1. Introduction
The most common way to stabilize a soil is with the use of binders, such as lime and cement [
1,
2,
3,
4,
5,
6]. These binders improve the strength and workability of the soils via ion exchange or by forming C–S–H gel and calcium carbonate [
4]. One of the disadvantages of using these binders is sulfate attack. Sulfate attack in concrete has been studied vastly in the concrete sector [
7,
8,
9], and it is well defined as the reaction between sulfate and certain compounds in concrete that leads to the expansion and formation of cracks in concrete. To be specific, sulfate in certain conditions reacts with cement compounds such as monosulfate, portlandite and C–S–H gel. The results of this reaction may be ettringite, gypsum, or thaumasite [
10,
11].
However, there are two completely different opinions regarding the use of these binders in soils. The first is that these expensive materials in soil are beneficial and can fill up the pores in soil particles, leading to better bonding of soil particles and an improvement of the final strength of soils [
12]. The other criticizes the formation of these expansive materials, which may compromise the strength. Needless to say, both of these points of view depend on the generated expansive material, conditions, and sulfate availability in the soil.
One of the sources of sulfate is the soil itself. Soils with some amount of sulfate are quite common all around the world [
13]. Gypsum, commonly composed of calcium sulfate dihydrate, is a primary source of sulfate in soils and can be found in gypsiferous soils. As reported by Verheye, gypsiferous soils cover approximately 1 million km
2 of the world’s surface [
14]. Gypsiferous soils can be found in countries in the Middle East (e.g., Iraq, Syria, and Iran) and in Europe (especially in Spain), as well as some parts of North Africa and the USA [
15]. According to Jara [
16], 7.2% of Spain is covered with gypsiferous soil, mainly located in the eastern part of the country as shown in
Figure 1.
As mentioned before, the existence of sulfate in soil can be beneficial or problematic due to its reaction with the binder. When lime is added to the soil, cation exchange and flocculation/agglomeration take place almost instantaneously, increasing the pH to around 12. This high pH makes the solution a suitable environment for alumina, silica, and other minerals to react with lime, thereby developing the silica gel (C–S–H) and alumina gel (C–A–H) [
3]. The reaction of producing C–S–H and C–A–H are as follows:
However, two phenomena may occur when the soil or underground water contains some amount of sulfate. The sulfate may react with alumina and form calcium aluminum sulfate hydrates, eventually leading to the formation of ettringite. Undoubtedly, certain parameters must be met in order for ettringite to form, such as a high temperature, a pH above 10, and enough water. All requirements are satisfied when gypsiferous soil is stabilized with cement or lime, except for water, which may come from underground water or rainwater. It should be pointed out that if the pH drops below 10, ettringite formation stops [
18]. The ettringite formation process was proposed by Harris et al. [
18] as follows:
The second phenomenon involves the formation of thaumasite. The sulfate may react with calcium silicate hydrate gel in the system and form thaumasite. The rate of reaction can increase at temperatures below 15 °C. Although ettringite and thaumasite have a similar structural arrangement, the expansive capability of thaumasite is less than that of ettringite as it occupies 45% less volume [
19]. Moreover, it was reported that ettringite could expand to as much as two times its original size [
20], by 250% [
21], or by 137% as calculated using molar volume [
19].
In cement hydration, it is believed that this expansion leads to better strength if it happens at an early age. However, if the expansion happens at a later age (delayed ettringite formation), some problems can occur in the structure [
22]. It has been reported that the soil expansion does not follow the same rate as Portland cement concrete [
19]. The authors of [
23] found no significant swelling at the early age. They believed this to be due to the void spaces within the stabilized soil, suggesting greater effects for coarse-grained soil [
4]. At a later age, the void spaces inside the soil are filled up, resulting in a more rigid product with fewer pore spaces. Nevertheless, as the ettringite swells and no more pore spaces are left to fill, the swelling pressure is applied to other parts of the structure and can lead to potential catastrophe [
23]. One example was reported by Chen et al. [
24], where an 8.8 mile section of a road in Texas, USA, which was treated with lime, was damaged and caused 12.7 million USD worth of damage.
Nevertheless, ettringite formation does not only depend on sulfate content [
19,
25], and is not always expansive. It depends on many factors such as composition, curing period time and temperature, water availability, reactive phase availability [
23,
25], compatibility with other cement phases [
9], and amount of lime. Studies have reported that the swelling potential of sulfate-rich soils is decreased when they are treated with low C
3A binders such as ground granulated blast furnace slag (GGBS) [
26]. Seco et al. [
27] found that stabilization with a byproduct from the calcination of natural MgCO
3 rocks (defined as PC-8) could significantly decrease swelling while maintaining a similar or better strength compared to stabilization with lime. Eyo et al. [
28] conducted a study using RoadCem (RC), an additive for nanotechnology manufacturing. It was concluded that using 1% RC and replacing cement with GGBS could decrease the swelling. Fly ash geopolymer has also been shown to be a viable solution by increasing pozzolanic reactions [
29]. However, it requires supplemental additives. The use of fly ash, particularly low-calcium fly ash, can reduce the rate of heat evolution and the magnitude of the temperature rise in concrete, especially at high replacement levels. Another way to reduce swelling, at least for soil treated with lime, is mellowing [
18,
30]. It has been shown that mellowing can significantly decrease swelling and double the sulfate content.
Another important factor influencing the structure and strength of stabilized soils in cold regions is represented by freeze–thaw cycles [
31]. Yan et al. [
32] investigated the characteristics of unconfined compressive strength and pore distribution of lime–fly ash loess mixtures under freeze–thaw cycles and drying–wetting cycles through a series of experiments in the laboratory. The authors showed that the freeze–thaw cycles caused frequent phase changes and water transference in samples, which continuously lowered the friction and bite forces between the soil particles, eventually leading to lower strength.
Waste paper ash (WPA), a byproduct of recycling paper, varies in terms of its chemical and physical properties, generally depending on the raw material used during incineration. However, in most cases, WPA contains cementitious properties [
33,
34] and, to some extent, follows the same pattern as cement. After mixing WPA with water, lime makes the solution alkaline (around 12) [
35].
In the previous study, the usability of WPFA as a binder was discussed [
35]. In summary, WPFA was successfully used as the sole binder to stabilize the given soil. However, the durability of WPFA in the presence of a sulfate source was not considered. The study of the durability of WPFA is essential because of the similarities between WPFA and cement. Similarly to cement, WPFA in the presence of sulfate could swell and eventually lead to structural damage.
Moreover, swelling in soils by the formation of ettringite depends on many factors such as temperature, water and sulfate content, and time [
23,
25]. Therefore, this paper studies the long-term effect of soil stabilized using WPFA in the presence of different sulfate concentrations, by means of measuring the mechanical performance and swelling. The study valued mineralogical changes using XRD, thermogravimetry analysis (TGA), Fourier-transform infrared spectroscopy (FTIR), and scanning electron microscopy (SEM) analyses. For this purpose, all tests were conducted in different conditions (at 5 °C and 20 °C with different sulfate solutions), and the results were compared with a commonly used binder (CEM IV). This study’s findings will further reveal the usability of WPFA as a binder even in a harsh environment and would be a major importance in assessing WPFA in comparison with traditional cement.
2. Materials and Methods
2.1. Soil, Stabilizers, and Reagent
The treated soil was collected from the suburbs of the city of Zaragoza, Spain (north of Spain), where the metropolitan area is predominantly composed of soils contaminated with sulfate. Given the low load-bearing capacity of these soils due to their physical–mechanical properties, they are not used in construction work; thus, the stabilization of soils with cement materials is a practical solution. The goal was to stabilize a 30 cm layer of this soil. Upon further inspection, the subgrade soil (below the treated soil, around 0.5 m depth) showed a high amount of sulfate concentration (1.4% according to EN 103201). Hence, the study also considered the underlying soil. The properties of tested soils such as particle size distribution, liquid limit, plastic limit, sulfate content, and pH values are shown in
Figure 2 and
Table 1.
The stabilizers included a pozzolanic Portland cement (CEM IV B(Q) 32.5 N) and WPFA. The cement consisted of a pozzolanic cement with additional calcined natural pozzolana (Q) and a resistance class of 32.5 N; and the waste paper fly ash was derived from paper manufacturing. The WPFA studied in this study was supplied by Saica (Sociedad Anonima Industrias Celulosa Aragonesa), a Spanish pulp and paper manufacturer that uses only recycled paper as raw material.
The chemical composition of all raw materials (both soils and stabilizers) was determined by X-ray fluorescence, using a Philips/PANalytical spectrometer, model PW2400. The main elements in both soils were calcium and silicon. The main elements in WPFA and cement were calcium, silicon, and aluminum. There were some traces of magnesium and chlorine in WPFA, as shown in
Table 2.
Figure 3 shows the particle size distribution of PC and WPFA. The as-received WPFA contained particles with a d
50 of ~6.4 μm, whereas the cement contained particles with a d
50 of 11.8 μm, showing a far coarser particle size than WPFA.
This study applied the powder diffraction technique to identify the crystalline phases in soils and stabilizers using a Philips X-ray diffractometer with a PANalytical X’Pert PRO MPD Alpha 1 diffractometer using Cu Kα radiation (λ = 1.5406 Å, 45 kV–40 mA). The results were interpreted using EVA software (database PDF-2).
The presence of calcite, lime, quartz, larnite, aluminum, and halite was recorded in WPFA, as shown in
Figure 4. Moreover, a tiny amount of portlandite could be found due to moisture in the environment. The minerals presented in cement were quartz, calcite, mayenite, brownmillerite, gypsum, tricalcium aluminate, larnite, and calcium magnesium aluminum oxide silicate. The soil and subgrade soil were composed of quartz, calcite, albite, biotite, chamosite and mica, as shown in
Figure 5.
Moreover, in addition to these materials, calcium sulfate 2-hydrate (CaSO
4·2H
2O) from Panreac was used as a reagent. It was mixed with water in order to facilitate sulfate attack of the test specimens. The amount of calcium sulfate is described in
Section 2.3.
2.2. Sample Preparation
The studied soil was part of an experimental trial, located in Villamayor de Gállego, a small village near Zaragoza, Spain. The goal was to treat the soil for heavy traffic usage, whereby a minimum of 2.5 MPa according to the unconfined strength test (UCS) was required at 7 days in this case.
Table 3 shows the test design parameters such as the binder content and UCS for both soil + WPFA and soil + cement.
2.3. Procedure for Measuring Swelling
An in-house experiment was designed to characterize the effect of swelling under sulfate attack on stabilized soils in the long term. The experiment measured one-dimensional swelling/shrinkage in the vertical direction of a confined specimen. This allows more flexibility and a greater experimental duration without damaging the sample. PVC molds were fabricated a thickness of 0.5 cm, height of 20 cm, and diameter of 10 cm. As the base, perforated PVC was also used, with a thickness of 0.5 cm. The base and the mold were glued together using eight screws.
The preparation of soil consists of grinding the soil to obtain a maximum particle size of 16 mm. The soil was weighed and mixed with binder (either 5 wt.% WPFA or 3 wt.% cement) and water according to
Table 3. After mixing thoroughly, the mix was then poured into the mold in five layers and compacted. For samples with WPFA, after mixing with water, a 30 min delay was considered before pouring and compacting. This delay time allowed the WPFA to gain better workability and performance and as well as reduced swelling to some extent. After fabricating the samples, a 125 g sphere was added to level the top layer of the soil. Finally, to facilitate measurements, a metallic plate was added to the top of the samples. For clarity, a schematic of the mold is shown in
Figure 6.
After fabrication, the samples were cured for 7 days. In total, 36 samples were fabricated. Two temperatures were considered (5 °C and 20 °C). To determine the effect of sulfate on the samples, three sulfate solutions were considered. In the first (W1), only tap water was used. For the second batch (W2), the samples were fabricated using subgrade soil to determine the effect of underlying soil on the stabilized soil. Additionally, to avoid any loss of sulfate concentration from subgrade soil to the water bath, 2.5 g/L calcium sulfate was added to the water bath. Lastly, for the third batch (W3), 20 g/L calcium sulfate was added to the water bath. Each experiment for WPFA (in terms of water batch and temperature) was conducted in quadruplicate, whereas experiments with cement were conducted in duplicate. Due to the lack of research and the possible greater heterogeneity on the behavior of WPFA exposed to sulfate concentrations, more samples were manufactured.
Table 4 shows the number of samples and the wetting/drying conditions for each sample batch.
The wetting/drying cycles were carried out immediately after day 7. The samples were placed inside a designated water bath for 1 week, and 2 weeks to dry. This cycle was repeated for 800 days. After each cycle, the weight and the swelling were measured.
A precise displacement device was used to measure the swelling/shrinkage in the soil samples as described in previous work [
36].
Samples were placed in a water bath at 20 °C, and 90 ± 5% humidity, or at 5 °C. The water level was maintained around 1.5 ± 0.5 cm above the samples to allow the water to be drawn up into the sample via capillary action.
Figure 7 shows a schematic view of wetting portion with different sulfate solutions.
2.4. Microstructural Studies of Stabilized Soil
To study the behavior of the soil stabilized with binders, 100 g of soil was ground to a particle size of 63 µm. Later, it was mixed with 5% WPFA or 3% cement and the three different solutions mentioned in the previous section (W1: tap water, W2: containing underlying soil, W3: containing 20 g/L sulfate). After mixing thoroughly, it was poured into a container and left to cure at two different temperatures (5 and 20 °C). This experiment was conducted to accelerate the hydration process and promote the appearance of the other phases such as ettringite or thaumasite.
The hydration of the samples was stopped at different curing ages (30, 180, and 360 days) using the solvent exchange method. Then, the samples were pulverized for characterization by XRD, thermogravimetry analysis (TGA), scanning electron microscopy (SEM), and FTIR. For TGA, a Mettler Toledo model TGA/DSC 1 Thermal Analyzer was used with 10 µg of material at temperatures within the interval 30–1000 °C, N2 flow of 50 mL/min, and a heating rate of 10 °C/min. Before each test, the samples were stabilized at 100 °C for 15 min. A Frontier FTIR spectrometer (Perkin Elmer) was used to acquire 16 scans with a spectral resolution of 4 cm−1 over a range of 4000–400 in attenuated total reflection (ATR) mode.
Scanning electronic microscopy (SEM) images of soil stabilized with WPFA or cement at 360 days were obtained to verify the formation of ettringite, using a FEI scanning electronic microscope equipped with an energy-dispersive X-ray microscopy device model ESEM Quanta 200, XTE 325/D8395.
2.5. Unconfined Compressive Strength
To evaluate the changes in strength of the stabilized soil under sulfate attack, a UCS test was carried out. The samples were prepared in accordance with EN13286 and cured for 7 days at 90 ± 5% RH. Then, the samples were placed inside a tray, and the water solution (W1, W2 or W3) was poured into the tray. For this test, the samples were exposed to the solution for 1 week, and then left to dry at room temperature for 2 weeks. The experiments were carried out over a 1 year period, and then the compressive strength test was performed.