Exploring the mechanical response of low-carbon soil improvement mixtures

Exploring the mechanical response of low-carbon soil improvement mixtures 1 Alessandro Fraccica1, Giovanni Spagnoli2*, Enrique Romero1,3, Marcos Arroyo1,3, 2 Rodrigo Gómez3 3 1 Geomechanics Group, International Centre for Numerical Methods in Engineering, 4 Campus Nord UPC, calle Gran Capità, S/N 08034 Barcelona, Spain 5 2 MBCC Group, Dr.-Albert-Frank-Straße 32, 83308 Trostberg, Germany, gio- 6 [email protected], spagnoli_giov[email protected], https://or- 7 cid.org/0000-0002-1866-4345 (corresponding author) 8 3 Department of Civil and Environmental Engineering, Universitat Politècnica de Cata- 9 lunya, calle Jordi Girona, 1-3, 08034 Barcelona, Spain 10 11 ABSTRACT 12 As society moves towards decarbonisation it is important to assess the hydro-mechan- 13 ical behaviour of binders that could offer a low-carbon alternative to Portland cement 14 in ground improvement technologies. This work considers two such alternatives: one 15 still largely unexplored (metakaolin-based geopolymers) and a better known one (col- 16 loidal silica). Results from unconfined compressive strength, permeability tests, un- 17 drained monotonic and cyclic triaxial tests on granular soils (sand and silty sand) 18 treated with those two binders are presented and discussed, emphasizing similitudes 19 and differences with the response of similar soils treated with other conventional and 20 unconventional binders. Effects of silt content, curing conditions and soil/binder ratios 21 are examined. Both colloidal silica and metakaolin-based geopolymer significantly im- 22 prove the mechanical properties of the treated soils, although the geopolymer results 23 in a stronger and stiffer material. Both treatments reduce much the permeability of the 24 treated soil, but the reduction achieved with CS is larger. 25 Keywords: sand-silt mixture; triaxial tests; permeability; metakaolin-based geopoly- 26 mer; colloidal silica 27 ABBREVIATIONS AND SYMBOLS 28 𝐶𝑖𝑣 Volumetric cement content 29 Dr Relative density 30 c’ Drained cohesion 31 e Void ratio 32 𝐸 Young modulus 33 𝐹0 Metakaolin volumetric filling 34 FA Fly-ash 35 𝐺𝑆 Specific gravity 36 𝑘𝑤 Water permeability 37 MK Metakaolin 38 OPC Ordinary Portland cement 39 𝑞 Deviatoric stress amplitude 40 TXCIU Monotonic triaxial compression test 41 CTXU Cyclic triaxial compression test 42 UCS Unconfined compressive strength test 43 𝜌𝑠 Density of the solid particles (components) 44 𝜌𝑠∗ Density of the solid particles in the mixture 45  Diameter 46 φ’ Drained friction angle 47 ν Poisson’s ratio 48 𝜂 Porosity 49 INTRODUCTION 50 Ground improvement techniques enhance the mechanical and hydraulic properties of 51 soils for engineering applications. Several ground improvement methods involve the 52 addition of binders to soils in place. These methods are usually classified according to 53 the level of soil disturbance associated with binder placement (e.g., Cambefort, 1977; 54 Mitchell, 1981; Karol, 2003; Spagnoli, 2021). Techniques based on mechanical mixing, 55 such as deep soil mixing, or fluid-driven erosion and mixing, such as jet-grouting, re- 56 quire total soil remoulding and occupy one end of this spectrum. At the other end are 57 located techniques that imply very low disturbance, such as permeation grouting. Re- 58 gardless of the technology selected, the binder that is currently most often employed 59 in ground improvement technology is ordinary Portland cement (OPC). 60 Cement manufacturing accounts for 8% of global carbon dioxide (CO2) emissions and 61 reducing the carbon dioxide footprint of concretes is increasingly seen as urgent (IEA, 62 2018). Low-carbon binders, in which OPC has been totally substituted, are key to 63 achieve significant long-term emission reductions (Lehne and Preston, 2018). This fits 64 well with the perception of material use as the dominant factor in life-cycle environ- 65 mental impact of geotechnical systems (Kendall et al 2017). Referring particularly to 66 ground improvement, life-cycle carbon emissions have been proposed as a means to 67 evaluate the global environmental impact of specific projects (Shillaber et al. 2016). It 68 is thus likely that the road towards sustainable ground improvement passes through a 69 much-increased use of low-carbon binders (Mohammed et al. 2021). 70 Alkali-activated binders (AAB) represent one important low-carbon alternative to OPC 71 (Provis and Deventer, 2014). AAB result from the reaction of a solid alumino-silicate 72 based material (precursor) and an alkaline metal (activator, e.g. sodium hydroxide). A 73 variety of products can act as precursors in AAB. Important examples include industrial 74 wastes, (such as ground blast-furnace slag (GBFS) or fly-ash residue from coal-fired 75 electricity generation), as well as natural products, (such as volcanic ash or a calcined 76 kaolinite known as metakaolin). The potential of AAB for ground improvement applica- 77 tions has been repeatedly highlighted from a variety of perspectives from strength im- 78 provement (e.g. Cristelo et al., 2011; Canakci et al. 2019) to environmental remediation 79 (Ji and Pei, 2019; Du et al. 2020). Finally, it is worth noting that the use of AAB is not 80 the only alternative to obtain low-carbon binders for ground improvement, as other 81 residues (e.g. calcium carbide, Du et al. 2016) and additives (e.g. superphosphate, Xia 82 et al. 2017, 2019) are also useful for that purpose. 83 84 Metakaolin (MK) and low-calcium fly-ash result in almost exclusively aluminosilicate 85 AABs, which are generally known as geopolymers (Davidovits, 1994, 2008; Provis and 86 Bernal, 2014). The term “geopolymer” describes an amorphous network of polymer- 87 ized silicoaluminates (Ma et al. 2018). Geopolymers are also used in the concrete in- 88 dustry as full or partial replacement for conventional cements (e.g. Singh et al. 2015). 89 90 Metakaolin is an industrial product, generally having more consistent properties than 91 residue-based AAB. The raw material (clay) is abundant (IEA, 2018) and, unlike some 92 residues, will not be limited by foreseeable changes in technology (as is the case of 93 GGBS) or in the energy production mix (as is the case of FA). On the other hand, the 94 current production of metakaolin is still small and it is currently marketed at significantly 95 higher prices per weight that OPC or other AAB precursors. 96 Metakaolin-based binders are highly viscous (Provis and Bernal, 2014). This makes 97 them unsuitable for soil permeation purposes, and better adapted for techniques such 98 as deep-mixing. A low-carbon alternative to OPC for permeation purposes is colloidal 99 silica (CS). Colloidal silica of interest in ground permeation takes the form of manufac- 100 tured aqueous suspensions of nanometric silica particles, solidifying as gel at a rate 101 controlled by pH and salt concentration (Bergna and Roberts, 2006). CS has several 102 inherent advantages for permeation treatments, such as small particle size, low vis- 103 cosity and non-toxicity. All these benefits have driven the uptake of CS for ground im- 104 provement in geo-environmental (Moridis et al. 1995; Wong et al. 2018) and liquefac- 105 tion mitigation applications (Gallagher et al., 2007; Zhao et al., 2020). It turns out that 106 CS treatments are also advantageous over OPC-alternatives from the carbon emission 107 viewpoint (Gallagher et al. 2013) 108 As with any other geomaterial, good mechanical and hydraulic characterization of 109 binder-soil mixtures is necessary to achieve sustainable design objectives, as the al- 110 ternative is overdesign and/or increased failure risk (Basu et al. 2015). Hydromechan- 111 ical characterization studies of soils improved with CS are well advanced (e.g., Persoff 112 et al, 1998; Gallagher and Mitchell, 2002; Díaz-Rodríguez et al., 2008; Porcino et al., 113 2011; Porcino et al., 2012; Vranna and Tika, 2015; Georgiannou et al., 2017; Salvatore 114 et al., 2020). Table 1 presents a brief summary of such work: one aspect that has not 115 been investigated previously is the effect of fines on CS treatment. 116 Mechanical investigations of soil treated with metakaolin-based geopolymer are lim- 117 ited. Some studies (e.g. Kolovos et al., 2013; Cyr et al., 2013; Deng et al., 2015; Wu 118 et al., 2016; Asteris et al., 2017) have assessed the use of metakaolin in soil treatment 119 as a partial Portland substitute, but they used no alkali activators and, therefore, the 120 resulting binders were not geopolymers. Furthermore, when metakaolin-based geo- 121 polymers have been used, mechanical testing was limited to unconfined compressive 122 strength (Zhang et al. 2013; Rong-rong and Dong-dong, 2020; Spagnoli et al. 2021a). 123 The situation is different for residue-based geopolymer-soil mixtures where initial scop- 124 ing studies (Verdolotti et al., 2008; Cristelo et al., 2013; Singhi et al., 2016; Yaghoubi 125 et al., 2018) have been followed by more in-depth mechanical studies (Rios et al. 2016; 126 2017; Abdullah et al. 2019, 2020). 127 Spagnoli et al. (2021b) presented a detailed study of the effect of curing conditions on 128 microstructural and hydraulic properties of metakaolin-soil mixtures. However, the me- 129 chanical response was only explored by means of unconfined compression tests. It is 130 thus necessary to perform more in-depth studies of the mechanical response of me- 131 takaolin treated soils, for instance with triaxial tests where effective stress can be con- 132 trolled and pore pressure is registered. 133 The purpose of this work is to partially fill that gap in the current knowledge and present 134 a study of the monotonic and cyclic triaxial strength of soils (a sand and a silty sand) 135 treated with a metakaolin-based geopolymer. For contrast, the results are presented 136 alongside those obtained with a CS treatment of the same soils: this had the added 137 interest of examining the effect of soil fines in the CS treatment, an aspect that was not 138 touched upon in previous studies. 139 MATERIALS AND METHODS 140 Base materials 141 Two reference granular soils were used for the treatment. The first one is Holcim quartz 142 sand (0.2-0.6mm). The second is silty sand obtained by mixing dry carbonate silt 143 (CaCO3) in a proportion of 10% by weight with the previous reference sand. The grain 144 size distribution of the soils is presented in Figure 1. Chemical and physical properties 145 of the materials, as well as their initial state, are summarized in Table 2. 146 The first binder employed in this study uses CS (MasterRoc MP 320®, Master Builders 147 Solutions) as precursor. This product is an aqueous dispersion (density 1.30 Mg/m3) 148 of silica particles of uniform nanometric size and silica concentration of 40% (Table 2). 149 The CS was mixed with a solution of NaCl (10% solution) at a volume ratio of 12% to 150 induce the gelation process. 151 For the geopolymer employed here the precursor material was a metakaolin powder 152 (Argical™-M 1000, Imerys) (Table 2), a dehydroxylated aluminium silicate 153 (Al2O3∙2SiO2) resulting from the calcination and micronization of kaolinitic clay and 154 having lamellar-shaped particles. The metakaolin powder was activated with an alka- 155 line water solution of potassium silicate (w(SiO2)/w(K2O) = 1). Potassium silicate was 156 selected to enhance workability, as it is known to result in less viscous binders than 157 the more frequently employed sodium-based activators (Provis and Bernal, 2014). 158 159 Sample preparation 160 The CS suspension was mixed with the accelerator saline solution (Table 3), stirred 161 for 1 minute (with a Robot 500, 500W (Taurus, Spain)) and permeated into the soils, 162 which had been previously dry poured in the moulds from a height of 200 mm. Perme- 163 ation took place through the porous steel base of the moulds, driven by an air-liquid 164 interface system applying an injection pressure of 3 kPa. Injection continued until the 165 injected fluid fully permeated the specimen, in a process that always lasted less than 166 5 minutes. Tomographic inspection of permeated specimens (Figure 2) showed that 167 the process resulted in a very uniform and complete filling of pore space by the hard- 168 ened colloid. 169 creasing the filling ratio for geopolymer resulted in more dilatant and stronger speci- 312 mens. Similar trends were observed by Wong et al. (2018) and Georgiannou et al. 313 (2017) in CS-treated soils. 314 Contrary to what happened with UCS the effect of curing time on the triaxial response 315 of dry-cured MK treated specimens was rather small, showing lesser or no clear decay 316 of peak mobilised strength with time (Figure 8). This suggests that confinement inhib- 317 ited the mechanical effect of MK micro-cracking that was visible on microscopic images 318 (Spagnoli et al. 2021a) 319 Undrained secant Young moduli are presented in Figure 9 (at a small strain level) and 320 Figure 10 (at intermediate strain levels). In these figures the strain level effect (de- 321 creasing stiffness as strain level increases) is larger than the stress level effect (stiff- 322 ness increases as confinement increases). As for the time effect, there is little to no 323 evidence of stiffness reduction due to dry-curing on the geopolymer-soil mixtures. On 324 the contrary, a certain curing-induced increase of stiffness at both strain levels is visi- 325 ble, although more for the silty sand (SO) based mixtures than for the pure sand (S) 326 mixtures. For the CS treated specimens the increase of stiffness with curing is even 327 more consistent. 328 Cyclic triaxial undrained compressions (CTXUC) 329 The evolution of normalized excess pore pressure during cyclic loading is shown in 330 Figure 11, along the axial strain and the number of cycles. The normalizing stress em- 331 ployed is the minor principal effective stress at consolidation, as is customary for triax- 332 ial conditions. The untreated specimens S and SO underwent high strains (i.e. εax > 333 5%) within the first 10 cycles of loading already, jointly with a fast increase of the pore 334 water pressures. They reached values of ru between 0.7 (clean sand) and 0.85 (sand 335 + fines), which are close to a condition of liquefaction of the material. This might reflect 336 an increased relative density in the silty sand (Polito and Martin, 2003) although it is 337 recognized that the effect of non-plastic fines on liquefaction is a complex issue, be- 338 yond the scope of this paper. 339 In all the treated samples the effect of the applied cycles was not dramatic and axial 340 strains remained lower than 1%. Soils treated with CS showed higher strains and ex- 341 cess pore water pressures than those treated with the metakaolin geopolymer. This 342 may be attributed to the combined effect of smaller stiffness and lower permeability of 343 the CS treatment. 344 The stress path of the treated specimens remains distant from the triaxial failure enve- 345 lope (Figure 12). Notwithstanding that, significant stiffness degradation took place dur- 346 ing cyclic loading, even for the stronger MK treated material. In Figure 13, the effect of 347 cycling on the normalized secant undrained Young modulus Eu/Eu,in. is presented along 348 the cycles. Eu,in. is the stiffness calculated on each CTXUC backbone curve, in corre- 349 spondence of the first shearing cycle. The final secant stiffness is around 5 MPa (2% 350 of Eu,in.) in the untreated samples, 55 MPa (13% of Eu,in.) in CS-treated samples, and 351 236 MPa (43% of Eu,in.) in MK-treated samples. 352 DISCUSSION 353 As pointed out before precursor materials for geopolymers not only include metakaolin 354 but also low calcium fly ash. Rios et al. (2016, 2017, 2017b) presented a mechanical 355 study of silty sand treated with a fly ash based geopolymer. The base materials were 356 similar to those employed here (see Figure 1). The binder dosage in the mixtures was 357 slightly higher and the porosity was slightly lower (see Table 7). 358 Figure 14 and Figure 15 compare the peak strength envelopes obtained with the dif- 359 ferent base soils alone and with the geopolymer treated soils, (see also Table 8). Note 360 that in the triaxial tests by Rios et al. (2017) the specimens were sheared in a drained 361 condition, after anisotropic consolidation, whereas here shearing is undrained and con- 362 solidation isotropic. Despite those differences, the two sets of results fit well together. 363 The envelopes obtained for the base soils are remarkably close, a fact that allows 364 better appreciation of the relative improvement obtained by each binder mixture. The 365 strength attained increases as the MK dosage increases and also when silt is added 366 to the sand. This last effect may be due to the silt reducing the soil porosity, although 367 a chemical effect -the strengthening of geopolymer by small doses of calcium car- 368 bonate (Yip et al. 2008)- may also play a role. The FA treatment results in higher en- 369 velopes which, again, may be first related to the smaller porosity/binder ratio (Table 7) 370 and perhaps also to the different binder chemistry. 371 Secant stiffness degradation during shearing, Eu/Eu,0.01, is compared in Figure 16 for 372 the different geopolymer improved soils. Eu,0.01. is calculated in correspondence of an 373 axial strain of 0.01%, as for Figure 9. The results suggest that the MK geopolymer and 374 CS are somewhat more fragile than the FA one, as the stiffness decay is faster. Again, 375 the effects of dosage and porosity may explain the difference, although other factors 376 may also play a role. One such factor is the harsher curing condition of the MK speci- 377 mens, where dry curing resulted in visible retraction microcracks in the binder joining 378 together the different grains (Figure 19). 379 By comparison with the geopolymers, the strength increase obtained with the CS treat- 380 ment is much smaller (Figure 17). The results are in good agreement with direct shear 381 results obtained by Wong et al. (2018) on a CS treated sand similar to the one in this 382 study (MP 320, silica concentration = 40% (w/w) in the solution). The limited effect of 383 CS on friction angle was also observed by Wong et al. (2018) and Vranna and Tika 384 (2015). It is noticeable that, contrary to what happened for the UCS, the addition of 385 carbonate silt did not induce significant effects on the triaxial envelope of the CS im- 386 proved soil. 387 The permeability values obtained in this work are compared with similar measurements 388 in Figure 18. The important role of the curing condition on this property is evidenced 389 by the results for CS which was more permeable when cured dry, even for relatively 390 more intense binder treatment (as measured by the w/c0 ratio). 391 The observed differences between CS and MK treated soils can be related to some 392 features of the induced microstructure. Both binders fill in the gaps between the soil 393 grains, cementing them. But the metakaolin geopolymer presents pervasive retraction 394 cracking at the microscale (Figure 19; see also Spagnoli et al. 2021a), whereas the CS 395 cement bridges, with nanoscale porosity, have a much smoother texture (see Wong et 396 al. 2018). 397 CONCLUSIONS 398 This paper investigated the influence of two different binders (i.e. metakaolin-based 399 geopolymer and colloidal silica) on the hydro-mechanical behaviour of loose sandy 400 soils under monotonic and cyclic stresses. The main observations may be summarized 401 as follows 402 • Treatment with metakaolin-based geopolymer is as effective to increase the 403 strength of the soils as OPC, attaining similar UCS values for similar poros- 404 ity/binder ratio values 405 • Although dry-curing reduces the UCS of metakaolin treated soil, it did not re- 406 duce its confined (triaxial) strength or stiffness, 407 • The strength improvements obtained with the metakaolin geopolymer and the 408 colloidal silica are generally well aligned with those obtained with Portland Ce- 409 ment at similar w/c0 dosage ratios. 410 • Metakaolin treated soils had larger stiffness and slower stiffness degradation 411 than those permeated with CS 412 • Although both treatments achieved significant reductions in permeability, treat- 413 ment with CS was more effective to reduce permeability than treatment with the 414 metakaolin geopolymer. 415 • Both treatments improved significantly the cyclic response of the sandy soils, 416 within the number of cycles investigated, exhibiting low pore-water pressures 417 and axial strains. 418 • The presence of carbonate silt within geopolymer-treated specimens resulted 419 in a slight increase of soil cohesion and stiffness, while negligible effects were 420 observed in CS-treated samples. 421 • Finally, the mechanical improvement obtained with the metakaolin-based geo- 422 polymer is well aligned with previous observations on fly-ash based geopoly- 423 mers. 424 The dominant role of OPC in ground improvement technologies results from various 425 important factors, including a well proven track record, economic considerations and 426 technical familiarity. Systematic and extensive geomechanical studies of low-carbon 427 alternatives are required for this situation to change. Given the large strength increases 428 obtained in this work, future work on the mechanical and hydraulic properties of me- 429 takaolin-based geopolymers should explore the possibility of using smaller binder dos- 430 ages. The complexities added by realistic curing scenarios must also continue to be 431 explored, for instance addressing also the possible effect of curing under stress. Fi- 432 nally, laboratory testing should be complemented by field testing and demonstration 433 projects of MK-based ground treatment. 434 ACKNOWLEDGEMENT 435 The authors wish to thank Master Builders Solutions for the permission granted to pub- 436 lish these results. 437 REFERENCES 438 Abdullah H.H., Shahin M.A., and Walske M.L. (2019) Geo-mechanical behaviour of 439 clay soils stabilized at ambient temperature with fly-ash geopolymer-incorporated gran- 440 ulated slag. Soils and Foundation 59, 1906-1920 441 Abdullah H.H., Shahin M.A., Walske M.L., and Karrech A. (2020b) Cyclic Behaviour of 442 Clay Stabilised with Fly-ash Based Geopolymer Incorporating Ground Granulated 443 Slag. Transportation Geotechnics (In Press) 444 ASTM (2017) Volume 04.08 Soil and Rock (I): D420 – D5876/D5876M. West Con- 445 shohocken, PA, 19428-2959 USA. 446 Axelsson, M. (2006). Mechanical tests on a new non-cementitious grout, silica sol: A 447 laboratory study of the material characteristics. Tunnelling and Underground Space 448 Technology, 21(5), 554-560. 449 Basu, D., Misra, A., and Puppala, A. J. (2015). Sustainability and geotechnical engi- 450 neering: perspectives and review. Canadian Geotechnical Journal, 52(1), 96-113. 451 BAUER Maschinen GmbH (2016). CSM-Cutter Soil Mixing. https://www.bauer.de/ex- 452 port/shared/documents/pdf/bma/datenblatter/CSM_Cutter_Soil_Mix- 453 ing_EN_905_656_2.pdf905-656-2_EN.pdf (last accessed December 8, 2020) 454 Bergna, H. E., and Roberts, W. O. (Eds.). (2005). Colloidal silica: fundamentals and 455 applications (Vol. 131). CRC Press. 456 Bonavetti, V., Donza, H., Rahhal, V., and Irassar, E. (2000). Influence of initial curing 457 on the properties of concrete containing limestone blended cement. Cement and Con- 458 crete Research, 30(5), 703-708. 459 Cai, G. H., Du, Y. J., Liu, S. Y., & Singh, D. N. (2015). Physical properties, electrical 460 resistivity, and strength characteristics of carbonated silty soil admixed with reactive 461 magnesia. Canadian Geotechnical Journal, 52(11), 1699-1713. 462 Cambefort, H. (1977). The principles and applications of grouting. Quarterly Journal of 463 Engineering Geology and Hydrogeology, 10, 57-95, 464 https://doi.org/10.1144/GSL.QJEG.1977.010.02.01. 465 Canakci, H., Güllü, H., and Alhashemy, A. (2019) Performances of Using Geopoly- 466 mers Made with Various Stabilizers for Deep Mixing. Materials, 12, 2542 467 Cyr, M., Trinh, M., Husson, B., Casaux-Ginestet, G. (2013) Design of eco-efficient 468 grouts intended for soil nailing. Construction and Building Materials 41, 857–867 469 Consoli, N. C., Cruz, R. C., and Floss, M. F. (2011). Variables controlling strength of 470 artificially cemented sand: influence of curing time. Journal of Materials in Civil En- 471 gineering, 23(5), 692-696. 472 Consoli, N.C., da Fonseca, A.V., Silva, S.R., Cruz, R.C. and Fonini, A. (2012). Param- 473 eters controlling stiffness and strength of artificially cemented soils. Geotechnique 474 62(2), 177–183, http://dx.doi.org/10.1680/geot.8.P.084. 475 Consoli, N. C., Ferreira, P. M. V., Tang, C. S., Marques, S. F. V., Festugato, L. and 476 Corte, M. B. (2016). A unique relationship determining strength of silty/clayey soils – 477 Portland cement mixes. Soils and Foundations 56(6), 1082–1088. 478 Consoli, N.C., Da Silva Carretta, M., Festugato, L., Batista Leon, H., Ferreira Tomasi, 479 L. and Salvagni Heineck, K. (2020) Ground waste glass–carbide lime as a sustainable 480 binder stabilising three different silica sands. Géotechnique (Ahead of Print) 481 Cristelo, N., Glendinning, S. and Pinto, A. T. (2011). Deep soft soil improvement by 482 alkaline activation. Proceedings of the ICE-Ground Improvement, 164, (2), 73–82. 483 Cristelo, N., Glendinning, S., Fernandes, L. and Pinto, A.T. (2013) Effects of alkaline- 484 activated fly ash and Portland cement on soft soil stabilisation. Acta Geotechnica 8, 485 395–405. https://doi.org/10.1007/s11440-012-0200-9 486 Davidovits, J. (1994). Properties of geopolymer cements. 1st International Conference 487 on Alkaline Cements and Concretes. Kiev, Ukraine: Kiev State Technical University, 488 131-149. 489 Davidovits, J. (2008). Geopolymer Chemistry and Applications. In: Saint Quentin, 490 F.G.I. (Ed.). 491 Díaz-Rodríguez, J.A.; Antonio-Izarraras, V.M.; Bandini, P.; López-Molina, J.A. (2008) 492 Cyclic strength of a natural liquefiable sand stabilized with colloidal silica grout. Cana- 493 dian Geotechnical Journal, 45, 1345–1355. doi:10.1139/T08-072 494 Deng, Y., Yue, X., Liu, S., Chen, Y., Dingwen, Z. (2015) Hydraulic conductivity of ce- 495 ment-stabilized marine clay with metakaolin and its correlation with pore size distribu- 496 tion. Engineering Geology 193, 146-152 497 Dong, C., Zhang, R., Zheng, J., and Jiang, W. (2020) Strength behavior of dredged 498 mud slurry treated jointly by cement, metakaolin and flocculant. Applied Clay Science 499 193, 105676 500 Du, Y.J., Jiang, N.J., Liu, S.Y., Horpibulsuk, S., and Arulrajah, (2016) Field evaluation 501 of soft highway subgrade soil stabilized with calcium carbide residue. Soils and Foun- 502 dations, 56(2), 301-314, https://doi.org/10.1016/j.sandf.2016.02.012 503 Du, Y.J., Wu, J., Bo, Y.L. and Jiang, N.J. (2020) Effects of acid rain on physical, me- 504 chanical and chemical properties of GGBS–MgO-solidified/stabilized Pb-contaminated 505 clayey soil. Acta Geotechnica, 15, 923–932, https://doi.org/10.1007/s11440-019- 506 00793-y 507 Ghadir, P., and Ranjbar, N. (2018) Clayey soil stabilisation using geopolymer and Port- 508 land cement. Construction and Building Materials, 188, 361-371. 509 Gallagher, P.M. and Mitchell, J.K. (2002). Influence of colloidal silica grout on liquefac- 510 tion potential and cyclic undrained behavior of loose sand. Soil Dynamics and Earth- 511 quake Engineering, 22, 1017–1026. 512 Gallagher, P.M., Pamuk, A. and Abdoun, T. (2007). Stabilization of liquefiable soils 513 using colloidal silica grout. Journal of Materials in Civil Engineering, 19(1), 33-40, 514 https://doi.org/10.1061/(ASCE)0899-1561(2007)19:1(33). 515 Georgiannou V.N., Pavlopoulou E.M., Bikos Z., (2017). Mechanical behaviour of sand 516 stabilised with colloidal silica. Geotechnical Research, 4(1), 1-11 517 International Energy Agency (2018) Technology Roadmap: Low-Carbon Transition in 518 the Cement Industry, IEA 519 Horpibulsuk, S., Miura, N., & Nagaraj, T. S. (2005). Clay–water∕ cement ratio identity 520 for cement admixed soft clays. Journal of geotechnical and geoenvironmental engi- 521 neering, 131(2), 187-192. 522 Ishihara, K., Yamazaki, A., Haga, K. (1985). Liquefaction of K0-consolidated sand un- 523 der cyclic rotation of principal stress direction with lateral constrains. Soils and Foun- 524 dations, 25(4), 63-74. 525 Ji Z., and Pei, Y. (2019). Bibliographic and visualized analysis of geopolymer research 526 and its application in heavy metal immobilization: a review. Journal of Environmental 527 Management 231, 256-267. https://doi.org/10.1016/j.jenvman.2018.10.041 528 Karol, R.H. (2003). Chemical Grouting and Soil Stabilization. Marcel Dekker, Inc., New 529 York, NY. 530 Littlejohn GS (1993). Underpinning by chemical grouting. In Underpinning and Reten- 531 tion (Thorburn S and Littlejohn GS (eds)). Springer, New York, 242-275. 532 Lehne, J. and Preston, F. (2018). Making concrete change. London, UK: The Royal 533 Institute of International Affairs. 534 Katsiki, A., Hertel, T., Tysmans, T., Pontikes, Y., and Rahier, H. (2019) Metakaolinite 535 Phosphate Cementitious Matrix: Inorganic Polymer Obtained by Acidic Activation. Ma- 536 terials 12, 442 537 Kendall, A., Raymond, A. J., Tipton, J., and DeJong, J. T. (2017). Review of life-cycle- 538 based environmental assessments of geotechnical systems. In Proceedings of the In- 539 stitution of Civil Engineers-Engineering Sustainability (Vol. 171, No. 2, pp. 57-67). 540 Thomas Telford Ltd. 541 Kolovos, K.G., Asteris, P.G., Cotsovos, D.M., Badogiannis E., and Tsivilis, S. (2013). 542 Mechanical properties of soilcrete mixtures modified with metakaolin. Construction and 543 Building Materials 47, 1026–1036. 544 TABLES 664 Table 1: Summary of previous geotechnical characterization studies on CS. 665 Year Authors CS Mixture porosity, 𝜼 * CS concentration, CCS, w/w (%) silica/ dry soil w/w (%) 𝑪𝒊𝒗 Soil Sample forming Mech tests 1999 Persoff et al., 1999 DuPont Ludox SM 0.38 5 to 27 1.4 to 7.5 0.008 to 0.047 sand Pluviation on grout UCS 2002 Gallagher and Mitchell, 2002 DuPont Ludox SM30 0.429 5 to 20 1.7 to 6.8 0.010 to 0.039 loose sand Pluviation on grout UCS CTX 2008 Diaz Rodriguez et al. 2008 n.a. 0.482 to 0.502 14.5 6.8 to 7.4 0.036 to 0.037 fine sand Pluviation on grout K0-cyclic simple shear 2011 Porcino et al., 2011 n.a. 0.417 10 2.9 0.017 sand Permeation from base (3 pore volumes) UCS Direct simple shear CSSU CTXU 2012 Porcino et al., 2012 TSG 0.417 10 2.9 0.017 sand Permeation from base TXCID cyclic simple shear Year Authors CS Mixture porosity, 𝜼 * CS concentration, CCS, w/w (%) silica/ dry soil w/w (%) 𝑪𝒊𝒗 Soil Sample forming Mech tests 2015 Vranna and Tika, 2015 Ludox SM30 0.408 to 0.432 10 3.1 to 3.5 0.019 to 0.020 sand Permeation from base (4 specimen volumes) TXCIU CTXCIU 2017 Georgiannou et al., 2017 Ludox SM30 0.355 to 0.432 10 2.5 to 3.5 0.016 to 0.020 sand Pluviation on grout UCS Direct shear test TXCID TXCAD 2018 Wong et al. 2018 MasterRoc MP 320 0.350 sand 0.480 kaolin 40 sand @ 11.7 % kaolin @ 61.7 % sand @ 0.076 kaolin @ 0.321 Sand / Kaolin Pouring CS on sand Hand mixing kaolin Direct shear test Oedometer test Year Authors CS Mixture porosity, 𝜼 * CS concentration, CCS, w/w (%) silica/ dry soil w/w (%) 𝑪𝒊𝒗 Soil Sample forming Mech tests 2020 Ciardi et al., 2020 MasterRoc MP 325 0.405 2 to 13 0.6 to 3.7 0.003 to 0.022 sand Pluviation on grout Direct shear test Oedometer test CTXU 2020 Salvatore et al., 2020 MasterRoc MP 325 0.429 3; 5; 10 0.9 to 3.1 0.005 to 0.018 sand Permeation from base (until immersion complete) Vane test TXCID CTXCIU * : As compacted/poured soil porosity 666 667 Table 2 Physical and chemical properties of the base materials 668 669 670 671 672 673 674 675 Holcim sand Carbonate silt Sand with 10% carbonate silt Metakaolin powder Colloidal Silica Quartz content (w/w, %) 92.1 - 82.9 55 40 CaCO3content (w/w, %) - 98.2 9.8 - - pH value 6.69 9.90 - 6 7-9 Maximum grain size, d100 (mm) 0.710 0.161 0.710 0.080* - Mean grain size, d50 (mm) 0.450 0.033 0.450 0.010-0.015 1.510-5 ** Grain size, d10 (mm) 0.336 0.003 0.172 - - Coefficient of uniformity 1.4 - 1.8 - - Grains fraction < 2 m (%) - 8.0 0.8 100 100 Density of solids,  s (Mg/m3) 2.65 2.71 2.66 2.40 2.11 ** Hygroscopic w/c (%) at RH=50% <0.3 0.1 <0.3 - - Bulk density as poured (Mg/m3) 1.34 1.10 1.47 - - Void ratio as poured 0.825 0.464 0.584 - Hydraulic conductivity as poured, kw (m/s) 7.6710-4 - 2.8510-4 - Maximum void ratio, emax 0.982 - - - - Minimum void ratio, emin 0.532 - - - - * d95 ** Wong et al., 2018 Table 3 Characteristics of the binders. 676 Geopolymer (MK) Colloidal Silica (CS) Binder Precursor material Metakaolin powder Aqueous dispersion of silica Activator (*) / Accelerator (**) Potassium silicate (K2SiO4) (*) De-aired water + 12% added to volume of NaCl (10% w/w) (**) Precursor : activator / accelerator (V/V) 1:1 8.3:1 Other fractions De-aired water - Water/binder (w/w) 1:2 - % of void’s volume filling 40% - 100% 100% Binder-soil mixing technique Hand-mixing Low-pressure permeation Curing conditions 50% RH / Under water (20°C) 50% RH (20°C) Base soils S / SO S / SO Density at slurry/liquid state (Mg/m3) 1.37 1.30 677 678 Table 4 Material proportions in the mixtures. 679 Geopolymer (MK) Colloidal Silica (CS) Size Soil 𝐹0 = Vfluid/Vpores (%) Geopolymer slurry/dry soil (w/w) Metakaolin powder/dry soil (w/w) 𝐶𝑖𝑣 = VMKpowder/Vtot (-) 𝜂 / 𝐶𝑖𝑣 (- ) w/c0 (-) Colloidal Silica slurry*/dry soil (w/w) Silica / dry soil (w/w) 𝐶𝑖𝑣 = Vsilicar/Vtot (-) 𝜂 /𝐶𝑖𝑣 (-) w/c0 (-) UC S 100 0.43 0.14 0.086 5.3 1 0.40 0.14 0.075 6.0 1.8 TX S 100 0.43 0.14 0.086 5.3 1 0.40 0.14 0.075 6.0 1.8 UC SO 100 0.27 0.09 0.061 5.3 1 0.28 0.09 0.053 6.0 1.8 TX SO 100 0.27 0.09 0.061 5.3 1 0.28 0.09 0.053 6.0 1.8 TX S 40 0.17 0.06 0.034 13.1 1 n.a. n.a n.a. n.a. n.a. TX SO 40 0.11 0.04 0.024 13.1 1 n.a. n.a. n.a. n.a. n.a. * aqueous dispersion + accelerant 680 Table 5 Overview of the tests performed and treated soils tested. Number of curing days in parenthesis 681 Unconfined compression tests (UCS) Consolidated-Un- drained Static Triaxial Tests (TXCIU) Consolidated-Un- drained Cyclic Triaxial Tests (CTXU) S X X SO X X SMK(100)D X (3,7,28) X (3,7,28) SMK(100)W X (3,7,28) SMK(40)D X (3,7,28) SMK(40)W X (7) SOMK(100)D X (3,7,28) X (3,7,28) SOMK(40)D X (3,7,28) SOMK(40)W X (7) SCS(100)D X (1,7,28) X (3,7,28) X (7) SOCS(100)D X (1,7,28) X (3,7,28) X (7) 682 683 Table 6 Density of grains in the mixture, dry density, water content and void ratio of CS-treated samples. 684 ρs* (Mg/m3) ρd (Mg/m3) w (%) as-treated void ratio, e (-) SCS(100)D28 2.47 1.70 0.12 0.453 SOCS(100)D28 2.49 1.72 0.13 0.448 ρs* calculated as weighed average of the constituent fractions 685 686 Table 7 Summary of mixture characteristics from the FA geopolymer studies of Rios et al. (2017a; 687 2017b) 688 Author Sample ID As poured/compacted porosity, 𝜂 (-) Binder powder/dry soil (w/w) % 𝐶𝑖𝑣 (-) 𝜂 / 𝐶𝑖𝑣 (-) Rios et al. (2017a) n.a. 0.248 20 0.150 1.65 Rios et al. (2017b) M1 0.298 18 0.124 2.41 M2 0.342 25 0.164 2.08 M3 0.382 33 0.206 1.86 689 690 Table 8 Strength parameters obtained by the different treatments. 691 SMK(100) SMK(40) SOMK(100) SOMK(40) SCS(100) SOCS(100) S SO c’ (kPa) 225 20 310 84 26 26 0 0 φ’ (°) 48 48 48 48 41 41 38 38 692 693 730 Figure 9 TXCIU Secant undrained Young Modulus at εax = 0.01% 731 732 733 734 735 Figure 10 TXCIU Secant undrained Young Modulus at εax = 2.50% 736 737 738 Figure 11 Evolution of the normalized excess pore water pressure as a function of axial strains (left) and 739 number of applied cycles (right). 740 741 742 743 744 745 746 747 748 749 750 Figure 12 Evolution of the stress state in CTXC tests and comparisons with failure envelopes obtained 751 by TXC. 752 753 754 Figure 13 Cyclic stiffness degradation in the untreated and treated specimens, jointly with some samples 755 at the end of test. 756 757 758 759 760 761 Figure 14 Peak strength envelopes for treated and untreated soils. Treated soils include the silty sand 762 treated with a Fly-ash (FA)-based geopolymer (Rios et al., 2017a) and Holcim sand treated with a MK- 763 based geopolymer (this study) 764 765 Figure 15 Peak strength envelopes for treated and untreated soils. Treated soils include the silty sand 766 treated with a Fly-ash (FA)-based geopolymer (Rios et al., 2017a) and Holcim sand treated with a MK- 767 based geopolymer (this study) 768 769 Figure 16 Secant stiffness degradation during monotonic triaxial shearing for different geopolymer im- 770 proved soils 771 772 773 Figure 17 CS treated sands (this study+literature). Comparisons between TXC and Direct Shear tests 774 at low vertical stresses (σ1 < 300 kPa) on samples treated with the same CS product (MP 320, silica 775 concentration in the solution Ccs = 40%). 776 777 Figure 18 Permeability vs water content / ponderal binder ratio for the materials tested in this study and 778 literature values for other CS and fly ash geopolymer treatments. 779 780 781 Figure 19 FESEM image of specimen SMKD(40). Sand grains are bound by cracked geopolymer 782 bridges 783 784 View publication statsView publication stats