Catalyzed Synthesis of Zinc Clays by Prebiotic Central Metabolites

How primordial metabolic networks such as the reverse tricarboxylic acid (rTCA) cycle and clay mineral catalysts coevolved remains a mystery in the puzzle to understand the origin of life. While prebiotic reactions from the rTCA cycle were accomplished via photochemistry on semiconductor minerals, the synthesis of clays was demonstrated at low temperature and ambient pressure catalyzed by oxalate. Herein, the crystallization of clay minerals is catalyzed by succinate, an example of a photoproduced intermediate from central metabolism. The experiments connect the synthesis of sauconite, a model for clay minerals, to prebiotic photochemistry. We report the temperature, pH, and concentration dependence on succinate for the synthesis of sauconite identifying new mechanisms of clay formation in surface environments of rocky planets. The work demonstrates that seeding induces nucleation at low temperatures accelerating the crystallization process. Cryogenic and conventional transmission electron microscopies, X-ray diffraction, diffuse reflectance Fourier transformed infrared spectroscopy, and measurements of total surface area are used to build a three-dimensional representation of the clay. These results suggest the coevolution of clay minerals and early metabolites in our planet could have been facilitated by sunlight photochemistry, which played a significant role in the complex interplay between rocks and life over geological time.

One of the major scientific questions that remains unresolved is how the origin of life occurred 1 . Within the prebiotic chemistry context of this interdisciplinary problem, recent work has shown the potential of ZnS promoted photocatalysis to harvest sunlight energy into chemical bonds 2 . The mechanism of semiconductor promoted photochemistry may have played a major role in cycling small organic compounds essential for the origin of life 3 . For example, illuminated ZnS has successfully driven several reactions of the reverse tricarboxylic acid cycle (rTCA) 3 , which is central to metabolism. Our latest work explored in detail the photoreduction of fumarate to succinate on the surface of ZnS 4 , which proceeds with a 95% yield 5 . The synergistic interaction between sunlight, photocatalysis, and organic acids in the prebiotic Earth 6, 7 could have promoted reactions otherwise not favoured, providing the foundation for present complex metabolism 3 .
Within the origin of life framework, a separate but relevant problem is the potential role of clay minerals, strong adsorbents of polar organic molecules 8 , to facilitate abiogenesis 9 . The catalytic power of clays can promote the polymerization of biomolecules and the conversion of fatty acid micelles into vesicles 9 . Indeed, the process of clay formation is key for developing an understanding of the possible roles of these minerals in the origin of life 10 . Recent work has proven the crystallization of saponite clays can proceed easily in only 20 h under relatively mild conditions in the presence of urea 11 as a catalyst. Interestingly, nowadays abundant oxalic acid, an endpoint oxidation product for all organic matter exposed to environmental oxidizers before their final conversion into formic acid and CO 2 12-14 , successfully substituted urea in the synthesis of saponite clays 15 . The role of oxalate as a chelating agent for aluminum atoms in the octahedral state has been previously studied 16 . Bare Al 3+ has been proposed to retain the hexacoordination for crystalizing the phyllite structure when its complex with oxalate is decomposed in the presence of silicate 16 .
Despite the previous knowledge in both fields, the unanswered question remaining is whether a relevant clay formation mechanism could have proceeded catalyzed by photogenerated central metabolites 3 of the rTCA cycle. In this work, succinic acid (and other organic acids) is used as a probe from the intermediates of the rTCA cycle to catalyze the synthesis of sauconite, a model zinc trioctahedral clay mineral from the smectite family similar to saponite. The work hypothesizes that succinate accommodates in the interlayer space after catalyzing the incorporation of Al 3+ into the precursor soluble gel, dominated by SiO 2 , to create the tetrahedral layer. The work shows that sauconite crystallization is related to the ability of succinate to chelate bare Al 3+ . Syntheses of sauconite are performed under variable initial concentration of sodium salts of carboxylic acids (e.g., [succinate] 0 ), pH, and temperature, to monitor during 20 h the nucleation process. Succinate is shown to catalyze the reaction of Al 3+ with silicic acid, zinc and sodium compounds, yielding an authentic 2:1 clay mineral of the smectite group. The synthesis of Zn-clay material proceeds at low temperature (<100 °C) and ambient pressure (~1 bar) as demonstrated from the structural characterization displaying the early stages of the synthesis. This characterization is performed by a combination of techniques including: 1) Powder X-ray diffraction (XRD), 2) diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy, 3) conventional transmission electron microscopy (TEM), and 4) cryogenic TEM (cryo-TEM).

Results and Discussion
The bottom-up synthesis of sauconite, with a theoretical full cell formula Na 1 , is performed from a silicic acid gel in contact with Zn 2+ , Al 3+ , and succinate 2− ions. The synthesis can be easily altered to vary the components integrated into the structure. Aiming at enhancing nucleation and to accelerate the formation of clay minerals, we designed a set of experiments to optimize the synthesis of a model sauconite.
To study the role of succinate, a central metabolite, as a catalyst promoting clay formation, we made several modifications to the gel. In the first set of experiments we varied the concentration of sodium succinate (Fig. 1). Alternatively, the nature of the organic salt is investigated by substituting succinate with sodium salts of formic acid, acetic acid, oxalic acid, and malic acid (Supplementary Figs S1-S4). In the second set of experiments, we performed the synthesis under variable temperature. In the third set of experiments, the initial pH was varied in the range 6-14. Effect of Varying the Concentration of Organic Acids. Figure 1 shows the powder XRD diffractograms after 20 h of synthesis of the dried gel with and without succinate added. When no succinate is added, an amorphous gel is observed as shown by the lack of XRD reflections and presence of an elevated background in the 2θ range 15-35° (Fig. 1A), which does not possess any features from the smectite group of 2:1 layer silicates. In contrast, the gel with succinate added at concentrations of 0.01, 0.10, 0.50, and 1.00 M (Fig. 1) shows an XRD peak emerge with a low-angle shoulder between 3.3 and 8.3° centered at 5.82° corresponding to a first-order basal reflection, d 0 0 1 of sauconite. The d 0 0 1 values, which corresponds to the thickness of the 2:1 phyllosilicate layer to layer distance (per half unit cell), increase for larger [succinate], varying between 14.2 and 15.1 Å (Fig. 1B). In addition to the (0 0 1) reflection labeled in each XRD diffractogram, six more appear at 2θ angles of 11.46 (7.72 Å), 19.68 (4.51 Å), 28.60 (3.12 Å), 33.96 (2.64 Å), 59.98 (1.54 Å), and 70.20° (1.34 Å) most likely representing the (0 0 2), (0 2 l), (0 0 5), (1 1 0), (0 6 0), and (2 2 0) basal reflections of sauconite (ICDD PDF No. 00-008-0445) 18,19 , respectively. Interestingly, the weak peak at 4.51 Å that should correspond to the 0 2 l reflection suggests the sauconite is turbostratic (irregular and not perfectly oriented) 20,21 . The peak (0 6 0) at 1.54 Å confirms the trioctahedral structure of sauconite. The reversal in the trend of d 0 0 1 with concentration above certain high [formate], or [acetate], [oxalate], or [malate] (Figs S1-S4) is associated to larger particle sizes limiting the process of expansion 22 .
The higher d 0 0 1 basal spacings for the series of sauconite in Fig. 1A with increasing succinate levels coincides with the growing total surface area (TSA) (Fig. 1B). The larger TSA with increasing [succinate] is likely attributed to greater formation of the 2:1 layered structure. It is also probable that succinate has accumulated in the interlayer space, as supported by both the increase in d 0 0 1 basal layer to layer distance and the lower than expected  Supplementary Fig. S5. In the gel, there is a broad band in the range 3100-3600 cm −1 corresponding to overlapping OH stretching bands due to water and the weaker corresponding OH-bending mode of water at 1645 cm −1 (Fig. S5, dashed line) 21 . In addition, there is a broad stretching band near 1000 cm −1 in the precursor gel assigned to Si-O. After addition of succinate, two new shoulders appear in the spectrum at ~3639 and ~3743 cm −1 , which are characteristic stretching vibrations of hydroxyl groups (ν OH ) attached to octahedral Zn ions located in the interior blocks of sauconite 23 . IR bands appear at 2951 and 2973 cm −1 in the presence of succinate and these are assigned as C-H stretching vibrations of succinate, which match the spectrum of the sodium succinate standard.
The new band at 1630 cm −1 in sauconite ( Supplementary Fig. S5) quickly grows from the stretching at 1645 cm −1 in the gel, due to OH bending vibrations of interlayer water 20,21 . This band at 1630 cm −1 due to the OH-bending mode of interlayer water is typical of 2:1 layered trioctahedral clay minerals 24 . After addition of succinate, the bands from the amorphous gel at 604 and 1009 cm −1 show the prominent development of sauconite shoulders from symmetric Zn-O vibration at 66 cm −1 , and asymmetric in plane Si-O-Si vibration at 1020 cm −122,

25
. The bands at 1404 and 1383 cm −1 in the amorphous gel are due to a small amounts of sodium nitrate left from the synthesis 26 . Spectra of sauconite also show bands at 1579 and 1296 cm −1 , corresponding to the stretching of C = O and C-O groups in the structure 19 , which are matched to sodium succinate standard ( Supplementary  Fig. S5). Interestingly, the absorbance from succinate incorporated in the interlayer is correlated to the amount added during the synthesis. The intercalation of aliphatic organic carbon in 2:1 layered clay minerals has been reported in pure standards and natural clays extracted from sediments 27,28 . Overall, two analytical methods (XRD and DRIFT spectroscopy) confirm a clay mineral is successfully synthesized in the presence of central metabolites such as acetate, malate, and succinate. Figure 2 shows the powder XRD diffractograms for 20 h syntheses with 0.10 M succinate under variable reflux temperature. While at 70 °C no sauconite formation is registered, the spectroscopic features of the mineral are observed at 75 °C. However, the first-order reflection (0 0 1) is only well-resolved for a minimum temperature of 80 °C, which allows the measurement of d 0 0 1 displayed in Fig. 2B. As the first-order reflection (0 0 1) becomes better defined with increasing temperature, its width decreases indicating the layer to layer distance expands. For example, d 0 0 1 increases in Fig. 2    Si ratios of 1:9 for the synthetized sauconite also discard the presence of tubular crystallites of imogolite in the samples 8, 30 . Seeding Induced Crystallization. The catalytic power of the synthesized clay is confirmed by repeating the procedure at 70 °C after spiking the gel with a single (macroscopic) sauconite particle obtained at 90 °C. As shown by the bottom trace labeled "seeded" in Fig. 2, after spiking the single particle, the synthesis is promoted at 70 °C in the same 20 h timespan, yielding 1.86 (±0.06) g of sauconite (a 54% of the material obtained at 90 °C). The fact that a simple particle of sauconite serves as a seed crystal to growth a larger amount of mineral serves as an outstanding example of the self-catalytic power of clays toward their crystallization. The surface of the seed particle facilitates heterogeneous nucleation at lower temperature by reducing the activation energy for crystallization. The observed acceleration in crystallization by seeding does not rely solely on random events but results from the surface interaction with chemical species in the precursor gel at 70 °C. The seed particle interacts with soluble free and complexed ions moving randomly in the gel, establishing intermolecular forces that are needed to form the crystal lattice. The addition of a seed particle to the gel provides a route to direct a process that does not depend on random interactions anymore.  Fig. S9). However, the previous phases were only observed in the untreated, air-dried condition for TEM and were likely unstable under the conditions required for plastic embedding or cryo-fixation. TEM images and energy dispersive X-ray spectroscopy (EDS) demonstrate that the morphology of the precursor gel corresponds to an amorphous material that has not incorporated aluminum in the structure ( Supplementary Fig. S9).

Effects of Temperature and pH.
After 6 h of synthesis the sample showed aggregates of sauconite nanocrystals, sometimes in association with an unknown gel-like phase ( Fig. 3A-C), and occasionally Zn-containing nanocrystals as seen in the sample at the zero timepoint. These latter crystals have an atomic structure close to Zn 2 SiO 4 .with lattice fringes spaced by ~2.3 Å (Fig. 3C). Sauconite is made up of Al, Si, H and O in ratios similar to other smectite-group minerals (Fig. 3D). Compared to the Zn-clay synthesized, the Zn-containing particles contain more Zn, much less Si and no Al. These nanoparticles were not observed in the TEM images from samples prepared for ultrathin section suggesting they are sensitive to the chemical treatment required for sample processing i.e. dilution, washing, dehydration, embedding. The objective of the TEM analysis of the samples in ultrathin section was to image the lattice fringes and observe the stacking order of individual 2:1 layers. High-resolution TEM (HRTEM) analysis of the cross-section through aggregates of sauconite showed disorganized and random arrangements of individual 2:1 layers. Most smectite-group minerals, including synthetic saponite 15 , consist of packets of short-range, turbostratic stacks of 2:1 silicate layers. Both the structure and chemical composition of sauconite resembles, however, other known smectite-group minerals.
The structure and chemical composition of the sauconite 2:1 layers after 20 h of synthesis is the same as the sample collected after 6 h. There is, however, a difference in the organization and aggregation of individual 2:1 layers as well as in the distribution and density of the particles. The TEM images for the 20 h synthesis in ultrathin section shows only isolated stacks containing a few individual 2:1 layers of sauconite (Fig. 4). The corresponding XRD diffractogram for this sample shows the 2:1 layer to layer distance expands a rate of 1.54 Å day −1 , which is accompanied by the incorporation of Al. Aluminum incorporation could take place by a 3 Zn 2+  2 Al 3+ +1 vacancy substitution mechanism, which would not introduce a positive charge to the octahedral sheet.
Electron Tomography and 3D Reconstruction. Electron tomography (ET) was carried out to determine the detailed 3D structure and mode of interaction of the sauconite 2:1 layers in ultrathin sections. The objective in ET is to reconstruct the 3D structure of an object from a series of 2D projections. Using the single-axis tilting method, a series of TEM images was collected from isolated aggregates of sauconite 2:1 layers in ultrathin section at every 2° tilt angle from −70° to +70°. No significant differences were observed between samples after 6 and 20 h of synthesis. A snapshot of the tomogram and its reconstructed image are shown in Fig. 5A,B. The images show the individual 2:1 layers are arranged in a random edge-to-face orientation within isolated aggregates.
Cryo-TEM was applied to study the size and shape of individual sauconite particles and their distribution in their natural dispersed state. The results of the extremely diluted (0.2 wt %) frozen samples show significant differences between samples collected after 6 and 20 h of synthesis (Fig. 6). Figure 6A shows the emergence of the sauconite particles from the gel-like precursor suggesting initial nucleation and growth of the sauconite phase. Surprisingly, the sauconite 2:1 layers were not dispersed in either sample confirming the features and morphology observed in the ultrathin sections (Fig. 5) is not an artifact, but a growth feature.  Figure 8A Fig. 8B are also presented. Remarkably, the expansion of the layer to layer distance (d 0 0 1 ), the total surface area, and the amount of succinate accumulated in the interlayer, all depend on time with a similar exponential correlation (Fig. 8C).
The peak at 918 cm −1 in the DRIFT spectra of synthesized sauconite at pH 0 9.0 from 0 to 6 h (Fig. 8B) corresponds to Al-OH bending from the reactant − Al(OH) 4 32 , the dominant hydroxy complex of aluminum in the starting gel at high pH 16 . As the previous peak disappears after 15 h of synthesis, a new peak is developed at 885 cm −1 assigned to Si-O-Al IV vibrations in the tetrahedral position of sauconite. The stronger intensity for the Si-O-Al IV vibration at 20 h in Fig. 8B, which is also present in experiments with variable [succinate] ( Supplementary Fig. S5), suggests the incorporation of Al to the tetrahedral layer 24 , which is key for crystallization. The role of succinate as a catalyst is proposed to facilitate the simultaneous insertion of two Al 3+ above and below a plane that becomes the interlayer of sauconite. Figure 9 displays a representation that takes into account the previous concept for the layers in 2:1 trioctahedral sauconite, which agrees with cryo-TEM, XRD, and DRIFT spectroscopy observations. In this model representation, succinate and the water molecules are slightly scaled-up relative to the centers in the crystalline structure to facilitate their visualization. The constrains that succinate experience to accommodate in the interlayer space is thought to be controlled by the slower free rotation of σ-bonds among carbon atoms in the gel compared to the freedom provided in a complete solution state. The 3D structure of succinate is represented in Fig. 9 for the staggered conformation that decreases the torsional strain in energy from eclipsing interactions by optimizing the rotation around carbon-carbon σ-bonds. Because d 0 0 1 grows over time (Fig. 8C), it is logic to assume that the sauconite structure in Fig. 9 reaches its maximum swelling as the conformer of succinate optimizes its geometry reaching a length of 5.6 (±0.7) Å. In addition, the selected succinate structure appears stabilized by interactions with two aluminum centers located on opposite sites of the interlayer space (Fig. 9). However, it would also be possible to envision interactions of succinate with a single zinc center of the interlayer. Figure 9 also depicts the tetrahedral-octahedral-tetrahedral sheets with Zn, Al and Si, as proved by EDS (Fig. 3D).
The peaks at 1438 and 1570 cm −1 (Fig. 8B) at 20 h of synthesis represent symmetric and asymmetric stretching vibrations for C = O (ν C=O ) in the carboxylate group of succinate, respectively 19 . In addition, the two undissociated -COOH groups of succinic acid (pK a1 = 4.21 and pK a2 = 5.64) 33 expected at 1297 cm −1 (ν C=O ) and 1691 cm −1 (v C-OH ) 34 are absence at pH 0 9.0 because only completely dissociated succinate is available in equilibrium. However, at pH 0 6.3 the fraction of dianion drops to 81.9% while succinate monoanion reaches an 17.9% justifying the observation of a peak at 1297 cm −1 for the symmetric v C-OH stretching of -COOH groups in Fig. 8B.  Therefore, DRIFT spectroscopy confirms the dianion of succinate depicted in Fig. 9 is the dominant species both participating in the synthesis and accumulating in the interlayer space of sauconite in experiments at pH 0 9.0.
Total surface area measurements, TEM and cryo-TEM together with XRD and DRIFT spectroscopies reveal that the slow initial swelling during sauconite growth becomes exponentially faster after aluminum incorporation to the structure. The total surface area of the synthesized clay is a direct function of the amount of a simple dicarboxylic acid, which is an adsorbate in the interlayer space (Fig. 9) and a catalyst for crystallization. Therefore, as the amount of organic salt increases, swelling occurs and the reactive surface growths exponentially. The associated retention of water in the interlayer space is enhanced for longer syntheses as indicated by the 2. with K = 1.84 36 , X-ray wavelength λ = 1.5418 Å, and a full width at half maximum β = 3.4534 × 10 −2 radians for the baseline corrected (0 0 1) peak centered at 2θ = 5.7892° can be used to estimate the mean crystallite size L hk = 8.2 nm for the two-dimensional random layer lattice at 20 h in Fig. 8C.

Conclusions
Because clay minerals contribute most of the inorganic surface to catalyze reactions in soil environments, these results provide new understanding of the continuous evolution of present agricultural fields. Overall, this laboratory study demonstrates an effective coupling between photochemistry mediated by semiconductor minerals and clay formation exists. The work also demonstrates the self-catalytic power of clays toward mineral formation in the geological past as well as the ability of simple central metabolic molecules to co-catalyze crystallization in short time scales. The co-catalytic role of succinate between layers is to provide soluble complexes of Al 3+ , which is the limiting reagent for sauconite crystallization. Once the likely intercalation of succinate (or the other organic species) balances the relative adsorption forces between the precursor gel and the aluminum species, due to its bidentate ability, it can overturn the attraction forces between adjacent sauconite whiskers. A detailed description of the nucleation and growth of model clay particles is supported by cryogenic and conventional TEM. Finally, a major outcome of this work is that photochemistry may have played an important role in the origin of life on early Earth and other rocky planets.   Company). The diffractogram of the standard matches the pattern of saponite-15 Å (ICDD PDF No. 00-030-0789) 19 .

Methods
The preparation of oriented sauconite slides for XRD analysis in Fig. S7 (Supplementary Information) was based on Drever's method 37 . Briefly, 25 mL of 0.50 M MgCl 2 was mixed in a 50 mL centrifuge tube with 200 mg of sauconite synthesized during 20 h with 1.0 M succinate at pH 0 7.0 and 90 °C. After 1 min sonication and 5 min centrifugation (3000 rpm), the supernatant was discarded. This washing procedure was repeated by triplicate before a final wash with DI water to remove extra Mg 2+ . The unflocculated suspension was filtered (0.45 μm pore size and 47 mm diameter) under vacuum. The sauconite sample was carefully transferred onto a glass slide (26 × 46 mm) as described in the filter-membrane peel technique 37 . The Mg-saturated glass slides were analyzed by XRD from 2 to 40° 2θ degrees with a step scan of 0.07° (2θ) and a time per step of 4 s. Finally, the Mg-saturated sample was exposed to a vapor saturated glycerol atmosphere in a desiccator for 24 h before scanning by XRD. While the 2:1 layer structure is confirmed, the reversal in the intensities expected for both samples suggests there is no oriented stacking of layers, explaining the difficulties encountered during sample preparation.
DRIFT Spectra. DRIFT spectra were recorded (200 scans) with a Nicolet 6700 FTIR spectrometer and analyzed with OMNIC32 software (both Thermo Fisher Scientific). A liquid N 2 cooled MCT detector was employed in the range 600-4000 cm −1 with 4 cm −1 resolution. Oven dried samples of synthesized sauconite (30 mg) were homogenized with spectroscopic grade KBr (500 mg), and then poured into the smart collector diffuse reflectance accessory, which optics were continuously purged with N 2 (g). The incorporation of succinate over time was monitored by integrating the area under the C-H stretching (ν C-H ) between 2937 and 2991 cm −1 .

Measurement of TSA.
TSA is determined using the ethylene glycol monoethyl ether (EGME) method 38   TEM, EDS, and Cryo-TEM. To image particle size and shape of the samples, aqueous suspensions (5 µL) from the synthesis were dispersed and pipetted onto a 200-mesh copper TEM grid with carbon support film (whole mount), and allowed to dry under air at room temperature. For high-resolution TEM (HRTEM), sampled from synthesis at different timepoints were embedded in EPON liquid epoxy resin. Approximately 10-20 mg of the sample was placed into 1.5 mL polypropylene Eppendorf micro-test tubes and dehydrated by adding 100% ethanol to remove adsorbed water. After centrifugation at 13000 rpm for 15 min, the supernatant was removed and the sample dispersed in a mixture of 10% EPON resin and 90% ethanol by ultrasonic treatment. These steps were repeated with mixtures containing 30, 50, 70, and 100% EPON.
To obtain complete dispersion of the material, the ethanol:resin mixtures were agitated in an ultrasonic bath for ~1 min. Each incubation step lasted 24 h with the samples placed onto a rotator to ensure continuous agitation. The resin-clay mineral mixture was transferred into embedding moulds after the fifth incubation step (100% EPON resin). After a settling time of 1-2 h, the samples were polymerized by placing the moulds into an oven at 65 °C for 48 h. Ultrathin sections (70-80 nm) were cut from the polymerized resin blocks using an ultramicrotome and transferred onto 200-mesh copper TEM grids with carbon support film. The samples on whole mounts and in ultrathin section were imaged with an FEI Tecnai G 2 F20 200 kV T EM equipped with a Gatan Ultrascan 4000 CCD Camera System Model 895 and EDAX Octane T Ultra W /Apollo XLT2 SDD and TEAM EDS Analysis System.
For cryo-TEM, 5 μL of the aqueous suspension of the synthesized sauconite (previous to the final drying and washing steps) was transferred onto a C-flat holey carbon sample support grid (R2/2; Protochips, Inc.). Excess fluid was blotted and the sample flash frozen hydrated by plunging into a bath of liquid nitrogen-cooled liquid ethane using the FEI Vitrobot Mk IV Grid Plunging System (FEI Co.). The grids were stored in liquid nitrogen until imaged in a FEI Titan Krios 300 kV Cryo-S/TEM equipped with a Falcon 2 direct electron detector (DED) (FEI, Inc). Images were collected at a magnification of 75 k× corresponding to a pixel size of 1.41 Å and a defocus level ranging from −2.0 to −3.0 μm, under low dose conditions. Tomograms from the cryogenic and epoxy embedded samples were collected using FEI Batch Tomography Software version 4.0. The cryogenic tilt series was collected at a magnification of 59 k× every 2° over a tilt range of ±70°. The nominal pixel size was 0.14 nm with defocus of −2 μm. Figure 9. Model displaying the stacking order of layers in a 2:1 trioctahedral sauconite structure al synthesized during 20 h using 1.0 M succinate at pH 0 6.3 and 90 °C. The layer to layer distance d 0 0 1 = 15.68 Å is represented to be about 3-times larger than the length of succinate that appears longitudinally oriented perpendicular to tetrahedral aluminum centers.