Gradient cement pastes with efficient energy dissipation and electromagnetic wave absorption

Introduction

The rapid expansion of 5G networks and the proliferation of wearable electronics have raised concerns over electromagnetic (EM) radiation, which is emerging as an environmental pollutant with potential risks to both sensitive electronic devices and human health^1,2,3,4. Additionally, EM pulses can severely disrupt or even incapacitate vital infrastructure, such as dams, power plants, and military installations⁵. Addressing these challenges requires the development of materials with efficient electromagnetic wave absorption (EMA) capabilities.

Cement, a fundamental material in modern infrastructure, is valued for its abundant raw materials, high compressive strength, fire resistance, and cost-effectiveness^6,7. As the demand for cement is projected to increase by 50% by 2050, driven by the growing deployment for critical infrastructure such as electricity, transportation, and sanitation in developing countries, and upgrading or replacing the ageing infrastructure in developed countries^8,9, integrating EMA performance into cement paste presents a promising solution to mitigate electromagnetic wave (EMW) contamination. However, traditional cement paste lacks effective EMA functionality due to its inherently non-conductive nature, which limits EM energy dissipation primarily to dipole polarization losses¹⁰. Moreover, its brittleness and low toughness hinder its application in advanced structures, such as long-span infrastructure and super-high-rise buildings^11,12,13. In demanding environments, cementitious materials must also exhibit exceptional impact resistance to withstand extreme conditions like collisions and explosions. This challenge is compounded by intrinsic trade-offs between mechanical properties, such as strength and toughness, and between functionality and structural integrity^14,15,16.

Natural materials provide a blueprint for overcoming these challenges, featuring hierarchically ordered structures with compositional or architectural gradients. These gradient structures enable excellent mechanical properties and functionalities while minimizing density^17,18,19. For example, the radial gradient of cellulose fibers in the honeycomb cell matrix of bamboo and palm stem, as well as the compositional gradient at the cementum-dentin junction in human teeth, endow these natural components with excellent toughness^20,21. Such gradients induce nonlinear stress and strain responses, which are accommodated and intercoordinated by the microstructure, potentially activating new dislocation structures²². Additionally, they enhance resistance to wear, impact, and fatigue through interfacial toughening and bonding optimization^23,24. These strategies inspire the creation of artificial functionally gradient materials by manipulating compositional features²⁵.

In cement-based materials, gradient systems have demonstrated significant potential to integrate exceptional mechanical and functional properties. For example, both structural and compositional gradients of graphene oxide have been shown to enhance mechanical strength and durability of cement pastes^26,27. Functionally gradient materials further exhibit superior EMA performance by optimizing impedance matching and facilitating EM energy dissipation^28,29. Among these, carbon-based nanomaterials, especially carbon nanotubes (CNTs), hold great promise for enhancing both the mechanical and EMA properties of cement-based materials due to their lightweight nature, high strength, and excellent electrical conductivity^{5,30,31,32,33,34}. Leveraging biomimetic design principles to create gradient-distributed CNTs in cement paste offers a transformative strategy for integrating mechanical robustness and functionality.

In this work, we present a biomimetic gradient cement paste fabricated by layering cement slurries with varying CNT concentrations on a pre-designed order using a bottom-up assembly scheme. This innovative gradient structure not only enhances mechanical properties, including strength, toughness, and impact resistance, but also boosts EMA performance through optimized impedance matching and multiple internal reflections within the gradient layers. This functionally gradient cement paste introduces a creative paradigm for designing advanced, multifunctional cement-based materials, with promising applications in complex and extreme service environments where mechanical robustness and EMW management are critical.

Results and discussion

Fabrication and characterization of gradient cement paste

The fabrication process of the gradient cement paste is illustrated in Fig. 1a. It begins with the preparation of cement slurries with varying CNT concentrations, while maintaining a consistent water-cement ratio of 0.4. These slurries are sequentially cast into a mold in descending CNT concentrations to create a compositional gradient. The samples are demolded after 24 h and cured in a deionized water tank for 27 days to ensure full development of their properties. This process successfully produces a structure-function integrated cement paste with a compositional gradient (Fig. 1a).

**Fig. 1: Preparation and characterization of gradient cement paste.**

Scanning electron microscopy (SEM) analysis reveals that regions with higher CNT concentrations form a denser matrix (Fig. 1b, c; Supplementary Fig. 1), with hydration products densely packed. This densification is attributed to CNTs acting as nucleation sites, promoting the growth, aggregation, and transformation of hydration products during the early stages of cement hydration^35,36. The interfaces between regions are continuous and well-bonded, with no observable gaps (Supplementary Fig. 2). Raman mapping further corroborates the compositional gradient, showing variations in the G-band peak intensity of CNTs across the interface (Fig. 1d; Supplementary Fig. 3). Additionally, energy dispersive spectroscopy (EDS) analysis reveals an increase in carbon content from 3.58% to 4.09%, aligning well with the CNT addition, while calcium content rises from 54.62% to 58.2%, indicating an increase in hydration products (Supplementary Fig. 4). The consistent distribution of carbon element in both CNT-doped and undoped matrixes suggests a uniform dispersion of CNTs throughout the matrix.

To analyze the effect of microstructure and porosity on macroscopic properties, such as mechanical robustness and EMA, mercury intrusion porosimetry (MIP) was conducted. The homogeneous cement paste with 1% CNTs (Homo-1) exhibits the lowest porosity of 17.88%, compared to 18.30% for cement paste with 0.75% CNTs (Homo-0.75) and 19.82% for cement paste with 0.5% CNTs (Homo-0.5) (Fig. 1e). This reduction in porosity with increased CNT content is attributed to the interactions between CNTs and hydration products, including hydrogen bonding, electrostatic forces, and the pore-filling effect of CNTs. The pore size distribution, analyzed by mercury content evolution and cumulative intrusion curves, shows that capillary pores (0.01–0.1 μm in diameter) are predominant, with fewer macro pores (>1 μm) and gel pores (<0.01 μm) (Supplementary Fig. 5a–c). The gradient cement matrix transitions from a porous structure to a dense one with increasing CNT content (Supplementary Fig. 5d), consistent with SEM observations. While the addition of CNTs reduces total pore volume, it does not significantly alter the pore size distribution pattern. This gradient design with CNT-induced compositional changes can effectively mitigate property mismatches, promising to integrate enhanced mechanical and functional properties into a single cementitious material.

Mechanical properties of gradient cement paste

This section evaluates the mechanical properties of heterogeneous cement paste with a gradient transition layer (HeCG) using both microscopic and macroscopic tests. Nanoindentation results reveal a progressive increase in indentation modulus, from 31.87 GPa in the porous region to 39.83 GPa in the denser region within the HeCG (Fig. 2a–c). Similarly, hardness values rise from 1.25 GPa to 1.72 GPa along the gradient layers (Fig. 2a–c). These buffering changes in microscopic mechanical properties are expected to enhance load transfer and alleviate stress concentrations within the material. Under a load-controlled regime, unrecoverable indentation depths vary across different regions of the HeCG. For instance, at a load of 1 mN, Homo-0.75 experiences an unrecoverable depth of 164.33 nm, between 172.33 nm for Homo-0.5 and 152.50 nm for Homo-1 (Fig. 2d, e). The stiffer surface layer (Homo-1) offers greater resistance to plastic deformation, resulting in smaller triangular indentation craters. The compositional gradient across the layers optimizes the material’s displacement response, enhancing energy dissipation efficiency. Notably, no visible fractures, nanocracks, or ancillary pits are observed around the indentation sites (Fig. 2d), indicating effective resistance to localized damage. This behavior can be attributed to the dense, stiff outer layer’s resistance to deformation and the efficient stress transfer facilitated by the gradient structure, aligning with principles of biomimetic design³⁷.

**Fig. 2: Microscopic mechanical properties.**

To evaluate the effect of the gradient design on macroscopic mechanical properties, a comparative heterogeneous cement paste with the absence of the gradient transition layer (HeCA) was prepared by replacing the middle transition layer containing 0.75% CNTs with one containing 0.5% CNTs. Three-point bending tests reveal that HeCG achieves a flexural strength of 9.68 MPa, surpassing HeCA’s 8.37 MPa and even Homo-1’s 9.10 MPa (Fig. 3a, b). Meanwhile, HeCG also demonstrates higher flexural toughness, reaching 10.31 kJ m⁻³, which is 1.9 times that of HeCA (5.50 kJ m⁻³) and 1.4 times that of Homo-1 (7.32 kJ m⁻³) (Fig. 3b). Numerical simulations under identical bending loads show a more uniform strain distribution in the x direction for HeCG compared to monolithic cement paste (Supplementary Fig. 6), indicating its ability to bear higher loads within the tensile strain limit.

**Fig. 3: Macroscopic mechanical properties.**

Single-edge notched bend (SENB) tests highlight that HeCG achieves higher initiation fracture toughness (K_IC) at initial crack growth and fracture toughness (K_JC) at the end of steady-state crack propagation compared to HeCA and Homo-1 (Fig. 3c). These improvements suggest that the gradient structure effectively activates both intrinsic and extrinsic toughening mechanisms, enhancing resistance to crack initiation and propagation. Unlike Homo-1, which suffers from catastrophic fracture with a nearly linear crack path originating at the bottom midspan, HeCG exhibits an obvious wavy crack deflection between layers (Fig. 3d). This deflection slows down the propagation speed of cracks and increases the energy required for progression, representing a prominent extrinsic toughening mechanism^38,39. In addition, crack branching observed in HeCG further extends the crack path and disperses stress at the crack tip (Supplementary Fig. 7), enhancing crack resistance.

Digital image correlation (DIC) strain measurements provide further insights into crack propagation and fracture processes, capturing crack characteristics in both horizontal and vertical directions⁴⁰. Strain cloud maps reveal that HeCG’s crack tips follow a hierarchical failure path with visible deflections, while Homo-1 shows evident stress concentration, resulting in unsteady crack growth at the crack tip and brittle fracture (Supplementary Fig. 8b, c, Movies 1 and 2). This stress concentration is even more pronounced in HeCA (Supplementary Fig. 8a). Moreover, despite elevated strain following initial crack propagation, HeCG shows reduced stress concentration at peak bending moments near the midspan, resulting in a more uniform strain distribution (Supplementary Fig. 8c). These findings align with nanoindentation results, confirming that the gradient structure enhances energy dissipation and mitigates crack progression.

In extreme conditions, engineering materials must be capable of withstanding both static and impact loads to ensure structural stability. To evaluate the impact resistance of gradient cement pastes, split Hopkinson pressure bar (SHPB) tests were conducted. While HeCG initially shares a similar modulus with cement paste and Homo-1 at high strain rates, its stress–strain curve shows multiple buffering strains with distinct fluctuating platforms (Fig. 3e), indicating efficient energy absorption during impact. The impact resistance (engineering stress) of HeCG reaches an impressive 96.61 MPa, surpassing the 57.19 MPa for cement paste and 79.53 MPa for Homo-1 (Fig. 3e, f). Correspondingly, HeCG’s energy absorption capacity reaches 72.57 kJ m⁻³, which is 1.66 times that of Homo-1 (43.70 kJ m⁻³) and 2.87 times that of cement paste (25.26 kJ m⁻³) (Fig. 3f).

High-speed camera recordings highlight distinct failure patterns among the samples. The cement paste fails catastrophically, shattering into fine powder under impact (Supplementary Fig. 9a, Movie 3). Homo-1, while demonstrating improved mechanical properties from CNT addition, fractures into small fragments similar to cement paste (Supplementary Fig. 9b, Movie 4). In contrast, HeCG exhibits a distinctive failure mode: it breaks into several large fragments with axial cracks during initial impact (Supplementary Fig. 9c, Movie 5). This failure mode allows the gradient structure to absorb and rebound impact force, reflected in the fluctuating platforms on the stress–strain curve. The enhanced impact resistance of HeCG is attributed to the interfacial interactions within its gradient layers, which strengthen the connections between heterogeneous layers, enabling efficient load transfer and stress delocalization. The gradient design effectively dissipates impact energy, preventing catastrophic failure and maintaining structural integrity under dynamic loads.

EMA performance of gradient cement paste

The EMA performance of the gradient cement paste is evaluated across the 2–20 GHz frequency range. Homogeneous cement samples exhibit a slight increase in both the real (ε’) and imaginary (ε”) parts of the relative permittivity with increasing CNT content (Supplementary Fig. 10a, b). HeCG, however, exhibits higher permittivity compared to Homo-0.5 and Homo-0.75 (Fig. 4a, b), indicating enhanced charge storage and EM energy dissipation. A notable transformation in permittivity is observed between 10 and 13 GHz, likely due to dipole relaxation polarization induced by the electric fields within the matrix. Reflection loss (RL) analysis reveals that HeCG achieves an RL_min of −59.24 dB, significantly outperforming Homo-0.5 (−34.64 dB) and Homo-0.75 (−44.03 dB) (Fig. 4c–e).

**Fig. 4: Electromagnetic wave absorption (EMA) performance.**

In homogeneous samples, increasing CNT content results in limited EMA enhancement. While higher CNT concentration improves the material’s permittivity (Supplementary Fig. 10a, b), excessive CNT concentrations disrupt impedance matching with air due to the abundance of free electrons on the surface, causing substantial reflection of incident EMWs and preventing effective EMW penetration and absorption⁴¹. For instance, the RL_min of Homo-1 is attenuated to only −24.39 dB (Supplementary Fig. 10c), underperforming compared to Homo-0.5 and Homo-0.75. In contrast, the gradient structure of HeCG minimizes surface reflection, allowing EMWs to penetrate into the inner layers with higher CNT content, thereby improving EMA performance. The effect of CNTs on the permittivity of cement paste follows a linear trend across different frequencies (Supplementary Fig. 10d), closely aligning with a modified linear model based on effective medium theory⁴². This provides a potential approach for predicting the permittivity of cement pastes with various CNT contents.

The EM energy absorbed by the sample is converted into heat, promoting a temperature rise. Microwave heating facilitates molecular alignment in the microstructure by forcing the electrode arrangements of the molecules to orient with the electric field, followed by ion vibration and friction between ions to generate heat¹⁰. Therefore, to verify the EMA capability of the sample, microwave heating experiments are conducted (Fig. 4f). After 1 min of 300 W microwave exposure, HeCG reaches a surface hot spot temperature of 94.26 °C, significantly higher than that of Homo-0.5 (80.89 °C) and Homo-0.75 (82.81 °C) (Fig. 4g). This temperature rise correlates with HeCG’s higher attenuation constant (3.68 dB m⁻¹) compared to Homo-0.5 (0.96 dB m⁻¹) and Homo-0.75 (1.39 dB m⁻¹) (Fig. 4h; Supplementary Fig. 11).

The simulation results closely align with experimental findings, underscoring the gradient structure’s role in improving EMA performance. The upper layer with lower dielectric properties guides more EMWs into the matrix (Fig. 5a). As EMWs propagate through the matrix, they undergo continuous attenuation, resulting in nearly zero electric field intensity at the bottom surface with enhanced dielectric properties. The electric field intensity decreases progressively from the top to the bottom across heterogeneous interfaces, with the high-CNT-content regions effectively absorbing EMWs. This progressive attenuation is visually represented by the gradual disappearance of power flow vectors (white arrows) from top to bottom. An isosurface diagram further illustrates this effect, depicting the diminishing electric field intensity throughout the material (Fig. 5b). Additionally, the relationship between the incident angle of EMWs and the S₁₁ parameter, which indicates impedance matching for the absorber, was investigated (Fig. 5c). At a 0-rad incident angle, HeCG exhibits a lower S₁₁ value than Homo-0.75, indicating reduced EMW reflection at the boundary and superior impedance matching. As the incident angle increases, the impedance matching improves and reaches a peak. Beyond a 1.0-rad angle, the dielectric properties have minimal effect on the S₁₁, with the S₁₁ responses of HeCG and Homo-0.75 converging. This pattern indicates that, similar to how antenna orientation affects signal reception, the EMA efficiency is influenced by the incident angle, with the gradient structure’s impact on impedance matching diminishing at higher angles.

**Fig. 5: Electromagnetic wave absorption (EMA) mechanism analysis in heterogeneous cement paste with a gradient transition layer (HeCG).**

The excellent EMA performance of HeCG stems from multiple loss mechanisms (Fig. 5d). A stable, continuous conductive network formed by uniformly distributed CNTs within the matrix facilitates the movement of free and polarized charges⁴³, enhancing electrical conductivity and conductive losses of EMWs. In addition, under alternating magnetic fields, induced currents in vortex-like patterns are generated in the conductive network, resulting in eddy current losses that dissipate EM energy as heat. The heterogeneous interfaces within the cement matrix with embedded CNTs enhance carrier migration and accumulation under high-frequency alternating EM fields, contributing to significant interfacial polarization losses⁴⁴. In particular, moisture in the matrix induces polarization losses of dipoles. The congenital defects, porous structure, and gradient interlayers in the material promote multiple EMW reflections and scattering, extending the propagation path and enhancing total absorption effectiveness.

In summary, the gradient structure effectively addresses the challenges of poor impedance matching and limited EMA within homogeneous cement matrices. As a result, HeCG offers a feasible solution for mitigating EMW pollution in the environment, enhancing stealth capabilities in military and medical equipment, and constructing anechoic chambers for scientific research. The integration of gradient design with nanomaterial enhancement demonstrates a comprehensive strategy for optimizing EMA performance, expanding the potential of HeCG for diverse EMW management applications.

Conclusions

In this work, we have developed a gradient cement paste by finely tuning the distribution of CNTs within the cement matrix using a bottom-up approach. This design significantly enhances interfacial interactions and reduces overall porosity, resulting in improved mechanical properties, such as increased flexural toughness, crack resistance, and impact resistance. These improvements stem from efficient load transfer and stress delocalization enabled by the gradient interlayers. Moreover, the introduction of heterogeneous transition layers enhances the material’s EMA performance. The gradient structure achieves optimal impedance matching with air and superior attenuation, outperforming homogeneous cement pastes. These results address key limitations of traditional cement-based materials, including low toughness, limited ductility, and poor EMA capability, while overcoming the impedance mismatch typically associated with highly conductive carbon-based materials. The gradient cement paste demonstrates considerable potential for applications requiring both impact energy absorption and EMA, particularly in military infrastructure.

Increasing the number of gradient layers from stepwise to continuous gradients within a defined thickness is expected to amplify mechanical functionality by optimizing the microstructure. This scalable design strategy offers an innovative pathway for constructing high-performance, multifunctional cement-based materials. The integration of such materials into advanced infrastructure and smart city initiatives can contribute to energy conservation and emission reduction, supporting the sustainable development of next-generation building materials.

Methods

Sample preparation

Commercial P·I 42.5 cement and multi-walled carbon nanotubes (Nanjing XFNANO Materials Technology Co., Ltd.) were used, with properties detailed in Supplementary Tables 1 and 2. The CNT solution containing a dispersant (25 wt% relative to CNTs) was magnetically stirred for 10 min, followed by ultrasonic treatment at 180 W and 40 kHz in an ice bath for 30 min. Fresh cement pastes containing different CNT contents (Supplementary Table 3) were formulated at a water-cement ratio of 0.4 by compounding cement with the CNT solution for 10 min to ensure thorough interaction between the CNT solution and cement particles. To construct the gradient structure, fresh cement pastes were poured layer-by-layer into a steel mold lightly coated with silicone oil, starting with the highest CNT concentration at the bottom and progressing to lower concentrations at the top. Each layer was vibrated individually, and a 10-minute casting interval was maintained between layers to ensure proper cohesion and structural integrity. Layer thicknesses were controlled by pouring pre-calculated volumes based on the desired thickness. After casting, the samples were demolded after 24 h and submerged in a water tank for curing for 27 days.

Sample characterizations

SEM images and EDS mappings of the 28-day samples were acquired using a Merlin Zeiss instrument operating at an acceleration voltage of 5 kV for imaging and 20 kV for EDS mapping. Samples were coated with a thin platinum layer to enhance imaging resolution. Raman spectroscopy and mapping were obtained using a confocal laser Raman spectrometer (XploRA PLUS, HORIBA) under ambient conditions. Porosity and pore structures of the samples were determined via MIP using an automatic porosimeter (AutoPore V9620). The MIP method, which relies on mercury’s non-wetting properties, estimates pore volume based on mercury intrusion content at varying pressures. Prior to testing, the samples were vacuum-dried in a vacuum oven at 60 °C for 48 h to remove residual moisture. Three samples were tested for each experimental group.

Mechanical testing and simulation

Nanoindentation tests were performed using a Hysitron TI-980 system under ambient conditions. Cubic samples (10 mm × 10 mm × 10 mm) were prepared by cutting, embedding in resin, grinding, polishing, and vacuum-drying at 60 °C for 48 h. Each sample was tested at a minimum of six different sites. Macroscopic mechanical properties were evaluated using three-point bending and SENB tests. For three-point bending, bulk samples (80 mm × 10 mm × 10 mm) were tested on a universal testing machine (CMT6104, NSS) following ASTM C293M-16 standards. The test setup included a 1 kN load cell, a 72 mm support span, and a displacement rate of 0.2 mm min⁻¹. SENB tests adhered to ASTM E1820-20b standards, with sample dimensions and support spans identical to those used in the three-point bending tests. Notches about half the width of the sample were created using diamond wire cutting and sharpened with a razor blade. Tests were performed using a 100 N load cell at a displacement rate of 1 μm s⁻¹. The DIC technique was employed to capture the evolution of strain fields at the crack tip during crack propagation under SENB testing⁴⁵, with a high-speed camera (S1310, Revealer) tracking crack expansion. Sample surfaces were polished and sprayed with a white undercoat followed by black paint to achieve a well-contrasted gray-scale speckle pattern for DIC analysis. Impact strength was evaluated using a modified SHPB test on samples measuring 10 mm by 10 mm by 10 mm at an initial driving pressure of 35 kPa. The dynamic fracture process was recorded using a high-speed camera (iX Camreas, i-SPEED 717-S). For heterogeneous cement paste samples, the bearing surface was comprised of a compacted cement layer during macroscopic mechanical tests. Each experimental group was tested with at least three samples. Finite element analysis of strain distribution under three-point bending was performed using COMSOL Multiphysics 6.1. A HeCG model with a support span of 72 mm was constructed, consisting of three layers: a top layer of 80 mm × 2 mm, a middle layer of the same dimensions, and a bottom layer measuring 80 mm × 6 mm. To streamline strain distribution analysis, displacement continuity conditions were applied at the interfaces between layers. Modulus values derived from nanoindentation results were set to 39.83 GPa, 35.79 GPa, and 31.87 GPa for the top, middle, and bottom layers, respectively, with a Poisson’s ratio of 0.22 across all layers.

EMA performance testing and simulation

EMA performance was evaluated in the 2–20 GHz range using a vector network analyzer (VNA, Keysight, E5080B) with the coaxial probe method. Samples were prepared as cylinders with a diameter of 50 mm and a thickness of 20 mm. Prior to testing, the VNA was calibrated with short-circuit (fitting), open-circuit (air), and load (deionized water) standards. For the HeCG, the cement layer with 0.5% CNTs was selected as the incident surface. More than three samples were tested for each component. Microwave heating experiments were conducted using a domestic microwave oven (Panasonic, NN-GF599M) operating at a frequency of 2.45 GHz for 60 s at 300 W. Surface temperatures were captured using a Fluke TiX580 infrared thermal imager (temperature range: −20 °C to 1000 °C, accuracy: ±2 °C or 2%, refresh rate: 9 Hz). The surface emissivity was set at 0.92, and the ambient temperature was about 27 °C. The temperature legend was standardized in Fluke Connect software. Finite element analysis of EMA at 10 GHz was performed in COMSOL Multiphysics 6.1. The HeCG model consisted of three layers (Supplementary Fig. 12): a bottom layer measuring 5 mm × 5 mm × 2 mm (ε_r = 5.96–1.26j), a middle layer of the same dimensions (ε_r = 5.75–1.12j), and a top layer measuring 5 mm × 5 mm × 6 mm (ε_r = 5.20–0.90j). A perfectly matched layer (PML) of 5 mm × 5 mm × 3 mm was positioned the top layer of model, with a 5 mm × 5 mm × 3 mm air-filled domain separating the sample and the PML. The PML’s upper surface was set to a scattering boundary condition, while the lower surface, adjacent to the air domain, had a port boundary condition. The bottom of the sample was defined as a perfect electric conductor (PEC) to block any noise from the outside. Floquet periodic boundary conditions were imposed on all sides of the model.