Properties of fresh and hardened mortar using citric acid to tailor its suitability for 3D printing

Abstract

In large-scale 3D concrete printing (3DCP), maintaining the material’s workability during the printing process is essential, as on-site delays and environmental conditions such as elevated temperatures can accelerate setting. Consequently, the ability to control the hydration kinetics in printable cementitious materials is of great practical importance. This study investigates the effect of citric acid, used as an organic retarder at dosages of 0, 2.5, 5.0 and 7.5 wt% by weight of water (bwow), on the hydration of a printable Portland cement-calcium aluminate cement binder system. Fresh- and hardened-state material properties were evaluated by measuring static and dynamic yield stress in the cast and printed material using penetration and slug tests, and by monitoring Young’s modulus evolution using ultrasound. In addition, thermogravimetric analysis (TGA) and scanning electron microscopy (SEM) were conducted to evaluate hydration kinetics and microstructural differences and shrinkage was measured using molds. Mechanical performance was evaluated, and the deformation of the printed objects was measured. The results demonstrate a clear dosage-dependent retardation effect: citric acid extended setting times, increased flexural and compressive strength and reduced the strength loss associated with a cold joint. At the same time, higher dosages also increased shrinkage and reduced buildability due to greater deformation of the lower filaments under the load of the subsequently deposited layers. Microstructural analyses indicated that citric acid affects the development of hydration products, particularly calcium silicate hydrates (C-S-H), thereby influencing long-term strength development. The results highlight the need to optimize the citric acid dosage to balance open time, buildability and structural performance in 3DCP.

1 Introduction and theoretical background

The advent of 3D concrete printing (3DCP) has revolutionized the construction industry by enabling automated, material-efficient, and geometrically complex structures. However, successful implementation requires precise rheological control to balance pumpability, buildability, and the interlayer bond. To meet these complex requirements, additives are commonly used to modulate reaction rates (Bos et al., 2016; Mechtcherine et al., 2020). Retarders are especially relevant for maintaining an open time to ensure pumpability while also reducing the risk of cold joints, which lead to poor interlayer bonding and decreased structural performance (Perrot et al., 2016; Wangler et al., 2019). While physical methods such as cooling can delay setting, chemical retarders such as citric acid are often preferred due to their cost-effectiveness, widespread availability, low toxicity, and environmentally benign characteristics (Nangare et al., 2021; Aïtcin, 2016; Mantellato et al., 2016; Singh et al., 1986; Ai et al., 2024; Byrd et al., 2021). In 3DCP, retarders are crucial to maintain workability and control structural build-up, thereby reducing the risk of cold joints between layers. Furthermore, citric acid has been reported to be suitable for calcium aluminate cements and ternary systems, as reported by Blask et al. (Blask et al., 2000), as well as for Portland cement, as Ramachandran and Lowery (Ramachandran and Lowery, 1992) established.

On a mechanistic level, Nguyen et al. (Nguyen et al., 2021) established that in ettringite-based binders, citric acid inhibits ettringite formation, leading to monosulfate and gypsum precipitation. More broadly, citric acid inhibits hydration by decreasing the zeta potential of cement particles through the adsorption of citrate ions onto the surface of cement grains, which leads to the retardation of alite and aluminate phases (Singh et al., 1986; Möschner et al., 2009). Studies by Möschner et al. (Möschner et al., 2009) suggest that potassium may function as a counter-ion to the negatively charged citrate on these surfaces. However, Möschner et al. (Möschner et al., 2009) concluded from thermodynamic analysis that citric acid has no significant long-term effect on the pore solution, as it was removed from the solution within hours. Though Möschner et al. (Möschner et al., 2009) as well as Smillie and Glasser (Smillie and Glasser, 1999) found minimal evidence of cation-citrate complexes, both Ramachandran and Lowery (Ramachandran and Lowery, 1992) and Vohburger et al. (Vohburger et al., 2025) concluded that citrate-induced retardation is linked to C-S-H precipitation, involving the complexation of calcium and silicon ions, either through their incorporation into the precipitated gel or within the solution. This effect becomes more pronounced with higher dosages, leading to an extended induction period and a broader, less intense hydration peak (Singh et al., 1986; Ramachandran and Lowery, 1992; Möschner et al., 2009).

With respect to fresh mortar properties, the effectiveness of citric acid is highly dosage-sensitive compared with that of other retarders such as tartaric acid and gluconates (Yu et al., 2021). According to Singh et al. (Singh et al., 1986), very low concentrations can accelerate the reaction and excessive amounts can potentially impair early-age structural performance. In contrast, Möschner et al. (Möschner et al., 2009) reported enhanced workability without compromising long-term strength also over the low investigated dosage range (0.1–0.5 wt% bwoc and 0.25 to 1.25 wt% bwow). Regarding the properties of hardened concrete, the influence of citric acid is complex and dependent on dosage and cement type. While higher dosages can noticeably decrease early-age strength by significantly delaying hydration, studies indicate that with optimal dosages, long-term strength can be maintained or even enhanced (Zhu et al., 2021; Mendes et al., 2020; Purnomo et al., 2019; Wang et al., 2018). For instance, research by Mendes et al. (Mendes et al., 2020) on Portland cement and by Wang et al. (Wang et al., 2018) on magnesium oxysulfate (MOS) cement has shown that specific citric acid additions can lead to increased compressive strength after 14 d and 28 d. Nguyen et al. (Nguyen et al., 2021) also showed that citric acid can increase ettringite formation as well as C-S-H precipitation and improve the degree of hydration. However, Zhu et al. (Zhu et al., 2021), Purnomo et al. (Purnomo et al., 2019) and Wang et al. (Wang et al., 2019) have found a less significant, yet highly dose-dependent impact. Beyond its strength, citric acid can also influence other critical hardened properties relevant for 3DCP, such as shrinkage and durability: Qin et al. (Qin et al., 2024) found that a reduction in shrinkage (up to expansion) at higher citric acid-modified chitosan dosages (0.6% CAMC) is due to a “bleeding-like” phenomenon, where delayed hydration results in excess free water. Its subsequent reabsorption leads to an expansion that effectively counteracts later drying shrinkage and minimizes overall volumetric change. Sun et al. (Sun H. et al., 2023) demonstrated that citric acid improves freeze-thaw resistance, reduces pore volume and mass loss when immersed in water, contributing to overall durability and a denser microstructure. Wang et al. (Wang et al., 2019) additionally showed a reduction in porosity for acid concentrations of 2% and 3%.

Despite this extensive chemical foundation, a significant research gap remains: while retarders offer a means to tailor hydration kinetics, their specific effects on the coupled requirements of pumpability, buildability, and structural stability in 3DCP remain insufficiently understood, particularly regarding the reliability of laboratory-scale measurements in predicting material behavior under realistic process conditions. Many of the studies discussed above primarily examine individual material responses—such as hydration kinetics, workability, strength, shrinkage, or durability—in isolation rather than in their combined relevance for 3DCP performance. In contrast, the present study adopts a more comprehensive perspective essential for optimizing admixture dosages to improve 3DCP efficiency by assessing the influence of citric acid across several interrelated fresh-state and hardened-state properties that together govern the performance of printable mortar. Accordingly, this work systematically investigates the effect of citric acid dosage (0, 2.5, 5, and 7.5% bwow) on a printable mortar mix, with the specific experimental framework and research objectives detailed in the following Scope and Concept section.

2 Scope and concept

To answer these research questions, an extensive experimental program was designed to determine suitable citric acid dosages for 3D printing and to validate the selected mixtures under realistic printing conditions (Figure 1). First, the constituent materials were selected, and the mortar mixtures were prepared according to the compositions described in Section 3. In the laboratory phase, mixtures containing 0, 2.5, 5.0, and 7.5 wt% citric acid bwow were characterized under controlled conditions. The mixes are called according to the citric acid content CA0, CA2.5, CA5.0 and CA7.5. This characterization included penetration tests to quantify the evolution of static yield stress and ultrasonic measurements to determine the development of Young’s modulus. Based on the laboratory results, those mixtures that provided a practical processing window and sufficient structural stability were selected for printing trials. During printing, static and dynamic yield stress were determined using a mobile penetrometer and the slug test, respectively, to compare in-line measurements with laboratory-derived values. The geometric deformation of printed elements was also quantified and related to the previously measured evolution of Young’s modulus.

FIGURE 1

Beyond the fresh-state characterization, the experimental program evaluated the final material performance and hydration mechanisms. Flexural and compressive strength were compared for cast and printed specimens, also investigating the effect of a 12-min cold joint. To understand the underlying chemical processes, thermogravimetric analysis (TGA) and scanning electron microscopy (SEM) were used to assess hydration kinetics and microstructural development, while shrinkage was monitored over 28 d using ASEM grooves. This combined approach made it possible to evaluate the predictive capability of laboratory-scale characterization for practical 3D printing applications, to characterize how citric acid affects fresh mortar properties that are specific to 3DCP and to assess how these retarded characteristics are modified by process-specific printing effects.

3 Materials and mixtures

The printable material investigated in this study was a commercially available mortar marketed as “3D print moertel” by RYGOL Baustoffwerk GmbH & Co. KG, Germany, and is based on quartz aggregates with a maximum diameter of 1.2 mm. The binder consisted of 10 wt% calcium aluminate cement and 15 wt% Portland cement (CEM I). The water-to-binder ratio was 0.74 for all samples, corresponding to a water-to-dry-premix ratio of 0.185, which was required to achieve printable consistency without additional admixtures. The citric acid was added relative to the water content, and four different concentrations were investigated: 0, 2.5%, 5.0% and 7.5% acid by weight of water (bwow), as shown in Table 1. The sample names were chosen according to the amount of citric acid, namely, CA0, CA2.5, CA5.0 and CA7.5. The acid was first mixed into the water, after which the liquid components were added to the dry pre-mix over a 15 s period. The compositions were then mixed in batches of 2 kg for an additional 30 s at ≈ 100 revolutions per minute (RPM), using a KitchenAid 5K45SSEOB in laboratory settings. During printing sessions, the material was then mixed using a ZM80 mixer from Baumaschinen Beckschulte KG, Germany, in batches of 50 kg for another 2 min at its maximum intensity of ≈50 RPM. The longer mixing time was chosen due to the lower mixing intensity of the ZM80 mixer and the lower effective mixing-tool-to-material-volume-ratio.

4 Experimental setup

4.1 Measurements in laboratory setting

4.1.1 Static yield stress evolution using a penetrometer

The penetration test was selected to evaluate buildability through static yield stress evolution, which governs resistance to deformation and thus structural stability during printing. Unlike manual methods, the automated penetrometer continuously records the resisting force, enabling detailed monitoring over time. After mixing, the material was placed in a cylindrical plastic cup (8 cm height, 5.7 cm diameter). The cup was filled halfway and compacted with 10 controlled impacts, then filled completely, compacted again, and leveled. The sample was tested using the ToniFORCE apparatus (Toni Technik Baustoffprüfsysteme GmbH, Germany) and covered with a plastic hemisphere to prevent surface drying and draught effects, see Figure 2.

FIGURE 2

Before measurement, a 25 mm diameter sphere was fully embedded in the material, and the force was reset after 2 min. During testing, the sphere penetrated at a constant rate of 0.015 cm/min while the required force was recorded every second up to 400 N as previously established by Pott et al. (2023); Pott and Stephan (2021). As material flow around the sphere was negligible, the yield stress τ₀ [kPa] was calculated Equation 1 according to Lootens et al. (2009).

Here, F [N] is the measured force and R [m] the sphere radius. Two measurements per composition were conducted under standardized climate conditions of 20 °C and 60% relative humidity (RH).

4.1.2 Young’s modulus progression using ultrasound

Ultrasonic testing was conducted to quantify early-age hydration over 72 h, extending the insight gained from penetrometer measurements. In the first hours, structural build-up was assessed by tracking ultrasonic wave velocity. Since wave velocity is directly related to material stiffness and density, it enables the determination of the dynamic elastic modulus (Young’s modulus) and can be correlated to structural deformation after extrusion, which is highly relevant for 3DCP. Moreover, velocity development can serve as an indicator of later strength, as wave transit time is inversely proportional to stiffness and density. Prior to mixing, the exact distance between the sender and receiver mounted on the molds was recorded. Immediately after mixing, the material was placed into silicone molds in two layers and compacted with 10 impacts per layer. Ultrasonic wave velocities were recorded once per minute over 72 h using the UltraTest IP-8 device of UltraTest GmbH, Germany. Measurements were performed at 20 °C and 60% RH with two repetitions per mixture. The Young’s modulus E was calculated according to Equation 2.

Here, ρ is the material density, v is the measured wave velocity and μ is the Poisson’s ratio. Though the Poisson’s ratio may change during early hydration as the material transitions from a fresh suspension to a progressively stiffening solid, a constant ratio of 0.3 was used as an approximation according to Swamy (Narayan Swamy, 1971).

4.1.3 Mechanical strength of cast samples

To evaluate the influence of citric acid on the mechanical performance of mortar, compressive and flexural strength tests were conducted after 1 d and 28 d according to DIN 196–1. However, the prisms were stored in a dry standard climate of 20 °C and 60% RH to ensure comparability with the printed samples. The flexural strength was measured with a three-point bending test under a loading speed of 640 N/s. The compressive strength was measured under an applied loading speed of 2400 N/s.

4.1.4 Shrinkage

Given the high surface exposure of printed specimens, measuring shrinkage is crucial to assess potential cracking, deformation, and reduced stability. To isolate shrinkage from printing-related variables, ASEM grooves (Advanced Shrinkage & Expansion Measuring System) of UltraTest GmbH, Germany, were used. Each mold had a length of 10 cm and a volume of 45 cm³. The shrinkage was measured with a resolution of 0.001‰. To ensure unhindered movement, the molds were lubricated and lined with a thin plastic film and, after filling, covered with foil. Compared to shrinkage of printed, exposed filaments this partly suppresses drying from exposed surfaces, so the recorded deformation reflects autogenous compared to plastic shrinkage more than would be the case for printed specimen. For each material composition, measurements were conducted in duplicate over 28 d at 1-min intervals.

4.1.5 Phase composition analysis using thermogravimetric analysis (TGA)

Thermogravimetric analysis (TGA) was employed to quantify hydration products, which allowed indirect assessment of hydration progress across different mortar compositions. Mortar samples were cast according to DIN EN 196–1, but after 24 h, cured under dry conditions (20 °C, 60% RH). The compositions were compared at an age of 1 and 28 days. Upon reaching the target age, hydration was halted by soaking the samples in isopropanol for 3 h, followed by oven drying at 40 °C for 3 days. After drying, the specimens were milled using a vibratory disc mill and then stored airtight for subsequent analysis. Four measurements were averaged for each composition, and two for the unhydrated material. 10 mg of each dried sample was placed into pre-weighed crucibles. The analysis was conducted using a TG 209 F3 Tarsus instrument from NETZSCH-Gerätebau GmbH, Germany. Samples were heated at a rate of 10 K/min up to 850 °C under a nitrogen atmosphere.

4.1.6 Phase composition analysis using scanning electron microscopy (SEM)

Microscopy was used to enable a detailed characterization of the hydration phases of the mortar samples. Samples for scanning electron microscopy (SEM) were prepared in the same way as those for TGA. However, the samples were not milled but broken to ensure a fresh fractured surface. For SEM analysis, a GeminiSEM 500 NanoVP instrument (of Carl Zeiss IQS Deutschland GmbH, Germany), equipped with an in-lens secondary electron (SE) detector, was used to capture high-resolution images of the structure.

4.2 Setup of printing experiments

The printing of the samples was conducted using a batch system consisting of a P20 screw pump and associated mixer and a UR10e robotic system, see Figure 3. The printing nozzle had a diameter of 2.8 cm and extruded the material with an extrusion rate of 5.4 dm³/min at a printing speed of 15 cm/s. The surrounding temperature at the time of printing was 28.7 °C, and the relative humidity was 41%. The differences to the laboratory testing climate of 20 °C, 60% RH are expected to influence the tests results through hydration kinetics and structural build-up. The printing geometries, which form the basis of the following analysis, are shown in Figure 3. These were printed once each for the compositions CA5.0 and CA7.5, so observations are interpreted as indicative trends rather than statistically validated process outcomes.

FIGURE 3

4.3 Measurements during the printing session

4.3.1 Static yield stress evolution using a mobile penetrometer

Static yield stress evolution post-extrusion was measured using a Mecmesin Shotcrete Penetrometer and then compared with laboratory measurements. The measuring head of the penetrometer consisted of a cylinder with a radius of 0.15 cm and a height of 1.25 cm, and a cone with a height of 0.25 cm. Measurements were taken at the center of the filament at 2, 5, 10, 15, and 30 min post-extrusion, with ten repetitions each. The applied force F [N] was converted into yield stress τ [kPa] using Equation 3 outlined by Lootens et al. (Lootens et al., 2009).

With F being the recorded force [N], R being the radius of the cone (here: R = 1.5 mm), h₂ being the height of the cone (here: 2.5 mm) and h₁ being the cylinder height (here: h₁ = 12.5 mm).

4.3.2 Dynamic yield stress determined via slug test

The dynamic yield stress was measured using the in-line slug test described by Ducoulombier et al. (Ducoulombier et al., 2020) based on the analysis of 100 individual slugs. Their total weight was recorded, from which the average mass per slug was calculated. Using the mass data, the yield stress τ was determined using Equation 4 according to Ducoulombier et al. (Ducoulombier et al., 2020).

Where g is the gravitational acceleration [m/s²], m_i is the average slug mass [kg] and S is the cross-sectional area of the nozzle [m²], which was 0.000616 m² given the diameter of 2.8 cm. Beyond a quantitative assessment, the extruded slug shape was qualitatively evaluated to characterize the material’s flow, given that the actual form often deviates from a perfect cylinder.

4.3.3 Mechanical strength of printed samples and impact of cold joints

To quantify the impact of interlayer bonds and cold joints, prisms were sawn from printed structures of both the CA5.0 and CA7.5 compositions. For each, one structure was printed consecutively, while a second included a 12-min cold joint at mid-height, shown in Figure 3. The 12 min delay was selected based on the yield stress evolution previously measured in the laboratory (Section 5.1.1), as it offers a comparison between the reference time of about 8 min (time it takes for the material to get from the mixer to the nozzle) and 20 min. At 20 min, structural build-up is already appreciable, with the yield stress of CA5.0 being nearly double that of CA7.5. All specimens were cured in a standardized climate (20 °C, 60% RH) until testing. This deviation from DIN EN 196–1 was intentional, as water curing is often impractical for 3DCP. After 28 d, three prisms per group were tested for flexural strength (640 N/s), and six prism halves were tested for compressive strength (2400 N/s).

4.3.4 Geometrical deformation via 3D scanning

The evaluation of the printed geometry aims to determine the angle between an ideal vertical build-up and the actually printed outer surface, as shown in Figure 4. A higher angle signifies greater deformation, while 0° indicates a perfect vertical profile. This deformation is expected to correlate inversely with the Young’s modulus obtained from ultrasonic wave transmission tests. This non-invasive, in-situ approach is highly advantageous as it quantifies structural stability during the printing process without physical sampling or disrupting the material’s rheological state.

FIGURE 4

Measurements are captured using a RealSense D415 RGB-D camera mounted on a UR10e robotic arm, which follows a predefined trajectory. This setup captures 51 RGB-D images over a specimen area of 36 mm (length) by 150 mm (height), saved as PLY files containing 3D coordinates and RGB data. Using a Python script, these files are merged into a single integrated model. The integrated data is divided into nine 4 mm wide sections. Within each section, the mean point cloud is computed to reconstruct the surface (Figure 4, middle). A linear regression line was then fitted to these mean points and compared to an ideal vertical line to calculate the deviation angle (Figure 4, right). This metric provides a direct assessment of shape stability; a high deviation angle suggests the lower filaments have widened under the weight of subsequent layers, likely due to a lower Young’s modulus and insufficient early-age stiffness.

5 Results

5.1 Measurements in laboratory setting

5.1.1 Static yield stress evolution using a penetrometer

The yield stress, shown in Figure 5 reveals a pronounced dependency on citric acid content. The composition CA0 exhibited a rapid increase in yield stress, reaching the maximum measurable force within approx. 14 min, indicating a rapid structural build-up and formation of hydration products. The compositions containing 2.5% and 5.0% citric acid demonstrated delayed responses, postponing the onset of structural build-up by 16 min from CA0 to CA2.5 and by a further 16 min from CA2.5 to CA5.0. CA7.5 exhibited the most extended delay in yield stress increase, reaching 272 kPa after approx. 69 min, a further delay of 23 min, slightly more than the previous acid dosages, indicating a non-strict linearity.

FIGURE 5

The retardation effect appears approximately linear up to 5.0% citric, with a stronger increase in retardation at 7.5%. All measurements showed an exponential increase in yield stress, corresponding to the progressive formation of hydration products (Perrot et al., 2016; Pott and Stephan, 2021; Reiter, 2019). The results are consistent with prior studies, for example, those of Purnomo et al. (Purnomo et al., 2019), who confirmed a linear to slightly exponential relationship between setting time and acid dosage using the Abrams cone and Vicat penetration. Similarly, using Vicat penetration and a penetration test, Soriano (Soriano, 2019) confirmed a near-linear relationship between retardation and retarder concentration.

5.1.2 Young’s modulus progression using ultrasound

The measurements in Figure 6 show ultrasound velocity and the calculated Young’s modulus depending on mixture density (see Table 1). Phase 1 is characterized by a short plateau (length increasing with acid concentration), followed by phase two, a steep increase in ultrasound velocity to approx. 2700 m/s. Between 2700–3000 m/s, there is then a transition into a third phase where the velocity significantly slows down and does not exceed 3500 m/s and 19 kN/mm² Young’s modulus, respectively. With increasing citric acid dosage, the increase in ultrasonic velocity and Young’s modulus during the first two phases became slower, and the curves showed a more pronounced deviation from linearity during the first 30–120 min. The linear approximation of the velocity increase (velocity = a * time) decreased from 0.74 m/s² for CA0 to 0.13 m/s² for CA7.5, while the time required to reach 2500 m/s increased from less than 1 h to approximately 2 h, 3 h and more than 5 h. This is consistent wi delayed the reaction more significantly than CA2.5 and CA5.0, and thus a not strictly linear retardation with dosage increase. The trend indicates a strong dosage dependence of setting retardation, which were confirmed by Purnomo et al. (Purnomo et al., 2019) and Soriano (Soriano, 2019), who used Vicat penetration among other methods to quantify citric acid retardation.

FIGURE 6

With increasing citric acid dosage, the progression of ultrasonic velocity and Young’s modulus became more pronounced: After an initially slower increase, a steep increase in velocity was observed, followed by a distinctive slowdown and a subsequent second increase. We hypothesize that this could be due to the progression of cement hydration and be attributed to the induction period (initially slower increase), the accelerating period (first steep increase), followed by the deceleration period (a distinctive slowdown) before the final steady state begins, all of which occur later and slower in retarded samples (Wangler et al., 2022). Above 3000 m/s, no correlation was measured between retarder and velocity, matching findings by Desmet et al. (2011) and Daake and Stephan (2016) that mainly the setting time can be quantified and that ultrasonic velocities stabilize in the asymptotic phase. Only ultrasonic measurements of materials with greater compositional differences exhibit more significant differences, as shown by Pott et al. (2023); Leusmann et al. (2017).

Furthermore, the stiffness in the elastic range (Young’s modulus) was compared to the onset of permanent deformation (static yield stress) in Figure 7 a key insight for design and material selection. The findings indicate a nearly linear correlation between yield stress and Young’s modulus in the cementitious materials, with a stronger correlation observed in compositions containing citric acid. This trend suggests that, within the investigated system, retardation may improve the balance between early-age stiffness development and resistance to deformation. The results show a strong correlation between both variables and therefore agree with Valentini et al. (2014), who found a significant relationship between Young’s modulus and load-bearing capacity. The direct correlation further aligns with Sanahuja et al. (2007), who described Young’s modulus as increasing with the degree of hydration, strongly affecting the yield stress as well. However, while a relationship between yield stress and Young’s Modulus exists, Wolfs et al. (2018) emphasized that there are early hydration products that impact the velocity of sound waves, but not the strength and stiffening behavior (De Belie et al., 2005), thereby indicating that this relationship is not universal.

FIGURE 7

5.1.3 Mechanical strength of cast samples

Mechanical strength tests were conducted on cast specimens at 1 d and 28 d, see Figure 8. For all compositions, flexural and compressive strength approximately doubled between 1 d and 28 d, mixtures containing retarder exhibiting superior strength. The flexural strength ranged from 1.3 to 1.5 N/mm² after 1 d, and from 2.8 to 3.3 N/mm² after 28 d. The compressive strength ranged from 21 to 27 N/mm² (1 d) and from 52 to 61 N/mm² (28 d). The mortar without citric acid (CA0) consistently exhibited the lowest strength. In contrast, CA5.0 and CA7.5 reached the highest strength after 1 d. After 28 d, all retarded compositions showed a similar mechanical performance, with CA2.5 having slightly lower compressive strength after 28 d compared to CA5.0 and CA7.5.

FIGURE 8

These observations mainly align with previous studies. Mi et al. (2023) concluded that the rapid formation of hydration products (in this study ettringite in CA0) can lead to expansive stresses. If the hydration products form too quickly and in large amounts, they may induce cracks due to expansion and lead to a porous structure. Further, the effect of citric acid on the hydration process and mechanical properties was investigated by (Wen et al., 2014; Wang et al., 2025; Wang et al., 2024), who concluded that the increases in both compressive and flexural strengths at later ages can likely be attributed to a more refined microstructure (Wen et al., 2014) and reduction in early hydration heat release (Wang et al., 2025; Wang et al., 2024). This was supported by Purnomo et al. (2019), who showed that moderate citric acid concentrations promote a denser development of hydration phases, thereby improving long-term strength. While they also reported that excessive dosage can lead to a decline in strength, other studies, such as Wang et al. (2025), who similarly reported both a higher early-age strength in samples with a high dosage (8%) and also found the later strength (14 d) to be equal for both 4% and 8% dosages of citric. Additionally, Wang et al. (2025) found no strict linear relationship between citric acid concentration and strength. Therefore, further analysis was conducted to quantify any differences between acid concentrations in hardened mortar.

5.1.4 Shrinkage

As shown in Figure 9, the magnitude of shrinkage generally increased with citric acid dosage, from approximately −0.6 to −1.2‰ over 28 d. Except for CA7.5, all mixes initially exhibited a volume increase (expansion) within the first 3–6 d before continuous shrinkage commenced. The control (CA0, dark blue line) expanded by 0.1‰ initially, followed by a secondary expansion peak at 4 d, and a final shrinkage of 0.6‰ at 28 d. Regarding CA2.5 (light blue line), lower data resolution resulted from a different gauge model, though identical molds ensured comparable results. The mortar quickly expanded within the first hours, reaching a much broader peak at about 5 h with an expansion of ≈0.15‰ compared to CA0. Further increasing the dosage to CA5.0 and CA7.5 suppressed the initial expansion; both mixtures then showed nearly identical shrinkage kinetics from day 2 onward and reached final values of approx. 1.2‰ at 28 d. However, due to the mold-supported and foil-covered setup these values cannot be applied to printed specimens, as in 3D-printed specimens, drying and plastic shrinkage likely play a much greater role due to the high surface-area-to-volume ratio and open exposure.

FIGURE 9

As the shrinkage remains active across all samples at 28 d, the recorded values likely reflect an intermediate stage, as the literature often reports a decrease in shrinkage with increasing retardation. For instance, Qin et al. (2024) found that shrinkage in concrete decreases when retarded with citric acid modified chitosan (CAMC) of up to 0.6%, which led to an expansion rather than shrinkage. Xu et al. (2023) concluded that the shrinkage could be reduced by more than a third when retarding the mortar with 2.5%, 5.0% and 7.5% acetic acid over 60 d. Conversely, Chen et al. (2022) also measured reduced shrinkage with an increase in acid concentrations, but the differences between mortars with no acid and 3%, 6% and 9% acid nearly diminished. The higher shrinkage observed in the present study may stem from a delayed water consumption due to strong retardation and from the high water-binder ratio (w/b > 0.7) required for printability, as it is well-established that a higher water content in concrete mixtures directly correlates with increased shrinkage due to the larger evaporable volume (Kristiawan and Aditya, 2015; Neville and Brooks, 2010).

5.1.5 Phase composition analysis using thermogravimetric analysis (TGA)

Thermogravimetric analysis (TGA) and derivative thermogravimetry (DTG) – the first derivative of TGA–revealed that mass loss increased with both specimen age and citric acid dosage, see Figure 9. All compositions exhibited a significant mass loss up to 150 °C, with the DTG minimum at ≈ 80 °C attributed mainly to ettringite and partially C-S-H phases (Möschner et al., 2009; Gruskovnjak et al., 2011). The gradual mass loss up to 600 °C likely corresponds to C-S-H decomposition, consistent with Song et al. (2018) and Scrivener et al. (2016), who used XRD and TGA to show that the decomposition of C-S-H occurs over a wider temperature range. The subsequent mass loss around 650 °C was nearly uniform for all samples, suggesting a non-reactive additive present in equal quantities across all mixes, possibly calcite (Scrivener et al., 2016). These observations, consistent with the previous results, indicate that extended hydration caused by citric acid allows the hydration products to grow more slowly but in greater quantity.

Although all compositions were similar in progression, CA0 showed the lowest mass loss (92.7% at 1 d and 91.9% at 28 d). CA2.5 showed a similar progression of the curve, with a total difference in mass loss of ≈0.8 and 0.9% for 1 d and 28 d samples, mainly due to a greater mass up to 150 °C. CA5.0 and CA7.5 at the age of 1 d are almost identical, but differ more significantly at 28 d between 150 °C and 600 °C, with CA5.0 showing the greatest total mass loss (residual mass 89.6%). That differently retarded samples start to diverge already below 150 °C was also acknowledged by Möschner et al. (2009), who found a lower mass loss up to 100 °C in faster setting Portland cement mortars using TGA and XRD.

The larger differences between 150 °C–600 °C suggest a greater amount of hydration products, likely including ettringite and C-S-H, possibly due to a later C-S-H growth–a behavior also quantified by Chen et al. (2020), who measured a higher water retention of the Portland cement when increasing the retarder (Hydroxypropylmethylcellulose–HPMC) content. However, the results of (Chen et al., 2020) also indicated that water retention increases with higher plastic viscosity, although most retarders rather decrease the viscosity. The degree of hydration and the specific phases formed are further elucidated by the DTG curves, which represent the rate of mass loss. All samples containing citric acid, irrespective of their specific dosage, exhibited a more pronounced deflection peak around 80 °C, suggesting a higher (bound) water content, possibly due to enhanced hydration product formation.

A rather inverse correlation was found between TGA mass loss and shrinkage, which contrasts with the relationship established by Bouasker et al. (2008). Their work demonstrated that a higher degree of hydration typically reduces shrinkage, as more water becomes chemically bound within the hydration products rather than remaining available for evaporation. While the higher acid content in the specimens promoted more extensive hydration, it did not lead to reduced shrinkage. We suspect this is due to the high initial water content of the printable mortar; the retarded samples likely allow for increased hydration product formation while still maintaining a significant volume of free water that remains susceptible to evaporation. However, the mass loss trends are consistent with the measured mechanical strength, as the increased quantity of hydration products in retarded samples directly correlates with enhanced mechanical strength observed in 5.1.3 and also by Qi et al. (2024).

5.1.6 Phase composition analysis using scanning electron microscopy (SEM)

Scanning electron microscopy (SEM) analysis of cast CA0, CA2.5 and CA7.5 (Figure 10) revealed significant microstructural changes depending on age and citric acid concentration. The observed hydration products are shown schematically in Figure 11, with identification based on the work of Scrivener et al. (2016). After 1 d, CA0 exhibited ettringite needles and weakly developed, unevenly distributed C-S-H phases with an overall open microstructure. In contrast, CA2.5 displayed a denser matrix with fewer unhydrated cement grains, finer pores and more homogeneous C-S-H coverage over the ettringite. The CA7.5 sample was the densest at 1 d, characterized by minimal porosity and very fine, evenly distributed C-S-H and portlandite phases. By 28 d, CA0 maintained a relatively open structure, as rapid C-A-H and monosulfate formation likely limited C-S-H growth due to reduced water and calcium availability (Goedeke, 2018). Conversely, CA2.5 and CA7.5 developed much denser microstructures with much less visible ettringite, likely because it is overgrown by the slower reacting C-S-H or because it has partly transformed into other phases over time.

FIGURE 10

FIGURE 11

Comparing the results, the open microstructure of CA0 after both 1 d and 28 d suggests that rapid ettringite precipitation and C-A-H development (40% calcium aluminate cement in the binder, 10% total) may have bound water limiting later C-S-H formation. This is evidenced by the chemical formula of ettringite (C₃A·3CaSO₄·32H₂O), given that 1 mol of Al₂O₃ binds to 32 mol of water (Bullard et al., 2011). Scrivener et al. (2019) used microscopy, calorimetry, Rietveld analysis and porosity measurements to demonstrate that fast-setting cement reaches a high degree of hydration early, leaving less unhydrated clinker and water-filled pores, necessary for later reactions and a dense structure. This aligns with the findings of Ghafoori et al. (2025), who showed that rapid early hydration in faster reacting cement types leads to an incomplete matrix with significant air voids.

In contrast, the CA7.5 sample initially exhibited the densest microstructure, though by 28 d, CA2.5 appeared slightly more compact. While SEM images suggest a similar density, TGA measurements (Section 5.1.4) confirmed a higher degree of hydration for the CA7.5 sample, indicating that the visual density in SEM may only represent specific local areas. Möschner et al. (2009) confirmed that retarded systems delay early ettringite formation, enabling a more balanced growth of other, later hydration products (like C-S-H and CH). Using XRD and calorimetry measurements, they showed that moderate to high citric acid levels eventually lead to greater quantities of ettringite than in acid-free systems.

That the formation of AFt phases, such as ettringite, becomes more pronounced in later stages of hydration in retarded samples was further confirmed by Ma et al. (2011). This aligns with findings by Purnomo et al. (2019), who used strength measurements and concluded that moderate citric acid concentrations likely lead to the densest development of hydration phases. Wang et al. (2025), who also tested higher contents of citric acid (up to 8%), found no evidence of a critical dosage threshold beyond which mechanical strength and structural integrity decrease. Instead, their results indicate a continuous retardation of hydration alongside a progressive densification of the microstructure with increasing citric acid. The findings in this study represent an intermediate position between these studies. While the SEM analysis confirmed a clear densification when retarding with citric acid, when considering the results in conjunction with TGA, shrinkage data and mechanical performance, a dosage of 5.0% may have the greatest impact and lead to the densest overall matrix development.

5.2 Conclusions of laboratory measurements

The laboratory investigations provided critical insights for selecting appropriate mortar compositions for the subsequent 3D printing experiments. Analysis of the yield stress evolution via penetrometer and the Young’s modulus development through ultrasonic tests showed a clear correlation between citric acid concentration and retardation, see Figure 7. Compositions with lower citric acid contents exhibited excessively rapid structural build-up, reaching 250 kPa within 30 min. This processing window rendered CA0 and CA2.5 impractical for batch-based printing. In contrast, the mixtures containing 5.0% and 7.5% citric acid showed prolonged dormant periods and a more gradual increase in stiffness. The reduced structural build-up rate provides the necessary workability window for extrusion and layer deposition, as previous studies have shown that increasing citric acid concentrations leads to extended induction periods and a broader, less intense hydration peak (Singh et al., 1986; Ramachandran and Lowery, 1992; Möschner et al., 2009; Huang et al., 2019). These findings align with the understanding that without sufficient retarder dosage, the immediate hydration of cement phases leads to rapid stiffening (Aïtcin, 2016; Fikeni et al., 2024; Khan and Muhammad-Ullah, 2004) and loss of workability, as described in Section 1, where citric acid is noted to prolong the dormant period by adsorbing onto cement particles, preventing immediate hydration (Singh et al., 1986). Notably, the initial stiffening in this system is likely driven by the highly reactive nature of the calcium aluminate cement (CAC). Unlike Portland cement, where early hydration kinetics are primarily governed by tricalcium silicate (C₃S) and tricalcium aluminate (C₃A) (Aïtcin, 2016; Fikeni et al., 2024; Khan and Muhammad-Ullah, 2004), CAC hydration is dominated by the monocalcium aluminate (CA) phase. This phase achieves significant strength within hours through the rapid formation of ettringite (Klaus et al., 2013; Noor et al., 2024), which is the likely cause of the accelerated structural build-up observed in the low-retarder mixtures.

The higher long-term strength of the retarded mixtures is consistent with TGA and SEM observations suggesting more extensive later hydration and a denser microstructure. These results align with literature suggesting that while retardation delays structural build-up, optimal dosages maintain or enhance long-term performance (Zhu et al., 2021; Mendes et al., 2020; Purnomo et al., 2019; Wang et al., 2018) and improve durability by reducing pore volume (Sun H. et al., 2023). This dual benefit of extended workability and superior mechanical performance solidified the decision to utilize 5.0% and 7.5% citric acid mixtures for the subsequent printing experiments, leveraging controlled retardation to optimize both printability and structural integrity.

5.3 Measurements on printed samples

5.3.1 Static yield stress evolution using a mobile penetrometer

Mobile penetrometer results (Figure 12) show a more pronounced increase in yield stress for CA5.0 than for CA7.5, consistent with the laboratory measurements (Figure 5). Within the first 5 min, the yield stress was below the device resolution. After 10 min, the yield stress of CA5.0 was approximately double that of CA7.5 at all measurement points. At 15 min, CA5.0 reached 40 kPa compared to less than 20 kPa for CA7.5; after 30 min, the values increased to > 120 kPa and ≈75 kPa, respectively. While these data points do not definitively distinguish between linear or exponential growth, the latter appears more likely given the previous laboratory-based results. The data clearly demonstrate that citric acid delays early hydration, resulting in a slower yield stress development and prolonged workability. This retardation effect is supported by Pan et al. (2021), who noted that citric acid prolongs the dormant period by adsorbing onto cement surfaces, forming a protective film that hinders initial reactions.

FIGURE 12

The observed trends also align with Purnomo et al. (2019), who reported that citric acid extends setting times, thereby maintaining workability. Moreover, the slower yield stress increase in CA7.5 relative to CA5.0 concurs with Wang et al. (2025), who attributed this dosage-dependent retardation to citrate adsorption, hindering the nucleation and growth of hydration products. The use of mobile penetration to monitor structural build-up is consistent with Wangler et al. (2022) and Lootens et al. (2009), who related such measurements to colloidal interaction thresholds, hydration product formation (Wangler et al., 2022) and setting time (Lootens et al., 2009). Furthermore, Pott and Stephan (2021) confirmed that penetration testing effectively quantifies early structural build-up, which is crucial for ensuring the stability of subsequently deposited filaments.

Although the mobile penetrometer data is not directly comparable to laboratory values in magnitude, the overall progression aligns with the results in Section 5.1.1. The higher yield stress measured on pumped and printed samples is in accordance with Rehmann et al. (2024), who observed increased penetration forces in processed versus unprocessed concrete. However, as noted by Rehmann et al. (2024) and Lootens et al. (2009), absolute values and progressions vary significantly with the measurement method and probe geometry (e.g., the sphere used in the laboratory versus the cone used during printing).

5.3.2 Dynamic yield stress determined via slug test

Using the slug test, the dynamic yield stress was determined from the slug mass. This in-line measurement captures the minimum stress influenced by printing parameters and required to maintain material flow immediately after extrusion–distinguishing it from the static yield stress measured by the penetrometer. Consistent with previous measurements, a lower yield stress was observed for the stronger retarded composition, visible in Figure 12. CA5.0 exhibited an average slug mass of 77.3 g with the shape classified as viscous-plastic. In comparison, CA7.5 had a lower mass (56.7 g) and a similar viscous-plastic appearance. Thus, the slug mass is 1.37 times higher for CA5.0 than for CA7.5.

Applying Equation 4 from Ducoulombier et al. (2020), the dynamic yield stress was calculated as 0.71 kPa (CA5.0) and 0.51 kPa (CA7.5). This 1.39-fold difference between the two compositions (Figure 12) is consistent with the laboratory data: higher acid concentrations slow the hydration reaction, resulting in a reduced structural build-up and, consequently, shorter, lighter slugs with lower yield stress. These trends indicate that citric acid acts both as a hydration retarder and as a rheology-modifying admixture, where higher dosages delay solid structure formation to yield a more easily extrudable material. These lower dynamic yield stresses are considered favorable for 3D concrete printing, as they ensure sufficient flowability for extrusion while maintaining shape stability, as discussed by Rehman and Kim (2021). In a summary of printable concretes, Rehman and Kim (2021) noted that most materials range between 0.1 and 0.7 kPa. Both CA5.0 and CA7.5 fall approximately within this range, whereas, based on the laboratory results, CA2.5 or CA0 would likely exceed the upper threshold.

The finding that the yield stress decreases with a higher retarder dosage aligns with the literature. Wang et al. (2025) observed that citric acid reduces yield stress and enhances workability by forming a protective layer on the clinker surface, inhibiting early hydration. Similarly, using slump tests and setting-time measurements, Purnomo et al. (2019) showed that increasing citric acid content prolongs hardening and increases workability, directly linked to its yield stress. The observed reduction in yield stress for retarded binders in this study is further supported by calorimetric studies of Velazco et al. (2014), which demonstrated that citric acid in calcium sulphoaluminate cement significantly reduces the heat release rate, indicating slower structure formation, corresponding to a lower yield stress. The mechanism was detailed by Singh et al. (1986), who reported that citric acid retards hydration by decreasing the mobility of charge carriers in cementitious systems, a conclusion supported by zeta potential measurements, showing the adsorption of citrate ions onto cement grains.

5.3.3 Mechanical strength of printed samples and impact of cold joints

Mechanical strength tests (Figure 13) revealed significant variations between cast and printed samples, alongside differences dictated by citric acid dosage and the inclusion of cold joints of 12 min. As the material of takes 8 min between mixer and subsequent deposition, a cold joint of 12 min was chosen (according to a material age of 20 min, as at this time, a measurable difference in structural build-up is present, with the yield stress of CA5.0 being almost double that of CA7.5 (Figure 5). While cast CA5.0 and CA7.5 performed similarly in both flexural strength (≈3.2 MPa), and compressive strength (≈60 MPa), printed specimens showed markedly lower strengths than their cast counterparts, with compressive-strength reductions of more than 10 MPa in all cases, likely due to process-related effects such as the lack of compaction, the layered build-up, and stiffness-related limitations on interlayer adaptation (Perrot et al., 2016; Wangler et al., 2019; Rehman and Kim, 2021; Sun X. et al., 2023). In consecutively printed samples, CA7.5 specimens consistently outperformed CA5.0 and showed smaller cast-to-print reductions for CA7.5 (25% in flexural and 22% in compression) than for CA5.0 (31% and 33%, respectively). With the implementation of a cold joint, this advantage increased to about 0.3 MPa in flexural and 9 MPa in compression tests. The additional strength loss caused by the cold joint was greater for CA5.0 (14% and 10%) than for CA7.5 (10% and 4%), indicating that the mixture with lower citric acid content was likely more affected by its faster hydration and earlier stiffening, which reduced geometric adaptation to the previously printed filament and thereby interlayer bonding. This becomes evident when comparing it to the yield stress shown in Figures 5, 12, that both show a much more advanced build-up stage for CA5.0 than for CA7.5 at roughly 20 min (material takes 8 min from mixer to extrusion and 12 min cold joint). Accordingly, the interface was formed after measurable stiffening had already started, consistent with the greater cold-joint sensitivity.

FIGURE 13

CA7.5’s higher performance aligns with Purnomo et al. (2019) and Xu et al. (2023), who found that optimal citric acid dosages enhance strength by refining hydration and microstructure. The broader reduction in printed specimen strength relative to cast equivalents aligns with Rehman and Kim (2021) and Mechtcherine et al. (2025), who attributed the observed loss of up to 25% to the anisotropy due to the layered structure. Furthermore, Rehman and Kim (2021) as well as Wolfs et al. (2025) noted that flexural strength varies based on loading orientation and cold joints. The larger reduction observed in this study, reaching up to 40% for the printed CA5.0 specimens with a cold joint relative to the cast CA5.0 specimens, may be related to the rapid hydration and stiffening of the material, as Reiter et al. (2018) and Roussel and Cussigh (2008) stated that materials with higher viscosity and yield stress resist deformation, limiting the capacity to intermix with subsequent layers, likely reducing interlayer bonding and thereby lowering strength. Han et al. (Han et al., 2014) investigated how the impact of reaction time of cementitious materials influences the porosity and found that a decreased fluidity (due to accelerated hydration) can increase internal pores, leading to reduced later strength. That “for extruded specimens the average air void content increases with delay time” was also shown by Kloft et al. (2020), and supports the hypothesis that the mechanical performance might have decreased due to a possible increase in air-void content as workability decreases.

5.3.4 Geometrical deformation via 3D scanning

Geometric deformation analysis shown in Figure 14 demonstrated a clear relationship between citric acid content, yield stress, Young’s modulus and the stability of printed structures. This non-invasive, in-situ measurement technique provides critical insight into shape stability without altering the material in its fresh or hardened state. The average regression angle of CA5.0 was ≈3.2°, and thus notably smaller than the ≈4.5° deviating angle of CA7.5.

FIGURE 14

The increased deformation with a higher citric acid content is consistent with Singh et al. (1986), who observed that citric acid delays the structural development by retarding hydration, prolonging the fluid-like state. Combined with results by Perrot et al. (2016), who measured the material’s deformation when loaded at different states of hydration and demonstrated that the structural build-up rate directly affects the stability of cement-based materials, this suggests that increased delay in hydration due to citric acid can be expected to result in greater deformation. The similarity between the ratio of geometric deviation (1.43) and dynamic yield stress (1.39) corresponds with Ducoulombier et al. (2020), who measured yield stress (via slug test), also expecting a direct correlation to the deformation during printing. Similarly, Pott and Stephan (2021) assumed correlations between yield stress and printing stability, noting that the force required to measure the yield stress reflects the structural development of hydration products. Jayathilakage et al. (2020) directly demonstrated this correlation by linking the induced stress and eventual failure to the horizontal displacement and yield strength of the bottom layer.

This difference in deformation also reflects the material’s varying Young’s modulus (Section 5.1.2). While both compositions exhibited sufficient early-age stiffness to bear the cumulative load of subsequent layers, a smaller Young’s modulus results in proportionally larger geometric deformation due to reduced resistance to self-weight. Since the Young’s modulus of CA5.0 is higher than that of CA7.5 immediately after printing and becomes more than three times higher within the first three hours of hydration (Figure 6), a correlation between geometric deformation and Young’s modulus is plausible. To mitigate geometric deformations in the future, incorporating reinforcements such as optimal-length glass fibers can enhance early-age shape stability by reducing the flowability as shown by Li et al. (2025).

5.4 Conclusions on printing measurements

The 3D printing experiments provided a direct validation and application of the material characteristics gained from lab-based tests, affirming their predictive value. Qualitatively, the yield stress measurements obtained in the laboratory and those conducted during printing with the mobile penetrometer exhibited comparable exponential trends. This agreement affirms the central research question of whether laboratory testing can qualitatively predict printability. This is evidenced by the rapid structural build-up observed in the lab-based experiments, which exceeded a yield stress of 250 kPa in under 30 min, rendering CA0 and CA2.5 impractical for batch printing. The evolution of Young’s modulus in the lab further serves as an indicator of the deformation of the material during printing. A higher Young’s modulus and measured dynamic yield stress translated to reduced geometric deformation in the printed objects, highlighting the importance of early-age stiffness for buildability.

In contrast, direct quantitative comparability between the mechanical strengths of cast and printed specimens was not established, likely because of process-specific effects during printing. These include the complex interplay of processability (e.g., air void incorporation), interlayer compaction (Perrot et al., 2016; Wangler et al., 2019; Rehman and Kim, 2021; Sun X. et al., 2023), and the potential destruction of early hydration products during mixing, pumping, and extrusion, which may hinder the formation of a dense, interlocking structure. The transferability of laboratory results to the printing process should be interpreted primarily on a qualitative basis, given that each print geometry was produced once per composition. In addition to differences in test method and probe geometry, the printing trials were conducted at higher temperature and lower relative humidity than the laboratory tests, which likely accelerated hydration and altered surface moisture conditions (Wang et al., 2026). However, the beneficial effects of citric acid–namely, a prolonged workability while maintaining or even enhancing long-term strength–were consistent across lab and print results and with the literature (Zhu et al., 2021; Mendes et al., 2020; Purnomo et al., 2019; Wang et al., 2018). Compressive strength results further aligned with TGA and SEM, which confirmed the presence of a greater quantity of hydration products in samples containing citric acid. SEM further suggested denser microstructures with acid addition, while the reference mixture without citric acid showed limited C-S-H formation, which is consistent with its lower strength. This is consistent with literature suggesting that citric acid promotes a denser microstructure and reduced pore volume (Sun H. et al., 2023). Overall, CA5.0 appears to offer the most favorable overall balance, including open processing window, printability, deformation resistance, microstructural density and long-term strength, though shrinkage increases with higher citric acid content. Collectively, although absolute values differ between in-line and off-line measurements, the stability of these trends enables reliable lab-based quantification and characterization of compositions for 3D concrete printing.

The transferability of laboratory results to the printing process should be interpreted primarily on a qualitative basis. In addition to differences in test method and probe geometry, the printing trials were conducted at higher temperature and lower relative humidity than the laboratory tests, which likely accelerated hydration and altered surface moisture conditions. Moreover, each print configuration was produced once only. Consequently, the print-scale results should be regarded as indicative trends that support the laboratory findings rather than as statistically validated process correlations.

6 Conclusion

Overall, the results show that citric acid is suitable for the targeted modification of open time, structural build-up, and strength development in 3DCP, provided that its dosage is balanced against geometric stability. Future work should include in situ-shrinkage measurements and monitoring of the long-term hydration process beyond 28 d, as shrinkage remains ongoing, and SEM suggests that very high citric acid dosages may induce microstructural destabilization. Implementing shrinkage monitoring on printed specimens–e.g., with Digital Image Correlation (DIC) as done by Cuevas et al. (2023) – is essential to mitigate potential cracking on-site. Additionally, the interaction between citric acid and accelerators relevant for 3DCP should be studied to enable targeted adjustment of rheology and setting behavior post-extrusion.

References

• 1

  AiC.YuA.LiuC.LiT. (2024). Analysis of physical characteristics and mechanism of retarder to stratified cemented backfill. Sci. Reports14, 13711. 10.1038/s41598-024-64507-5

  • CrossRef
  • Google Scholar

• 2

  AïtcinP.-C. (2016). “Retarders,” in Science and Technology of Concrete Admixtures (Elsevier), 395–404.

  • Google Scholar
  • View reference in article

• 3

  BlaskO.JohnE.StephanD. (2000). “Construction chemistry,” in Ullmann's Encyclopedia of Industrial Chemistry (Wiley), 1–80.

  • Google Scholar
  • View reference in article

• 4

  BosF.WolfsR.AhmedZ.SaletT. (2016). Additive manufacturing of concrete in construction: potentials and challenges of 3D concrete printing. Virtual Phys. Prototyp.11, 209–225. 10.1080/17452759.2016.1209867

  • CrossRef
  • Google Scholar
  • View reference in article

• 5

  BouaskerM.MounangaP.TurcryP.LoukiliA.KhelidjA. (2008). Chemical shrinkage of cement pastes and mortars at very early age: effect of limestone filler and granular inclusions. Cem. Concr. Compos.30, 13–22. 10.1016/j.cemconcomp.2007.06.004

  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  BullardJ. W.JenningsH. M.LivingstonR. A.NonatA.SchererG. W.SchweitzerJ. S.et al (2011). Mechanisms of cement hydration. Cem. Concr. Res.41, 1208–1223. 10.1016/j.cemconres.2010.09.011

  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  ByrdN.LloydJ. R.SmallJ. S.TaylorF.BagshawH.BoothmanC.et al (2021). Microbial degradation of citric acid in low level radioactive waste disposal: impact on biomineralization reactions. Front. Microbiology12, 565855. 10.3389/fmicb.2021.565855

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  ChenN.WangP.ZhaoL.ZhangG. (2020). Water retention mechanism of HPMC in cement mortar. Mater. Basel, Switz.13, 2918. 10.3390/ma13132918

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 9

  ChenL.JiangL.JinW.GuoM. Z.NiuY.ChuH.et al (2022). Effect of citrate retarders on the expansion property and hydration behavior of light-burned magnesia cement-based materials. J. Build. Eng.61, 105265. 10.1016/j.jobe.2022.105265

  • CrossRef
  • Google Scholar
  • View reference in article

• 10

  CuevasK.StrzałkowskiJ.KimJ. S.EhmC.GlotzT.ChouganM.et al (2023). Towards development of sustainable lightweight 3D printed wall building envelopes – experimental and numerical studies. Case Stud. Constr. Mater.18, e01945. 10.1016/j.cscm.2023.e01945

  • CrossRef
  • Google Scholar
  • View reference in article

• 11

  DaakeH. vonStephanD. (2016). Combination of Isothermal Heat Flow Calorimetry and in situ Ultrasonic Testing to Monitor the Early Cement Hydration in Presence of Chemical Admixtures. Dar es Salaam.

  • Google Scholar
  • View reference in article

• 12

  De BelieN.GrosseC. U.KurzJ.ReinhardtH. W. (2005). Ultrasound monitoring of the influence of different accelerating admixtures and cement types for shotcrete on setting and hardening behaviour. Cem. Concr. Res.11, 2087–2094. 10.1016/j.cemconres.2005.03.011

  • CrossRef
  • Google Scholar
  • View reference in article

• 13

  DesmetB.AtitungK. C.Abril SanchezM. A.VantommeJ.FeysD.RobeystN.et al (2011). Monitoring the early-age hydration of self-compacting concrete using ultrasonic p-wave transmission and isothermal calorimetry. Mater. Struct.44, 1537–1558. 10.1617/s11527-011-9717-x

  • CrossRef
  • Google Scholar
  • View reference in article

• 14

  DucoulombierN.CarneauP.MesnilR.DemontL.CaronJ. F.RousselN. (2020). “the slug test”: inline assessment of yield stress for extrusion-based additive manufacturing”. Second RILEM International Conference on Concrete and Digital Fabrication. Editors BosF. P.LucasS. S.WolfsR. J.SaletT. A. (Cham: Springer International Publishing), 28, 216–224. 10.1007/978-3-030-49916-7_22

  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  FikeniK. G.PangX.YuY.XiaX.SunF.WangH.et al (2024). An overview of oil well cement retarders and the retardation mechanisms. Geoenergy Sci. Eng.241, 213116. 10.1016/j.geoen.2024.213116

  • CrossRef
  • Google Scholar
  • View reference in article

• 16

  GhafooriN.MalerM. O.NajimiM.HasnatA.GbadamosiA. (2025). Role of cement type on properties of high early-strength concrete. J. Compos. Sci.9, 3. 10.3390/jcs9010003

  • CrossRef
  • Google Scholar
  • View reference in article

• 17

  GoedekeH. K. (2018). Lösender Angriff Auf Zementgebundene Baustoffe – Veränderungen Der Porenstruktur Und Folgen Für Transport-Und Korrosionsprozesse. Doctoral Thesis. TUHH Universitätsbibliothek.

  • Google Scholar
  • View reference in article

• 18

  GruskovnjakA.LothenbachB.WinnefeldF.MünchB.FigiR.KoS. C.et al (2011). Quantification of hydration phases in supersulfated cements: review and new approaches. Adv. Cem. Res.23, 265–275. 10.1680/adcr.2011.23.6.265

  • CrossRef
  • Google Scholar
  • View reference in article

• 19

  HanJ.WangK.ShiJ.WangY. (2014). Influence of sodium aluminate on cement hydration and concrete properties. Constr. Build. Mater.64, 342–349. 10.1016/j.conbuildmat.2014.04.089

  • CrossRef
  • Google Scholar
  • View reference in article

• 20

  HuangT.LiB.YuanQ.ShiZ.XieY.ShiC. (2019). Rheological behavior of Portland clinker-calcium sulphoaluminate clinker-anhydrite ternary blend. Cem. Concr. Compos.104, 103403. 10.1016/j.cemconcomp.2019.103403

  • CrossRef
  • Google Scholar
  • View reference in article

• 21

  JayathilakageR.RajeevP.SanjayanJ. (2020). Yield stress criteria to assess the buildability of 3D concrete printing. Constr. Build. Mater.240, 117989. 10.1016/j.conbuildmat.2019.117989

  • CrossRef
  • Google Scholar
  • View reference in article

• 22

  KhanB.Muhammad-UllahM.-U. (2004). Effect of a retarding admixture on the setting time of cement pastes in hot weather. Eng15, 63–79. 10.4197/Eng.15-1.5

  • CrossRef
  • Google Scholar
  • View reference in article

• 23

  KlausS. R.NeubauerJ.Goetz-NeunhoefferF. (2013). Hydration kinetics of CA2 and CA—Investigations performed on a synthetic calcium aluminate cement. Cem. Concr. Res.43, 62–69. 10.1016/j.cemconres.2012.09.005

  • CrossRef
  • Google Scholar
  • View reference in article

• 24

  KloftH.KraussH. W.HackN.HerrmannE.NeudeckerS.VaradyP. A.et al (2020). Influence of process parameters on the interlayer bond strength of concrete elements additive manufactured by Shotcrete 3D Printing (SC3DP). Cem. Concr. Res.134, 106078. 10.1016/j.cemconres.2020.106078

  • CrossRef
  • Google Scholar
  • View reference in article

• 25

  KristiawanS. A.AdityaM. T. M. (2015). Effect of high volume fly ash on shrinkage of self-compacting concrete. Procedia Eng.125, 705–712. 10.1016/j.proeng.2015.11.110

  • CrossRef
  • Google Scholar
  • View reference in article

• 26

  LeusmannT.KraussH. W.BudelmannH. (2017). “The hardening process of cement based materials observed by calorimetry and ultrasonic tests,” in 2nd International RILEM/COST Conference on Early Age Cracking and Serviceability in Cement-based Materials and Structures (EAC2), 12–14.

  • Google Scholar
  • View reference in article

• 27

  LiL. G.XiaoB.KouS. (2025). Influences of fiber length on the printability and strength of glass fiber-reinforced 3D-Printed mortar. J. Intell. Const.3, 1–14. 10.26599/JIC.2025.9180093

  • CrossRef
  • Google Scholar

• 28

  LootensD.JoussetP.MartinieL.RousselN.FlattR. (2009). Yield stress during setting of cement pastes from penetration tests. Cem. Concr. Res.39, 401–408. 10.1016/j.cemconres.2009.01.012

  • CrossRef
  • Google Scholar

• 29

  MaB. G.XiaoJ.TanH. B. (2011). Effect of citric acid on cement hydration. AMR393-395, 49–53. 10.4028/www.scientific.net/AMR.393-395.49

  • CrossRef
  • Google Scholar
  • View reference in article

• 30

  MantellatoS.EberhardtA. B.FlattR. J. (2016). “Formulation of commercial products,” in Science and Technology of Concrete Admixtures (Elsevier), 343–349.

  • Google Scholar
  • View reference in article

• 31

  MechtcherineV.BosF.PerrotA.da SilvaW. L.NerellaV.FataeiS.et al (2020). Extrusion-based additive manufacturing with cement-based materials – production steps, processes, and their underlying physics: a review. Cem. Concr. Res.132, 106037. 10.1016/j.cemconres.2020.106037

  • CrossRef
  • Google Scholar
  • View reference in article

• 32

  MechtcherineV.MuthukrishnanS.Robens-RadermacherA.WolfsR.VersteegeJ.MennaC.et al (2025). Mechanical properties of 3D printed concrete: a RILEM 304-ADC interlaboratory study – compressive strength and modulus of elasticity. Mater. Struct.58, 181. 10.1617/s11527-025-02688-9

  • CrossRef
  • Google Scholar
  • View reference in article

• 33

  MendesB. C.AlvarengaR. d. C. S. S.FassoniD. P.PedrotiL. G.de AzevedoA. R. G.PortoA. B.et al (2020). Citric acid effect in high early type Portland cement pastes and mortars. MSF1012, 284–288. 10.4028/www.scientific.net/MSF.1012.284

  • CrossRef
  • Google Scholar
  • View reference in article

• 34

  MiT.YangE. H.UnluerC. (2023). Investigation of the properties of reactive MgO-based cements and their effect on performance. Cem. Concr. Compos.138, 104984. 10.1016/j.cemconcomp.2023.104984

  • CrossRef
  • Google Scholar
  • View reference in article

• 35

  MöschnerG.LothenbachB.FigiR.KretzschmarR. (2009). Influence of citric acid on the hydration of Portland cement. Cem. Concr. Res.39, 275–282. 10.1016/j.cemconres.2009.01.005

  • CrossRef
  • Google Scholar
  • View reference in article

• 36

  NangareS.VisputeY.TadeR.DugamS.PatilP. (2021). Pharmaceutical applications of citric acid. Futur J. Pharm. Sci.7, 54. 10.1186/s43094-021-00203-9

  • CrossRef
  • Google Scholar
  • View reference in article

• 37

  Narayan SwamyR. (1971). Dynamic Poisson's ratio of portland cement paste, mortar and concrete. Cem. Concr. Res.1, 559–583. 10.1016/0008-8846(71)90060-3

  • CrossRef
  • Google Scholar
  • View reference in article

• 38

  NevilleA. M.BrooksJ. J. (2010). Concrete Technology. 2nd edn.Harlow: Prentice Hall.

  • Google Scholar
  • View reference in article

• 39

  NguyenH.KuntherW.GijbelsK.SamynP.CarvelliV.IllikainenM.et al (2021). On the retardation mechanisms of citric acid in ettringite-based binders. Cem. Concr. Res.140, 106315. 10.1016/j.cemconres.2020.106315

  • CrossRef
  • Google Scholar
  • View reference in article

• 40

  NoorL.RostP.KirchbergerI.Goetz-NeunhoefferF.IdekerJ. H. (2024). Hygrothermal stability of ettringite in blended systems with CAC-OPC-CŠ. Cem. Concr. Res.184, 107590. 10.1016/j.cemconres.2024.107590

  • CrossRef
  • Google Scholar
  • View reference in article

• 41

  PanJ.WangL.WangP.WangH.CaoK.ShenW. (2021). “Dormancy” and “Awakening” method used for fresh mortar waste recycling and reuse. Adv. Civ. Eng.2021, 8896277. 10.1155/2021/8896277

  • CrossRef
  • Google Scholar
  • View reference in article

• 42

  PerrotA.RangeardD.PierreA. (2016). Structural built-up of cement-based materials used for 3D-printing extrusion techniques. Mater. Struct.49, 1213–1220. 10.1617/s11527-015-0571-0

  • CrossRef
  • Google Scholar
  • View reference in article

• 43

  PottU.StephanD. (2021). Penetration test as a fast method to determine yield stress and structural build-up for 3D printing of cementitious materials. Cem. Concr. Compos.121, 104066. 10.1016/j.cemconcomp.2021.104066

  • CrossRef
  • Google Scholar
  • View reference in article

• 44

  PottU.JakobC.WolfJ.StephanD. (2023). Comparison of physical and physicochemical methods for 3D printing application with the focus on the unconfined uniaxial compression test. Constr. Build. Mater.395, 132260. 10.1016/j.conbuildmat.2023.132260

  • CrossRef
  • Google Scholar
  • View reference in article

• 45

  PurnomoJ.SumarniS.SaputroI. N. (2019). Effect of citric acid on setting-time and compressive strength of concrete. IOP Conf. Ser. Mater. Sci. Eng.578, 012077. 10.1088/1757-899X/578/1/012077

  • CrossRef
  • Google Scholar

• 46

  QiG.ZhangQ.SunZ. (2024). Mechanical properties and hydration mechanism of super-sulfated cement prepared with ordinary Portland cement, carbide slag, and sodium silicate. Front. Mater.11, 1406045. 10.3389/fmats.2024.1406045

  • CrossRef
  • Google Scholar

• 47

  QinZ.WuJ.HeiZ.WangL.LeiD.LiuK.et al (2024). Study on the effect of citric acid-modified chitosan on the mechanical properties, shrinkage properties, and durability of concrete. Mater. Basel, Switz.17, 2053. 10.3390/ma17092053

  • CrossRef
  • Google Scholar

• 48

  RamachandranV. S.LoweryM. S. (1992). Conduction calorimetric investigation of the effect of retarders on the hydration of Portland cement. Thermochim. Acta195, 373–387. 10.1016/0040-6031(92)80081-7

  • CrossRef
  • Google Scholar

• 49

  RehmanA. U.KimJ.-H. (2021). 3D concrete printing: a systematic review of rheology, mix designs, mechanical, microstructural, and durability characteristics. Mater. Basel, Switz.14, 3800. 10.3390/ma14143800

  • CrossRef
  • Google Scholar

• 50

  RehmannA. U.KimI. G.KimJ. H. (2024). “Exploring the impact of the pumping process on the structural Build-up in the context of 3D concrete printing,” in Digital Concrete 2024 - Supplementary Proceedings. 10.24355/dbbs.084-202408151202-0

  • CrossRef
  • Google Scholar

• 51

  ReiterL. (2019). “Structural build-up for digital fabrication with concrete – materials, methods and processes”. ETH Zürich.

  • Google Scholar

• 52

  ReiterL.WanglerT.RousselN.FlattR. J. (2018). The role of early age structural build-up in digital fabrication with concrete. Cem. Concr. Res.112, 86–95. 10.1016/j.cemconres.2018.05.011

  • CrossRef
  • Google Scholar

• 53

  RousselN.CussighF. (2008). Distinct-layer casting of SCC: the mechanical consequences of thixotropy. Cem. Concr. Res.38, 624–632. 10.1016/j.cemconres.2007.09.023

  • CrossRef
  • Google Scholar

• 54

  SanahujaJ.DormieuxL.ChanvillardG. (2007). Modelling elasticity of a hydrating cement paste. Cem. Concr. Res.37, 1427–1439. 10.1016/j.cemconres.2007.07.003

  • CrossRef
  • Google Scholar

• 55

  ScrivenerK.SnellingsR.LothenbachB. (2016). A Practical Guide to Microstructural Analysis of Cementitious Materials. A Spon Press Book (Boca Raton, London, New York: CRC Press Taylor & Francis Group).

  • Google Scholar

• 56

  ScrivenerK.OuziaA.JuillandP.Kunhi MohamedA. (2019). Advances in understanding cement hydration mechanisms. Cem. Concr. Res.124, 105823. 10.1016/j.cemconres.2019.105823

  • CrossRef
  • Google Scholar

• 57

  SinghN. B.SinghA.Prabha SinghS. (1986). Effect of citric acid on the hydration of portland cement. Cem. Concr. Res.16, 911–920. 10.1016/0008-8846(86)90015-3

  • CrossRef
  • Google Scholar

• 58

  SmillieS.GlasserF. P. (1999). Reaction of EDTA, oxalic acid and citric acid with Portland cement. Adv. Cem. Res.11, 97–101. 10.1680/adcr.1999.11.2.97

  • CrossRef
  • Google Scholar

• 59

  SongH.JeongY.BaeS.JunY.YoonS.Eun OhJ. (2018). A study of thermal decomposition of phases in cementitious systems using HT-XRD and TG. Constr. Build. Mater.169, 648–661. 10.1016/j.conbuildmat.2018.03.001

  • CrossRef
  • Google Scholar

• 60

  SorianoE. (2019). The Influence of Citric Acid on Setting Time and Temperature Behavior of Calcium Sulfoaluminate-Belite Cement. Civ. Eng. Undergrad. Honors Theses.

  • Google Scholar

• 61

  SunH.ShiF.WangH. (2023a). Influence of citric acid on the fundamental properties of CO2 cured magnesium oxysulfate paste. Mater. Basel, Switz.16, 1315. 10.3390/ma16031315

  • CrossRef
  • Google Scholar

• 62

  SunX.MazurM.ChengC. T. (2023b). A review of void reduction strategies in material extrusion-based additive manufacturing. Addit. Manuf.67, 103463. 10.1016/j.addma.2023.103463

  • CrossRef
  • Google Scholar

• 63

  ValentiniL.ParisattoM.RussoV.FerrariG.BullardJ. W.AngelR. J.et al (2014). Simulation of the hydration kinetics and elastic moduli of cement mortars by microstructural modelling. Cem. Concr. Compos.52, 54–63. 10.1016/j.cemconcomp.2014.05.005

  • CrossRef
  • Google Scholar

• 64

  VelazcoG.AlmanzaJ. M.CortésD. A.EscobedoJ. C.Escalante-GarciaJ. I. (2014). Effect of citric acid and the hemihydrate amount on the properties of a calcium sulphoaluminate cement. Mater. Constr.64, e036. 10.3989/mc.2014.03513

  • CrossRef
  • Google Scholar

• 65

  VohburgerA.CollinM.BouissonniéA.NicoleauL.GädtT. (2025). Citric, tartaric, and succinic acid effects on C3S dissolution and the nucleation kinetics of C S H and portlandite. Cem. Concr. Res.197, 107964. 10.1016/j.cemconres.2025.107964

  • CrossRef
  • Google Scholar

• 66

  WangN.YuH.BiW.TanY.ZhangN.WuC.et al (2018). Effects of sodium citrate and citric acid on the properties of magnesium oxysulfate cement. Constr. Build. Mater.169, 697–704. 10.1016/j.conbuildmat.2018.02.208

  • CrossRef
  • Google Scholar

• 67

  WangS.XuC.YuS.WuX.JieZ.DaiH. (2019). Citric acid enhances the physical properties, cytocompatibility and osteogenesis of magnesium calcium phosphate cement. J. Mechanical Behavior Biomedical Materials94, 42–50. 10.1016/j.jmbbm.2019.02.026

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 68

  WangL.QinZ.WuJ.ShengG.WangH.LiuK.et al (2024). Effect of citric acid-modified chitosan on hydration regulation and mechanism of composite cementitious material system. Buildings14, 41. 10.3390/buildings14010041

  • CrossRef
  • Google Scholar

• 69

  WangH.LiangS.ZhouX.HouP.ChengX. (2025). Regulating hydration and microstructure development of reactive MgO cement by citric acids. Cem. Concr. Compos.155, 105832. 10.1016/j.cemconcomp.2024.105832

  • CrossRef
  • Google Scholar

• 70

  WangS.ChenY.GongF.FanY.HuangB. T.ZhangH. (2026). Intelligent framework of temperature and humidity field coupling of early-age concrete considering hydration based on physics informed neural network. Int. Commun. Heat Mass Transf.172, 110203. 10.1016/j.icheatmasstransfer.2025.110203

  • CrossRef
  • Google Scholar

• 71

  WanglerT.RousselN.BosF. P.SaletT. A.FlattR. J. (2019). Digital concrete: a review. Cem. Concr. Res.123, 105780. 10.1016/j.cemconres.2019.105780

  • CrossRef
  • Google Scholar

• 72

  WanglerT.FlattR. J.RousselN.PerrotA.SonebiM.WolfsR.et al (2022). “Printable cement-based materials: fresh properties measurements and control”. Digital Fabrication with Cement-based Materials. Editors RousselN.LowkeD. (Cham: Springer International Publishing), 36, 99–136. 10.1007/978-3-030-90535-4_4

  • CrossRef
  • Google Scholar

• 73

  WenJ.YuH.LiY.WuC.DongJ. (2014). Effects of citric acid on hydration process and mechanical properties of thermal decomposed magnesium oxychloride cement. J. Wuhan. Univ. Technol.-Mat. Sci. Ed.29, 114–118. 10.1007/s11595-014-0877-8

  • CrossRef
  • Google Scholar

• 74

  WolfsR.BosF.SaletT. (2018). Correlation between destructive compression tests and non-destructive ultrasonic measurements on early age 3D printed concrete. Constr. Build. Mater.181, 447–454. 10.1016/j.conbuildmat.2018.06.060

  • CrossRef
  • Google Scholar

• 75

  WolfsR.VersteegeJ.SanthanamM.BhattacherjeeS.BosF.Robens-RadermacherA.et al (2025). Mechanical properties of 3D printed concrete: a RILEM TC 304-ADC interlaboratory study — flexural and tensile strength. Mater. Struct.58, 182. 10.1617/s11527-025-02687-w

  • CrossRef
  • Google Scholar

• 76

  XuZ.JiY.ZhaoY.XuS.ZhangZ.GaoF.et al (2023). Effect of acid activation on the mechanical and shrinkage properties of cement-based materials. J. Build. Eng.68, 106184. 10.1016/j.jobe.2023.106184

  • CrossRef
  • Google Scholar

• 77

  YuY.ZhangC.GuT.XuW.ZhangJ.ZhangG.et al (2021). SCnd evaluation of a new type of oil-well cement temperature-resistant retarder. Constr. Build. Mater.302, 124153. 10.1016/j.conbuildmat.2021.124153

  • CrossRef
  • Google Scholar

• 78

  ZhuH.YuK.LiV. C. (2021). Citric acid influence on sprayable calcium sulfoaluminate cement-engineered cementitious composites’ Fresh/hardened properties. 10.14359/51733103

  • CrossRef
  • Google Scholar