Air Bubbles as an Admixture for Printable Concrete: A Review of the Rheological Effect of Entrained Air

Abstract

This article presents a review of the current solutions for the rheological challenge of three-dimensional concrete printing (3DCP), providing a rheological definition for printability, and an overview of the current techniques for obtaining a printable concrete, placing special emphasis on understanding structural build-up and the current mixture proportions and admixtures used to improve it. A promising alternative for improving structural build-up is the use of entrained air (EA), as bubbles, whose effects are reviewed in generic yield stress fluids and then specifically in concrete. After revision of micromechanical models and experimental trials from literature on yield stress fluid bubble suspensions and concrete, EA appears to be ideal for 3DCP when generated by anionic surfactants, as it increases static yield stress and decreases plastic viscosity. Cationic surfactants, however, maintain or slightly decrease static yield stress. It is proposed that the lubricating or stiffening property of the bubbles determines their ability to deform under the shear stress generated by the surrounding fluid. The ability to deform depends on the surfactant used to entrain the bubbles and the mixture design of the concrete. Further experimental research must be carried out for the advantages of EA to be fully realized.

Introduction

Concrete is the most used material in the world.¹ In the past few decades, there have been many advances in the development of material technologies and construction methodologies to make the use of concrete more environmental friendly and more efficient.² Although productivity in other industrial sectors has grown exponentially in the past decades, global labor productivity growth in construction lags far behind that of the total economy.³ Concrete and its constituent materials are under the global spotlight and scrutiny due to their environmental impact⁴; as a result, many efforts have been directed to the research and development to mitigate impacts, such as the use of supplementary cementitious materials (SCMs), also known as mineral admixtures (e.g., fly ash, slag), and recycled concrete aggregate,⁵ or through supply chain integration.⁶

Among the efforts made to improve the concrete industry, one of the most promising technologies is three-dimensional concrete printing (3DCP), also known as additive manufacturing. 3DCP is an extrusion-based manufacturing process, in which layers of fresh concrete are placed sequentially on top of each other by a printer that is fed by three-dimensional (3D) computer-aided design or other digital models.⁷

3DCP presents a threefold sustainability opportunity, from economic, environmental, and social perspectives.⁸ This process eliminates the need for any type of formwork, reducing the cost of a concrete structure by 35–60%,⁹ and significantly reducing the necessary man labor. The integration of digital design, collaboration, and Building Information Modeling and cost and time reduction of a robotically fabricated concrete have large potential for the increase of productivity.^10,11 3DCP also permits a more precise concrete placement and allows the design of structurally optimal constructions,¹² considerably reducing the amount of concrete needed. In addition, many companies have begun using 3DCP for fast and low-cost construction of social housing with a durable material.^12–14

Although it has been under development for the past 10 years, 3DCP still presents many challenges for a broad implementation in the construction and building industry, requiring substantial investment in research.⁸ Among the challenges for the development and scale-up of 3DCP are structural resistance,^7,12,14 durability,^7,12 and rheology of the printable material.^7,12,15–17

The mixture design for 3DCP must be sufficiently fluid and workable to flow within the piping of the printer, but sufficiently firm to support its own weight and that of subsequent layers once deposited and extruded. These rheological characteristics require a material with structural build-up, that is, that increases its static yield stress (${\tau }_0^s$) with time. Although concrete changes ${\tau }_0^s$ in time by nature, the rate of change is not fast enough to be used without formwork in 3DCP if no help of chemical admixtures and SCMs is provided.¹⁸

Concrete in its fresh state is a non-Newtonian fluid, more specifically, a yield stress fluid.¹⁹ Roussel²⁰ defines these fluids as materials with two natures: They behave like solids when the applied stress is below a critical shear stress value, and like liquids when the applied stress is above this value. This value is called yield stress. Among yield stress fluids, some exhibit thixotropic behavior, which is defined as a reversible, time-dependent viscosity variation.²¹ For concrete, however, it has been commonly defined as the increase of ${\tau }_0^s$ in time.²²

From the moment that cement is in contact with water, reversible and irreversible processes occur at the same time,²³ and it is not simple to separate these two effects. This is why for cement-based materials the term structural build-up (also referred to as A_thix) is more accurate than thixotropy,²⁴ as it considers both reversible and irreversible processes. Structural build-up is the strength increase of fresh cementitious materials due to flocculation, colloidal interactions, and calcium silicate hydrates (C–S–H) nucleation at the contact points between cement grains.^23,25–27 In this study, thixotropy refers to only the early-age, reversible part of structural build-up, which represents an increasing resistance to flow, precisely what is required for 3DCP.

To obtain the required increase in structural build-up, the effects of many concrete constituents and admixtures have been and are being explored. Researchers that have obtained a printable concrete mixture agree on a relatively high cement dosage (i.e., between 400 and 600 kg/m³),^15,28–32 which makes those mixtures unsustainable for their high environmental impact and cost. In addition, these printable concretes use chemical admixtures whose effects depend heavily on the chemical composition of the cement used, which varies importantly among regions, as well as unconventional SCMs unavailable in every region and the temperature of the mixture.^{12,18,28,30,33,34}

An unexplored and promising chemical admixture for increase of structural build-up is proposed in this study, as entrained air (EA) in the form of bubbles. To the best of the authors' knowledge, the rheological effect of bubbles in concrete has not been studied in detail, but mechanical theory and experimental trials suggest that it could help concrete printability, as previous studies have shown that it could increase ${\tau }_0^s$^35–37 and decrease plastic viscosity (μ_p)^36,37 within certain circumstances, thereby promoting the required rheological properties for 3DCP. This counterintuitive twofold effect could be attributed to the deformability of the bubbles under shear stress.

The development of 3DCP depends heavily on the research of chemical admixtures and SCMs that promote the required rheological properties. In this article, first, the material requirements for a printable concrete are reviewed, along with a rheological definition of printability and structural build-up. Then, gathered data are organized to generate information on how to obtain a printable mixture, describing the existing measurement methods for printability. Also, the study goes into detail over the components of a printable mixture, specifically chemical admixtures and SCMs (also referred to as mineral admixtures).

The second half of this study proposes EA as a potential admixture for structural build-up in 3DCP, based on what exists in literature, starting with a review of the effect of bubbles in yield stress fluids, followed by an analysis of the rheological effect of EA in concrete. Finally, a description is presented of the characteristics and effects of available air-entraining agents. This is used to further explain the apparently counterintuitive twofold effect of this admixture.

Material Requirements for 3DCP

Definition of printability

Printability in the context of cementitious materials was first defined by Lim et al.¹⁶ as the ease and consistency of depositing material through a deposition device. This, in turn, means the material would require: adequate buildability, defined as the resistance of the deposited fresh material to deformation under load; pumpability, defined as the ease and reliability with which the material is moved through the delivery system; and open time, the period within which these properties are consistent within acceptable tolerances. Le et al.¹⁵ later introduced the term extrudability, which would constitute the capacity of the material to pass through the small pipes of a printer and the nozzle at the printing head.

Most researchers have synthesized these properties into one definition of printability, making it the ability for a fresh material to be conveyed, extruded into layers with acceptable quality that is able to maintain themselves and the weight of layers subsequently deposited without deformation.^27,31,38,39

More specifically, buildability has been referred to as the ability to withhold geometry under pressure or retain its shape post-extrusion,^16,29,40,41 but has also been considered as a determinant for vertical printing rate or a limit to number of layers or height.^15,27,42 In this study, the former definition will be considered as shape stability (self-weight resistance), and the latter will be referred to as building rate (subsequent layer weight resistance). Pumpability is given by the flowability of the mixture, or the ease with which it flows, with emphasis in the requirement for resistance to segregation and bleeding. Extrudability, in turn, is considered to be the ability for the material to be extruded into continuous, tear-free filaments.^7,39,41

In addition to these fresh state requirements, an adequate printable material must also have the sufficient open time for the operation to be finished, the sufficient compressive strength of a cementitious structural material, and the required interlayer bonding for the printed structure to withstand the shear stress of a conventional concrete structure.

All these properties are deeply correlated; one is difficult to attain without modifying the other, which outlines the main difficulty of designing the mixture proportions and constituents of a printable concrete. Pumpability and extrudability are interrelated, whereas pumpability and buildability are opposed.²⁷ On the other hand, maintaining a consistent level of pumpability, extrudability, and interlayer adhesion requires a long open time, which could result detrimental to the buildability.¹⁵

According to Roussel,⁴³ the required combination of rheological aspects for a printable, buildable mixture will result in a very narrow printable window, raising serious problems related to quality control and robustness. For this reason, these novel concepts must be translated into existing, well-studied rheological properties, so that printability can be more easily assessed and obtained.

Relevant rheological concepts

Before relating the printability requirements with existing rheological concepts, these must be listed and defined for a successful printing process:

• Static yield stress (${\tau }_0^s$): This property is defined as the stress required for initiating flow,⁴⁴ and it will contribute to the buildability of the mixture. The increase of ${\tau }_0^s$ is related to the structural build-up for 3DCP, and it is directly responsible for the resistance to deformation under the weight of printed layers.^27,45,46

• Dynamic yield stress (${\tau }_0^d$): ${\tau }_0^d$ is the minimum stress required for maintaining flow, and it is often lower than ${\tau }_0^s$.⁴⁴ For a better pumpability, the material should have a low ${\tau }_0^d$, as this allows higher flow rates without an increase in pressure.⁴² However, the material requires a high ${\tau }_0^s$ once extruded. The discrepancy between both types of yield stress is what researchers call thixotropy,^45,47 and a high level of thixotropy is required for 3DCP.

• Young modulus (E₀): Also referred to as elastic modulus or stiffness, it is the ratio of principal stress in one direction to the strain in the same direction,⁴⁸ and it contributes directly to compression strength, in the hardened but also fresh state.^49,50 A minimum E₀ of the fresh material contributes to a resistance to deformation of each layer,^50,51 which can be a more indicative property than ${\tau }_0^s$ in the bottom layers, due to the stiff consistency of the material in later ages. More importantly, E₀ becomes a relevant parameter when monitoring stability against elastic buckling of a printed structure.^51,52

• Plastic viscosity (μ_p): μ_p describes a fluid's resistance to flow; a printable mixture must have a low μ_p to guarantee an adequate pumpability,⁴² but a minimum μ_p is required to prevent segregation of the components of the mixture, thereby assuring a good extrudability (smooth filaments), and avoiding blockage while pumping.⁵³

Although all the previous properties are required for a successful, printable concrete, they are not needed simultaneously, which somewhat simplifies the design, given the time-dependent properties of concrete. Marchon et al.¹⁸ have defined the different stages of printing and their respective rheological characteristics, which are related to printability properties in Figure 1.

FIG. 1.

Rheological requirements for each stage of three-dimensional concrete printing (adapted from Marchon et al.¹⁸) and associated with chronological printable properties.

During pumping and extrusion, adequate fluidity is required (i.e., adequate ${\tau }_0^d$ and μ_p) to allow pumping within the pipes without risk of blockage, maintaining the mixture stability and avoiding segregation, but also maintaining its shape during extrusion.^18,54

During deposition, the mixture requires shape stability, meaning a rapid structural build-up to sustain its own weight, meaning an induced peak in ${\tau }_0^s$ to prevent the filament from flowing or losing shape once deposited on a flat surface or a previous filament.^18,42

After this stage, green strength is necessary, which is the “fresh resistance” of the concrete, to augment the building rate, such that the initial layers can withstand the weight of the upper layers. As well as ${\tau }_0^s$, one would expect to see an increase in elastic modulus and a slow transition from a viscoelastic to elastic behavior and eventually brittle failure mode.

This transition would give way to the final phase of rapid strength gain, meaning an accelerated hardening of the concrete, such that the layers obtain sufficient structural capacity. Between this phase and the previous, the open time window must be engineered such that there is sufficient resistance and therefore hydration of the material so as to gain strength, but also retain the fresh state so as to guarantee an adequate adherence between layers and avoid cold-joints.^18,55

A review of structural build-up and its mechanisms

The importance of structural build-up has been stated by several researchers in 3DCP. There are two singularities of cementitious properties that make this behavior possible, which are colloidal interactions and C–S–H nucleation at contact points between cement grains.^22,55 This ultimately means that at rest or for low shear rates, the shear stress to initiate or sustain flow increases because of the structural build-up of a stress-transferring structure, whereas it breaks down as flow rates increase.⁵⁶

Structural build-up can, therefore, be separated into two types, depending on the mechanism that triggers it. These assumptions are based on the clarifications on thixotropy and structural build-up of cement pastes made by researchers.^22,56 In Figure 2, the mechanisms are illustrated as an increase in ${\tau }_0^s$:

FIG. 2.

Early- and later-age structural build-up described in time scale. On the x-axis, time, and on the y-axis, ${\tau }_0^s$. The solid line represents a normal concrete material, and the discontinuous line represents the required structural build-up for a printable material.

• Early age structural build-up: This is obtained with thixotropic or flocculating agents, without strong bonds between the particles, during the first tenths of seconds.^22,57 This is what will contribute to short-term buildability, or shape stability, right after extrusion.

• Later age structural build-up: This is obtained by accelerating the hydration (i.e., chemical reactions) of the cementitious particles in the mixture, creating strong C–S–H bonds that cannot be easily undone. This is during the first tenths of minutes,^22,57 and it will contribute to long-term buildability and permit a faster building rate.

The main mechanisms responsible for early age structural build-up are immediate colloidal interactions, which can be thixotropic, and therefore reversible by submitting it to shear via mixing or pumping,^22,43,56 or irreversible, which is the case of early C–S–H bridges.⁴³ These effects are shown within the first tenths of seconds after mixing, up to hundreds of seconds.^22,55

Flocculation is given by the number of contacts between particles and the distance separating them, making the solid volume fraction a determinant, and by the Van der Walls attraction and double-layer repulsion energies.⁵⁸ Roussel et al.²² showed that flocculation was the combination between the attractive colloidal forces between cement particles and the viscous dissipation in the interstitial water between the cement grains, generating a structure of interacting particles that are able to withstand stress. This can be controlled by varying the water–cement ratio, or by adding admixtures that are adsorbed onto the cement particles, creating steric hindrance.⁵⁶

At the same time, another effect takes place, which occurs at the pseudo-contact points between cement grains, where nucleation of hydrates begins, without exiting the dormant period. These bonds are stronger than the colloidal bonds, forming solid hydrate bridges that further give resistance to the particle network.⁴³

Later-age structural build-up is due to two main mechanisms: the acceleration of the main binder hydration, and the precipitation of a secondary phase of additional hydrates.⁵⁹ The former is a product of the nucleation and growth of C–S–H, which generates C–S–H bridges, at the pseudo-contact points. This is irreversible and is given by a time-dependent surface growth.⁵⁶ Roussel⁴³ refers to this phase as structuration, and it is defined as a non-reversible hydrate bond between particles. It is attributed to an increase in the strength and number of hydrate bridges. The latter is a faster formation of a small amount of additional hydration products, obtained by precipitating aluminates and/or silicates, namely ettringite, much like shotcrete accelerators, which can be more practical for 3DCP, where a limited but rapid initial strength gain is required.⁵⁹

Wangler et al.⁶⁰ state that to reach structural scale heights in a continuous printing process, 3DCP requires an active hydration control, as early-age structural build-up is not enough for the stability required for heights of more than 1 m.

It is important to remark that although hydration reactions are irreversible, the process of C–S–H bridge formation could be reversed to some extent, as some bonds are weak enough to be broken by shear and may reappear at rest if the reaction is still ongoing.^22,43

Although these mechanisms are very different, many authors sustain that both are necessary for the required structural build-up. Jarny et al.⁶¹ sustains that at early ages, thixotropy plays a leading role in structural build-up, whereas at later ages, hydration is the more required effect. Researchers^25,27 also sustain that structural build-up is obtained by both physical and chemical effects, and that the combination of both mechanisms is the most effective.

Structural build-up in the past has been quantified by measuring the growth ${\tau }_0^s$ in time, thereby obtaining A_thix with the following equation⁶²:

$${\tau }_0^s\left(t\right)={\tau }_{0,0}^s+A_{thix}t.$$ (1)

This method has been used by various researchers, but it would appear to be insufficient as it does not measure both aspects and ages of structural build-up. As the term suggests, A_thix considers only flocculation rate and nucleation of cement grains,⁶² which is descriptive for early age structural build-up, but for later age, solid volume increase due to hydration is also important. The equation describes a linear increase in ${\tau }_0^s$, which is the case for ages between 10 and 60 min after mixing, but research shows that after 1 h, the evolution of ${\tau }_0^s$ becomes exponential or power law.^59,63 Many researchers have begun combining yield stress measurements with tests more related to setting time, such as penetration tests and calorimetry. This will be further discussed in the following section.

Obtaining a Printable Mixture

Measuring printability

There is not a measurement method universally agreed on for quantifying printability, buildability, or structural build-up. As was explained, cement-based materials change their properties in time; therefore, it is ideal to have a constant input of the material's property at every moment, from the moment the water is added to the mixture, to the moment where the printing is finished.

Many researchers^{27,39,41,51,64,65} have taken it upon themselves to evaluate and define the optimal testing methods for each printability property. The most used methods are qualitative and depend highly on the criteria of the researcher. Nerella et al.⁶⁵ listed the ideal attributes for an extrudability test, specifying that it must be quantitative, inline, continuous and with potential automation. Considering that existing printers do not have the technology to give this information, researchers must search for and evaluate the most adequate testing methods for 3D printing, considering mainly three test characteristics:

• Continuous: As the material evolves continuously the best measurement would be as such, and not discrete like most rheological tests, enabling an integral monitoring of the material's rheological development.⁶⁶

• Non-destructive or locally destructive: Many rheological tests destroy the existing bonds in the material and, therefore, do not represent that actual state of the material in the printing cycle. It is also impractical to use a different sample for each measurement. A non-destructive or only locally destructive test allows efficient, frequent, and digital measurements.⁴⁹

• Online and inline: The measurement will ideally be in real time and fed directly to the printer's computer, permitting the necessary modifications on the printing or mixing parameters.^49,65

Researchers have employed conventional and more innovative test methods to characterize the different aspects of printability.

Some popular methods among researchers are the slump, slump flow, and flow table tests for characterizing pumpability and buildability. Slump and slump flow are easy test methods to implement, as they are widely used and are easily accessible.^15,41,67 Tay et al.⁶⁴ used slump and slump flow for locating an initial “printability region,” which they made more accurate in a second phase with more specific laboratory pumping and printing tests. Flow table tests were also used^38,39,68 as a buildability measuring method, recreating “self-weight” stresses with the drops of the table. Although these tests are easily reproducible, they provide no specific value of static or ${\tau }_0^d$ and depend highly on the performer of the tests; however, they are a good starting point for exploring the effect of certain additions or admixtures when designing a printable concrete.

Empirical tests are practical, whereas rheological tests provide more information on parameters such as static and dynamic yield stress and plastic viscosity.⁶⁹ These are tests that require a rheometer, and the more widely used geometry for printable concretes are vane in cup or concentric cylinders. A wide array of publications include studies with rheometers.^{15,26,27,29,41,53,66–68,70} These tests are used with different routines (hysteresis loops, stress growth) and provide a specific reading of shear stress or torque, which can be used for obtaining ${\tau }_0^s$, ${\tau }_0^d$, and μ_p. These tests are mostly destructive and discrete, but small-amplitude oscillatory shear testing has been proposed for continuous monitoring of structural build-up, via very small, non-destructive excitations.²⁴

Rheometer tests are, however, only possible in a laboratory, are very costly equipment, and become much less dependable as aggregate size grows, due to the broadening of the particle size distribution, though for printable concrete aggregate sizes are no larger than 4 mm, allowing the use of some mortar rheometers. Further, some researchers⁶⁴ sustain that readings vary from one rheometer to the other, and they depend highly on measurement protocols and subsequent data analysis.

Due to the consistency of printable concrete, some researchers believe that fresh-state test methods such as slump or rheometry are not the most adequate, and have developed more suitable tests for measuring “green strength,” such as the squeeze flow method,^29,55,65 or the unconfined uniaxial compression test,^49–51,66 more widely used in geotechnics. These tests provide a measurement of compressive strength before setting, which can be traduced into ${\tau }_0^s$. A test has also been developed for measuring extrudability by using a ram extruder, which obtains a measurement of both extrusion force and friction force acting on the interior walls of the device.^65,71

Some authors have taken the approach of setting time or heat of hydration tests to characterize structural build-up, such as Vicat penetration resistance,^{15,27,39,41,65,70} calorimetry,^26,27,39,70 and ultrasonic wave transmission tests.^41,49,72 These tests, though sometimes offering a continuous, locally destructive measurement, only consider the structural build-up generated by hydration mechanics, and not those due to flocculation (early-age structural build-up).

Some of these tests are said to describe the transition from early- to later-age structural build-up, showing the resistance given by flocculation and then hydrates. This is the case of the penetration test or penetrometry, which is different from the Vicat penetration resistance test, consisting of a penetrating head being inserted at a constant quasi-static speed into a fresh sample.^66,72 This reading in force units can be converted into ${\tau }_0^s$ when modeled as a geo-mechanical soil stability problem, and more importantly is continuous and only locally destructive, with only a single sample required to monitor static yield stress evolution.^59,66

Finally, many researchers^{27,38,64,73,74} have developed personalized pumpability, extrudability, and buildability tests in their laboratories, using custom-made equipment and techniques. Many of these include visual inspection, image data analysis for deformation quantification and pressure reading from pumps.

In Table 1, most methods have been listed and compared depending on their unit and type of measurement.

Table 1.

Measuring Methods and Tests for Printability

Tests	Property measured	Type of measurement
Mini-slump cone test	${\tau }_0^s$	Discrete
		Destructive
		Not online
Hand vane test	${\tau }_0^s$	Discrete
		Destructive
		Not online
Static yield test	${\tau }_0^s$	Discrete
		Destructive
		Not online
Small amplitude oscillatory shear test	Shear modulus or rigidity modulus	Continuous
		Destructive
		Not online
Unconfined uniaxial compression	Young modulus	Discrete
		Destructive
		Not online
Vicat penetration test	Setting time	Discrete
		Locally destructive
		Possibly online
Calorimetry	Heat flow or energy released	Continuous
		Non-destructive
		Possibly online
Penetrometry	${\tau }_0^s$	Continuous
		Locally destructive
		Possibly online
Ultrasonic wave transmission test	Dynamic elastic modulus	Continuous
		Non-destructive
		Possibly online

Successful printable mixtures

Composition of printable mixtures

The type of binder is very important when designing a printable concrete. Cement percentage in the binder can vary between 40% and 100%,^{7,15,31,38,72,75} mostly being Portland Type I. As part of the binder, most researchers^{15,31,38,73,75} include a percentage of SCMs, mainly due to their effects in the rheology of the fresh material: Due to its spherical shape, fly ash increases workability,⁷⁶ but it has been found that class F fly ash can significantly decrease μ_p compared with class C fly ash^42,77; the fine particle size of silica fume has been shown to increase ${\tau }_0^d$ and μ_p^42,77; ground blast furnace has a reducing effect on μ_p, but the competing effects of micro-filling and increased water demand generate variable results in ${\tau }_0^s$^42,77; limestone filler is also widely used in binder proportion to replace cement; and fine limestone filler has been shown to have no considerable effect, whereas coarse limestone filler increases μ_p.⁷⁸

Other researchers have used the benefits of blended cements for their accelerating properties, such as calcium sulfoaluminate cements,^79,80 since it enhances ettringite precipitation and fast surface development, meaning rapid strength gain without the risk of flash setting⁸¹ and calcium aluminate cements that generate the same effect.⁵⁹ In addition, research has been made on the printability of mixtures with 0% cement, such as geopolymers (i.e., a combination of slag and fly ash or pure fly ash paste), successfully printing non-structural building components,⁸² and showing the potential of greatly decreasing the carbon footprint of a printed structure.⁸³

Aggregate, much like in conventional concrete, is a key component for printable concretes. However, a more accurate term would be printable mortars, as aggregates for printable mixtures generally never exceeds 5 mm.^7,38,51 Since 3DCP uses mainly fine aggregate in the mixture, the aggregate fraction comes close to the maximum packing density, having an important effect on rheology, namely μ_p and density.^18,34

Equally as important as the type of binder and aggregate is the amount and ratios of these components. Water-to-binder ratio becomes a fundamental parameter, as well as cement-to-binder and binder-to-aggregate ratio. The commonly ratios used by 3DCP researchers are exposed in Table 2. The combination of different components will pave the way for the possibility to further modify these mixture design parameters, in virtue of optimizing structural performance and cement proportion for a lower carbon footprint.

Table 2.

Printable Concrete Mixtures According to Different Researchers

Researcher(s)	Le et al.15	Sonebi et al.31	Kazemian et al.38	Malaeb et al.85	Rushing et al.30	Bentz et al.33	Weng et al.34	Nerella et al.29	Chen et al.73	Perrot et al.55	Yuan et al.27
Binder proportions (by mass)
Cement	70%	68%	90%	100%	90%	66%	48%	62%	40%	50%	100%
Fly ash	20%	24%			5%		48%	24%
Silica fume	10%	8%	10%		5%		5%	14%
Calcined clay									40%	25%
Limestone						33%			20%	25%
Water-to-binder ratio	0.26	0.4	0.43	0.4	0.43	0.375	0.14	0.26	0.3	0.2	0.35
Binder-to-aggregate ratio	0.66	0.56	0.44	0.66	0.3	—	4.2	0.55	0.66		0.66
Admixtures	SP Ret. Acc. PP	SP VMA PP	HRWRA VMA PP	SP	VMA HRWRA	—	SP	SP	SP VMA	SP	Acc.a Ret. VMA HRWRA

^a This researcher used sulfoaluminate cement and attapulgite clay as a percentage of the binder (10%; 1%, respectively).

SP, superplasticizer; Ret., retarder; Acc., accelerator; PP, polypropylene fiber; HRWRA, high-range water-reducing agent; VMA, viscosity modifying agent.

Chemical admixtures and SCMs have gained importance and innovation for the design of printable concrete. Researchers^18,38,84 have listed admixtures for each stage of the 3D-printing process:

• Flowability during pumping and extrusion: In this stage, superplasticizers (SPs), also known as high- range water reducing agents, are required to obtain pumpability without compromising the compressive strength gain.^15,18 Retarders are also used when in the presence of accelerators to maintain this workability.^15,41,85

• Structural build-up and shape stability after deposition: For this stage, viscosity modifying agents (VMAs) are widely used to thicken the mixture and increase μ_p.^{38,39,56,67,73,82} Nano-particles such as nano-clay^38,39,50 or nano-silica²⁶ have also been proven useful to improve ${\tau }_0^s$ in time, due to their size and the fact that they maintain their structure at rest but break down under shear.^{15,17,31,38,85} Accelerators are also used,^15,18,60 but the mixtures must be carefully designed to trigger setting once the mixture has been deposited. In this stage, some unconventional additions such as polymeric fibers and EA are being evaluated.

• Controlling setting and maintaining open time: Open time is often confused with setting time, but it is the interval beyond which the material is no longer extrudable.⁸³ For controlling open time, retarders such as sugar derivatives are widely combined with accelerators.¹⁵ The pozzolanic properties of some blended cements and SCMs have also been evaluated for this effect, such as slag and fly ash, which decrease yield stress and therefore maintain workability.⁸²

• Rapid strength gain and building rate: The chemical admixtures used for this stage are mainly accelerators, many times in the form of blended cements with calcium sulfoaluminate, calcium aluminate, calcium sulfate,^27,59,70 and also inorganic salts, portlandite, and C–S–H seeds.^18,56

As can be seen in the previous list, some of the used components have more than one desired effect, making it difficult to destine each type of chemical admixture or SCM for one only stage of printing. This makes design more challenging, as with each additional component, there are many simultaneous, competing mechanisms, but it also makes the required rheology obtainable. In Table 2, printable mixtures from different researchers have been collected and compared.

Chemical admixtures and SCMs for structural build-up

One of the more challenging aspects of printable mixture design is the required structural build-up in the different stages of printing, mainly due to the novelty of this requirement and the lack of standardized methods for mixture proportioning and monitoring the required rheology for printing. As such, in this section the admixtures used for obtaining this will be explored and classified.

Reiter et al.⁵⁶ indicate three different types of chemical admixtures for structural build-up: Those that are chemically inert and reversible (CIR), those that are chemically inert and irreversible (CII) (both used for early age structural build-up), and accelerators (used for later-age structural build-up).

CIR are usually VMAs,⁵⁶ which thicken the mixture by increasing the effective solid content through particle bridging. The main effects are often not only modification of viscosity, but of ${\tau }_0^s$ and ${\tau }_0^d$ as well.⁷⁷ VMAs, also known as anti-washout agents, modify rheology by: (i) raising the viscosity of the interstitial fluid, (ii) bridging flocculation between grains, and (iii) creating polymer–polymer interaction via entanglement or association,^86,87 resulting in a threshold shear stress, under which the structure behaves elastically and over which the material flows.

Some types of VMAs are: water-soluble synthetic and natural polymers that increase the viscosity of the mixing water (cellulose ethers)^77,87; organic flocculants that become adsorbed onto cement grains and increase viscosity due to enhanced interparticle attraction (natural gums such as diutan and weilan)^77,87; emulsions of organic materials that while increasing interparticle attraction, also supply super-fine particles⁸⁷; swellable inorganic materials with high surface area that increase water retaining capacity (silica fume, bentonites)⁸⁷; and inorganic materials with high surface area that increase the fine particle content (fly ash, hydrated lime).⁸⁷

CII also relies not only on increasing the effective solid content, but also on the consumption of organic admixtures. Some examples of VMAs with their consumable organic admixtures are swelling clays for mixtures with polycarboxylate SPs, montmorillonite for polymethacrylic acid SPs, and Portlandite for sucrose.⁵⁶

Finally, for later-age structural build-up there are accelerators, which increase the rate of nucleation by increasing the growth rate of C–S–H bridges. There are three different types of accelerators: The first type increases the concentration of ions in the pore solution, and these can be inorganic salts, calcium chloride, or calcium nitrate; the second type acts on the aluminates and leads to fast formation of ettringite, unleashing a very rapid stiffening, providing additional surface for consumption of retarders or formation of hydrates; and the last type are seeds that offer C–S–H surface for this phase to grow faster and strongly change the slope of the acceleration period.⁵⁶

On the other hand, Roussel et al.⁸¹ classified chemical admixtures by their effects and mechanisms into five groups:

• (i)repulsive inter-particles forces that decrease ${\tau }_0^s$, where one can place SPs;
• (ii)attractive inter-particle forces that increase ${\tau }_0^s$, which can be subdivided into three subgroups: bridging forces, generated by VMAs; attractive depletion forces, which are the result of an excess in SP; and hydrophobic forces, generated by surfactants, which can be adsorbed on the surface of cement particles. Hydrophobic surfaces are bound to attract each other, creating flocculation in the suspension and an increase in ${\tau }_0^s$;
• (iii)competitive adsorption between admixtures, which can be taken advantage of depending on the required rheology;
• (iv)the collateral effect of some admixtures that are meant to modify colloidal interaction, but that ultimately modify also the hydration reaction, either delaying or accelerating the setting; and
• (v)the effect obtained by introducing nanoparticles of C-S-H (calcium-silicate-hydrate), ettringite, and nano-clays. These can generate a combination of increase of the nucleation sites, thereby promoting hydration and electrostatic interactions that enhance ${\tau }_0^s$ recovery after shear.

In the context of actual applications of 3DCP, both physical and chemical mechanisms should be used because relying only on flocculation and early C–S–H bonds is not enough; one must count on setting exactly when it is necessary, for which chemical reactions are required.⁶⁰ For this, admixtures that generate both physical and chemical effects are ideal.

Among the less-researched and less-conventional component for rheology modification is EA, added via an air-entraining agent (AEA) or via mixing. This addition could potentially influence early and later age structural build-up. The AEAs are widely used to prevent damage from freeze-thawing cycles,⁸⁸ but from a rheological point of view, its benefits have not been seized to the fullest. The rheological effect of air bubbles in yield stress fluids and specifically in concrete has not been deeply investigated, but most authors agree with the counter-intuitive, twofold character of these additions^77,89: When in a flowing regime, the bubbles tend to deform and act as a lubricating agent, whereas when at rest, the bubbles act as solid inclusions, contributing to the viscosity and “stiffness” of the fluid. Unlike other chemical admixtures, AEA action is not highly affected by the chemical composition of cement.⁹⁰

It is proposed in this article that AEA could potentially have an effect in flocculation, due to the shape of the bubbles and their resistance to deformation under certain circumstances. In the next section the effects of air entrained admixtures in yield stress fluids and concrete will be studied and evaluated for their use in 3DCP.

Effect of EA on Rheological Properties

Air entrained yield stress fluids

To fully understand the behavior of EA bubbles in concrete rheology, the first step is to understand the behavior of bubbles in general yield stress fluids. Yield stress fluids are found in many industrial areas,^91,92 as well as in natural phenomena,⁹³ which makes understanding the rheology of yield stress fluids fundamental for the development in several sectors.⁹⁴ Some examples of these fluids include foams,⁹⁵ toothpastes,⁹² muds, and lava,⁹³ many of which undergo processes during which air bubbles inevitably get entrapped within them.

Although the behavior of suspensions of rigid particles in Newtonian fluids has been studied extensively and can be fairly well described by a Krieger-Dougherty law,^95–97 the behavior of non-rigid particles such as air bubbles, in non-Newtonian fluids, such as yield stress fluids, becomes very difficult to predict.

For the case of bubbles in Newtonian fluids, researchers^96,98 have introduced the viscous capillary number (Ca_μ), which is the ratio between the shear stress of a bubble and the surface tension that maintains their shape, to describe its behavior. Ca_μ is governed directly by the bubbles' deformability and can be calculated with the following equation⁹⁷:

$$C{a}_{\mu }=\frac{{\mu }_p\left(0\right)\dot{\gamma }}{\frac{\sigma }{R}}$$ (2)

where μ_p(0) is the viscosity of the suspending fluid in is the applied shear rate in s⁻¹, $\sigma$ is the surface tension between the gas and fluid in N · m⁻¹, and R is the bubble radius in m. Ca_μ value indicates whether the bubble is very deformable (i.e., Ca_μ >> 1) or not deformable (i.e., Ca_μ << 1). At values close to 1, the transition between both behaviors occurs.^95,96

Llewellin et al.⁹⁹ developed semi-empirical models for monodispersed bubble suspensions in a Newtonian fluid, with bubble size ranging from 1 to 100 μm. They found that at low deformation rates, a rise of the air volume fraction (${\varphi }_{air}$) increases the viscosity. In contrast, at high deformation rates, a rise of ${\varphi }_{air}$ reduces the viscosity.

This effect was experimentally demonstrated by Rust and Manga,⁹⁶ who also studied the effects of bubbles' deformation on the viscosity of suspensions, concluding that an air bubble in a viscous liquid flow tends to deform by the shear stresses generated around the bubble; however, the surface tension stresses tend to maintain its original spherical shape. They discovered that in a simple shear flow,¹⁰⁰ streamlines are deformed and deflected around the bubble's caps, increasing the viscosity; however, when the bubble starts to deform, the deflection of the streamlines decreases and so does the viscosity. In other words, if the bubble maintains its shape, it contributes to increase the viscosity of the fluid, whereas if it deforms, it acts as a lubricating agent decreasing viscosity of the fluid.

For the case of yield stress fluids, bubble deformability is governed more by static yield stress than by plastic viscosity, making the Bingham capillary number ($C{a}_{\tau 0}$)^35,96:

$$C{a}_{\tau 0}=\frac{{\tau }_0^s\left(0\right)}{\frac{\sigma }{R}}$$ (3)

where ${\tau }_0^s\left(0\right)$ is the static yield stress of the suspending fluid in Pa. A very deformable bubble has $C{a}_{\tau 0}\to \infty$ and a non-deformable bubble has $C{a}_{\tau 0}\to 0$.

A more recent experimental study,⁹⁷ carried out with monodispersed bubble suspensions in concentrated emulsions (silicones of different viscosities), proposed that the change in behavior of the bubbles described in previous studies^96,98 was the result of two opposing physical effects that are precisely described by $C{a}_{\tau 0}$. The viscous stress in the fluid tries to elongate the bubbles in the flow whereas the capillary stress tries to minimize the bubble surface, thereby maintaining its spherical shape.

Ducloué et al.⁹⁷ also concluded that the macroscopic response of the suspensions under shear depends on ${\varphi }_{air}$ and the bubbles' stiffness, which is quantified by the $C{a}_{\tau 0}$. At the yield of the suspending fluid, bubbles are generally stiffer than the suspending fluid, so they help to increase ${\tau }_0^s$ of the bubble suspension. If the bubbles become soft and deformable in the bubble suspension, ${\tau }_0^s$ of the bubble suspension decreases.⁹⁷

Models have been proposed by different researchers to predict the behavior of yield stress fluid suspensions. Micro-mechanical computations¹⁰¹ and experiments with bead suspensions in yield stress fluids¹⁰² have shown that relative static yield stress ${\tau }_{rel}^s={\tau }_0^s\left(\varphi \right)∕{\tau }_0^s\left(0\right)$—obtained by comparing the properties of the suspension with those of the suspending fluid—can be a function of relative elastic modulus, $g\left(\varphi \right)=G\left(\varphi \right)∕G\left(0\right)$^{95,97,101,103}:

$${\tau }_{rel}^s=\sqrt[]{\left(1-\varphi \right)g\left(\varphi \right)}$$ (4)

where ϕ (%) corresponds to the volume fraction of the suspended particles/bubbles. When the suspending fluid behaves as a Herschel-Bulkley fluid ,¹⁰⁴ a relationship can also be established for relative plastic viscosity ${\mu }_{rel}={\mu }_p\left(\varphi \right)∕{\mu }_p\left(0\right)$, where n corresponds to the flow index (n = 1 results in a perfect Bingham fluid).^95,97,101

$${\mu }_{rel}=\sqrt[]{{\left(1-\varphi \right)}^{1-n}g{\left(\varphi \right)}^{1+n}}$$ (5)

$g\left(\varphi \right)$ can be modeled differently depending on whether it is a suspension of rigid particles,^{95,101,102,105} of non-deformable bubbles or deformable bubbles.^{35,97,101,103} For the case of rigid particles, $g\left(\varphi \right)$ is well described by the Krieger-Dougherty law [Eq. (6)]¹⁰⁵:

$$g\left(\varphi \right)={\left(1-\frac{\varphi }{{\varphi }_m}\right)}^{-2.5{\varphi }_m}$$ (6)

However, for non-deformable and deformable bubbles, $g\left(\varphi \right)$ is described by the Mori-Tanaka scheme, depending on the value of $C{a}_{\tau 0}$, for a suspension of bubbles with known radius in a non-Newtonian fluid.⁹⁷

$$g\left(\varphi \right)=1-\frac{\varphi \left(4C{a}_{\tau 0}-1\right)}{1+\frac{12}{5}C{a}_{\tau 0}-\frac{2}{5}\varphi \left(1-4C{a}_{\tau 0}\right)}$$ (7)

The $g\left(\varphi \right)$ models [Eqs. (6) and (7)], if combined with Equations (4) and (5), result in models for both ${\tau }_{rel}^s$ and ${\mu }_{rel}$ in function of the volume fraction of the suspension. For the case of rigid particles, the resulting equations are:

$${\tau }_{rel}^s=\sqrt[]{\left(1-\varphi \right){\left(1-\frac{\varphi }{{\varphi }_m}\right)}^{-2.5{\varphi }_m}}$$ (8)

$${\mu }_{rel}=\sqrt[]{{\left(1-\varphi \right)}^{1-n}{\left(1-\frac{\varphi }{{\varphi }_m}\right)}^{-2.5{\varphi }_m\left(1+n\right)}}$$ (9)

where ${\varphi }_{m}eqnoopen\left(\%\right)$ is the maximum packing fraction of the particles (whereas φ is the actual volume percentage of the dispersed phase, and ${\varphi }_m$ is the maximum attainable volume fraction). Equation (8), also known as the Chateau-Ovarlez-Trung model,¹⁰¹ and Equation (9), which becomes the Krieger-Dougherty model¹⁰⁵ when n = 1, compare well with experimental data obtained by Mahaut et al.¹⁰² on yield stress fluid bead suspensions.

For the case of non-deformable bubbles, or Ca →0, the theoretical properties are given by^{35,97,101,103}:

$${\tau }_{rel}^s=\sqrt[]{\left(1-\varphi \right)\frac{5+3\varphi }{5-2\varphi }}$$ (10)

$${\mu }_{rel}=\sqrt[]{{\left(1-\varphi \right)}^{1-n}{\left(\frac{5+3\varphi }{5-2\varphi }\right)}^{1+n}}$$ (11)

However, for the case of deformable bubbles, or Ca → ∞, the equations become^{35,97,101,103}:

$${\tau }_{rel}^s=\sqrt[]{\left(1-\varphi \right)\frac{3-3\varphi }{3+2\varphi }}$$ (12)

$${\mu }_{rel}=\sqrt[]{{\left(1-\varphi \right)}^{1-n}{\left(\frac{3-3\varphi }{3+2\varphi }\right)}^{1+n}}$$ (13)

The curves for both deformable and non-deformable bubbles have also been obtained experimentally^95,97 with bubble suspension in yield stress fluids.

The main conclusions obtained are that the behavior of bubble suspensions in these types of fluids will vary mainly depending on ${\varphi }_{air}$ and $C{a}_{\tau 0}$, meaning the stress that the fluid is subject to, the size of the bubbles, and the rheological properties of the suspending fluid.

Air entrained concrete

A few authors^{36,37,89,97,106–110} have investigated the effect of air entrainment in fresh concrete rheology. The AEAs have traditionally been used to protect concrete from freeze-thaw damage due to the generated bubble network that can accommodate water expansion on freezing.⁸⁸ However, the benefits of AEA in workability of concrete have also been identified,^111,112 and it is precisely this bubble network that might grant the concrete mixture the requirements for digital fabrication.

There is considerable controversy surrounding the rheological effect of AEAs. This may be because there are many different types of AEAs, and many methods of measuring rheological properties. Most researchers who mention the AEAs' effect in rheology usually refer to an “improvement in workability,” in general terms,⁸⁸ defined by the American Concrete Institute as the ease of placement of concrete and can be quantified by the slump cone test.⁷⁶ Tattersall and Banfill¹⁰⁸ sustain that air entraining reduces both ${\tau }_0^s$ and ${\mu }_p$. In contrast, Kurdowski⁹⁰ found an increase in ${\mu }_p$ due to the adsorption of the AEA on cement grains. Edmeades and Hewlett⁸⁹ further explore into this mechanism, describing the formation of “bubble bridges” that are mutually repulsed and therefore stabilized.

Experimental results in literature show that rheological measurements with AEA produce a large variability of the results. The results of Nehdi and Rahman¹¹⁰ showed that a change of the geometry of the rheometer considerably affected the results of each experiment. It is also possible that the result varies depending on whether the agent is anionic or cationic, as was suggested by Feneuil et al.³⁵ Whatever the reason, it is an effect worth studying as the majority of the experiments show an increase in ${\tau }_0^s$^{35,37,110,113} and a decrease in μ_p^{37,107,113,114} showing that EA bubbles might be an ideal admixture for 3DCP, considering the fluidity required within the printer (i.e., under shear), and the structural build-up outside the printer (i.e., at rest).

It is worth noting that when $C{a}_{\tau 0}$ tends toward 0, the bubbles have low deformability, whereas when it tends toward infinite, the bubbles are very deformable. Feys et al.,³⁷ after conducting experiments on air entrained concrete by pumping, concluded that if $C{a}_{\tau 0}$ << 1, the surface tension overcomes the shear stresses, and therefore the bubbles remain spherical, and they move at a different velocity than the surrounding fluid, increasing the flow resistance; however, if $C{a}_{\tau 0}$ >> 1, the shear stresses overcome the surface tension, and therefore the bubbles deform in the flow direction, moving with the flow and decreasing the flow resistance. Therefore, they concluded that the deformability of the bubbles depends on the surface tension between the gas and fluid, the size of the air bubble. and the yield stress of the cement paste or concrete mixture.

Accordingly, the effect that EA has on concrete rheology depends on the ability of the bubbles to deform, which is determined by surface tension, bubble size, and rheology of the surrounding fluid. Previous studies^18,97 have shown that if the bubbles deform, the air entrainment reduces the viscosity and therefore increases the flowability of concrete; however, if the bubbles maintain their shape, they act like solid inclusions, much like normal, round aggregates, and, consequently, increase the μ_p and ${\tau }_0^s$.

Other authors⁷⁷ provide a different hypothesis for the increase in ${\tau }_0^s$: The air bubbles are attracted to the cement particles, forming “bubble bridges” and creating a bond between them. They concluded that without shear, the bubble bridges act like mass particles and increase the yield stress, whereas under shear, the bridges might break and the bubbles act as a lubricant reducing viscosity. These bubble bridges were also described by Kurdowski,⁹⁰ making the hypothesis that the AEA's molecules are adsorbed by the cement grains, forming these bridges, and increasing the consistency of the mixture if they are not broken.

Further, this bubble behavior was also observed by Aïtcin,¹¹¹ and the explanation provided by this author is that the small air bubbles act like ball bearings within the paste, which is exactly the behavior reported by Feys et al.³⁷ at low $C{a}_{\tau 0}$. Aïtcin¹¹¹ also mentioned that EA modifies the viscosity of the paste and its cohesiveness at the same time.

A recent study conducted by Feneuil et al.³⁵ measured the ${\tau }_0^s$ variation in cement paste by mixing it with aqueous foam. They obtained a decrease in ${\tau }_0^s$ when using a cationic, low-adsorption ability surfactant, but a significant increase in ${\tau }_0^s$ when using an anionic, high-adsorption surfactant, attributing the behavior to a hydrophobization induced by the surfactants. They sustain that the effect of EA is comparable to that of added solid particles, and that it is governed by $C{a}_{\tau 0}$. The results of the study incorporate two additional factors that may influence the rheological effect of air-entraining concrete: the charge of the surfactant, and the adsorption ability of the surfactant with respect to the cement grains.

One approach to understand the behavior of bubbles in concrete is numerical modeling, which researchers^96,103 have attempted to do with yield stress fluid models or rigid particle models. To incorporate the ${\varphi }_{air}$ into the model, Rust and Manga⁹⁶ suggested using the Krieger-Dougherty model [Eq. (9) with n = 1] that was proposed to describe the flow of a yield stress fluid suspension of rigid spheres¹⁰⁵ and is widely used to correlate the relative viscosity with solid volume fraction in rigid particle suspensions.¹¹⁵

Although some authors^20,37,77,93 have proposed this model for yield stress fluid air bubble suspensions, it has several limitations for its application in a concrete or mortar with EA. Specifically, the Krieger-Dougherty equation was proposed for (i) a Newtonian suspending fluid, with (ii) only two phases, and with (iii) a solid particle suspension. A concrete mixture with air bubbles does not comply with any of these three requisites.

Despite these limitations, many researchers have reported a valid representation of ${\mu }_{rel}$ of cement paste.^94,115–117 If it can be assumed that air bubbles—under certain circumstances—act like rigid spheres, it can be assumed that reliable fitting can be achieved with a proper adaptation of the parameters to mortar and concrete.

For further understanding of the rheological effect of AEA on cement paste, the existing experimental data for ${\tau }_{rel}^s$ and ${\mu }_{rel}$ have been plotted in Figures 3 and 4 against the models given by Equations (8)–(13). The value chosen for ${\varphi }_m$ was 0.56 as measured for a bead suspension in cement paste,⁹³ whereas the flow index n for ${\mu }_{rel}$ models was set at 0.5.

FIG. 3.

Relative static yield stress of cement paste as a function of air volume fraction. Black and gray points are experimental results obtained by Feneuil et al.³⁵ for an anionic (AEA-A-1) and cationic (AEA-C-1) surfactant, respectively. Black continuous line, gray continuous line, and dotted continuous line are the theoretical limits given by Equations (8), (10), and (12), respectively.

FIG. 4.

Relative plastic viscosity of cement paste as a function of air volume fraction. Gray points are experimental results obtained by Yun et al.¹⁰⁷ (AEA-A-2) and Chia and Zhang¹¹⁴ (AEA-A-3) for anionic surfactants. Black continuous line, gray continuous line, and dotted continuous line are the theoretical limits given by Equations (9), (11), and (13), respectively.

For ${\tau }_{rel}^s$, the data have been taken from Feneuil et al.³⁵ It can be seen that for the anionic, high-adsorption surfactant, the results follow the trend of the rigid particle model, whereas for the cationic, low-adsorption surfactant, the results remain below the non-deformable bubble model ($C{a}_{\tau 0}\to 0$), and mainly above the deformable bubble model ($C{a}_{\tau 0}\to \infty$). The authors attribute the rigid-particle behavior of the bubbles generated by the anionic surfactant to an armor-like particle coverage around the bubble, caused by the irreversible adsorption of the cement grains onto the bubble's surface. This armored bubble behavior is not observed for the cationic surfactant; the bubbles generated are bare, explaining their deformable behavior.

For ${\mu }_{rel}$, the data have been obtained from experiments conducted by Chia and Zhang¹¹⁴ and Yun et al.¹⁰⁷ on air-entrained concrete with anionic surfactants (the data have been adapted to the cement paste proportion in the concrete formulation, considering that EA is only in the paste). Contrary to what was observed with ${\tau }_{rel}^s$, the experimental data for ${\mu }_{rel}$ follow the trend set by the deformable bubble model. This behavior was also observed by Struble and Jiang,³⁶ who obtained a decrease in μ_p when using two different neutralized surfactants and an anionic surfactant. Feys et al.³⁷ also observed a decrease in μ_p when entraining concrete via pumping. These last two experimental trials, however, do not follow the trends set by the models previously mentioned, which could be due to different measuring methods, or only a behavior obtained by surfactant-generated bubbles.

The experimental results available suggest that in a flowing regime, the bubbles—even if they are armored bubbles—are deformed by the surrounding shear, whereas at rest, armored bubbles would seem to contribute to structural build-up.

In summary, among the reasons mentioned in previous studies,^{35,37,95,97,103} the relevant factors that determine whether the bubble will deform are:

• the shear rate applied; the larger the force, the more likely the bubbles “squeeze” and deform.

• the size of the bubble; the larger the bubble, the more likely the bubbles deform.

• the air volume fraction (${\varphi }_{air}$); the larger the volume fraction, the more likely the bubbles are pressed against each other, increasing the shearing force.

• the rheology of the suspending fluid; the lower the μ_p and ${\tau }_0^s$ of the suspending liquid, the less likely the bubbles deform.

• the charge and adsorption ability of the surfactant; which will depend directly on the type of AEA used.

Although other methods of air entraining exist, AEA seems to be the ideal one for 3DCP. Feys et al.³⁷ experimented with direct air inclusion via pumping to avoid any additional effects apart from the bubbles, as they sustained that AEAs have a rheological effect on concrete due to surfactants present in these agents and apart from the bubbles themselves.³⁷ However, air inclusion via pumping is unviable for 3DCP, as the processing stages involved, particularly pumping, have been shown to decrease the base air content in concrete, as shown by Das et al.¹¹⁸ The authors expect, however, that air inclusion via AEAs would create a stabler air-void system, perhaps resistant to these processing stresses.

Indeed, AEAs have the advantage of stabilizing the bubbles¹¹⁹ by avoiding coalescence (i.e., the action by which when two or more bubbles touch, they form into one)¹²⁰ and, therefore, size increment that would make them more prone to deformation or being lost by the action of buoyancy. Since the surfactants of AEAs act at the air–water interface in cement paste, they stabilize the air within the concrete, creating bubbles in the size range of 50–150 μm. The polar reaction reduces the surface tension of the water, and a negatively charged cover is generated around each bubble, leading to mutual repulsion and therefore preventing the coalescence of the bubbles.⁸⁹ In the following section, the different types of AEAs will be described to further understand the rheological effect of these admixtures, as well as their mechanisms and possible effects on the mixture in question.

Types of air-entraining agents

The AEAs are surfactants, meaning that their molecules, when incorporated into a heterogeneous solution (i.e., bubbly water), are abstracted from it and adsorbed at the solid–liquid interface, or in this case, the gas–liquid interface.^119,121 The surfactant's molecules are amphipathic, which means that they have a polar head and a non-polar tail.^119,121 Depending on the electric charge of the polar head, surfactants can be ionic (i.e., cationic when positively charged, anionic when negatively charged), non-ionic, or uncharged.¹¹⁹ The molecule becomes oriented naturally at the interface, leaving the charged or uncharged head in the water, and the tail in the air bubble, as shown in Figure 5. The adsorption of the surfactant creates a series of consequences in the solution, mainly, the reduction of surface tension,^89,90,122 which is directly proportional to the concentration of the surfactant.

FIG. 5.

Orientation of surfactant molecule in air–water solution.

There are many types of AEAs. Previous researchers^119,122 have specified seven types, depending on the origin of the AEA, which are: (i) salts of wood resins (i.e., phenolics, carboxylic acids, abietic acids), (ii) synthetic detergents (i.e., alkyl aryl sulfonates), (iii) salts of sulfonated lignin, which is a by-product of the paper industry, (iv) salts of petroleum acids (i.e., sulfonates), (v) salts of proteinaceous materials, from the animal-processing industry, (vi) fatty and resinous acids and their salts (i.e., tall oils, alkali metal salts), and (vii) organic salts of sulfonated hydrocarbons. The salts of wood resins are the most widely used.^88,119,122

Mechanisms for stability of air bubbles

The AEAs have been used commonly to improve the freeze-thaw resistance,¹²³ and in some cases to reduce bleeding and segregation of concrete.^88,119 The mixing process of concrete incorporates air naturally, defined as entrapped air, and can vary from 1% to 2% in volume.¹ The AEA stabilizes this air, dividing these large, coarse bubbles into a network of small, equally dispersed air bubbles.^89,90,111

Air is originally entrapped during mixing, either by infoldings due to vortex action, or coming from fine aggregate, which holds air during mixing.¹¹⁹ These bubbles would normally tend to coalesce and become larger, therefore destabilizing the concrete, which is reduced by vibrating the concrete, or dissipating due to the action of gravity. The AEA prevents the coalescence of the small bubbles generated, by stabilizing them via several mechanisms:

• Hydration sheath: Water molecules orient themselves around the bubble, forming a layer and deflocculating the network. This occurs in both ionic and nonionic agents.¹²⁰

• Marangoni effect: The bubbles are stabilized against deformation and rupture due to the decreased surface tension; the gradient causes the liquid to flow away from regions of low surface tension, which in this case is radial, maintaining the shape of the bubble.^89,120

• Electrostatic repulsion: If the surfactant molecule is charged, the bubble will acquire this charge and therefore repel other bubbles.^89,119

• Steric effects: In anionic AEAs, the insoluble calcium salts precipitate and accumulate at the air–liquid interface, creating a protective film or colloid.^88,119,122 This does not reduce the surface tension of the liquid but, nonetheless, stabilizes the air bubbles.

• Adsorption onto hydrated cement grains: This is perhaps the strongest and most reported of effects, present in anionic AEAs. The hydration products created when cement is mixed with water have a positive surface (calcium ions from calcium silicate hydrates). The negatively charged AEAs coating the bubbles can then be adsorbed onto the surface of the cement particles due to electrostatic bonds, consequently attaching the bubbles to the cement (Fig. 6), and making the cement particles hydrophobic.^{88,90,119,122}

FIG. 6.

Adsorption of anionic surfactant onto cement grains, based on Khudhair et al.¹²³

Although some of these effects are exclusive to anionic AEAs, it has been demonstrated that cationic and nonionic surfactants also cause air entraining, depending solely on the first three mechanisms for cationic, or the first two for nonionic. However, nonionic agents usually generate poorer air entraining with larger and less stable bubbles.¹¹⁹

The theory and experiments exposed in this section suggest that a higher stability of the air in concrete could lead to an increase in ${\tau }_0^s$ and μ_p. Further, the experiments conducted by Feneuil et al.³⁵ showed a decrease in the ${\tau }_0^s$ of cement paste with a cationic, low-adsorption capacity surfactant, but an increase of ${\tau }_0^s$ when using an anionic, high-adsorption surfactant. These results suggest that the determinant of the rheological effect of AEA in cement paste could depend completely on the type of surfactant used. A more protected, less deformable bubble acts like a solid inclusion, generating what may be considered a filler effect, with the benefit of no added weight or material.

In the context of 3DCP, some factors may have to be considered regarding AEAs, as some components or properties of 3DCP have been shown to interact with the surfactants. It is known that the presence of fine SCMs used widely for printable concretes, such as fly-ash and fillers, reduces the effectiveness of the AEAs and, therefore, ${\varphi }_{air}$. For this reason, a larger amount of AEA is required in those mixtures that use fine SCMs.^112,124 The presence of a high proportion of fine aggregate promotes the air content. Mixtures with high ${\tau }_0^s$, which is usually the case in 3DCP, are more difficult to air entrain.¹¹⁹

An important factor to consider is that some types of AEA interact with other chemical admixtures, particularly SPs, which are used widely for printable concretes. The main issue is with organic SPs, where a competitive adsorption at the liquid–gas interface generates less stable air bubble systems.¹²⁵ This is more pronounced for certain combinations of AEAs and SPs, such as lignosulfonate and hydroxycarboxylate SP's with neutralized wood resin AEA, but less pronounced with AEAs from salts of fatty acids and sodium sulfates, which have shown good compatibility with these SPs.¹²² The negative effect generated by the combination of naphthalene or melamine-based SPs can usually be solved by adding more AEA.¹²⁵

Though some combinations have been studied,⁸⁸ we believe that it is impossible to generalize the interactions of AEAs and SPs due to their complexity, and we recommend an experimental examination of combinations before implementation.

Effect of AEA on Printability

It has been shown in this review that the rheological effect of AEAs on concrete could be beneficial in the context of 3DCP. The effect reported by most authors, given certain circumstances, is that the addition of AEA to a concrete mixture would increase ${\tau }_0^s$ and decrease μ_p, directly improving the properties of buildability, extrudability, and pumpability. In addition to the rheological benefits exposed in the article, AEA has the potential to further improve buildability since it reduces the density of the concrete when a high ${\varphi }_{air}$ is used. Researchers^27,43 have calculated that maximum building height, printing velocity, minimum static yield stress, and other printing parameters depend directly on the density of the concrete, so any considerable reduction in density would contribute to such results.

It was indicated that one aspect of printability cannot be modified without affecting another, and the addition of AEA is no exception. It is well known that air-entraining concrete reduces its compression and flexural strength. However, within the context of 3DCP, where a printed element is not monolithic like conventional concrete, the structure is more likely to fail at the plain between layers,^126,127 making interlayer bonding a more limiting factor than the mechanical strength of the material itself, and allowing a higher air incorporation without a detrimental effect in the overall resistance of the printed element.

Conclusions and Further Research

Conclusions

Within the rheological requirements for 3DCP, special emphasis has been made on buildability, namely, early and later age structural build-up, and the challenges of obtaining this property for an all-around printable mixture, in other words, without negatively affecting other properties, such as extrudability, pumpability, and open time. The different approaches followed by previous researchers have been explored, from a more physical approach due to flocculation and thixotropy, to a more chemical approach obtained from accelerated hydration. It can be concluded that a combination of the two is required to obtain the optimal results in 3DCP, whether by using different, compatible chemical admixtures for each effect, or chemical admixtures with both effects.

This literature review has shown that AEAs have great potential for aiding early-and later-age structural build-up. In the right circumstances, they could be an ideal mechanism for enhancing concrete printability, given that they generate an increase in static yield stress (${\tau }_0^s$) and a decrease in plastic viscosity (μ_p). This review has demonstrated this behavior theoretically, using different rheological models, and experimentally, as shown in data obtained by other researchers.

Although it must be further proven experimentally, it can be hypothesized that for the case of armored bubbles (anionic AEAs), air bubbles could act as solid inclusions and contribute to ${\tau }_0^s$ of fresh concrete, whereas for other types of AEAs, the behavior remains between the theoretical limits of a non-deformable fluid particle [capillary number ($C{a}_{\tau 0}$) that tends to 0) and a deformable one ($C{a}_{\tau 0}$ that tends to infinity)]. For μ_p, however, the behavior seems to follow the trend of a deformable fluid particle, even in the case of armored bubbles.

Further research

Although this review has demonstrated promising potential for AEAs as a rheological enhancer for printability, further experimental research is required to completely validate it. All three types of AEAs should be submitted to rheological tests to characterize their effect experimentally. In addition, experimental trials should be held to verify interactions between AEAs and other components usually present in printable concretes, such as other admixtures or SCMs.

An approach for further research is the development of models that predict the behavior of bubbles in a printable mortar. The combination of models that incorporate the air volume fraction (${\varphi }_{air}$) as a parameter, such as the Krieger-Dougherty and Chateau-Ovarlez-Trung, with the boundaries of $C{a}_{\tau 0}$, could have potential in describing this behavior, if further validated with experimental trials.

The use of alternative solutions to obtain the required structural build-up could ultimately permit a reduction of the cement dosage in these mortar designs, making 3D printing with mortar even more sustainable, in an environmental and financial aspect. 3DCP has been proven viable and could have many advantages for the concrete industry and could help to reduce and optimize the use of involved natural resources and the emissions generated, meaning that 3DCP has the potential to contribute importantly to improve the productivity and sustainability of the construction industry.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

The work included in this article is supported by the National Agency of Research and Development under grants Fondecyt No. 1190641 and by the Center for Sustainable Urban Development, FONDAP No. 15110020.