IntroductionThe pore structure of cement-based materials significantly influences their compressive strength. Studies have shown that the nanoscale pore structure significantly changes with variations in water content. Traditional pore measurement methods like mercury intrusion porosimetry and gas adsorption require the drying of specimens, potentially altering or even destroying the pore structure. The accuracy and representativeness of the compressive strength-pore structure relationship model established on this basis are therefore questionable. To accurately describe the important correlation between compressive strength and pore structure of cement-based materials, low-field magnetic resonance relaxation technique is utilized to carry out in-situ, nondestructive testing of saturated cement mortar doped with air-entraining agent. This is then combined with compressive strength measurement results to validate and refine the model describing the relationship between compressive strength and pore structure characteristics.Methods(1) Specimen preparation To avoid the ambiguous effects of ferromagnetic substances when using by low-field nuclear magnetic resonance (LF-NMR) technique, white Portland cement with low Fe₂O₃ content was used to prepare cement mortars with water-to-cement ratio of 0.4. Cement to sand ratio is 1:2. And air-entraining agents (SJ-2 type) were incorporated. Chemical composition of white cement were 0.50% Fe₂O₃,64.60% CaO,21.71% SiO₂,4.60% Al₂O₃,2.80% SO₃,2.43% MgO,0.48% R₂O. The mass ratio of air-entraining agents to cement was 0%, 0.05%, 0.10%, and 0.15%. Considering different mass ratios, they are named as WP-BLK, WP-SJ05, WP-WP-SJ10, and WP-SJ15, respectively.According to the designed proportions, the white cement, sand, air-entraining agents and water were mixed and casted into prisms of size 20 mm×20 mm×50 mm. After demounting at 24 h, they were all cured in saturated lime water at (20±2) ℃ for 180 d. After reaching the specified age, ten specimens were taken from each group of mortar and put into the high-pressure water saturation equipment for 3 d to ensure that the specimens were completely saturated and set aside.(2) Pore structure test The pore structure of saturated mortars were tested by LF-NMR technology. Utilising a 2 MHz NMR analyser manufactured by Limecho Ltd., China, transverse relaxation test was performed through the Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence at controlled room temperature about 20 ℃. In detail, the echo time τ_e (s) was adopted as short as 60 μs to detect the water confined in nanoscale pores of fast relaxation. Preliminary experimental tests had revealed that echo time τ_e ⩽100 μs was short enough to make the contribution of diffusion relaxation negligible due to the controlled low concentration of Fe₂O₃ in white cement pastes. Moreover, A long sequence of N = 40 000–60 000 echoes generated at time τ= nτ_e (n=1, 2, · · · , N) (s) was recorded to capture the water stored in large capillary pores and even air voids with long relaxation time. 256–1024 scans with enough long repetition time 5–15 s were averaged. Except for the echo time, the rest of the parameters were flexibly adjusted according to water content of specimen. One can see detailed steps for testing pore structure by low-field magnetic resonance techniques in the relevant literature.(3) compressive strength test The compressive strength test of specimens were carried out by YAW-300 microcomputer automatic cement folding testing machine, which was produced by Jinan Hengruijin Testing Machine Co., Ltd., with a maximum capacity of 100 kN. The size of the compression surface of the specimen is 20 mm×50 mm, and the loading rate is fixed at 0.2 kN/s.Results and discussionLF-NMR technology accurately captures the pore structure characteristics of saturated mortar. Empirical models using total porosity or graded porosity as variables are generally effective, but the empirical parameters obtained from model fitting are not stable enough, limiting the model’s applicability. An improved empirical model, based on the pore size distribution curve and assuming a linear relationship between the weights of different pores on compressive strength and their logarithmic pore sizes, offers a concise form, significantly improves fitting accuracy, and accurately reflects the differences within the specimen groups. Furthermore, building on Griffith’s fracture theory, the theoretical model, which uses total porosity and logarithmic mean radius as independent variables, exhibits complex expressions and low predictive accuracy. If pores with a radius above 10 nm are regarded as defects, their influence can be explained by Griffith’s fracture theory. Meanwhile, pores with a radius of 10 nm or less are regarded as internal pores of the gel, and its effect on the strength of the matrix is approximated using the model description proposed by Powers. The resulting modified Griffith fracture theory model is very high in prediction accuracy, stable, and interpretable, accurately reflecting how minor differences in pore structure lead to small changes in compressive strength.ConclusionsThe incorporation of air-entraining agents mainly increases the capillary pores above 10 nm in radius but also decreases the matrix pores below 10 nm in radius. LF-NMR technology can obtain the full-scale pore structure of cement mortar, aiding in the description of the fundamental correlation between pore structure characteristics and compressive strength.