The Role of Different Clay Types in Achieving Low-Carbon 3D Printed Concretes

Abstract

Concrete 3D printing, an innovative construction technology, offers reduced material waste, increased construction speed, and the ability to create complex and customized shapes that are challenging to achieve with traditional methods. This study delves into the unique fresh-state performance required for 3D printing concrete, discussing buildability, extrudability, and shape retention in terms of concrete rheology, which can be modified using admixtures. Currently most 3D printing concretes feature high cement contents, with little use of secondary cementitious materials. This leads to high embodied carbon, and addressing this is a fundamental objective of this work. The research identifies attapulgite, bentonite, and sepiolite clay as potential concrete admixtures to tailor concrete rheology. Eight low-carbon concrete mixes are designed to incorporate GGBS at a 50% replacement level and are used to measure the influence of each clay on the concrete rheology at varying dosages. A comprehensive rheological test protocol is designed and carried out on all mixes, together with other tests including slump-flow and compression strength. The objective is to determine the applicability of each clay in improving the printability of low-carbon concrete. The findings reveal that at a dosage of 0.5%, sepiolite was seen to improve static yield stress, dynamic yield stress, and rate of re-flocculation, resulting in improved printability. The addition of attapulgite and sepiolite at a dosage of 0.5% by mass of binder increased compressive strength significantly; bentonite had very little influence. These gains are not repeated at 1% clay content, indicating that there may be an optimum clay content. The results are considered encouraging and show the potential of these clays to enhance the performance of low-carbon concrete in 3D printing applications.

1. Introduction

Three-dimensional concrete printing is an emerging concrete construction process in which fresh concrete is pumped through a robotic arm where it is continuously extruded, or ‘printed’, at a constant rate as the extrusion head moves along a programmed path, typically a closed loop [1]. The printing of concrete along the path produces a ‘layer’. Vertical elements can be constructed by programming the printer to continuously move along the same path and increasing the height of the printer head as it returns to the start of its loop, such that the following layer of concrete is extruded atop the previously deposited layer. The concrete employed must rapidly gain strength after deposition in order to sustain the weight of successive layers [2]. Construction through this process has the potential to reduce the cost, time, and labor associated with vertical concrete elements by eliminating the need for formwork and a formwork contractor. Three-dimensional printing also allows for the simple construction of elaborate geometries that would traditionally involve complex formwork [3]. As the success of a 3D printed concrete structure is dependent upon the fresh state behavior of concrete [4], it is the fresh state properties of concrete that must be engineered. Rheology is the study of the properties that determine a material’s deformation and flow and, thus, is highly relevant for contexts such as this [5]. The selection of an appropriate model that describes the behavior of the material in the fresh state then becomes an important consideration. The Bingham plastic model, given in Equation (1), is typically used to characterize the stress–strain relationship of fresh concrete [6]. This model predicts that concrete behaves as a viscoelastic material, which behaves as an elastic solid until imposed strain induces shear stress τ greater than the static yield stress τ₀, at which point flow begins [7]. During flow, the material exhibits a constant plastic viscosity μ_p that relates the strain rate $\dot{\gamma }$ linearly to the induced shear stress.

$$\tau ={\tau }_0+{\mu }_p\dot{\gamma }$$

(1)

Roussel [8] has shown structuration to occur at a linear rate for approximately the first 30 min, during which Equation (2) characterizes this behavior. The growth of the static yield stress occurs at a rate A_thix, the structuration rate (Pa/s), τ_0,i is the initial static yield stress at the reference state, t = 0, immediately after the concrete has undergone mixing. Beyond 30 min of resting time, the effect of hydration results in a more rapid and irreversible loss of workability [8].

$${\tau }_0\left(t\right)={\tau }_{0,i}+A_{thix}t$$

(2)

Kruger et al. [9] propose a bilinear model for structuration, in which the increase in yield stress from the initial dynamic yield stress to the initial static yield stress after shearing occurs at a rate R_thix, the short-term re-flocculation rate, which is greater than A_thix. It has been seen that this growth rate is characterized by Equation (3) for approximately the first 5 min of rest following shearing, after which A_thix predicts the rate of yield stress growth [10].

$${\tau }_0\left(t\right)={\tau }_{D,i}+R_{thix}t$$

(3)

Within the context of 3D printing, the terminology has developed to describe the desired behavior of concrete. ‘Extrudability’ describes the ability of the concrete both to be pumped and to pass through the smaller diameter nozzle at a constant and desirable rate [11]. A concrete that cannot be successfully extruded cannot form a continuous filament for printing. ‘Shape retention’ describes the ability of a 3D printed concrete to retain the cross-section as that of the extrusion nozzle [11]. Shape retention is a critical factor in the 3D printing process due to the cumulative effect of deformation as successive layers are printed [2]. A filament that cannot retain its shape cannot successfully sustain the load of additional layers or successfully deliver the programmed shape. ‘Buildability’ is used to describe the ability of a concrete mix to successfully sustain the weight of successive layers without flow occurring and is typically quantified as the number of layers printed until collapse [12]. Desirable shape retention is dependent upon the static yield stress of the concrete mix, as this property determines how resistant the filament is to deformation under its own self-weight. It follows that the buildability of a mix is dependent upon the development of the yield stress over time as it is required that the material can resist the increased stress of successive layers. Thus, the structuration and re-flocculation rate is relevant. It is essential to optimize the mix composition and processing techniques to enhance the aforementioned properties for achieving better 3D printed concrete. This can be performed by adjusting the different types and quantities of additives/admixtures used to achieve desired properties as per printed concrete requirements. Superplasticizers and viscosity modifying agents have been adopted in numerous studies for the purpose of adjusting the viscoelasticity of the fresh concrete [13]. Retarders and accelerators have also seen use for the purpose of increasing or decreasing open time [13].

A barrier to the adoption of concrete 3D printing is the high carbon embodied by 3D printed structures due to the typically high cement content [14]. Different supplementary cementitious materials (SCM) have been used with varying content to reduce this impact [15]. GGBS, silica fume, and fly ash have all seen use as SCM due to the low carbon footprint associated with these waste products [13]. Further efforts have also been made to develop 3D printable geopolymer mixes, in which cement is absent from the binder [16]. In order to retain desirable rheological properties, admixtures often have to be incorporated into the mixes [17]. One group of admixtures that has received extensive research is nano-clays [18]. These are clay particles with at least 1 dimension in the nanoscale, which have been used to compensate for the poor printability of low cement mixes [2]. Therefore, it is essential to study different types of clays and identify their role in improving the properties of 3D printed concrete. Attapulgite/palygorskite is a clay that has seen significant use as a rheology modifier in concrete [19]. The addition of attapulgite nano-clay has repeatedly been shown to result in a linear increase in static yield stress proportional to the amount of nano-clay added, resulting in an increase of over 300% from 2% cement being replaced with nano-clay (by weight) [11]. A smaller increase could also be seen in the dynamic yield stress in proportion to the amount of nano-clay added [11]. It was also shown that following shearing, the rate of strain decline exhibited by nano-clay cement pastes was greater than those without the nano-clay [20]. This demonstrates that the nano-clay increases the flocculation rate within the cement paste, resulting in a faster recovery of the static yield stress following shearing. This is also reflected in the results of Arunothayan et al. [21], in which the static yield stresses of cement paste samples containing the 0.01% and 0.2% nano-clay (by mass of binder) were shown to recover rapidly within the first 5 min after shearing after which it continues to grow at a slower rate. This is in line with the bilinear static yield stress growth model proposed by Kruger et al. [9]. For binders of high fly ash content, the structuration rate A_thix is shown to increase for cement pastes where attapulgite nano-clay replaces 0.5% of the binder by mass [22]. Incorporating the nano-clay into a cement paste has also been shown to cause a minor increase in the apparent viscosity [11]. These results all describe a significant stiffening of the material at low volumes of nano-clay. The impact of these rheological modifications is reflected in 3D printing performance tests, where samples incorporating attapulgite show greater buildability [22,23]. Bentonite clay has also been utilized for similar purposes. A study by Kaci et al. [24] tests cement mortars containing between 0 and 1% bentonite and reported that the characteristic time for yield stress recovery is shorter for those containing greater quantities of bentonite. This is in agreement with Aydin et al. [23] measurement of the rate of structural recovery, which is shown to increase within the same range of bentonite quantities. Bentonite also increases the static yield stress of a mix that has not rested and accelerates the structuration rate of the concrete [25]. The static yield stress was shown to increase greatly for cement mortar containing more than approximately 5% bentonite by weight of cement [26]. Reales et al. [27] measured the shape stability of concrete samples dosed with bentonite by comparing the max height that each printed concrete mix was able to sustain and found that samples composed of 1% bentonite and beyond could sustain heights of 14 mm compared to the control which could sustain a 2 mm height. While there exists extensive research into the application of attapulgite as a rheology modifier in concrete, there is limited research into the use of sepiolite nano-clay for the same purposes despite their physical similarities. Replacing cement binder with sepiolite in cement mortars has resulted in increased static yield stress and apparent viscosity [28]. From a study by Aydin et al. [25] into the application of sepiolite nano-clay for the purposes of 3D printing, it was reported that a 1% mass of cement replacement resulted in a dynamic yield stress greater than 4 times that of the control and that the rate of structuration was consistently greater than that of the control during rest. The structural recovery of cement mortars following shearing is the greatest in samples containing 0.5% of the nano-clay [25]. In printing tests, the high dynamic yield stress associated with 1% sepiolite resulted in the tearing of the filament. However, when sepiolite was combined with viscosity modifying agents, the samples showed high levels of extrudability [25,28]. Therefore, a comparative study is necessary to evaluate the difference between all these various types of clay to provide a better understanding of using it for 3D printed concrete. It has been observed that the inclusion of nano-clay in 3D printed concrete can have a positive influence on tensile bond strength [20]. The issue of tensile strength in 3D printed concrete arises during curing due to shrinkage. Moreover, 3D printed structures are more susceptible to cracks arising from shrinkage due to the lack of formwork [29]. Tensile strength has been improved in previous studies through the addition of fibers [29]. While success has been shown with steel fibers, alternative materials such as carbon, polypropylene, and aramid fibers have been utilized to avoid issues of corrosion [30]. As fibers primarily serve to improve the mechanical properties of hardened concrete, they are not included in this investigation, which focuses on printability.

As discussed above, one of the barriers to adopting the mix for 3D printed concrete is higher cement content. Therefore, in this study, a low-carbon 3D concrete printing mix is designed using GGBS to replace cement. The influence of these different types of clays in a 50% GGBS binder 3D printable mix has not been investigated previously, and it is still unknown if these clays can improve the properties of a low-carbon mix. The objectives of this paper are as follows: (1) To model the fresh state behavior of low-carbon concrete mix in terms of rheological properties; (2) To determine the printability of a low-carbon concrete mix based on said rheological properties; (3) To determine the influence of sepiolite, attapulgite, and bentonite clays on the printability of a low-carbon concrete mix; (4) To measure the impact of sepiolite, attapulgite, and bentonite clays on the compressive strength of a low-carbon concrete mix. The results of rheological tests and slump-flow tests judge the printability before and after the addition of clays. The compressive strength of mixes with and without clays was evaluated. The outcomes of this will contribute to the body of knowledge on the development of low-carbon 3D printable concrete. This will bring concrete 3D printing closer to the low carbon footprint required in modern construction.

2. Materials and Methods

2.1. Materials

The binder was composed of Irish Cement Type II Portland cement (CEM), which was compliant with I.S. EN 197-1 [31], and Ecocem ground granulated blast slag (GGBS), which was compliant with I.S. EN 15167-1 [32]. The binder that was composed of these materials is equal in proportion by mass. Silica sand compliant with I.S. EN 196-1, with a max diameter of 2 mm, was used as the fine aggregate [33]. A SikaPlast Polycarboxylate (PCE) based super-plasticizing agent (SP) in line with I.S. EN 934-2 was included in all mixes [34]. Following trial mixes, a water-to-binder ratio (w*/b) of 0.252 was adopted, a sand-to-binder ratio (s/b) of 1 was adopted, and SP was included at a dosage of 1.2% by mass of binder. ‘w*’ represents the mass of water and SP.

Table 1. Mass-to-binder and mass-to-cement ratios of every constituent of each mix.

Mix Name	Constituent Ratios
	CEM/b	GGBS/b	w*/b	s/b	BEN/b	ATT/b	SEP/b
REF	1	0	0.252	1	-	-	-
CLAY 0%	0.5	0.5	0.252	1	-	-	-
BEN 0.5%	0.5	0.5	0.252	1	0.005	-	-
BEN 1%	0.5	0.5	0.252	1	0.01	-	-
ATT 0.5%	0.5	0.5	0.252	1	-	0.005	-
ATT 1%	0.5	0.5	0.252	1	-	0.01	-
SEP 0.5%	0.5	0.5	0.252	1	-	-	0.005
SEP 1%	0.5	0.5	0.252	1	-	-	0.01

below lists the constituent proportions for each of the eight mixes produced and their mix names.

Attapulgite (ATT), bentonite (BEN), and sepiolite (SEP) powders were purchased from Kremer Pigmente and were included in mixes at dosages of either 0.5% or 1% by mass of binder. The purpose of this study is to explore the potential of various clay types to contribute to 3D printed concrete. As such, the clays were used as supplied, and no treatments were utilized to modify their activity. While it has been observed that calcined clays can improve thixotropy and buildability, these clays were not included in the investigation as the higher carbon footprint associated with the calcination process would be detrimental to the goal of developing a low-carbon solution [35]. The chemical compositions of ATT and BEN were determined through X-ray fluorescence (XRF) analysis and are given in

Table 2. Chemical composition and bulk density of clays and GGBS.

Constituent	ATT (%)	BEN (%)	SEP (%)	GGBS (%)
CaO	3.51	2.49	0.5	42.4
SiO2	52.7	46.3	60.5	36.5
Al2O3	11.3	18.2	2.4	10.4
MgO	8.02	2.72	23.8	8.1
Fe2O3	3.96	5.72	0.9	0.7
K2O	0.894	1.5	0.5	-
Na2O	<0.0001	0.238	0.1	0.5
P2O5	0.476	0.0385	-	-
TiO2	0.616	0.485	-	0.5
MnO	0.0345	0.054	-	0.4
SO	-	-	-	0.1
S2−	-	-	-	0.7
Cl	-	-	-	0.01
L.O.I	-	-	11.3	-
Bulk Density (kg/m3)	497	750	60 ± 30	900–1200

below. This was performed using a Rigaku NEX CG ED-XRF machine calibrated to Standard Reference Material 2701. The composition of GGBS and SEP was provided by the supplier. The bulk density of these materials is also included. As per the supplier, SEP particles had an average fiber length of 2 μm and a diameter of 20 nm. The SEP particle size distribution is as follows: >90% of particles are less than 5 μm in size, <5% are retained by a 10 μm mesh, and <0.5% of particles are retained by a 44 μm mesh. The ATT particles had an average size of 31 μm, with 92% of particles passing a 100 μm mesh. The BEN particles were stated to have 10–40% of mass retained by a 63 μm mesh. While existing literature has focused on nano-sized clay particles [18], larger particles on the micro-scale were used in this investigation. The monetary cost of highly processed and purified clay particles poses a barrier to their adoption in concrete 3D printing applications, and as such larger clay particles were used at a fraction of the cost of nano-clays.

Table 3. Carbon factor of each mix constituent excluding water.

Constituent	CEM	GGBS	SAND	SP	CLAY
Carbon Factor (kgCO2/tonne)	678	34	33 [36]	369	140 [37]

below provides the carbon factor associated with phase A1–3 of each mix component. The carbon factors of CEM, GGBS, and SP are taken from the product EDPs. The carbon footprints associated with BEN, SEP, and ATT are assumed to be equal, and the carbon factor of water is considered negligible. The carbon factors are used with the mix proportions described in Table 1 to calculate the embodied carbon associated with one tonne of each mix.

2.2. Mixing

Mixing was conducted in an ELE International mixer, compliant with I.S. EN 196-1 [33]. CEM, GGBS, sand, and clay (if present) were dry mixed first at 140 ± 5 rpm for 60 s. Prior to mixing, water and SP are combined. Following dry mixing, the water-SP mixture was added to the mixer using a funnel for over 30 s. The wet mix was then mixed at 285 ± 10 rpm for 1 min and finally at 140 ± 5 rpm for 2 to 3 min, depending on the apparent consistency of the mix. Figure 1 below shows the mixer with dry constituents, and Figure 2 illustrates the steps for the mixing procedure.

2.3. Test Methodology

2.3.1. Slump-Flow

A sample of each mix was filled in a slump-flow apparatus to test its consistency. For this test, a slum-flow table and mold in line with ASTM C230 were used [38]. The slump-flow apparatus was adapted with a Perspex table to accommodate the wider flow diameter of the mixes, as shown below in Figure 3. In the test, the mold was fully filled without tamping. Compaction was deemed unnecessary during trial mixes due to the high flowability of the mixes. The mold was scraped level, and any excess concrete was cleaned off the table. The mold was immediately removed, and any concrete remaining on the mold was allowed to drip onto the sample. The diameter of the resulting undisturbed mortar flow was recorded in two perpendicular directions. The average was taken as the slump-flow result. During trial mixes, it was deemed unnecessary to use the table’s striking mechanism to agitate the samples, given the high flowability seen in the samples.

2.3.2. Compressive Strength Test

To test the compressive strength of each mix, 50 mm cubic samples were prepared by filling molds of the same size with concrete from each batch. These molds were coated with de-molding oil prior to being filled. Upon being filled, mixes were deemed flowable enough that no vibration was necessary. The specimens were wrapped with plastic sheeting in order to retain moisture and allowed to set. The specimens were removed from their molds the following day and placed in a water tank, maintained at 20 °C until they reached their designated testing age. Three samples were tested after 28 days. Specimens were placed into an Instron 6800 series universal testing machine and subjected to uniaxial unconfined compression at a constant rate of 2 mm/min until failure. Figure 4 below shows a specimen before and after testing. The compressive strength $f_c$ (MPa) of each specimen is determined using Equation (4) below where F_f denotes the compressive load at failure (N) and A_c (mm²) denotes the cross-sectional area subjected to loading. The result is considered the average strength of the three specimens crushed at each testing age.

$$f_c=\frac{F_f}{A_c}$$

(4)

2.3.3. Rheological Tests

Determination of the rheological properties of concrete mixes was carried out using an Anton Paar MCR 102e SmartPave, shown in Figure 5 below. Immediately upon the completion of the mixing process, the bowl was unloaded from the mixer, and the mix was poured into an Anton Paar Building Material Cell 90 (BMC) and loaded into the rheometer. The BMC was filled to a volume that reached 50 mm below the top surface. The BMC features a ribbed interior surface to prevent wall slip and has an interior diameter of 70 mm and depth of 100 mm. A construction material stirrer was used to perform shearing on the sample. The geometries of the stirrer and BMC are shown in Figure 6 below. The stirrer was positioned 5 mm from the bottom surface of the BMC to prevent blockages due to the max particle size of the mixes being 2 mm. The rheometer heating plate was maintained at 20 °C throughout all tests, and upon beginning the rheological testing, the temperature was allowed to reach equilibrium over 1 min.

A hysteresis loop test protocol was adopted to test the samples. As the structuration of each sample would not be the same upon being loaded into the rheometer, each sample is initially subjected to a pre-shear at a rate of 100 (1/s). By inducing a high shear rate in each sample, any strength that may have developed due to flocculation before the rheological testing begins is eliminated by breaking down the network of flocculated cement particles in the samples. A high shear rate will not entirely eliminate structuration in each sample, but by maintaining the shear rate for 100 (1/s), each sample will be brought to an approximately equal level of structuration to allow for comparison of the following tests’ results.

The pre-shear phase was followed by a 30 s rest period for each sample, allowing the structure to re-develop. As per a hysteresis loop test protocol, a linearly increasing shear rate is imposed on each sample and then decreased linearly until the sample returns to rest. The shear rate increases from rest to 100/s over 60 s (the ‘up’ phase) and decreases at the same rate over the following 60 s (the ‘down’ phase). This shear rate cycle is illustrated below in Figure 7. During this phase of testing, the sample is forced to undergo flow, and thus, the vane must apply stress greater than the static yield stress of the sample to accomplish this. As a result, the static yield stress of the sample can be determined by measuring the torque required by the rheometer to initiate flow. The static yield stress corresponds to the stress at the point of highest torque.

The stress–strain rate results of a hysteresis loop protocol highlight any thixotropy within the material tested. As the shear rate rises, so too does shear stress, as per Equation (1), while simultaneously, as the shear rate rises and persists over time, the structuration within the sample is disrupted, and viscosity is decreased. This reduces the stress required to generate the shear rate. Consequentially, the rate of de-flocculation and increasing shear rate compete to decrease and increase the stress required for the imposed shear rate. Whichever of these factors dominates will determine if the shear rate-stress relationship will be positive or negative during the up phase of the shear rate loading. The sample has been de-flocculated to its greatest extent upon reaching the peak shear rate. The stresses measured during the down phase of the shear rate loading will thus be lower than those recorded at the same shear rate during the up phase. The resulting difference between the results of the up and down curves has been used to measure the thixotropy of a material. This test, however, reveals nothing of the thixotropic strength recovery characteristics of a material at rest. This controlled shear rate protocol was carried out three times on each sample. The proceeding test began immediately upon the end of the flow curve test. The protocol was varied by increasing the rest period between the pre-shear and flow curve phases. The 30 s rest period was increased to 3 min and 5 min for tests that followed. By making each sample undergo pre-shearing and resting for longer periods of time, the thixotropic strength recovery of each sample can be observed by comparing the yield stress during each flow curve. The full protocol is shown below in Figure 8.

3. Results

3.1. Carbon Footprint

Analysis of the Environmental Product Declarations (EPDs) of the materials used in this work shows that the cement used has an embodied carbon of 678 kg CO₂ eq. per tonne for modules A1 to A3; the GGBS has an embodied carbon of 34 kg CO₂ eq. per tonne. This means that all concrete mixes with 50% GGBS replacement will have a significant reduction in the embodied carbon arising from the binder. Given the high binder contents associated with concrete printing, the use of GGBS has significant potential to reduce embodied carbon.

Table 4. Embodied carbon of 1 tonne of reference mix and low-carbon mix at clay dosages of 0, 0.5, and 1%.

Mix Label	REF	CLAY 0%	CLAY 0.5%	CLAY 1%
Embodied Carbon (kgCO2)	398.932	215.142	215.541	215.941

illustrates the embodied carbon of one tonne of each group of mixes studied in this investigation. As the carbon factor of each clay is considered identical, the embodied carbon of all mixes that contain the same clay dosage will be identical. Thus, CLAY 0.5% and CLAY 1% represent the carbon emissions from BEN 0.5%, ATT 0.5%, SEP 0.5% and BEN 1%, ATT 1%, SEP 1% respectively. It can be seen that CLAY 0% represents a 46% reduction in embodied CO₂ and that the addition of clay up to 1% does not cause a significant change in the magnitude of this reduction.

3.2. Slump-Flow

Figure 9 shows the slump-flow results for each mix, and results are also tabulated in

Table 5. Average max flow diameter achieved by each mix.

Mix Label	REF	CLAY 0%	BEN 0.5%	BEN 1%	ATT 0.5%	ATT 1%	SEP 0.5%	SEP 1%
Flow (mm)	143	358	328	298	313	263	273	183

below. All mixes were highly flowable except for REF. This is in line with the expected effect of GGBS substitution on cement-based materials to increase workability [39]. In the past, the slump-flow diameter has been used as an estimator for printability, as it has been seen that the static yield stress is related to the max diameter of flow [40]. At all times within the sample, the stress at the bottom is proportional to the mass of the material above. This stress is typically greater than the static yield stress of the sample, and thus, as the mold is removed, flow occurs. Upon spreading, the mass of material above any point diminishes, and flow is expected to stop once the stress induced by this diminishing mass is less than the dynamic yield stress. For this reason, a small slump-flow diameter represents a mix with high resistance to flow and strong shape retention. The greatest result observed was that of CLAY 0% at 358 mm. In all cases, the addition of clay can decrease the max flow radius from that of CLAY 0%, thus increasing the consistency of the mixes.

The influence of each clay on the flowability of mixes is illustrated below in Figure 10. The slump-flow diameter decreases linearly with the addition of BEN and ATT clays. A linear relationship has been fit to all clays to illustrate this. It can be seen that SEP dosage also induces a decrease in slump-flow diameter, but the experimental results do not support a linear relationship. It is possible that, similar to how some properties begin to diminish at high doses of clay, as seen by Douba and Kawashima [11], the effectiveness of SEP dosage decreases beyond a certain dosage, and this dosage is between 0.5% and 1%. At a dosage of 0.5%, it can be seen that SEP induces a reduction in slump-flow of 125 mm, over four times that of BEN (30 mm) and over twice that of ATT (45 mm). The higher potency of SEP can be attributed to particle characteristics. The SEP particles were significantly less dense than BEN or ATT. SEP particles were also significantly smaller than BEN or ATT. As a result, a given mass of SEP particles will consist of a larger volume of smaller-size particles than the same mass of BEN or ATT. Consequentially, the SEP particles can likely disperse within the concrete mix more effectively than ATT or BEN, and thus more effectively influence the flowability.

3.3. Compressive Strength

Table 6. 28-day compressive strength of each concrete mix.

Mix Label	REF	CLAY 0%	BEN 0.5%	BEN 1%	ATT 0.5%	ATT 1%	SEP 0.5%	SEP 1%
28-day Compressive Strength (MPa)	86.55	100.92	101.31	91.81	135.60	113.03	126.00	87.11

below shows the 28-day compressive strength results of each mix. It can be seen that the addition of GGBS resulted in a 16% increase in compressive strength from REF. Figure 11 below illustrates the influence of each clay on the 28-day compressive strength. It can be seen that a 1% dosage of any clay resulted in a decrease in compressive strength when compared to the results at 0.5% of the same clay. A 0.5% dosage of SEP or ATT resulted in an increased compressive strength of 25% and 35%, respectively, while BEN did not result in a meaningful increase at this dosage. These results suggest that there exists a percent dosage of ATT and SEP less than 1% that results in the greatest improvement in mechanical strength. The improvement seen from ATT and SEP dosages that is not seen in BEN samples is possibly due to the similarly fibrous shape of ATT and SEP particles, while BEN is platelike in comparison. The smaller dimensions along one axis may allow these particles to better disperse between the larger CEM and GGBS particles, leading to a denser structure and improved strength.

3.4. Rheological Tests

3.4.1. Shear Thickening

Figure 12 below depicts the stress response of each sample in flow curve 2. Figure 12a depicts the stress–shear rate relationship of the stiffest mixes, REF, CLAY 0%, SEP 0.5%, and SEP 1%, while Figure 12b depicts the same relationship for BEN 0.5%, BEN 1%, ATT 0.5%, ATT 1%, and GGBS. CLAY 0% is present in both sub-figures for comparison. It can be seen that in all mixes except REF, shear thickening behavior is present. Shear thickening describes the phenomenon of materials that exhibit a viscosity that increases with the shearing rate. This is presented in flow curve diagrams as a positive non-linear relationship between shear rate and shear stress. REF does not display shear thickening but exhibits minor shear thinning behavior, which describes a negative non-linear relationship between shear rate and shear stress. Shear thickening behavior has been observed in ultra-high-performance concretes (UHPC) and self-compacting concretes (SCC) [41,42]. UHPC mixes typically feature high SCM content in order to maintain a low w*/c ratio and high SP dosages to maintain sufficient workability [43]. Moorthi et al. [44] have observed shear thickening behavior in cement pastes with high GGBS/CEM ratios, low w/b ratios, and relatively low PCE-based SP content (0.3% by weight of binder). Cyr et al. [45] have also shown that shear thickening behavior increases with SP content.

While several theories describe the origin of shear thickening in suspensions, Shen et al. [41] have argued that in cement-based materials, it is the result of the formation of clusters of particles that cause jamming in the suspension. Shen et al. [41] explained that at high shear rates, particles in suspensions can overcome the repulsive forces between each other and the aggregate. Cyr et al. [45] have used the Herschel–Buckley model, given in Equation (5) below, to model shear thickening and shear thinning cement-based materials, which has similarly been adopted in this paper.

$$\tau ={\tau }_0+k{\dot{\gamma }}^n$$

(5)

This model resembles that of Equation (1) but now includes an exponent ‘n’ with the shear rate. n models a Bingham plastic, as described earlier, when equal to 1. n models a shear thickening material when greater than 1, and a shear thinning material when less than 1. ‘k’ is a constant of proportionality and relates to a materials’ consistency. The degree of shear thickening that each material exhibits is compared through the magnitude of n, with a material of higher n showing greater shear thickening. A non-linear regression was used to fit the parameters n and k to the data. The static yield stress τ₀ is typically measured as the initial peak stress, but as results typically did not feature an initial peak, τ₀ for each mix was taken to be equal to the second stress measurement recorded by the rheometer, at which point the shear rate was equal to 1.7 (1/s). The second value was chosen as, in all cases, the first stress measurement was at a shear rate too close to 0 to provide a meaningful result. It is possible that a more accurate yield stress could be determined if the resolution of measurements were to be increased.

Table 7. Static yield stress, consistency coefficient k, and thickening/thinning exponent n for each mix tabulated for the up phase of each flow curve.

Mix Label	Up Curve 1			Up Curve 2			Up Curve 3
	τ0 (Pa)	k (Pa s)	n	τ0 (Pa)	k (Pa s)	n	τ0 (Pa)	k (Pa s)	n
REF	266.22	48.84	0.90	345	2.33	1.37	124.66	0.25	1.76
CLAY 0%	6.45	0.002	2.88	18.26	0.001	2.98	4.05	#	#
BEN 0.5%	11.35	0.002 ‡	2.92 ‡	18.08	0.002	2.76	4.79	#	#
BEN 1%	15.37	0.018 ‡	2.5 ‡	11.52	0.001	2.85	3.31	#	#
ATT 0.5%	14	0.012	2.59	21.66	0.005	2.62	3.53	#	#
ATT 1%	20.75	0.045	2.25	8.96	0.004	2.61	2.96	#	#
SEP 0.5%	39.56	0.284	2.07	24.66	0.032	2.29	3.29	#	#
SEP 1%	102.62	3.884 ‡	1.68 ‡	66.32	0.165	2.11	18.95	0.086	2.17

: The sample reached the torque limit during testing. Stress was not allowed to exceed 2814 Pa at the torque limit, and the flow curve could not be mapped for the highest shear rates. k and n for these flow curves were calculated using only stress–shear rate measurements below the torque limit. ‡: The sample showed deviation from the model; parameters k and n were calculated using only stress–shear rate measurements below the deviation. #: Sample could not be fit by Equation (5).

below gives the values of τ₀, k, and n calculated for each mix. The difference in behavior between REF and other samples is immediately apparent. REF’s flow is characterized by high yield stress, consistency, and both minor shear thinning and thickening during different flow curves.

In several mixes (BEN 0.5%, BEN 1%, and SEP 1%), the stress–shear rate plot features a point during the ‘up’ curve at which the viscosity and thus stress decreases rapidly for a few measurements, followed by a resumption of the stress–shear rate relationship seen before the sudden change. It is suspected that this is due to the sudden separation of an agglomeration of sand particles along the sides of the BMC, partially blocking the gap between the stirrer and the BMC. These agglomerations would produce an exaggerated apparent viscosity and, upon dissolution, would induce a rapid descent in viscosity, as seen in the ‘up’ curves shown below in Figure 13. This theory is supported by the fact that this behavior is only seen in the first flow curve, as these clusters of sand particles likely would form as the BMC was filled and not persist further into the shearing protocol.

Flow curve 2 was used to illustrate the mixing behavior in Figure 12 as this set of measurements best illustrates the shear thickening behavior within the material. The non-representative response of flow curve 1 has been described, but a different response was also observed in flow curve 3. Figure 14 below illustrates the behavior displayed by BEN 0.5%, BEN 1%, ATT 0.5%, ATT 1%, SEP 0.5%, and CLAY 0% in flow curve 3. During this flow curve for shear rates below approximately 50 (1/s), these mixes exhibit an approximately linear stress–shear rate relationship with an apparent viscosity significantly greater than that displayed during their flow curve 2. This apparent viscosity diminishes gradually once the shear rate passes 50 (1/s), after which point the mixes return to exhibiting shear thickening. This enhanced viscosity is likely due to the early influence of hydration. Flow curve 3 begins 17 min after the rheological test begins, and as the structuration of the mix has been greatly diminished by the end of flow curve 2, the early hydration products would dominate the behavior of these weaker mixes. These mixes, however, all developed smooth curves during the ‘down’ phase.

While the stress–shear rate relationship did not always present a smooth stress–shear rate curve for numerous flow curves, all down curves formed smooth curves. Smooth curves are also formed for samples that reached the rheometer torque limit if the measurements at the max torque are excluded. As a result, Herschel-Bulkley parameters could be calculated for every flow curve, which are given below in

Table 8. Dynamic yield stress, consistency coefficient k, and thickening/thinning exponent n for each mix tabulated for the down phase of each flow curve.

Mix Label	Down Curve 1			Down Curve 2			Down Curve 3
	τD (Pa)	k (Pa s)	n	τD (Pa)	k (Pa s)	n	τD (Pa)	k (Pa s)	n
REF	131.78	3.87	1.41	48.58	0.828	1.63	13.86	0.304	1.74
CLAY 0%	1.24	0.0002	3.35	0.82	0.00004	3.69	0.62	0.0007	2.89
BEN 0.5%	2.03	0.0134	2.45	0.71	0.00009	3.46	0.59	0.0005	2.96
BEN 1%	2.17	0.0025	2.89	0.76	0.00005	3.59	0.68	0.0002	3.29
ATT 0.5%	2.45	0.0008	3.17	0.88	0.0001	3.4	0.74	0.0008	2.89
ATT 1%	1.61	0.0010	3.06	0.94	0.0002	3.28	0.84	0.0002	3.3
SEP 0.5%	4.77	0.0141	2.72	1.25	0.0019	2.91	1.06	0.0005	3.14
SEP 1%	8.85	0.0728	2.52	2.41	0.0078	2.78	1.95	0.0041	2.83

. It is for this reason that the down curves are used to compare the linear viscosity and shear thickening behavior of each mix. For these results, the static yield stress is replaced with the dynamic yield stress, ‘τ_D’. By the same reasoning, as explained for the static yield stress, τ_D is taken to be the second last stress measured by the rheometer.

Figure 15 below illustrates the change in factors n and k during successive flow curves. Two behaviors can be observed among the mixes, which have been separated accordingly. Mixes REF, SEP 0.5%, SEP 1%, and ATT 0.5% feature a consistency and thickening exponent, which decrease and increase respectively with each consecutive flow curve. This means these mixes become less viscous and thicken more as testing proceeds. This suggests a weakening of the material with successive shear cycles. The development of shear thickening can be seen in REF where it was initially not present. This indicates that the observed thickening may result from the w/b ratio and SP content. The consistency of CLAY 0%, BEN 0.5%, BEN 1%, and ATT 0.5% decreases during the second flow curve before increasing again during the third flow curve. The opposite is true of the thickening exponent of these mixes. These mixes become less viscous in flow curve 2, in the same manner as the mixes described previously. The increase in viscosity among these mixes in flow curve 3 has been discussed earlier.

3.4.2. Behavior Development between Flow Curves

As k and n both rise and fall between each proceeding flow curve of a mix, it is not obvious if a material is becoming stiffer or weaker. Equation (6) below is the integral, which describes a relative measurement of the mechanical energy per unit volume required to proceed through one phase of a flow curve protocol. A full derivation can be found in Appendix A. This gives a single comparable figure that describes the total work required to carry out the same shearing on each sample and thus describes the mix’s stiffness, as stiffer mixes will require more energy. This was calculated for the up and down phases of each flow curve using the trapezoidal rule and the recorded stress and shear rate data from each measurement. The sum of the calculated work for each phase of the flow curve is plotted below in Figure 16 and illustrates how, for all samples, the total work each mix required changed for each repetition of the shear protocol.

$$W_V={\int }_0^{60}\tau \left(t\right)\dot{\dot{\gamma }\left(t\right)dt}$$

(6)

It can be seen that all mixes become weaker with each successive flow curve, except for ATT 1%, which stays approximately equal between flow curves 1 and 2. This is contrary to the expected behavior, which is that pre-shearing would return a sample to approximately the same state and that, as the samples were allowed to rest for longer periods, they would become flocculated and develop greater static yield stresses. It is expected that this strength growth would be reversible, and upon flow initiating, the material would return to a relatively similar flow curve, leading to a greater energy required with each successive flow curve. The behavior measured indicates that the material is not reaching an equilibrium stress during pre-shear, and the internal structure in each mix is breaking down progressively with each shear cycle. This may be a result of sand segregating from the mix, which was observed in all mixes. Sand settlement within the BMC is a consequence of the low viscosity of the mixes, as settlement is the result of the gravitational force of the particles overcoming the viscous resistance. The settlement of sand naturally leads to a paste with a lower volume fraction and, thus, a lower viscosity.

3.4.3. Thixotropy

In flow curve measurements, a material’s thixotropy is measured through the area captured between the up and down curves of a stress–shear rate plot [20]. As a result of some mixes reaching the rheometer’s torque limit during testing, the shape and area of these curves were distorted. The work integral in Equation (6) was instead used to compare thixotropy. The difference in work required between the up phase and down phase of a flow curve was calculated for each curve of each mix. Figure 17 below illustrates the change in this result per consecutive flow curve. It can be seen that for all samples, the thixotropy measured in flow curve 2 is lower than in flow curve 1. The difference in work between the up and down curve is a measure of the de-structuration of the material during shearing. The decrease in work from Flow Curves 1 to 2 indicates that when flow curve 2 begins, the sample is more disturbed than it was before the start of flow curve 1, meaning the pre-shear phase was not able to return the samples to the same level of structuration. This trend continued in flow curve 3 for both mixes containing SEP and for REF, while other mixes noted a rise in thixotropy during this flow curve. As discussed earlier, during this flow curve for shear rates below approximately 50 (1/s), these mixes exhibit an approximately linear stress–shear rate relationship with an apparent viscosity significantly greater than that displayed during their flow curve 2, which diminishes beyond a shear rate of 50 (1/s). This period of greater viscosity results in a greater thixotropy measurement, as once the mixes return to their shear thickening behavior, much lower stresses are generated at low shear rates.

3.4.4. Static and Dynamic Yield Stress

Figure 18 below illustrates the relationship between clay dosage and static yield stress during the first flow curve’s up phase. It can be seen that mixes containing BEN and ATT have a minor positive influence on τ₀, while 1% SEP increases it by a factor greater than 5. This result is in line with the slump-flow results observed, and the origin of this effect can be explained similarly. As per Table 8 above, the dynamic yield stress of every mix except for REF was found to be less than 10 Pa, which means these samples had negligible resistance to flow upon the end of shearing.

Figure 19 below illustrates the change in each mix’s static yield stress during each of the three flow curves. An almost 30% increase in τ₀ can be seen in REF between flow curves 1 and 2. Mixes, CLAY 0%, BEN 0.5%, and ATT 0.5% also saw increases in static yield stress by factors of 2.8, 1.6, and 1.5, respectively, but these all constituted lesser increases in strength than REF displayed, given how minor the flow resistance of these mixes was during the first flow curve. By flow curve 3, all mixes are exhibiting a lower τ₀ than they had earlier in the shearing despite the longer period of rest between flow curves 2 and 3.

As the rest periods following pre-shearing are all below 5 min, the rate of yield stress increase between flow curves was due to re-flocculation, and thus Equation (3) was used to calculate R_thix. R_thix was calculated as the difference between the τ₀ measured for a flow curve n and the τ_D measured for flow curve n-1 divided by the duration of the rest period between each flow curve. The results are provided below in

Table 9. Rate of re-flocculation for each mix between flow curves 1 and 2 and curves 2 and 3.

Rest Period	Rthix (Pa/s)
	CLAY 0%	BEN 0.5%	BEN 1%	ATT 0.5%	ATT 1%	SEP 0.5%	SEP 1%	REF
1 → 2	0.284	0.267	0.156	0.320	0.123	0.332	0.958	3.554
2 → 3	0.011	0.014	0.009	0.009	0.007	0.007	0.055	0.254

. The results for dosages of 0.5% are in line with those of Varela et al., who similarly find that SEP results in the greatest increase in the rate of growth in the static yield stress, followed by ATT and BEN [28].

Table 10. Rheological and hardened properties of CLAY 0% mix, and the relative increase in the same properties measured in other mixes.

Mix	τ0 (Pa)	τD (Pa)	k (Pa s)	n	Rthix (Pa/s)	Flowability (mm)	Compressive Strength (MPa)
CLAY 0%	6.45	1.242	0.002	2.88	0.284	358	100.92
REF	+259.77	+130.5	+48.838	−69%	+1151%	−60%	−14%
SEP	+33.11	+3.526	+0.282	−28%	+17%	−35%	+25%
ATT	+7.55	+1.208	+0.01	−10%	+13%	−13%	+34%
BEN	+4.90	+0.787	+0	+1%	−6%	−8%	0%

below summarises the change induced by adding 0.5% clay in each mix on both rheological properties and hardened strength, compared to CLAY 0%

4. Discussion

4.1. Limitations

Measurements of rheological properties are built upon the assumption that the test subject is experiencing laminar flow. This assumption is typically valid when the material is contained between two parallel surfaces in close proximity (typically < 1 mm gap) and subjected to a shearing [46]. This assumption is not valid for the geometric arrangement used in these rheology tests as the gap between the stirrer is 5 mm, and as such, turbulence is generated within the sample. As such, the results from these measurements are all relative. In conducting trials, the BMC was initially filled to a volume sufficient to submerge the vanes of the stirrer. It was seen that in all samples, the maximum allowable torque of the rheometer was not high enough to generate the shear rates programmed. The BMC was filled to approximately 50 mm down from its top surface to overcome this limitation. This meant the top portion of the stirrer would be exposed. As a consequence, additional turbulence on the sample’s surface was generated during shearing. Despite efforts to maintain a torque below the torque limit, it was still reached during the ‘Flow Curve 1’ phase of the REF mix and during numerous pre-shear phases, among other mixes. It has been seen that temperature significantly impacts the rheology of cement-based materials [47]. While the Peltier plate of the rheometer is programmed to maintain a constant temperature during testing, the large volume of the sample being tested meant it would be impossible to maintain a constant temperature within the BMC. The exothermic nature of cement hydration also furthers the inconsistency of temperature in the sample over the course of testing. The impact of this effect can be minimized through the use of a parallel plate testing apparatus due to the small volume of material required and its position relative to the temperature plate.

4.2. Expected 3D Printing Performance

4.2.1. Introduction

To compare the effectiveness of each clay’s ability to improve mix applicability for 3D printing, the expected performance of CLAY 0% is assessed under the following criteria: extrudability, shape retention, buildability, and mechanical performance. The difference in the influence of a 1% clay dosage rather than 0.5% was typically one of magnitude rather than effect. This, in combination with the relative nature of the measurement system, makes comparison of dosage levels difficult, and as such, the discussion identifies which clays performed best rather than an ideal dosage.

4.2.2. Extrudability

The shear thickening behavior exhibited by CLAY 0% poses issues for extrusion. The high stresses and shearing present during pumping would generate thickness within the sample. It has been seen that this thickening effect is reversible and dissipates quickly. This, in combination with the unpredictable peaks and drops in viscosity, as seen in Figure 13, suggests that this mix would generate localized and brief blockages in regions of high shear during pumping. It can be expected that such a mix could not be successfully extruded as a continuous filament but would spit out of a print head. The segregation of sand particles noted during rheological tests would also add to the non-uniformity of flow during pumping.

It can be seen from Table 5 that clay addition introduced an increase in linear-viscous behavior rather than thickening behavior during the first shear cycle. This is most relevant given the short period of pumping required during printing. While mixes containing clays still displayed shear thickening, in all mixes except BEN 0.5%, it was less than that of CLAY 0%. SEP mixes were able to reduce the exponent n more than ATT or BEN and were able to improve consistency k more than ATT or BEN. This suggests SEP will form a more uniform plug during pumping, resulting in less filament tearing when extruded. Despite this, SEP 1% still reached the torque limit of the rheometer at the lowest shear rate, and the exact dosage of clay may have to be varied in order to suit pump conditions. Segregation of sand particles was observed in all mixes, but in the context of fully developed plug flow, this means the migration of sand toward the plug from the lubrication layer would increase, leading to a stronger filament.

4.2.3. Shape Retention

It can be seen from the slump-flow results that CLAY 0% showed the greatest flow diameter and, thus, the worst shape retention. As discussed earlier, SEP was the most effective clay in improving this, but SEP 1%’s improvement was 20 mm less than that of REF. The 20 mm difference in flow radius between REF and SEP 1% represents a greater improvement in shape retention than that of the 50 mm difference between SEP 0.5% and SEP 1% due to the quadratic relationship between the height and radius of a slump. This is reflected in the static yield stress results seen in Table 5. The shear thickening behavior observed also negatively influences the shape retention of the mix. The thickening of the mix would be most extreme along the sides of the pipe where the stress is highest. Increasing the viscosity in this region would damage the mix’s ability to form a lubrication layer and, in turn, disrupt the development of plug flow. Shear thickening could be reduced by reducing the SP content and maintaining flowability by increasing the water content to compensate. Given the excessive compressive strength achieved, the water content can safely be increased.

4.2.4. Buildability

Buildability is assessed through a mix’s static yield stress and its rate of increase, as seen in Equation (2). The greatest increase in R_thix relative to CLAY 0% due to clay was in SEP 1%, which increased R_thix by a factor of 3.4. However, this, and by extension the influence of the other clays, is minor compared to the increase by a factor of 12.5 measured in REF. A low rate of increase in yield stress is acceptable, given the initial yield stress is high. REF also achieved the highest yield stress, but among the clay mixes, SEP 1% shows the best improvement. ATT and BEN are seen to be less effective in improving the buildability of the mix than SEP in the low-carbon context. It was expected that SEP would perform better than ATT and BEN, but the inability of ATT or BEN to reproduce the improvements in thixotropy seen by Panda et al. [22] and Chen et al. [48] suggests the significance of the material size. Given that there is minimal research into the use of SEP in concrete, the question of how much buildability can be enhanced through increased SEP dosage still stands.

4.2.5. Mechanical Performance

The influence of clays on compressive strength is of interest because its development in relation to dosage is negative. This means that increasing the dosage of clay to improve the fresh state performance of a mix will have a negative impact on the compressive strength of the hardened element. This is particularly significant as SEP clay produces the greatest improvement in printability but has a relationship with compressive strength that cannot be fully understood from the results collected. It is possible that the negative influence produced by the introduction of SEP into a mix is overcome at higher dosages of SEP as the greater volume of nanoscale particles produces a denser structure. Further research into this relationship is needed. Panda et al. [22] report a 10% increase in 28-day compressive strength due to the addition of 0.5% ATT, while an increase in compressive strength of 28% is observed at this dosage of ATT in this investigation. Chen et al. [48] report an increase in 1-day compressive strength for dosages of up to 3%, at which point a decrease is noted, and a decrease in compressive strength is observed as dosage increases. Similar behavior is observed in this investigation, which shows BEN resulted in no increase in compressive strength at a dosage of 0.5% and a decrease at 1%.

4.2.6. Carbon Emissions

The addition of clays to a concrete mix composition is seen to not have a significant impact on the embodied carbon of the entire mix due to the low mass required to cause a significant change in both fresh and hardened properties. In the case of the 50% CEM GGBS replacement binder adopted in this study, a 46% reduction in embodied carbon is achieved. Further reduction in a mix of this nature would necessitate a further reduction in the CEM content. The effectiveness of further CEM replacement by GGBS relies on the ability of additives like clays to maintain printability in the mix. Alternatively, Li et al. [16] have observed a reduction in embodied carbon of 70% through the development of wholly waste material-based geopolymers. The applicability of clays in geopolymers has previously been attested by Zhong and Zhang [49].

4.2.7. Summary

In the case of all clays, the mix constituent proportions are such that a reduction of 46% in embodied carbon is achieved. Due to the low mass of clay required to induce behavioral change in the concrete, it can be asserted that all mixes succeed in reducing the carbon footprint, and thus, the success of each clay should be judged on their efficacy at improving printability criteria. It can be seen from Table 10 that REF exhibited the greatest capacity for shape retention in its flowability result, and as discussed, it is expected that this mix would also exhibit the greatest extrudability and buildability. It is clear that BEN and ATT fail to improve the extrudability, shape retention, or buildability of a low-carbon 50% GGBS mix when compared to SEP. SEP shows an improvement in all rheological properties of over 1.3 times that of ATT, relative to the performance of CLAY 0%. In the low-carbon mix, at this dosage, ATT, BEN, and SEP all fail to raise the yield stress sufficiently high to retain shape after shearing or after rest, as evidenced by the flowability results. Simultaneously, the compressive strength can be seen to improve considerably at this dosage. ATT and BEN also exhibited a greater R_thix at 0.5% dosage. While the increase in compressive strength diminishes as clay dosage grows, τ₀ and R_thix were both seen to rise. It can be asserted that SEP should be used to improve printability instead of BEN or ATT, and an ideal percent dosage depends on printer variables and the induced decrease in compressive strength. Changes in water or SP content may be more effective in improving the printability of a 50% GGBS mix. Future research should focus on developing AI-based models [50] and a constitutive model [51] for 3D printed concrete to enhance material optimization and construction efficiency.

5. Conclusions

This paper investigates the influence of bentonite, attapulgite, and sepiolite clays on the rheological and mechanical properties of 3D printable low-carbon concrete. Buildability, extrudability, and shape retention are identified as criteria for the success of a 3D printable concrete mix, and the influence of rheological properties on these criteria is discussed. From rheological testing, it was seen that high GGBS and SP content in mixes produced a material that exhibited little resistance to flow under low shear rates but thickened at high shear rates. The carbon factor of each clay is considered equal, and thus, the applicability of each clay to 3D printing was judged by how effectively it could influence the rheology of high GGBS concrete toward a rheology that satisfied the success criteria. The following conclusions are made:

• Slump-flow diameter shows a linear decrease as the dosage of all clays increases, with SEP resulting in a diameter reduction of 2.8 times that of ATT at 0.5% dosage. This decrease is indicative of an improved capacity for shape retention.
• SEP and ATT both have the capacity to improve the compressive strength of low-carbon mixes by a minimum of 25% and 34%, respectively. There exists a clay dosage between 0 and 1% at which this compressive strength increase is greatest. It is possible this behavior is also applied to BEN.
• Shear thickening behavior dominates in 50% GGBS binder mixes and is detrimental to printability. SEP is the most effective of the clays tested in diminishing the extremity of this phenomenon, as evidenced by a reduction in n of 28% at 0.5% dosage. However, a reduction in SP content may be more effective in returning the stress to shear rate relationship to a linear one.
• Shear thickening behavior grows more prominent through successive flow curves due to the continued breakdown of the internal structure and settlement during rest periods. SEP 0.5% shows the greatest capacity to resist this effect, evidenced by the lowest increase in n between flow curves 1 and 2, of only 7%. This is indicative of poor extrudability.
• Static yield stress grows approximately linearly with the % dosage of all clays tested. SEP outperforms ATT and BEN by a factor of 4.4 and 6.8, respectively, at a dosage of 0.5%. This is linked to an improved expected buildability, which is supported by the slump-flow results.
• For all clay types, the mix proportions result in a significant reduction in embodied carbon. Given the minimal clay required to alter the concrete’s behavior, all mixes effectively reduce the carbon footprint. Therefore, the success of each clay should be assessed based on its ability to enhance printability criteria.

In all rheological properties, the addition of 0.5% SEP resulted in a greater improvement than the same dosage of ATT or BEN, and this is more successful in improving printability. The rheological results of SEP mixes are closer to one that is buildable, extrudable, and retains shape, but in neither mix containing SEP could these criteria be satisfied. This work raises the opportunity to further refine mix proportions and improve fresh state buildability while maintaining an SEP dosage of 0.5% and the positive influence on the rheological properties and compressive strength associated with it.