Effect of alteration on the geochemistry and mechanical properties of granite from Pingjiang, Hunan Province, China

The effect of alteration on the geochemistry and mechanical properties of granite from Pingjiang, Hunan Province, China was investigated. Six weathered and 14 hydrothermally altered samples in 3 adits were collected for mechanical strength tests, mineralogical and geochemical analysis. The types of alteration observed within the samples were chloritization and argillization. Samples taken from a silicified fracture zone were enriched in quartz. Weathering was observed to significantly weaken the granite whereas the effects of hydrothermal alteration on strength were more complex. The porosity increased with the enrichment of the altered minerals, indicating that the formation of altered minerals degrades the strength of physical bonds between minerals within the granite. The granite Na2O, CaO, K2O, MgO and SiO2 contents decreased while Fe2O3T increased due to weathering. Variations of major elements within the hydrothermally altered granite were distinguished from those observed in weathered samples—notably Mg was removed whilst Si and Fe were generally stable during the hydrothermal alteration. Whereas the quartz-enriched samples gained Si with slight depletions in Mg and Fe. Trace and rare earth elements were both removed in hydrothermal alteration. The variable behavior of major element was quantified by the mobility index which indicated that the different geochemical changes were attributed tochloritization and argillization. Furthermore, the mobility index of Mg was used to identify the dominated alteration in granite and evaluate the effects of chloritization and argillization. Generally, chloritization was found to be more dominant than argillization in weakening the granite.


Introduction
The mineralogical, geochemical, and mechanical properties of rocks can be significantly changed by alteration (del Potro and Hürlimann 2009;Huang et al. 2011;Julia et al. 2014;Moon and Jayawardane 2004;Pola et al. 2012Pola et al. , 2014Wang et al. 2015;Wyering et al. 2014). Almost all alteration occurs in two ways: (1) rocks interacting with water and other agents of atmosphere, which is called weathering (Fritz and Mohr 1984;Moon and Jayawardane 2004;Wang et al. 2013); and (2) hydrothermal fluids coming into contact with rocks causing chemical reactions. The latter process is referred to as hydrothermal alteration (Browne 1978;Wyering et al. 2014).
The influence of weathering on the mechanical properties of rocks has been well studied (Arikan et al. 2007;Ceryan et al. 2008;Julia et al. 2014;Pola et al. 2012Pola et al. , 2014Wyering et al. 2014). Weathering processes generally cause a reduction in the mechanical strength and durability properties of rocks-thereby defining a negative correlation with weathering degree (Arikan et al. 2007;Pola et al. 2012Pola et al. , 2014. According to Julia et al. (2014) and Wyering et al. (2014), the relationships between some mechanical parameters of rock (e.g., uniaxial compressive strength, compressional wave velocity, Young's modulus) and degree of weathering could be defined by exponential functions. The mineralogical and chemical changes are also recognized during weathering, whereby the presence of alumina-silicate minerals such as feldspars convert into clay minerals which further reduces rock strength (Arikan et al. 2007;Coggan et al. 2013;Wyering et al. 2014;Columbu et al. 2019Columbu et al. , 2020. While, the previous studies were mainly focused on the weathered volcanic rocks rather than intrusive igneous rocks, such as granite (Chigira et al. 2002;Duzgoren-Aydin et al. 2002;Sumner and Nel, 2002;Moon and Jayawardane, 2004;Yıldız et al. 2010;Wang et al. 2015;Columbu et al. 2019Columbu et al. , 2020. Typically, hydrothermal alteration occurs at higher temperatures and pressures compared with weathering (Browne 1978;Fritz and Mohr 1984). Therefore, the effects of weathering and hydrothermal alteration on rocks strength will be significantly different. Given the generally high-quality mechanical strength properties of granite, hydroelectric dams are often situated on such rocks, including the Three Gorges Dam in China (Chen 1999). The strength of the altered granite is directly related to dam stability. Due to the affinities between altered granite and the presence of metallic ores, the geochemical effects of alteration on the granite have been well studied (Baker 1985;Farmer and DePalol 1987;Meller et al. 2014;Xu et al. 2020). However, the mechanical strength of such hydrothermally altered granites has received little attention (Lan et al. 2003;Chen et al. 2018;Qin et al. 2019).
In this study, a Mesozoic altered granite intrusion in Hunan Province, China has been investigated. The site location is to be developed as a pumped storage hydroelectric station, which will be fed by dammed upper and lower reservoirs. Six weathered and fourteen hydrothermally altered samples were collected from three adits, which were marked as PD2, PD3, and PD4. Mechanical strength tests, major and trace element analyses were undertaken to: (1) investigate the effects of alteration on the mechanical strength properties of the granite; and (2) identify the geochemical changes during weathering and hydrothermal alteration and examine the differences between these two processes.

Geological background
The study area is located in the southern Yangtze Block where the regional structure is controlled by the deep Xinning-Miluo and Changsha-Pingjiang faults (Fig. 1). From north to south, this area is divided into four regions by these two faults, namely (1) Dongting rift basin, (2) Mufu Mountain-Ziyun Mountain uplift, (3) Pingjiang-Changsha rift basin, and (4) Lianyun Mountain-Hengyang uplift (Fig. 1). The basement strata comprise the Lengjiaxi (Mesoproterozoic) and Banxi (Neoproterozoic) Group. Due to tectonism, the granite was widely intruded in this area from the Mesoproterozoic to Mesozoic, particularly during the late Mesozoic.
Granite samples were collected from Fushou Mountain, which is located in the northern part of the Pingjiang-Changsha rift basin (Fig. 2). This granite was intruded into the Lengjiaxi Group at ca. 165 Ma (Xu et al. 2009;Zhang 1991). Field investigations have shown that the structure of this area is controlled mainly by seven faults (Fig. 2) and joints developed in the intrusion. Quartz veins and pegmatite dykes are distributed around the faults.

Sampling and analytical methods
The samples in this study were collected according to the specifications for rock testing in water conservancy and hydroelectric engineering of the People's Republic of China (DL/T 5368-2007). The weathered samples (WA) were collected at the entrances of the adits, whereas hydrothermally altered samples (HA) were collected along the adits. Five samples were collected at the end of the PD2, where there is a silicified fracture zone (SZ) (Fig. 2). The volumes of the collected samples were representative and adequate for analytical testing. The mechanical strength tests were carried out first, whereby small sub-samples were retained for the determining the bulk density, particle density and mineralogy. The powered samples were prepared by the agate mortar for geochemical analysis.

Mechanical strength tests
Ultrasonic P-wave velocities (V p ) were measured on all intact samples pioro to strength testing. Size-corrected point load strength index (I s(50) ) testing was undertaken on irregularly shaped specimens, based on previously published methodologies (Kahraman et al. 2005;Moon and Jayawardane 2004;Yang 2007). The rock strength test was carried out in situ using the rebound method with a hammer, whereby strength values were represented by the rebound value (R) (Ma 2014). The dry bulk density and particle density of samples were determined in the laboratory following the procedures presented in previous studies (Lv et al. 2011;Wang et al. 2013;Miao, 2017).

Mineralogical analysis
Thin sections of the granite samples were cut and subjected to observational analysis under a binocular petrographical microscope. Mineral compositions of the samples were determined with X-ray diffraction (XRD) at the Key Laboratory of Mineralogy and Metallogeny in the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou, China. A Bruker D8 Advance diffractometer with Ni-filtered CuKα radiation was employed, consistent with the analytical conditions and procedure used by Ma et al. (2016).

Whole-rock major and trace elements
Major and trace element concentrations were analyzed at the State Key Laboratory of Geochemistry in the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou, China. A Rigaku ZSX100e X-ray fluorescence (XRF) spectrometer was used for major element determination, following the analytical procedures of Li et al. (2006). The analytical precision was generally better than ± 2%. Trace elements were determined with a Thermo X Series II inductively coupled plasma-mass spectrometer (ICP-MS) following the procedures of Li et al. (2006), whereby the analytical precision was ± 5%.

Mechanical strength
The physical properties of the granite are summarised in Table 1. The weathered (WA) granite yielded lower dry bulk densities (2.19-2.32 g/cm 3 ) compared with the fresh (F) and hydrothermally altered (HA) samples (2.65 g/cm 3 and 2.41-2.58 g/cm 3 , respectively). The particle density values ranged between 2.58 and 2.67 g/cm 3 . Total porosity was calculated from knowledge of dry bulk density and particle density using the equation of Brown (1981). The mechanical test data are listed in Table 2. The fresh sample yielded 4.33 km/s for V p , 4,991 kPa for I s(50) and 30 for rebound value. The WA samples exhibited the lowest mechanical strength. The mechanical strength for most HA samples were lower than fresh sample except for one sample, namely XP4-200. Compared with the fresh samples, SZ samples in the silicified fracture zone possessed higher I s(50) (e.g., PD 2-2) and V p values (e.g. PD2-6).

Whole-rock major and trace elements
The major and trace element data are summarised in Table 2. Compared with the hydrothermally altered samples, weathered samples are characterised by higher concentrations of MgO (0.60-2.55 wt%) and total Fe 2 O 3 (Fe 2 O 3T ) (2.13-9.90 wt%), with lower Na 2 O (0.32-3.81 wt%), K 2 O (1.57-3.22 wt%), and CaO (0.67-2.97 wt%) concentrations. Due to the enrichment of quartz, the samples from the silicification zone are associated with high SiO 2 concentrations. The loss-on-ignition (LOI) values, which is an important parameter used to define the alteration degree of magmatic rocks, varied from 2.35 to 8.14 wt% and 0.53 to 1.58 wt% for weathered and hydrothermally altered samples, respectively. Generally, the fresh sample have the highest concentrations of trace elements. SZ samples have the variable contents of trace elements compared with HA samples (Table 1).

Effects of alteration on mechanical strength properties
Previous studies have demonstrated that weathering significantly reduces the mechanical strength of rocks, especially for the highly and completely weathered samples (Julia et al. 2014;Moon and Jayawardane 2004). The WA samples in this study were considered to be highly weathered, yielding the lowest recorded mechanical strength and dry bulk density values. According to the mineralogy analysis, the altered mineral content of the WA samples, including chlorite and clay minerals, were higher than that of the HA and fresh samples. The porosity and the altered mineral contents show a positive correlation (Table 1), indicating that the formation of altered minerals weaken the physical bonds between minerals within the granite. Therefore, the reduction of the mechanical strength of the WA granite may be attributed to chloritization and argillization.
The mechanical parameters of HA samples were variable compared with the fresh samples, thereby highlighting the complex effects of hydrothermal alteration (Fig. 4). SZ samples in PD2 were noted to have similar or even higher I s(50) and rebound values compared with the fresh sample (Fig. 4). Chloritization and argillization also occurred in the HA and SZ samples, but to a lower degree compared with the WA samples. The higher mechanical parameters   Table 2 Results of geochemical analysis and mechanical tests for granite samples in Pingjiang  of the SZ samples indicate that they were less affected by chloritization and argillization compared with WA samples and most HA samples (Fig. 4). The SZ samples were subjected to the replacement reaction during the formation of quartz veins, as a result of interactions with Si-rich hydrothermal fluids (Lv et al. 2011). Therefore, the SZ samples have higher quartz contents compared with the other samples (Table 1). Due to the high frictional resistance of quartz, SZ samples were strengthened rather than weakened by the enrichment of quartz. The LOI value was assumed to represent the water retained within a rock and its minerals, which is regulated by the degree of weathering (Moon and Jayawardane 2004; Pola et al. 2012). Water can be incorporated into the structures of secondary minerals during the hydrothermal alteration (Hurwitz et al. 2002;Pola et al. 2012Pola et al. , 2014Wyering et al. 2014;Julia et al. 2014). Therefore, the LOI value is also an indicator of the degree of hydrothermal alteration. The mechanical parameters of all samples declined rapidly when the LOI values range from 0 to 2 wt%, beyond which they were relatively constant (Fig. 5). According to previous studies, the relationships between mechanical strength and the degree of alteration could be defined by the exponential equations (Julia et al. 2014;Wyering et al. 2014). However, such expressions were not appropriate to this study (Fig. 5). The unclear relationships between mechanical strength parameters and the degree of alteration may be attributed to two reasons: (1) the heterogeneity of the granite-meaning that samples have inconsistent initial strength before alteration, and (2) the complexity of the alteration effects. As mentioned above, the chloritization and argillization could weaken the granite while the enrichment of quartz could strengthen them.
In addition, the reported correlation coefficients for the relationship between mechanical strength and degree of alteration were in the range of 0.6-0.8 (Julia et al. 2014;Wyering et al. 2014). Instead of LOI, porosity was used as a parameter to quantify the degree of alteration within samples from previous studies (Julia et al. 2014;Wyering et al. 2014). In this study, the porosity of granite increased with the enrichment of the altered minerals (Table 1). Although the porosity has a strong influence over the mechanical strength of rocks, it can also be affected by various other factors including grain size, mineral composition and microstructure (Ulusay et al. 1994;Li and Aubertin 2003;Palchik and Hatzor 2004;Baud et al. 2014;Ündül 2016). Therefore, porosity was partly partially controlled by the alteration degree of rocks, which is especially pertinent for volcanic rocks, which are usually more porous due to their vesicular structure (Saar and Manga 1999;Shea et al. 2010;Heap et al. 2014). The variable effects of hydrothermal alteration within HA samples indicates that hydrothermal alteration could affect not only porosity, but also other controlling  factors of mechanical strength of granite. A further study including more investigations on texture and porosity of hydrothermally altered granite is necessary to better understand the conditions that are responsible for the different effects of hydrothermal alteration.

Geochemical changes due to alteration
Generally, mineral alterations are accompanied by geochemical variations. Many studies have documented geochemical changes resulting from weathering of magmatic rocks, especially the volcanic and pyroclastic rocks (Aiuppa et al. 2000;Chigira et al. 2002;Columbu et al. 2019Columbu et al. , 2020Duzgoren-Aydin et al. 2002;Fritz and Mohr 1984;Guan et al. 2001;Lan et al. 2003;Moon and Jayawardane 2004;Sharma and Rajamani 2000). From these previous studies, the alteration of volcanic and pyroclastic rocks is considered to be mainly due to the devitrification of their glassy matrices, resulting in the formation of new minerals (e.g., phyllosilicates). Furthermore, weathering increased the mineralogical and geochemical changes of rocks, favoring the formation of clay minerals (Columbu et al. 2019(Columbu et al. , 2020. Na, Ca, Si, K, and Mg are mobile, whereas Zr, Ti, and Al are immobile during weathering. Thus, Na 2 O, CaO, SiO 2 , K 2 O, and MgO would be leached from rocks during weathering. FeO would be oxidized to Fe 2 O 3 during weathering, thereby increasing the total Fe 2 O 3 content and decreasing the FeO content (Aiuppa et al. 2000;Chigira et al. 2002;Guan et al. 2001;Moon and Jayawardane 2004;Columbu et al. 2019Columbu et al. , 2020. Most trace element concentrations increase during early weathering stages but decrease as weathering progresses further. Such increases in concentration during the early stages are likely to be a reflection of the loss of major elements, whilst concentration reductions likely represent the mobilization of trace elements during argillization (Aiuppa et al. 2000;Moon and Jayawardane 2004). Rb and Sr behave similarly to Ca during weathering due to their similar ionic radii properties (Moon and Jayawardane 2004).
The results for weathered samples in this study are similar to those of previous studies. Na 2 O, CaO, K 2 O, and SiO 2 contents in the weathered samples showed negative  (Fig. 6a-d), meaning that contents of these elements decreased during weathering. Fe 2 O 3T concentrations increased significantly with increasing LOI (Fig. 6f). However, MgO concentrations were different than those described in some previous studies (Guan et al. 2001;Moon and Jayawardane 2004). MgO contents in the weathered samples showed a positive correlation with LOI values, indicating that the samples gained MgO during weathering (Fig. 6e). A similar increase in MgO within weathered samples was reported by Chigira et al. (2002). The inconsistent behavior of MgO may be attributed to variations in alteration during the weathering (Aiuppa et al. 2000;Chigira et al. 2002;Guan et al. 2001;Moon and Jayawardane 2004).
The geochemical changes of the granite during hydrothermal alteration were complex and mainly depend upon the types and conditions of alteration (Baker 1985;Farmer and DePalol 1987;Wang et al. 2013;Meller et al. 2014;Wyering et al. 2014;Xu et al. 2020). The loss of Ca, Na and Fe is reported when biotite and plagioclase are altered to muscovite and clay minerals (Xu et al. 2020). During albitization of granite, Na, Si and Mg are added, while K and Fe are lost (Baker 1985). In this study, major elements of the hydrothermal altered samples showed no clear relationship with LOI and the variation trends are distinguished from weathering (Fig. 6). Na and K can be depleted but also enriched during the hydrothermal alteration (Fig. 6a, c). Ca was largely removed from granite (Fig. 6b). Si and Fe were generally stable during the hydrothermal alteration (Fig. 6d, f). The quartz-enriched samples yielded the addition of Si (Fig. 6d) and small losses of Mg and Fe (Fig. 6e, f). The trace elements and rare earth elements (REE) in hydrothermally altered samples were depleted; in only one sample (PD2-2) were heavy REE introduced (Fig. 7a, b). REE depletion during the chloritization and argillization was reported, whereas the enrichment of heavy REE were attributed to the affinities of heavy REE in chlorites (Alderton et al. 1980;Baker 1985;Dawood et al. 2005;Xu et al. 2020). In addition, Fig. 7 Primitive mantle normalized trace element patterns (a) and chondrite-normalized rare earth elements patterns (b) for the samples from PD2. Compositions of chondrite and primitive mantle are from Sun and McDonough (1989) the trace elements and REE of the SZ samples varied in a wider range compared with the HA samples (Fig. 7a, b).
The mobility of major elements could be quantitively evaluated with the mobility index (MI), which was described by Guan et al. (2001). Based on fieldwork and petrological observations, samples from PD2 and PD3 were noted to have similar mineral compositions. Therefore, these samples were ideal for assessing element mobility due to weathering and hydrothermal alteration. The calculated activities of Al species in weathering and many hydrothermal systems are very low, and Al is conserved during the conversion of biotite to chlorite (Helgeson 1970;Parry and Downey 1982). Therefore, Al 2 O 3 was taken to be immobile elements, and the unaltered sample PD2-7 was used as the baseline sample to assess alteration effects. The MI of major element was then calculated as follows: where R i a is the weight percentage of mobile element i in the altered sample, R a is the weight percentage of immobile elements in the altered sample, R i p is the weight percentage of mobile element i in the unaltered sample, and R p is the weight percentage of immobile elements in the unaltered sample.
The calculated MI for major elements normalized to Al 2 O 3 are shown in Fig. 8. The enrichment of Mg and Fe distinguish weathered samples from hydrothermally altered sample. The Si is generally immobile but could be mobilized during the quartz-enrichment. As mentioned above, the alterations in weathered and hydrothermally altered granite were both chloritization and argillization. Therefore, the mobility of elements is supposed to be similar in weathered and hydrothermally altered granite. The variable mobility of the major elements indicates the different geochemical changes in chloritization and argillization.

Differences in element mobilities due to alteration
Argillization and chloritization occur under different thermal conditions and by distinct processes. Argilliaceous rocks are formed at low temperatures (50-150 ℃) (Julia et al. 2014). During the early stages of argillization, plagioclase reacts with K-rich acidic hydrothermal fluids, forming smectite/illite. As the reaction progresses, H + replaces K + in K-feldspar and smectite/illite, forming kaolinite. The major compositional change in the rock is the removal of Ca, Na, and some Mg. K tends to remain constant, or slightly increases and then decreases (Hemley and Jones 1964). Thus, changes in K and the loss of Ca and Mg in the altered samples of this study are attributed mainly to argillization (Fig. 8). However, changes in Na content are not consistent with argillization, as the HA and SZ samples were observed to increase in Na content (Fig. 8). Thin-section observations revealed that orthoclase was partially altered to albite (Fig. 3a). During this process, K was replaced by Na, thereby increasing the Na concentration. Therefore, the formation of albite might be responsible for the increases in Na in the altered samples. The temperatures of chlorite formation are 150-300 °C (Huang 2017). Previous studies have concluded that the alteration of biotite to chlorite in granite conserves Al, resulting in the loss of K and Ti, along with the gain of Mg and Mn (Parry and Downey 1982). Therefore, the enrichment of Mg can be attributed to chloritization. Quartzenrichment occurs due to reactions with high-temperature fluids (300-550 ℃) (Julia et al. 2014). This process significantly increases SiO 2 concentrations of samples in two ways: (1) SiO 2 derived from the hydrothermal fluids and (2) by decomposition of feldspar to clay minerals (e.g., argillization; Lv et al. 2011). In this study, the MI of SiO 2 in the non-silicified samples are all near unity (Fig. 7). As such, silicification associated with argillization was negligible. Therefore, the SiO 2 added to the silicified samples was derived from hydrothermal fluids. The variable trace elements and REE in the SZ samples also indicate that the chloritization and argillization may be motivated in this process (Fig. 7a, b).
As mentioned above, Mg is noted to be enriched by chloritization while depleted by argillization. Therefore, the MI of Mg could be used to identify the dominated alteration when these two types of alteration both occurred. The chloritization-dominated samples were expected to have a high MI of Mg (higher than 1) while the MI of Mg within the argillization-dominated samples should be lower than 1. Generally, the argillization-dominated samples have the higher mechanical strength parameters (e.g., average of I s(50) = 3555 kPa) compared with the chloritization-dominated samples (e.g., average of I s(50) = 1150 kPa). This may indicate that the chloritization could weaken the granite more effectively.

Conclusions
The effects of alteration on the mechanical properties of granite were investigated using mechanical tests and geochemical data. Due to the high grade of weathering, weathered monzogranites have a low mechanical strength whereas the strengths measured for hydrothermally altered monzogranites were more variable. The porosity increased with the enrichment of the altered minerals, indicating that the formation of altered minerals weaken the physical bonds between minerals within the granite. Monzogranite was weakened by argillization and chloritization, but strengthened by the enrichment of quartz. A further study on hydrothermally altered granite is necessary to identify and better understand the conditions corresponding to the different effects of hydrothermal alteration on mechanical strength.
Granite Na 2 O, CaO, K 2 O, MgO and SiO 2 contents decreased while Fe 2 O 3T increased due to weathering. Variations of major elements within the hydrothermally altered granite were distinguished from those within weathered samples. Ca was removed from granite significantly while Si and Fe were generally stable during the hydrothermal alteration. The quartz-enriched samples gained Si, whilst losing some Mg and Fe contents. The trace and rare earth elements were both removed in hydrothermal alteration. The variable behavior of major element was quantified by the mobility index. The variable mobility of the major elements indicates the different geochemical changes in chloritization and argillization.
Argillization and chloritization occur under different thermal conditions and by distinct processes, leading to the different characteristics of the elements. Mg could be enriched by chloritization while depleted by argillization. This observation was used to identify the dominant type of alteration in granite and evaluate the effects of chloritization and argillization. Generally, the chloritization was noted to weaken the granite more than argillization.