Weathering features of a remineralizer in soil under different land uses

– The objective of this work was to analyze the mineralogical, morphological, and compositional modifications resulting from the weathering of diabase grains buried into soil under different land uses for up to 378 days. Samples of comminuted diabase were put into polyester bags and buried into soil under corn crop, elephant grass, and woods, being unburied and evaluated after four time periods. The samples of the remineralizer (RM) were analyzed by X-ray diffractometry, total chemical analysis, scanning electron microscopy, and Al (Al d and Al o ) and Fe (Fe d and Fe o ) contents extracted by sodium dithionite-citrate-bicarbonate (DCB) and ammonium oxalate (AAO) solutions. Plagioclases and pyroxenes were the most weathered minerals in all three land uses and showed the same pattern of elemental gains and losses. The characterization of Fe and Al solubility in DCB and AAO showed that the greatest change in these elements was from the lithogenic and crystalline to the pedogenic and amorphous phase, when compared with their total content. Plagioclases and pyroxenes were the most weathered minerals, and Fe and Al show a great transfer from the crystalline to the amorphous phase, with values up to 26 and 175, respectively, for the ratios between bags with RM/Fe o and RM-control and bags with RM/Al o and RM-control.


Introduction
Remineralization, the application of powdered rocks into soils, is an agricultural practice to improve edaphic properties and increase biomass productivity (Harley & Gilkes, 2000;Leonardos et al., 2000;Van Straaten, 2006;Manning, 2010;Silva et al., 2017).It is also a low-cost technology to sequester atmospheric carbon through enhanced weathering and, potentially, lead to carbon mineralization (Manning et al., 2013).The effect of remineralizers (RMs) varies according to several factors, but, ultimately, it is related to the dissolution rate of their minerals (Manning, 2010;Ramos et al., 2015).
Mineral dissolution is a complex phenomenon.Its characterization by standard chemical parameters, such as thermodynamic stability constants and dissolution rates, is determined in laboratory conditions (Garrels & Christ, 1965;Palandri & Kharaka, 2004), which is very useful as a starting point, but may deviate from the dissolution rate in field conditions by a factor as great as 10 2 (White & Brantley, 2003;Yokoyama & Matsukura, 2006;Ganor et al., 2007).This is so because variables as the mineral:solution ratio, bioweathering, and seasonal variations in moisture and temperature, among others, greatly diverge from those in laboratory conditions (Benedetti et al., 1994;Yokoyama & Matsukura, 2006;Ganor et al., 2007).Therefore, to properly evaluate and manage the use of RMs in soils, the dissolution of these rocks in field conditions must be well understood and parametrized.Daval et al. (2018), when analyzing feldspar grains buried into soil for 4 years and comparing the results with those found for minerals weathered under controlled conditions in a laboratory, reported a smaller dissolution rate in the field, which was attributed to the formation of amorphous silica surface layers.Despite this interesting result, the authors did not detail the release of elements from the feldspar nor the formation of secondary phases, possibly because their work was carried out with a single mineral.Gill (2014) also concluded that feldspars of the plagioclase group and pyroxenes are the major minerals in basalts and diabases, and that their dissolution in soil increases pH and cation exchange capacity, but decreases exchangeable Al, depending on the applied rate and particle size (Anda et al., 2015).
Grains of the diabase RM are basic rocks rich in plagioclases and pyroxenes, minerals that weather fast when compared with other ones such K-feldspars and muscovite (White et al., 1990).In Brazil, those grains are abundant in the Paraná basin, being a potential source of silicon, iron, aluminum, calcium, magnesium, and potassium, particularly in the western region of the state of São Paulo.
The objective of this work was to analyze the mineralogical, morphological, and compositional modifications resulting from the weathering of diabase grains buried into soil under different land uses for up to 378 days.

Materials and Methods
The used RM is a diabase mined from the Limeira sill, located in the state of São Paulo,Brazil (22º36'31.2"S,47º21'45.7"W).The most abundant minerals found were plagioclases and pyroxenes (Figure 1) as expected (Gill, 2014).Faria (2008) describes the Na-Ca solid solution in these plagioclases as An 19 at the edges and An 40 into the core, i.e., as more sodic and more calcic, respectively.Other minerals -ilmenite, magnetite, k-feldspar, and apatite -were also identified, but in smaller amounts.
For the experiment, polyester bags of 0.15x0.15m were made using a 200 mesh cloth and then filled with 300 g RM.The particle-size distribution of the RM was: 30 g kg -1 greater than 2.00 mm, 258 g kg -1 from 2.00 to 1.00 mm, 270 g kg -1 from 1.00 to 0.50 mm, 143 g kg -1 from 0.50 to 0.25 mm, 166 g kg -1 from 0.25 to 0.125 mm, 77 g kg -1 from 0.053 to 0.002 mm, and 56 g kg -1 smaller than 0.002.
In January 30, 2014, the bags were buried in a farm in the municipality of Pirassununga, in the state of São Paulo (21°58'52.32"S,47°22'44.27"W).The climate of the region is tropical seasonal, of the Cwa type, with dry winters from June to August and hot and rainy summers from October to March (Alvares et al., 2013).The annual rainfall is 1,410 mm, and the average annual temperature is 22°C, with a maximum and a minimum average of 30.8 and 10.6°C, respectively.Data were obtained from the weather station at the campus of Universidade de São Paulo, also located in the municipality of Pirassununga, in the state of São Paulo.
The experimental areas had the same soil type, classified as a Latossolo Vermelho ácrico típico according to the Brazilian system of soil classification (Santos et al., 2018a), which is equivalent to an Anionic Acrudox (Soil Survey Staff, 2014).However, the areas had three contiguous land uses, less than 1,000 m apart: corn (Zea mays L.) fields cultivated under a no-tillage system since 2009 and before that with passion fruit (Passiflora edulis Sims); elephant grass (Pennisetum purpureum Schumach), cultivated as a protective barrier for the field for at least 10 years; and a conservation area, with a forest fragment (woods).In each land use, four bags were buried horizontally, at 0.10 m depth, totaling 12 bags, with the greater axis of the bag parallel to the soil surface.
In the corn field, the bags with the RM (RM-corn) were buried approximately 1.0 m apart from each other and 0.10 m from the corn stalks, being unburied 59, 86, 119, and 154 days later.In the elephant grass site, the bags with the RM (RM-grass) were buried at about 0.10 m from the tussocks and 2.0 m apart from each other.In the woods site, the distribution of the bags with the RM (RM-woods) was random, depending on the distribution and location of the tree trunks, ensuring equidistance from the trees around each bag.In both the grass and woods areas, the bags were unburied after 86, 182, 274, and 378 days, differing from the unburying times for corn due to the crop's cycle.
Soil samples were collected in each area, at 0.0-0.2m depth, to characterize soil fertility (Table 1) using Table 1.Chemical attributes of soil samples collected at 0.0-0.2m depth from the experimental areas under corn (Zea mays) crop, elephant grass (Pennisetum purpureum), and woods.the standard methods of the state of São Paulo (Raij et al., 2001).In both the unburied samples and in the pristine RM (RM-control), pedogenic iron oxides and amorphous metal oxides were solubilized, respectively, in sodium-dithionite-citrate-bicarbonate (DCB) (Mehra & Jackson, 1958) and in acid ammonium oxalate (AAO) (McKeague, 1978) solutions; the Fe and Al in the extracts were measured by atomic absorption spectroscopy.

Land use pH water
Aliquots of 5 g from the last unburied bags -at 154 days for the corn site and 378 days for the grass and woods sites -were dry-sieved into: coarse sand (from 1.0 to 0.5 mm), fine sand (from 0.25 to 0.125 mm), and silt (smaller than 0.05 mm), ground until passing through a 60 mesh sieve, then powder mounted and irradiated into the Miniflex II benchtop X-ray diffractometer (XRD, Rigaku Corporation, Tokyo, Japan), with a CuKα radiation of 0.154056 nm, operated at 30 kV and 15 mA, with a graphite monochromator and Ni filter.The angular range was 3 to 60 °2θ, at 0.02 °2θ step, and 1 s per step.The XRD patterns were interpreted using the Match!2.1 software (Crystal Impact, Bonn, Germany).
Total chemical composition was determined only for the longest buried RM and the RM-control, comprising Al 2 O 3 , BaO, CaO, Cr 2 O 3 , Fe 2 O 3 , MgO, MnO, P 2 O 5 , K 2 O, SiO 2 , Na 2 O, SrO, and TiO 2 .Aliquots were fused in LiBO 2 /Li 2 B 4 O 7 , at 1,000°C, cooled, and then dissolved in a solution of 4% nitric acid + 2% hydrochloric acid.The extract was analyzed by atomic emission spectrometry using inductively coupled plasma (ICP-AES); solution readings were sent to a commercial laboratory, which is why the used equipment has not been disclosed.
Based on the obtained results, the elemental mass balance (EMB) was calculated with the equation: EMB = [(RM-corn or RM-grass or RM-woods) - Negative values imply loss of elements in the weathered RM, and positive values imply accumulation.
The RM samples were also fixed on an adhesive carbon metallic support coated with gold film and observed under the LEO 435 VP scanning electron microscope (Carl Zeiss Industrielle Messtechnik GmbH, Oberkochen, Germany), in order to determine the fine and coarse sand fractions.The grains were not subjected to cleaning procedures, so that weathering features and biological attachments were preserved on grain surface.
Regarding grain morphology, plain surfaces and strait edges, associated with the cleavage of minerals, were frequently observed in the fine and coarse sand fractions of the RM-control (Figure 2 A and B), as well as occasional features of mechanical stress resulting from the comminution process (Figure 2 C and D).In general, etch pits and corroded depressions were identified in the buried grains (Figure 2 E and  F), which are associated with the initial stages of weathering (Heindel et al., 2018).Filamentous material resembling fine roots or fungal hyphae were also frequent (Figure 2 D and F).During unburying, it was common to observe roots wrapping around the bags, particularly in the grass land use.Biological corrosion was inferred from some images (Figure 2 F), and similar structures were found in gneiss saprolites in the state of Pernambuco, Brazil (Santos et al., 2018b).Some grains seemed intensely dissolved, as inferred by the lack of sharp edges, when comparing the RMcontrol with the RM-woods after 378 days of being buried (Figure 2 B and E).However, at this point, it was not possible to conclude that such morphology is the result of weathering into the soil only.Krahl et al. ( 2020) cultivated corn seedlings in pure basalt during 315 days and did not report any changes in rock grain morphology; however, the pH of the pots in their experiment was close to 9, while that of the soil used in the present study was between 4 and 5.
Changes in the intensity of diffracted X-rays are influenced by several factors (Moore & Reynolds, 1997), including the amount of the diffracting phase in the sample.The changes in the relative XRD intensities of plagioclases (0.425, 0.344, and 0.321 nm peaks) and pyroxenes (0.299, 0.294, 0.289, and 0.257 nm peaks) suggest that these are the most weathered phases in the coarse and fine sand fractions, when compared with the RM-control of each fraction (Figures 3, 4,  and 5).Krahl et al. (2020) also noted that diopside (pyroxene) and andesine (plagioclase) samples were   the most affected by dissolution in the analyzed samples.Besides the decrease observed for the relative intensities, an increase in peak width (full width at half maximum) and asymmetry seemed to have occurred, suggesting the advancement of weathering and the loss of crystallinity (Figures 3, 4, and 5), which is in agreement with the losses in elemental mass balance (Figure 6).Of the found minerals, the most dissolved were Mg and Na, since pyroxenes and plagioclase feldspars were the source of Mg and Na, respectively, and the elemental mass balance showed that the loss of Mg and Na was the greatest in all land uses, which roughly agrees with the changes observed in the XRD scan.It is interesting, though, that all land uses had the same pattern and magnitude of gains and losses of elements.
The amount of iron determined in the dithionitecitrate-bicarbonate and oxalate ammonium acid solutions (Fe d and Fe o , respectively) that was extracted from the RMs in the bags was from 9.5 to 13.3 and from 15.9 to 26.1 times greater, respectively, than that in the RM-control (Table 2).Even if some soil particles may have entered the bag, the amount of Fe d and Fe o , respectively, was roughly 2 to 3 and about 40 times greater in the RM bags (that is, in RM-corn, RMgrass, and RM-woods) than in the surrounding soil.This suggests contamination should not have been a major process to increase Fe d and Fe o in the bags.Interestingly, Krahl et al. ( 2020) found precipitates they supposed were amorphous phases on the surface of the studied grains due to the rapid oxidation of Fe(II).Despite the great increase observed for Fe d and Fe o , their values were mostly erratic along the timeline.For RM-grass and RM-woods, the greatest Fe o value was verified in the first unburied bag at 86 days, suggesting that the Fe release was fast and occurred in the first three months.In fact, the incongruent dissolution of silicates results in a great mass of chemical elements released at the beginning of dissolution, followed by a smaller, roughly constant rate of release ( WildFigure 5. X-ray diffraction scans of the silt fraction of the remineralizer used as a control (RM-control), after being buried for 154 days in soil cultivated with corn (Zea mays) (RM-corn) and for 378 days in soil under elephant grass (Pennisetum purpureum) (RM-grass) and woods (RM-woods).Qz, quartz; Pl, plagioclase; Cpx, clinopyroxenes; Ilm, ilmenite; and Mag, magnetite.
The values of Al determined in the dithionitecitrate-bicarbonate solution (Al d ) were also erratic, varying from 1.9 to 2.8 times the content in the RMcontrol, with no apparent relationship with Fe d .This could be attributed to the fact that the source of Al d is mainly the Al in isomorphic substitution in the atomic structure of reducible metal oxides, such as Fe and Mn oxides; the total Fe 2 O 3 and MnO in the RM-control was 14.4 and 0.24%, respectively.The most impressive increase in the ratio between the RM-bags and RMcontrol was found for Al determined in oxalate ammonium acid solution (Al o ), which was 120 to 175 times greater, indicating an extensive transformation in the Al forms.However, the elemental mass balance for Al 2 O 3 was about zero (Figure 6).Together, the elemental mass balance and the Al d and Al o RM-bag/ RM-control ratios suggest that, despite the loss of Al 2 O 3 being in the same order of the overall mass loss of the RM, the Al atoms had a massive transfer from crystalline and lithogenic minerals to amorphous ones.This could be attributed to the fact that the elemental mass balance accounts only for the mass of elements that leaves the "volume of control" of the system, i.e., the volume of the bags.Therefore, the obtained result is indicative of the loss of elements, as observed for MgO and Na 2 O, or of their residual concentration, as for Fe 2 O 3 and CaO, but does not inform about their phase transformations in situ -from the crystalline to the amorphous phase, for example.The large increase in Al o is certainly related to plagioclase weathering, the main and abundant source of Al in the RM, with 11.90% total Al 2 O 3 in the RM-control.Plagioclase was also the source of Na and Ca, with 3.06% total Na 2 O  ) solutions that were found in the remineralizers both in pristine conditions (RM-control) and after being buried in soil under different land uses (1) and for different periods (number of days).(Hudson, 1995) and the albitic (Na-rich) composition of the borders of the plagioclase grains that should weather before their more anorthitic (Ca-rich) core that should weather later.In addition to these processes, the mineralization of carbon or mineral carbonationwhich causes the carbon in atmospheric CO 2 to react with alkaline earth metals (usually Ca) and precipitate CaCO 3 (Moosdorf et al., 2014;Kantola et al., 2017;Taylor et al., 2017;Beerling et al., 2018) -may have occurred inside the bags, since it was observed at a much lower RM concentration in the soil (Manning, 2008;Manning et al., 2013;Washbourne et al., 2015).
If this process happened inside the bags, some amount of Ca would accumulate as CaCO 3 .In summary, the results of the elemental mass balance calculated in the present study with a very simple approach agreed roughly with those of the XRD scans.The great loss in MgO and Na 2 O fits well with the expectation that Fe-Mg silicates (mainly pyroxenes) and plagioclases should weather first.Potassium was enriched into the bags because it should be present in the K-feldspars, which are more resistant to weathering than the plagioclases (Ca-feldspars) (Bakker et al., 2019).Fe 2 O 3 accumulated residually, but, as shown by the results for DCB and AAO, has gone through a great transfer from the lithogenic to the pedogenic and amorphous forms.Aluminum went through a similar transformation from primary minerals to amorphous ones, which was up to 175 times greater for Al o in the RM-bags than in the RM-control (Table 2), possibly explaining why its loss was comparatively greater than that of Fe 2 O 3 (Figure 6).Silicon, the most abundant element in the RM (52.10% SiO 2 in the RMcontrol) should also have gone through extensive transformations since Si is a major element in the composition of silicates, pyroxenes, and plagioclases that have undergone extensive weathering (Stefánsson, 2001); however, its loss was not significant in the mass balance because its mobility is intermediary (Hudson, 1995) and it forms amorphous silica layers on the mineral's surface (Daval et al., 2011(Daval et al., , 2018;;Hellmann et al., 2013).
These findings are interesting because they serve as a reference (including some order of magnitude) to infer mechanisms of diabase dissolution and to elaborate hypotheses to test the role of RMs applied to the soil beyond the regular release of plant nutrients and potentially toxic elements.Further research could be carried out on the potential to capture atmospheric carbon, the protection of organic matter by the formation of organic-mineral complexes between organic substances and recently precipitated amorphous metal oxides, the impact of recently precipitated amorphous metals in the "fixation" of phosphate applied to agricultural soils, and the role of soil microbiomes and the rhizosphere in the dissolution rate of RMs, among others.

Conclusions
1. Plagioclases and pyroxenes are the most weathered minerals found in the diabase remineralizer (RM) buried in bags into the soil.
2. The three land uses evaluated show the same pattern of gain and losses of elements.
3. The ratios iron determined in oxalate ammonium acid solution (Fe o ) in the RM-bags/Fe o in the RMcontrol and aluminum determined in oxalate ammonium acid solution (Al o ) in the RM-bags/Al o in the RM-control express a massive transfer of these elements from the crystalline and lithogenic to the pedogenic and amorphous phase, with values from 16 to 26 and 120 to 175, respectively, but a small mobility of aluminum.

Figure 2 .
Figure2.Scanning electron microscopy images of remineralizer (RM) grains, showing: A and B, fine and coarse sand fraction of the grains of the RM used as a control (RM-control), respectively; C, etched pits and a crack probably caused by mechanical stress during comminution (arrow) in the coarse sand fraction of the RM after being buried for 154 days in soil cultivated with corn (Zea mays); D, cleavage surfaces (possibly of a feldspar) and biological filament entangled to it in the fine sand fraction in the RM after being buried for 378 days in soil with elephant grass (Pennisetum purpureum); E, rounded surfaces, possibly due to preferential edge dissolution in the fine sand fraction in the RM after being buried for 378 days in soil in the woods, contrasting with the coarse sand fraction of the RM-control; and F, biological structure (possibly biofilm) at the edge of the grain, whose corrosion seems more intense around the biological structure in the coarse sand fraction in the RM after being buried for 378 days in the woods.

Figure 6 .
Figure 6.Elemental mass balance based on the total chemical analysis of the remineralizer (RM) samples unburied after 154 days in soil cultivated with corn (Zea mays) and after 378 days in soils under elephant grass (Pennisetum purpureum) and woods, in comparison with the pristine RM (control).