 h followed by 13 CO 2 for 1 h followed by 15 NH 3 for 1 h followed by 12 CO 2 for 1 h followed by 15 NH 3 for 1 h followed by 13 CO 2 for 1 h (py-ox wood NH 3 + CO 2 ) (Table S1). After each type of gas exposure, the TGA was purged with argon for 1 h. The weight of adsorption from gas exposure was calculated as the difference between the end weight after the final argon purge and the initial weight of the degassed sample, prior to CO 2 or NH 3 exposure.
Material Characterization. Heat of Adsorption. The effect of amine functional groups on CO 2 uptake was determined by measuring the heat of adsorption (Q st ) of CO 2 to py-HSW and pyox wood before and after exposure to NH 3 . Samples were exposed to unlabeled NH 3 for 1 h within a TGA at 30°C. Three adsorption isotherms were measured at three temperatures: 0, 25, and 35°C on a porosimeter (model-ASAP 2020, Instrument Corp., Norcross, GA) ( Figure S1), and details on Q st measurements are provided in the Supporting Information.
Enthalpic profiles showing the Q st as a function of CO 2 surface coverage (θ) at a given temperature and pressure (P, T) were calculated for py-ox and py-HSW with the Clausius Clapeyron Adsorption Kinetics. Avrami's fractional order model (eq 2) was used to describe adsorption of CO 2 and NH 3 onto py-ox wood and py-HSW The model describes measured CO 2 sorption (q t ) over time (t) as a logarithmically increasing function of the equilibrium adsorption (q e ), the rate parameter (k A ), and the exponent (n A ). Originally developed to model phase transitions and crystal growth, the Avrami model has been recently applied to describe CO 2 adsorption onto aminefunctionalized surfaces. 38,39 The materials used in this research are comparable to amine-functionalized surfaces on account of the ambient N enrichment of py-HSW and the artificial enrichment of pyox wood through the first exposure to NH 3 . We also evaluated the performance of the Avrami model in describing NH 3 uptake. We inversely solved for shape parameters k A and n A through regression using eq 3, the linearized version of eq 2 (Table S2). Further information regarding adsorption kinetics is provided in the Supporting Information.
Fourier Transform Infrared (FTIR) Spectroscopy. Functional group chemistry was analyzed with attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR) on a Vertex 70 FTIR spectrometer (Bruker Corp., Billerica, MA) equipped with a deuterated L-alanine-doped triglycine sulfate (DLaTGS) detector. Powdered samples were automatically scanned 64 times in the midinfrared region from 4000 to 550 cm −1 with a resolution of 4 cm −1 . Atmospheric correction was performed using OPUS 7.2 (Bruker Corp., Billerica, MA) while normalization was carried out in R studio, 40 as described in the Supporting Information.
To determine changes in functional group chemistry between unexposed "control" samples, py-HSW and py-ox wood, and samples exposed to NH 3 or NH 3 + CO 2 , difference spectra were calculated. For the difference method, samples exposed to NH 3 or NH 3 + CO 2 are normalized relative to the minimum and maximum values of respective control samples. Wavenumber assignments and functional group assignments 41 are presented in the Supporting Information (Table S3).
Isotope Ratio Mass Spectrometry by Combustion. Total NH 3 −N and CO 2 −C uptake were determined using isotope ratio mass spectrometry (IRMS), based on total C and N stoichiometry, AT % 13 C/ 12 C and AT % 15 N/ 14 N ratios, and changes in sample mass before and after gas exposure. Subsamples from each experimental treatment were ground and weighed into tin capsules. Total 12,13 C and 14,15 N of samples were measured by combustion on an isotope ratio mass spectrometer (Thermo Finnigan MAT Delta Plus, Thermo Electron Corporation, Waltham, MA) coupled to an elemental analyzer (NC2500, Carlo Erba, Egelsbach, Germany). To assure complete combustion, less than 0.5 mg of the sample was mixed with the 3-fold greater weight of vanadium pentoxide (Sigma-Aldrich, St. Louis, MO).
Nitrogen and C uptake were calculated according to eq 4 (shown for N), relying on the 15,14 N and 13,12 C AT % of samples before and after gas exposure and the AT % of gas cylinders. . Thermodynamics and kinetics of CO 2 and NH 3 adsorption onto py-HSW and py-ox wood. (A) Enthalpic profile showing the heat of adsorption before and after exposure to CO 2 followed by NH 3 . The uncertainty associated with the heat of adsorption calculations is represented by shaded regions. (B) Thermograms depicting the weight change of the first replicate of py-HSW and py-ox wood exposed to CO 2 (orange) followed by NH 3 (blue), separated by an argon purge (gray). (C) The first replicate of gravimetrically measured CO 2 and NH 3 adsorption at three sequential exposure intervals in py-ox wood (indicated in gray, orange, and blue lines for intervals 1, 2, and 3, respectively) overlaid with modeled adsorption curves using Avrami's fractional order model (indicated by solid, narrow dashed, and wide dashed lines for intervals 1, 2, and 3, respectively). (D) The first replicate of gravimetrically measured CO 2 and NH 3 adsorption onto py-HSW (blue line) overlaid with modeled adsorption curves using Avrami's fractional order model (black line). Avrami's model parameters are presented in Table S2; measured and modeled gravimetric CO 2 and NH 3 adsorption is presented in Figure S3 in the Supporting Information.

ACS Sustainable Chemistry & Engineering
Nanoscale Secondary Ion Mass Spectrometry. The co-location of 15 N and 13 C in samples before and after exposure to NH 3 and CO 2 was spatially resolved using nanoscale secondary ion mass spectrometry (NanoSIMS; Cameca NanoSIMS 50L, Gennevilliers Cedex, France). Measurements were carried out at the Environmental Molecular Sciences Laboratory of the Pacific Northwest National Lab (EMSL-PNNL). Samples were measured as whole particles within the identical size range used for adsorption experiments, 150−850 μm. To achieve flat topography, particles were pressed with a glass slide into indium foil covering silicon wafers. All samples were sputtercoated with 15 nm of iridium to minimize charging. 42 Secondary ions were measured after high-current Cs + sputtering, ∼1.2 pA, at a dosing rate of 2 × 10 16 ions cm −2 , to assure that sputtering equilibrium was achieved. 42,43 Secondary ions 12 C 12 C − , 12 C 13 C − , 12 C 14 N − , and 12 C 15 N − were measured. Sputter time was converted to sputtered depth 44 using eq 5.
Additional information on the NanoSIMS analysis is presented in the Supporting Information.
Near-Edge N-ray Absorption Fine Structure. The N K-edge bonding environment of py-ox wood and py-HSW before and after gas exposure was measured with near-edge X-ray absorption fine structure (NEXAFS) spectroscopy, conducted at the Canadian Light Source SGM beamline in Saskatoon, Canada. Partial fluorescence counts were normalized by the beamline incidence flux on a gold mesh 45 using Jupyter Notebook software (Jupyter Notebook, IPython Project, 2014). 46,47 Normalized spectra were deconvoluted into constituent N species through iterative fitting to curves associated with known N-containing compounds 26,29,48,49 with Fityk software (Fityk 0.9.8). 50 Peak assignments for N form in standard compounds 26 and are presented along with the sample analyses and deconvolution method in the Supporting Information (Table S4).
X-ray Photoelectron Spectroscopy (XPS). Surface C, N, and O bonding structures on the whole particle were determined using X-ray photoelectron spectroscopy (XPS). While the probing depth for XPS is approximately 10 nm, electrons reaching detectors are from the first few atomic layers. Measurements were carried out at EMSL-PNNL using a Physical Electronics Quantera scanning X-ray microprobe (Physical Electronics, a division of ULVAC PHI). Spectral deconvolution was based on peak assignments for C, N, and O forms in five standard compounds: ammonium carbonate ((NH 4 ) 2 2+ CO 3 2− ), ammonium bicarbonate (NH 4 + HCO 3 − ), urea, proline, and valine ( Figure S2, Tables S5 and S6), along with online and published literature. 51−55 Instrument specifications and data analyses are presented in the Supporting Information.
Statistical Analysis. Data analyses were carried out with ggplot2 56,57 within R Studio. 40 Significant differences (p > 95%) in C and N uptake between py-HSW and py-ox wood were calculated using a t-test with the lm() function and associated summary.lm() method. Significant differences between samples were marked with a compact letter display of pairwise comparisons, calculated with the Tukey method for comparing a family of three estimates within the cld() function in the multcomp package. 58 The kinetics of CO 2 and NH 3 adsorption were modeled using the nonlinear regression function nls(). Additional R packages employed for this work are listed in Table S7.

■ RESULTS
Thermodynamics and Kinetics of Adsorption. Containing 4.6% (w w −1 ) total N, py-HSW is naturally enriched in organic N 59 even before exposure to NH 3 . Py-ox wood can be artificially enriched with N following a single NH 3 exposure event (py-ox wood NH 3 ), during which the total N content increased from 0.21 to 3.74% (w w −1 ). Materials naturally enriched (py-HSW) and artificially enriched (py-ox wood NH 3 ) with amine functional groups reached the CO 2 chemisorption threshold, 50 (kJ mol −1 ), 60,61 as shown by the CO 2 Q st of py-HSW, 49.4 (kJ mol −1 ) and py-ox wood exposed to NH 3 , 50.4 (kJ mol −1 ) ( Figure 1A). The sorbed CO 2 facilitated NH 3 uptake in py-HSW, as observed by the 4.3% increase in sample weight corresponding to the uptake of 3.69 mg N g −1 during NH 3 adsorption (Figures 1B and S3A). Argon purges (gray line) lowered the weight, but never to the stable, degassed weight.
The CO 2 "priming" or "regeneration" mechanism appears repeatable, as observed by the weight increase in py-ox wood following repeated exposure to CO 2 followed by NH 3 . The end weight of our model sorbent, py-ox wood NH 3 + CO 2 , exposed to three cycles of NH 3 (1 h) followed by CO 2 (1 h) was 14% greater than the degassed initial weight. The weight increase in py-ox wood was of a similar magnitude after each CO 2 exposure interval, 4.8% (w w −1 ). Exposure of py-ox wood to NH 3 had a greater effect on the sample weight than exposure to CO 2 , but the magnitude of weight increase with NH 3 exposure decreased with repeated exposure, from 8.9% (w w −1 ) after the first exposure to 5.5% (w w −1 ) after the third exposure. Gravimetrically measured NH 3 adsorption to py-HSW was 2.4-fold less (3.7% w w −1 ) than in py-ox wood sorbed with NH 3 for the first exposure due to the lower CO 2 surface coverage within py-HSW, in spite of the high CO 2 Q st of the material (Figures 1 and S3B,C). Thus, in principle, the CO 2 regeneration mechanism on basic, amine-rich surfaces increased subsequent NH 3 adsorption but was lower in py-HSW than in py-ox wood.
A good fit of the fractional order Avrami model was observed for CO 2 adsorption in both py-HSW and py-ox wood ( Figure 1C,D; Table S2, and Figure S3B,C). Values for the shape parameter n A were lower than unity for all CO 2 exposure events, 0.54−0.85, in a similar range as reported for CO 2 sorption onto a zeolite bed. 62 While the fit of the Avrami model was also good for NH 3 sorption, it was unable to model continued NH 3 sorption on py-ox wood after the first exposure event, but rather leveled off at a plateau.
Isotope Ratio Mass Spectrometry. Total N uptake was 1.5-fold greater in py-ox wood exposed to three intervals of NH 3 + CO 2 than py-ox wood only exposed once to NH 3 (pyox wood NH 3 ), 40.57 (mg g −1 ) vs 27.04 mg g −1 ( Table 1). The molar ratio of N/C uptake in py-ox wood exposed to three intervals of NH 3 + CO 2 was 4.99 mol N mol C −1 , while that in py-HSW exposed to one interval of CO 2 + NH 3 was only 0.79 mol N mol C −1 .
The extent of 15 N isotope enrichment between py-ox wood and py-HSW corroborated gravimetrically measured adsorption (Table 1 and Figure 1B). NanoSIMS measurements across ROIs and depth profiles for a single ROI revealed an order of magnitude greater 15 N uptake in both py-ox wood exposed only to NH 3 and py-ox wood exposed to three intervals of NH 3 + CO 2 (15 145−19 094‰), compared to py-ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Research Article HSW exposed to CO 2 + NH 3 (2,635‰) (Figure 2A,B; Table  S8 and Figures S4−S9). Enrichment of 15 N in py-ox wood following exposure to 15 NH 3 did not diminish with sputtering depth, even at the micron scale, and remained greater than 15,000‰ at a minimum depth of 7.5 μm (Figure 2A). Evidence for CO 2 uptake, observed in gravimetric measurements, was less apparent by the IRMS data because of the relatively high natural enrichment levels of 13 C/ 12 C (1.10−1.11 AT %) compared to 15 N (0.38−0.39 AT %) for both py-ox wood and py-HSW (Table 1). No trend in 13 C enrichment with sputtering depth was evident for py-ox wood or py-HSW samples exposed to NH 3 + CO 2 ( Figure 2B; Table S8 and Figure S4).
Spectroscopic Investigation of the Nitrogen and Carbon Bonding Environment. FTIR. The adsorption peak at 1040 cm −1 in the py-HSW spectra confirms the presence of amines 41 (peak #7 in Figure 3A and Table S3), which, in addition to other factors such as oxidation of alkali metals, contributes to surface alkalinity, as shown by its pH of 10.3 in water (Table S9). We have ruled out silicon or clay minerals as the interpretation for this peak, contrary to their presence reported for animal manures 63 and compost teas; 64 HSW was freshly collected from waste buckets and had not been mixed with soil or residual plant material. Moreover, no silicon was detected in wide-scan XPS spectra of milled py-HSW, while a prominent N peak was evident ( Figure S10). We have also ruled out ethers or primary alcohols (C−O stretch), which are more common in woody biochars. 65 Corroborating our interpretation, a peak at 1082 cm −1 has previously been interpreted as NH 3 adsorbed to Lewis acid sites in oxidized graphene. 17 We expected a high affinity between the basic amine surface functional groups in py-HSW and CO 2 , the central C atom of which behaves as a Lewis acid in binding with the free electron pair on N in amine functional groups. 66,67 In contrast to py-HSW, py-ox wood has an acidic pH (3.8 in H 2 O; Table S9), low total N content (0.21% w w −1 ), and strong IR absorbance in regions corresponding to acidic functional groups such as carboxyls at 1709 cm −1 (peak #2) and phenols at 1215 cm −1 (peak #6) ( Figure 3A and Table S3). The high affinity between acidic surface functional groups in py-ox wood and the Lewis base NH 3 was therefore expected.
FTIR difference spectra rather than nondifference spectra provided a clearer distinction of chemical changes in py-ox wood and py-HSW following exposure to NH 3 or NH 3 + CO 2 . Following exposure of py-HSW to CO 2 + NH 3 , a decrease in the secondary NH amine bend at 1570 cm −1 (peak #3) and the tertiary aromatic amine CN stretch at 1362 cm −1 was observed in the difference spectrum (peak #5; Figure 3A and Table S3). These amine functionalities are binding sites for CO 2 . Peak #3 appears right shifted in the nondifference spectrum for py-HSW CO 2 + NH 3 , causing the dip in the position of peak #3 in the py-HSW difference spectrum. The appearance of peak #4 at 1435 cm −1 in the py-HSW CO 2 + NH 3 difference spectrum indicates the presence of NH 4 + . Unlike the difference spectrum for py-HSW CO 2 + NH 3 , peak positions #3 and #5 were sharp and prominent for py-ox wood NH 3 . In the py-ox wood nondifference spectrum, peak #3 lies just right of and on the shoulder of the CC−C aromatic ring stretch peak (1615−1580 cm −1 ). The shoulder to the left of peak #3 in the nondifference spectrum of py-ox wood NH 3 and py-ox wood CO 2 + NH 3 is likely an overlap of the CC−C aromatic ring stretch and newly incorporated amines (NH bend; 1650−1550 cm −1 ). Also noteworthy is the disappearance of peaks #2 and #6 in py-ox wood NH 3 and pyox wood NH 3 + CO 2 nondifference spectra, corresponding to carboxylic acid and Lewis acid phenol that serve as NH 3 binding sites. NEXAFS. Exposure of py-ox wood to NH 3 created new spectral features in the N K-edge not detectable in unexposed samples ( Figure 3B). The lack of features in the corresponding N 1s XPS spectra ( Figure 3C) for unexposed py-ox wood may indicate significant noise in the NEXAFS N K-edge spectra due to low N contents of 0.21% (w w −1 ) ( Table 1). The 18-fold increase in total N upon exposure of py-ox wood to NH 3 (3.74% w w −1 ) and NH 3 + CO 2 (4.05% w w −1 ) engendered the formation of protonated amines at 397.27 eV 68,69 (peak #1), new pyridinic compounds at 398.76 eV, 399.54−399.81 eV, and 404.11 eV (peak #3,5,11), and new pyrroles at 402.40 eV (peak #9, Figure 3B, Tables S10 and S11). Repeated exposure of py-ox wood to CO 2 following NH 3 introduced one new spectral feature compared to py-ox wood exposed only once to NH 3 , namely, keto-substituted pyridine rings at 399.20 eV (peak #4). Following exposure of py-ox wood to NH 3 and NH 3 + CO 2 , the proportion of pyridinic N decreased by 18− 22%, while the proportion of pyrrolic N increased by 48−52%. The proportion of N in primary amines bonded to pyridinic rings increased by 10−12%.
XPS. Significant N uptake on py-ox wood surfaces exposed to NH 3 and NH 3 + CO 2 was apparent from the sizable N 1s peaks in the XPS spectra collected from exposed samples compared to XPS spectra collected from unexposed py-ox wood. Pyridinic N (peak #14, 15) comprised more than 21% of the curve area in spectra collected from both py-ox wood NH 3 and py-ox wood NH 3 + CO 2 , indicating the formation of heterocyclic N compounds with NH 3 exposure (Tables S12 and S13). Additionally, 56−60% of N forms in these samples were detected as electrostatically sorbed N (C−O−NH 4 + ) and 12−17% as NH 4 + . As no KCl extraction was performed due to the limited sample size, the XPS-based assessment of electrostatically sorbed NH 4 + is putative. The center of the N 1s spectrum for py-ox wood exposed to both NH 3 + CO 2 (399.8 eV) is shifted toward higher energies compared to py-ox wood exposed to only NH 3 (399.3 eV). The contribution of a Gaussian curve representing primary amine compounds (C−NH 2 , 399.0−399.5 eV) shifts the N 1s curve center for py-ox wood NH 3 to lower energies. Pyridinic three replicate samples of py-HSW before and after exposure to NH 3 or NH 3 + CO 2 . The marker color differentiates between gas exposure treatments (blue = py-HSW CO 2 + NH 3 and py-ox wood NH 3 + CO 2 ; yellow = py-ox wood NH 3 ; and brown = unexposed py-HSW and py-ox wood), while the marker shape shows differences in measurements in replicate ROIs (circle = 1st replicate; triangle = 2nd replicate; and square = 3rd replicate). Replicate measurements are displayed individually in the Supporting Information ( Figure S4).  Table S3. (B) Normalized N K-edge NEXAFS spectra of experimental samples. Points show measured spectra, while the black line is the modeled spectra after deconvolution. Features marked with dotted lines and numbers and relative area of Gaussians used for deconvolution are provided in Table S4. (C) Normalized intensity of counts within C 1s, N 1s, and O 1s regions of experimental samples, measured with XPS. Points show measured spectra, while the black line is the modeled spectra after deconvolution. Features marked with dotted lines and numbers and relative areas of peaks used for deconvolution are provided in Table S6. ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Research Article N (peak #14,15) and aromatic N bonded to ketone groups (peak #11) comprised over 44% of N in unexposed py-HSW surfaces. Traces of mineral N as NH 4 + were also evident, comprising about 7% of the initial N compounds in py-HSW. Following exposure of py-HSW to CO 2 + NH 3 , three new features appear, nonaromatic amidic N (peak #10), electrostatically sorbed NH 4 + (peak #12), and amine-N (peak #13). Carbon 1s peaks were identical for py-HSW before and after exposure to CO 2 + NH 3 , showing a large CC feature (peak #6) and smaller contributions from C−N (peak #5) and C−O (peak #4). Amide peaks (peak #1) increased 2.8-fold in py-HSW CO 2 + NH 3 compared to py-HSW. Spectral variation in the C 1s energy region is more evident in py-ox wood before and after exposure to NH 3 and between py-ox wood exposed to NH 3 and NH 3 + CO 2 . The curve associated with C−N (peak #5) is largest in py-ox wood NH 3 , while the higherenergy CC (peak #2) is greatest for py-ox wood exposed to NH 3 + CO 2 . The upward slant in O 1s peaks for py-ox wood exposed to NH 3 and NH 3 + CO 2 is caused by an increase in CO functionalities (peak #20) relative to C−O (peak #18).

■ DISCUSSION
Weight Increase through Adsorption. Adsorption of NH 3 and CO 2 caused a non-negligible weight increase in both py-ox wood and py-HSW. It was essential to account for the mass change to determine N and C uptake from NH 3 and CO 2 that could not be obtained with either elemental or isotope analyses. The weight gain originates from adsorption on outer surfaces as well as within micropores, as seen by the NanoSIMS depth profiles that reach at least ∼7 μm in depth. Despite the relatively low flow rate of gases within the TGA, gases penetrated into inner cavities.
The experimental conditions significantly affect the NH 3 adsorption obtained in different studies. The lower NH 3 −N uptake in py-ox wood NH 3 + CO 2 in our study compared to that in differently oxidized py-ox wood based on their O/C ratios 26 (Table S14), 40.6 vs 90.3 mg N g −1 , can be explained by differences in time of exposure and pressure. NH 3 uptake to differently oxidized py-ox wood 26 was obtained under equilibrium conditions in a vacuum of 80−800 Torr at 35°C . We did not reach NH 3 pressures corresponding to equilibrium adsorption and do not report potential NH 3 uptake, but rather N uptake under a specified flow rate and time of exposure. A similar magnitude of discrepancy in measured vs expected NH 3 uptake based on the O/C ratio was shown for 500°C pine biochar exposed to NH 3 from a vaporizing ammonium sulfate solution. 70 Ammonia uptake on py-ox wood reported here, 8.4 mg N g −1 , was comparable to values reported for low-temperature woody biochar mixed into the soil and treated with cattle urine, 25 8.6 mg N g −1 (the O/C ratio estimated based on its N and C values 70 and the ash content of a similar biochar 71 ).
Effects of Surface Chemistry on Adsorption. Strong sorption of NH 3 on acidic functional groups such as carboxyl C is well documented in the literature, 10,67,72,73 as is chemisorption of CO 2 in liquid amines 74,75 or onto aminefunctionalized scaffolds. 34,61,67,76 Unlike activated carbon, the materials studied here were not engineered for high surface area through physical or chemical activation 77 and therefore had lower surface coverage of CO 2 . Nevertheless, surface chemistry of py-HSW, naturally enriched in amine functional groups, and py-ox wood, artificially enriched in amine groups following exposure to NH 3 , facilitated strong Q st , greater than 50 kJ mol −1 . Surface-stabilized CO 2 molecules within py-HSW and py-ox wood enhanced the surface affinity for new NH 3 molecules. Thus, for py-ox wood, each subsequent round of NH 3 + CO 2 exposure resulted in incremental N enrichment. While oxidized graphene or oxidized activated carbon can sorb greater quantities of N than our materials, 22−24 our interest was the NH 3 -enrichment potential of pyrolyzed waste, py-HSW, for use as a fertilizer, benchmarking its sorption against pyrolyzed and oxidized woody biomass as an upper threshold for N uptake of such materials.
The CO 2 heat of adsorption increased with increased surface coverage for py-HSW before and after NH 3 exposure and for py-ox wood after NH 3 exposure, unlike the expected trend for monolayer gas adsorption of noninteracting gas molecules. Varying degrees of a positive dependence of Q st on surface coverage have been reported for amine-functionalized scaffolds. 78−81 Adsorbate−adsorbate interactions indicative of increasing Q st with increasing surface coverage 80 have been explained as the cooperative binding of CO 2 molecules, whereby a higher heat of adsorption for a second incoming CO 2 molecule is observed, if an adjacent binding site is occupied by a CO 2 molecule. 78,79 Altered Surface Properties with NH 3 and CO 2 Adsorption. The initial surface chemistry of unexposed pyox wood included carboxylic, ketone, and phenolic groups, promoting a high affinity for NH 3 . When adding CO 2 after this NH 3 exposure, the appearance of CN pyrimidine bound to a keto group (py-ox wood NH 3 + CO 2 in Figure 3B) demonstrates that CO 2 chemisorption on py-ox wood introduces C-moieties similar to those found in unexposed py-ox wood such as carboxylic and ketone groups. Furthermore, N compounds in py-ox wood after 1 h of NH 3 exposure were similar to N compounds generated after 3 h of NH 3 + CO 2 . Thus, sequential exposure of py-ox wood to NH 3 and CO 2 resulted in an extension of the surface chemistry outward apparently without significantly introducing new functional group types.
We anticipated greater C (from CO 2 exposure) and as a consequence N uptake (from the subsequent NH 3 exposure) in py-HSW because of its high initial N content, 4.57% (w w −1 ) or 3.26 mmol N g −1 , and because of the results from CO 2 exposure to py-ox wood than actually observed. If each CO 2 molecule is sorbed to a single amine or a pair of amines, C uptake would have been at a ratio of 0.5−1 mol CO 2 per mol initial N, 1.63−3.26 mmol CO 2 g −1 , or 19.6−39.2 mg C g −1 . If NH 3 uptake continued according to the molar C/N uptake ratio of 1.2 measured for py-HSW in Table 1, we would have expected 1.4−2.7 mmol N g −1 or 16.3−32.6 mg N g −1 . We suggest that the high ash content of py-HSW (39% w w −1 ) compared to py-ox wood (0.7% w w −1 ; Table S9) may have impeded access to pore spaces, lowering CO 2 uptake to 4.0% (w w −1 ), far below expected values (Table 1). Reduced diffusion into pore spaces may have limited access of CO 2 to amine groups, also lowering the expected effect of CO 2 exposure on subsequent NH 3 uptake.
Furthermore, residual water in py-HSW coupled with a high ash content may have "consumed" CO 2 through formation of bicarbonates and subsequent ion pair reactions with ash minerals such as K + . Indeed, oxygen XPS spectra indicated that there were small amounts of H 2 O in py-HSW before and after exposure to CO 2 + NH 3 , even after degassing, but not in py-ox wood ( Figure 3C and Table S12). Residual water has been shown to catalyze the formation of bicarbonate, 32 can interact with cationic species including NH 4 + or K + . The molar ratio of NH 3 −N vs CO 2 −C uptake in py-ox wood NH 3 + CO 2 was 2-fold greater than in py-HSW CO 2 + NH 3 , even when considering the difference in the number of exposures (the repeated exposure of py-of wood decreases NH 3 uptake compared to CO 2 ). Taken together, these results suggest that HSW's high ash content may have reduced NH 4 + adsorption following CO 2 exposure.
Metal Effects on CO 2 and NH 3 Adsorption. The increase in surface K + concentrations of py-HSW following exposure to CO 2 + NH 3 is possible evidence of potassium bicarbonate ion pairs precipitating on the surface. XPS data revealed a 2.7-fold increase in the K 2p1 peak area of py-HSW exposed to CO 2 + NH 3 compared to unexposed py-HSW ( Figure S11), as well as 4.5% AT greater surface K in py-HSW CO 2 + NH 3 compared to unexposed py-HSW (Table S14). Concentrations of other metals remained unchanged between py-HSW before and after exposure to CO 2 + NH 3 . While bulk total K + was unaffected by NH 3 and CO 2 exposure, it is possible that K + ions within the HSW migrated to fill empty sites, as has been observed for crystalline solids such as glass. 83 It is unclear whether the presence of HCO 3 − created energetically favorable conditions for K + migration to surfaces, although this may be the only explanation for the 4.5% AT point increase in surface K + following CO 2 and NH 3 exposure. Thus, while repeated, sequential NH 3 uptake occurs in py-ox wood NH 3 + CO 2 , the nature of CO 2 interactions with the ash fraction in py-HSW restricts NH 3 adsorption, as explained above in our estimated CO 2 −C and subsequent NH 3 −N uptake in py-HSW judging from initial N contents. Improved adsorption might be possible by leaching the ash from py-HSW, although rinsing with water will also lower phosphorus (P), K + , magnesium (Mg 2+ ), and calcium (Ca 2+ ) concentrations, possibly lowering the overall agronomic value of the fertilizer. Since the ash fraction in biochar generally increases with higher pyrolysis temperatures, 30,71 lowering the pyrolysis temperature of HSW may provide an avenue to change adsorptive properties for CO 2 and subsequent NH 3 .

■ CONCLUSIONS
Significant sequential adsorption of NH 3 followed by CO 2 is possible in solid porous sorbents made from organic waste materials under dry conditions. Sorption kinetics of py-ox wood demonstrated that surface affinity can switch between NH 3 and CO 2 , allowing for material accretion through repeated chemisorption. These insights may provide a path toward conversion of untreated HSW and NH 3 , volatilizing from urine into N-rich soil amendments by decentralized waste management systems. The combination of pyrolysis-based sanitation producing a biochar that can be exposed to CO 2 emitted during pyrolysis merits further evaluation as a retention pathway for NH 3 that can be returned to soil. The process of CO 2 and NH 3 accretion through gaseous chemisorption points to the possibility of growing a fertilizer.
Method, conversion of wood chips and human solid waste into biochar, adsorption experiment with NH 3 and CO 2 , preparation of materials for heat of CO 2 adsorption, fitting the Avrami model to adsorption data, FTIR spectral processing, processing of NanoSIMS data, NEXAFS spectral processing, XPS instrument settings and data analyses, Table S1, experimental treatment structure; Figure S1, adsorption isotherms of CO 2 on py-HSW and py-ox wood before and after exposure to NH 3 at three temperatures 0, 25, and 35°C; Table S2, Avrami model parameter values; Table S3, FTIR wavenumber range assignments; Table S4, nitrogen forms and corresponding Gaussian peak assignments for NEXAFS; Figure S2, normalized XPS spectra for standard compounds; Table S5, binding energies and full-width at half-maximum values of standard compounds used for the deconvolution model; Table S6, the binding energy assignments for XPS based on standard compounds; Table S7, R software packages used; Figure  S3, thermograms of the weight of change with exposure to CO 2 and NH 3 ; Table S8, nanoscale secondary ion mass spectrometry ion ratios; Figure S4, isotopic enrichment at depth using NanoSIMS; Figure S5, NanoSIMS images of py-ox wood; Figure S6, Nano-SIMS images of py-ox NH 3 ; Figure S7, NanoSIMS images of py-ox wood NH 3 + CO 2 ; Figure S8, NanoSIMS images of py-HSW controls, unexposed to NH 3 or CO 2 ; Figure S9, NanoSIMS images of py-HSW CO 2 + NH 3 ; Table S9, chemical and physical properties of py-ox wood and py-HSW; Figure S10, wide-scan XPS spectra; Table S10, proportion of nitrogen bond forms determined by NEXAFS; Table S11, nitrogen forms measured with N K-edge NEXAFS and full-width at half-maximum values; Table S12, proportion of C, N, and O measured with XPS and determined from a deconvolution model using standards; Table S13, binding energies and full-width at half-maximum values of samples determined with XPS; Table S14, carbon, nitrogen, and oxygen contents determined by IRMS and XPS; Figure S11, narrow scan XPS spectra in the C 1s region; Supplementary references (PDF)