Cemented stabilized rammed earth (CSRE) is a building material used to build load bearing walls from locally available soil. The article analyzes the influence of soil mineral composition on CSRE compressive strength. Compression tests of CSRE samples of various mineral compositions, but the same particle size distribution, water content, and cement content were conducted. Based on the compression strength results and analyzed SEM images, it was observed that even small changes in the mineral composition significantly affected the CSRE compressive strength. From the comparison of CSRE compressive strength result sets, one can draw general qualitative conclusions that montmorillonite lowered the compressive strength the most; beidellite also lowered it, but to a lesser extent. Kaolinite lightly increased the compressive strength.

Predicting the compressive strength of cement-stabilized rammed earth (CSRE) using current testing machines is time-consuming and costly and may harm the environment due to the samples’ waste. This paper presents an automatic method using computer vision and deep learning to solve the problem. For this purpose, a deep convolutional neural network (DCNN) model is proposed, which was evaluated on a new in-house scanning electron microscope (SEM) image database containing 4284 images of materials with different compressive strengths. The experimental results show reasonable prediction results compared to other traditional methods, achieving 84% prediction accuracy and a small (1.5) oot Mean Square Error (RMSE). This indicates that the proposed method (with some enhancements) can be used in practice for predicting the compressive strength of CSRE samples.

Zinc-rich fluor-elbaite from Piława Górna, Poland, was studied by electron microprobe (EPMA), single-crystal X-ray diffraction (SREF), and Raman spectroscopy (RS) to check the possibility of the application of RS to draw crystal-chemical conclusions for Al-rich and Li-bearing tourmalines on basis of the O–H stretching vibrations in the spectral range 3400–3800 cm–1. This tourmaline, forming a thin metasomatic zone around gahnite, features varying compositions with a ZnO content reaching in the studied fragment of 5.70(12) wt%. The crystal structure of this Zn-rich fluor-elbaite [a = 15.921(1), c = 7.127(1) Å] was refined with a R1 value of 1.67%. Its formula was determined on the basis of electron-microprobe and structure refinement as (Na0.84☐0.14Ca0.01)XΣ1.00(Al1.06Li0.84Zn0.69Fe0.322+Mn0.09)YΣ3.00AlZ6(BO3)3(Si6TO18)(OH)3V(F0.65OH0.26O0.09)W. The deconvolution of the O–H stretching vibration bands, performed by fitting of an input model of component bands with Gaussian function shapes for the empirical spectrum, indicates that each of the three maxima assigned for VOH bonded to YAl3+, Y2+, and YLi+ and with the total integral intensity of at least 75% of the total OH content could be resolved into 1 to 3 bands, depending on the X-site occupation (vacancies, Na+, and Ca2+). The deconvolution indicates further that several low intense bands of WO–H modes above a Raman shift of 3600 cm–1, totally reaching ≤25%, are dependent on the occupation of triplets of YYY cations bonded to the hydroxyl. These WO–H modes are also influenced by the X-site occupation. Due to ordering of all octahedral cations (except Al) at the Y site and a complete occupation of the Z site by Al and the V site by OH, it seems possible to evaluate the Li and OH contents in a Al-rich and Li-bearing tourmaline directly from the Raman spectrum. By using the ratio VOHIYAlZAlZAl/(VOHIYZZ + WOHIYYY) as evaluated from RS, corresponding to the ratio YAl/V+WOH in the crystal, the formula of the Zn-rich fluor-elbaite can be calculated as (Na0.85☐0.14Ca0.01)XΣ1.00(Al1.11Y1.112+Li0.78)YΣ3.00AlZ6(BO3)3(Si6O18)(OH)3(F0.65OH0.13O0.22), where Y 2+ = Zn + Fe + Mn. The formula, determined only on basis of EPMA and deconvolution of RS in the O–H stretching bands, corresponds very well (≤1 SD range of EPMA) to the formula determined on basis of EPMA and SREF. This result implicates that the O–H stretching vibrations, measured by Raman spectroscopy, could be applied for Al-rich and Li-bearing tourmalines as a useful tool for providing additional information for determining the crystal-chemical formula. It is also very helpful when crystal structural data are not available.

The mature transgressive Albian quartz sands in the Miechów Synclinorium contain a poor (<1%) heavy mineral suite consisting of tourmaline, rutile, garnet, staurolite, ilmenite, zircon, monazite, kyanite, and gahnite. The other minerals, especially those containing Fe and Ti (e.g., biotite), are subordinate. Over 512 tourmaline grains from seven outcrops in the Miechów Segment were analysed using electron probe microanalysis (EPMA). The majority of grains belong to the alkali tourmaline group in which the X-site is dominated by Na (0.4 to 0.9 apfu). Detrital tourmalines are mainly dravite with a prevalent schorl end-member with average X_Mg values over 0.6. Apart from Mg and Fe, the other Y-site cations rarely exceed 0.1 apfu. Fluorine content is usually below the detection limit. Their chemical composition suggests that most tourmaline grains were sourced from Al-rich and Al-poor metapelites/metapsammites. The main source rocks for the Albian sands were rocks from low- to medium-grade metamorphism, probably from Al-rich quartz-muscovite schist and/or muscovite rich gneisses. Additional minor source rocks were granites and pegmatites coexisting with them. A comparison of the examined tourmaline to tourmaline from possible source areas indicates that these areas were located in the eastern part of the Bohemian Massif and/or eastern Sudetes.

Heflikite, the first Sc-dominant epidote-supergroup mineral, was discovered in two occurrences. The holotype was found in a granitic pegmatite associated with rodingite-like calc-silicate rocks and metasomatised granitic bodies exposed in a serpentinite quarry at Jordanów Śląski near Sobótka, Lower Silesia, SW Poland. The cotype comes from the Heftetjern pegmatite, Tørdal region, Norway. The holotype is composed of (in wt.%): 35.69 SiO₂, 0.22 TiO₂, 21.98 Al₂O₃, 6.12 Sc₂O₃, 0.07 V₂O₃, 1.10 Fe₂O₃, 0.11 Y₂O₃, 1.55 La₂O₃, 4.05 Ce₂O₃, 0.31 Pr₂O₃, 1.53 Nd₂O₃, 0.40 Sm₂O₃, 0.11 EuO, 0.56 Gd₂O₃, 0.14 MnO, 3.56 FeO, 0.16 MgO, 19.16 CaO and 1.78 H₂O⁽⁺⁾_calc.; total 98.60. The cotype contains: 34.92 SiO₂, 0.44 TiO₂, 0.82 SnO₂, 19.13 Al₂O₃, 4.79 Sc₂O₃, 1.96 Fe₂O₃, 2.55 La₂O₃, 7.39 Ce₂O₃, 0.48 Pr₂O₃, 0.67 Nd₂O₃, 0.12 EuO, 0.61 Gd₂O₃, 0.13 MnO, 5.97 FeO, 17.66 CaO and 1.73 H₂O⁽⁺⁾_calc.; total 99.37. The compositions correspond to the following empirical formulae: (Ca_1.729Ce_0.125La_0.048Nd_0.046Gd_0.016Sm_0.012Pr_0.010Y_0.005Eu²⁺_0.003)_Σ1.994[(Al_2.182Sc_0.449Fe³⁺_0.070V³⁺_0.005)_Σ2.706(Fe²⁺_0.251Mg_0.020Mn_0.010)_Σ0.281Ti_0.014]_Σ3.001(Si_3.006O₁₁)O(OH) and (Ca_1.644Ce_0.235La_0.082Nd_0.021Gd_0.018Pr_0.015Eu²⁺_0.004)_Σ1.019[(Al_1.958Sc_0.362Fe³⁺_0.128)_Σ2.448(Fe²⁺_0.434Mn_0.009)_Σ0.443(Ti_0.029Sn_0.029)_Σ0.058]_Σ2.949(Si_3.033O₁₁)O(OH), respectively, and to the ideal formula Ca₂(Al₂Sc)(Si₂O₇)(SiO₄)O(OH). The crystal structure of the holotype was refined in the monoclinic system with an R₁ index of 8.62%. The crystal-structure refinement indicates exclusively Si occupied T sites, Al occupied M1 and M2 sites, and a Ca occupied A1 site. The M3 site is filled predominantly by trivalent cations, mainly Sc³⁺, with divalent cations (mainly Fe²⁺) as minor occupants. The A2 site is filled mostly by Ca with minor amounts of rare earth elements (REE). The holotype heflikite crystallised from metasomatic fluids that infiltrated a contact between the granitic pegmatite and the surrounding rodingite-type calc-silicate rocks and serpentinites. The fluids that introduced Sc into the pegmatite could have been either hydrothermal or related to low-grade regional metamorphism that postdated the formation of the pegmatite. The cotype heflikite formed during the late-stage hydrothermal crystallisation of the Sc-enriched granitic pegmatite.