Natural aggregate originated from the river Danube was separated, and following fractions were used for SCC synthesis: I (0/4), II (4/8), and III (8/16). The particle size distribution of the aggregate determined according to the standard procedure (35) is shown in Figure 1. Contents of fine particles in fractions II and III were 1.84% and 0.94%, respectively, while the contents of coarser grains (> 4 mm) in fractions I, II, and III were 1.96%, 5.81%, and 0%, respectively. Content of fine particles in fraction I determined according to the standard procedure (36), was 0.59% (particle size <0.063 mm) and 1.68% (particle size < 0.09 mm). The grading curve of the fractions mixture is shown in Figure 1, as well.

Table 1 lists physico-mechanical and chemical properties of used aggregate fractions determined according to the standards (37–43). All determined properties meet the quality requirements (44–45).

The mass ratio of fractions II and III was determined on a laboratory test that made known the highest value of density for the mixture of these fractions in the dry compacted state, Figure 2. Consequently, the mass ratio of these fractions was adopted to be 1:1.

The content of fraction I in an aggregate mixture was determined based on the pilot mixtures and it was 31.8 % of SCC total volume.

Cement without additives (CEM I 42.5R, Lafarge, Serbia) was used as a binder. Physico-mechanical characteristics of cement are realized by standard procedures (46–48) shown in Table 2. Chemical composition was obtained by standardized procedure (49) and given in Table 2.

Limestone powder, carbonate with a medium grain size of 250 μm, (Granit Pescar, Ljig) was used as filler. Physico-mechanical properties of limestone are done on natural stone– limestone, which was ground to obtain limestone powder using standard procedures (47, 50–55) and given in Table 3. Chemical composition was provided by the producer and shown in Table 3.

Secondary sulfur used as a partial substitution for usual mineral filler, limestone, originates from the oil refining process by Claus`s procedure in the Oil Refinery Pancevo and its purity was 99.9%. The density of sulfur was 2050 kg/m³ and was done by standard procedure (50) while the specific surface area 2600 cm²/g was realized by Blaine method. Elemental sulfur is hydrophobic and insoluble in water under normal conditions (26-27, 32-33).

A new generation of chemical admixture, polycarboxylate based super-plasticizer, Glenium Sky 690 (BASF Construction Chemicals, Italy), was used to provide self-compacting abilities and enhance flowability of concrete. Its properties were provided by the producer and shown in Table 4.

Compressive strength test measurements were done on three samples (cubes of 10 cm edge length) for each composition and curing age, while the average values are shown in Table 7. As expected, strength was weakened with the increase of replaced limestone portion with sulfur while it was enhanced by concrete ageing. So the lowest value of 44.1 MPa was developed in the specimen with the highest replacement percentage (S20) at the age of 3 days, while the highest value of 62 MPa was reached after 28 days for E. Compressive strength of tested SCC samples during the ageing is given in Table 7. It is evident that the higher replacement ratio negatively affects compressive strength, more precisely strength decreases up to 12 % regarding the reference sample. Strength decrease depends both on the portion of limestone replaced with sulfur and ageing period, probably due to the two effects. Firstly, sulfur has a lower hardness value (1.5–2) than limestone (3) in the Mohs scale, so SCC with higher sulfur content would have lower strength.

Secondly, explication of obtained results regarding the compressive strength should be based on considering the facts that the compressive strength of any concrete is usually lower for the lower density (69), while connected to lower porosity as well (70). Accordingly, the lower compressive strength of the samples with sulfur can be explained by the presence of a certain amount of waste sulfur. This is because the used waste sulfur has both lower density and lower mechanical properties, in comparison to the used limestone. Also, the results of true density measurements of hardened concrete samples show that the samples with sulfur have lower true density comparing with the samples prepared only with the limestone.

Thirdly, it is well known that lower strength is obtained in a higher w/c ratio. Contents of water and cement are the same in all prepared mixtures. However, waste sulfur that is sulfur in its elemental form, almost does not absorb water, unlike limestone. Therefore, the content of free water in such fresh mixtures with sulfur (and thus the effective w/c ratio) is higher compared to the mixtures with limestone (26–29). According to the current knowledge and studies of concrete technology, the greatest impact on the concrete strength is the water/cement factor of the fresh concrete mixture. Namely, concrete with a lower water/cement factor will have higher strength than one with a higher water/cement factor (71,72). In this case, mixtures with sulfur have lower compressive strength. Similarly to the use of porous aggregates, such as recycled concrete aggregate (73), difference in porosity of the used filler directly affects the water volume available for the chemical reactions, i.e., lower the porosity of filler, higher the ‘free’ water content, and, consequently, the higher effective water to cement ratio, and finally lower compressive strength (74).

Compressive strength enhanced by ageing from 15 % to 26 %. It is noticeable that the strength increase with age is higher in the reference sample and the samples with a small percentage of limestone replacement, while in the samples with 10 and 20 % of replaced limestone, the strength increase did not exceed 15.88 %. Also, compressive strength change during ageing was not directly proportional to the increase of limestone replacement. Namely, strength decrease for SCC with 10 % and 20 % of sulfur was almost similar at the late-age stages. More precisely, differences in strength decrease between S10 and S20 are becoming smaller with the increasing ageing time. In addition, the strength values are probably influenced by the filling effect of different filler particles and their various specific surface areas. Therefore, the higher strength of the specimens without sulfur or with a lower portion of replaced limestone with sulfur can be attributed to higher specific surface area of limestone than sulfur.

Figure 5 shows the values of the compressive strength of each sample with sulfur divided by the value of compressive strength of the reference sample at a certain age.

As can be seen from Figure 5, all samples containing sulfur reached at least 88% of the corresponding strength of the reference sample, at the same ages. Two groups can be distinguished. The first group is consisted of samples S2 and S5 which reached almost the same values of compressive strength of the reference E (94% or more of the reference). The second group represents samples S10 and S20, with a substantial drop of compressive strength of the reference E (a drop of 6–12 %).

Flexural strength was measured according to the three-point loading test in the middle of the span, at the age of 180 days. The test was carried out on three 12×12×36 cm specimens, while the average values of three measurements are shown in this study.

The bond strength determined by pull-off strength test is often used for successful evaluation of concrete resistance and verification of the adhesion strength of the concrete repairing structures (75). The tensile bond strength was not the issue in this study because such strength of used adhesive was several times greater than the bond strength between aggregates and cement paste in the tested hardened concrete. It should be emphasized that in all conducted tests, the failure occurred usually inside the bulk of concrete, neither in the adhesive nor at the contact of adhesive and concrete. Therefore, obtained strength values refer to the strength of the bond among the concrete components bearing in mind the presence of different sulfur content. Quality of SCC was determined at the age of 180 days, on the 50×50×8 cm plates. Test dollies of two different diameters, Ø20 mm and Ø50 mm, were applied. Five tests/measurements were conducted for each series and both dolly diameters, while the average values are presented in this study.

Flexural strength values were in the range of 6.9 (S20) and 10.5 MPa (E), while the adhesion strength values were within the limits of 5.53 MPa (E) and 7.50 MPa (S10) in the case of Ø20 dolly and within the limits of 5.77 MPa (E) and 6.28 MPa (S10) for the Ø50 dolly.

Diagrams in Figure 6 show the decrease of flexural and increase of adhesion or bond strength of SCC with sulfur in relation to the reference SCC.

Flexural strength decreases for samples with sulfur in comparison to the reference sample from 18.1 % (S2) to 34.3 % (S20), Figure 6 a). It can be concluded that this decrease is directly, almost linearly, proportional to the increase of limestone replacement by sulfur. Similarly to compressive strength, flexural strength decrement of SCC with sulfur in comparison with SCC without sulfur can be explained by higher effective water to cement ratio and lower sulfur hardness.

Figure 6 b) shows a change in adhesion strength of SCC with sulfur compared to the reference sample. Results of the pull-off test show that the increase of adhesion strength is lower for the samples with higher content of sulfur for both dolly diameters. It can be concluded that the use of sulfur has no negative influence on the adhesive strength regardless of the used dolly diameter. In addition, bond strength increment is more pronounced for the dolly diameter of Ø20 compared to Ø50.

Ultrasonic pulse velocity measurements provide important information about the structural, mechanical, anisotropic, and elastic properties of a medium it passes through. Since the theory proves that ultrasonic pulse travels with lower velocity through the material of higher porosity and lower density (76–78), measuring ultrasonic velocity can be used for monitoring changes in the materials structure such are homogeneity, existence and evaluation of cracks, voids, holes and pits, estimation of compressive strength and determination of elastic modulus.

Table 8 shows that the increasing sulfur content induces the decrease of dynamic elasticity modulus (E_d) for all ageing stages while these values are increasing during the ageing per each series. It can be observed that E_d values are in the range of 40 and 44.12 GPa. However, a more comprehensive approach of dynamic elasticity modulus changes with adding sulfur can be shown through the decrement in comparison with the reference, Table 8. It is obvious that with increasing the sulfur content, that decrement is becoming higher. Changes between those two dynamic elasticity moduli were from 0.1% to 5.7%. Depending on the amount of replaced limestone, the largest decrement of elasticity modulus compared with the reference sample E was observed in S20 sample (2.8–5.7%) and the smallest in S2 sample (0.1–1.9%). Mutual variations among E_d values of the SCC with sulfur were characterized as relatively small, ranging 0.1–5.1% compared to the reference sample.

The analysis of E_d changes shows a similar trend as compressive strength, E_d decrease with the sulfur content increase and E_d increase with the ageing. This behavior can be explained not only by the influence of less hard and hydrophobic sulfur but also with entrapped air content. Namely, the concrete component with lower hardness, higher water content or w/c ratio, as well as higher entrapped air content would induce lower E_d values. Moreover, fresh concrete investigation shows that entrapped air content decreases with increased sulfur content. Therefore, it can be assumed that the influence of the entrapped air is not dominant but the filler hardness and free water content.

Since SCC has a uniform consistency, compressive strength and dynamic modulus of elasticity should have consistent correlation shown in Figure 7. This correlation is linear with excellent coefficient (R²> 0.9) and the equation y = 28.842 + 0.23x. Such a good correlation between strength and dynamic modulus of elasticity proves the uniform consistency of SCC.

The results for ultrasonic velocities are shown in Table 9. Concrete quality classification proposed by Whitehurst (79) categorizes the concrete as ‘’excellent’’, ‘’good’’, ‘’doubtful’’, ‘’poor’’ and ‘’very poor’’ for the UPV values of 4500 m/s and above, 3500–4500 m/s, 3000–3500 m/s, 2000–3000 m/s and 2000 m/s and below, respectively. Since measured UPV ranges from 4478 to 4676 m/s, all mixtures have excellent quality. Table 9 shows that UPV values tend to increase with age, mostly followed a similar trend for compressive strength. However, the rate of UPV development was slower than that concluded in the strength. Overall analysis shows that ultrasonic velocity is increasing during the ageing and with sulfur content decrease. Concerning decrement/increment approach, changes in relations between UPV of SCC with sulfur and the reference SCC are shown in Table 9.

It is evident that the largest decrease (1.7%) was at the age of 14 days, for the S10 sample. Also, mutual deviations of UPV per examined series of SCC with sulfur compared to the reference one are small and within the relatively narrow range (- 0.15% - 1.7%). Obtained results showed that kind of SCC behavior is induced by several factors: mixture composition, effective w/c ratio, porosity.

Since the mixture compositions mutually differ only in a portion of fillers type, the influence of density of fillers and therefore hardened concrete on UPV is as obvious. Concrete containing a higher content of sulfur has lower UPV value.

Effective w/c ratio is a higher in the samples with higher content of sulfur as discussed before, causing higher porosity and therefore lower UPV value. Also, this can be proved with measured values of true porosity and density (E sample has porosity of 7.9 % while the sample S20 has true porosity of 10.1 %).

Therefore, the higher UPV values of the reference SCC is probably due to its reduced porosity.

The effect of different fillers on the correlation between compressive strength and UPV was studied by many researchers (80–84). Since the influence of sulfur inclusion on compressive strength and UPV was assessed in this study, the correlation between compressive strength and ultrasonic pulse velocity of SCC at ages of 7, 14, and 28 days is shown in Figure 8. Strong linear correlation degree (R²=0.83127) is obtained, so the compressive strength of SCC with sulfur can be accurately predicted.

The results in Figure 8 are relatively close to the values obtained using functional dependencies proposed by Khatib (80) and Liu (81) for SCC with fly ash. On the other side, a similar slope of correlation line and lower ultrasonic velocity for the same compressive strength can be noticed compared to the results of linear correlation obtained by Mardani et al. (82) for SCC with different superplasticizers. The relationship between compressive strength and UPV for SCC with sulfur are lower compared to the values based on correlations for NVC given by Neville and Jones (83,84). Since there are no published results of similar tests for SCC with sulfur, these dependencies have a descriptive approach.

Based on the measured UPV and E_d, the dynamic Poisson ratios (μ_d) were calculated, Table 10. The values are ranging from 0.239 for the reference sample at the age of 7 days up to 0.257 for S20 at the age of 180 days. Measured values of dynamic Poisson ratios for all samples increased with the aging. Comparing the measurements made on studied series, dynamic Poisson coefficients for SCC with sulfur were very close to the calculated values of the reference sample E. Considering those observations, the dynamic Poisson coefficients of limestone and waste sulfur have similar values.

Relevant information about microstructure (homogeneity, porosity, properties of hardened cement matrix and the interface between aggregate particles and cementitious materials in hardened SCC) were examined on the reference and S20 samples surfaces. Aggregate grains (quartz, quartzite, dolomite), cement matrix, and fillers (limestone and sulfur) can be identified showing the compact internal structure of SCC, Figure 9.

Figure 10 shows the clearly evident interfacial structure between the aggregate particle and cementitious matrix of E and S20 specimens.

According to Figure 10, the surface porosity degree of S20 compared to E did not visually vary. However, in order to confirm the visual appearance, image analysis was applied for measuring the portion of cementitious matrix surface porosity (marked red in Figure 10) and the thickness of Interfacial Transition Zone (ITZ) (marked green in Figure 10). Micrographs of the sample`s surface were analyzed by Image-Pro Plus (IPP), a special software program that enables work in all known formats of images and automatically measures, counts, and classifies data on analyzed objects. The program communicates directly with Excel, thus enabling statistical and graphical treatment of data. Results are shown in Table 11.

The interfacial transition zones are the weakest bond of concrete and the cracks development commonly reveal in the interfacial zone between the aggregate particles and cementitious matrix (85). Therefore, ITZ thickness measuring at about 30 spots between aggregate grains and matrix and average values in both specimens was done by IPP and it was less than 5 μm. More precisely, it is evident that ITZ of S20 is thinner than in E. Difference between surface porosity of two examined samples is very small. Considering the Image Pro Plus results (Table 11), the number of pores in S20 is smaller while the porosity is higher than in E. These results correspond to the compressive strength decrease in the case of S20 compared with E. The presence of soft sulfur particles additionally contributes to compressive strength decrease.

With the aim of characterizing the distribution of phases in the matrix, EDS analysis was performed on several surface places. Appropriate EDS analysis results for the sample shown in Figure 11 are given in Table 12.

Based on the results, the matrix has following composition: S grains (Spectrum 1) and quartz/sand (Spectrum 5) while the Spectrums 2, 3, and 4 indicate the presence of CH and CSH phases. The contact zone between sulfur grains and the surrounding matrix (Spectrum 2) remained almost chemically unchanged since EDS analysis shows slightly higher sulfur content with other similar constituents. On the other side, the contact zone between quartz and the matrix (Spectrum 4) indicates higher Si content.

On the basis of image analysis, it can be concluded that E shows a presence of higher content of smaller pores, while S20 has a lower content of larger pores, which cannot be visually distinguished. According to Hg porosimetry, further information regarding pore size was obtained. The measurements showed that sample S20 had higher average pore size diameter (45.8 nm) than E (43.5 nm), which approved the image analysis. This measurement also approved discussion regarding effective w/c ratio (higher porosity of S20).

Sulfur in an alkaline matrix such as cement paste is a chemically stable material (86,87). Considering the SEM and EDS analysis, it can be concluded that the reaction of sulfur with hydrated cementitious phases does not happen. Therefore, unlike sulfates, the sulfur can be considered as a chemically inert component in such mixtures, so the sulfate attack phenomenon is practically impossible in these conditions. Even, by visual inspection of the samples sulfur can be detected with its characteristic yellow color.

Mapping by elements S, Ca, and Si (qualitative analysis) was performed at several matrix surface places, Figure 12. Based on the results, it is evident that main particles of the matrix are uniformly distributed showing good homogeneity of components inside the matrix. Regarding the dispersion of limestone and sulfur performed at several matrix surface places, grouping/clustering of ground particles was not observed and good distribution of all matrix components is evident.