Late Cretaceous Activity Record of the Guangsan Fault—Insights from Zircon U-Pb and Apatite Fission-Track Thermochronology

Abstract

The timing of fault activity is a concern for geologists. This study used zircon U-Pb and apatite fission-track dating of fault breccia to determine the upper and lower limits for the time of faulting. The Guangsan fault in South China was taken as an example, and zircon U-Pb and apatite fission-track thermochronology were applied to the surrounding rock and fault breccia. The surrounding rock and fault breccia demonstrated 74.9–91.8 Ma and 73.9–93.5 Ma zircon U-Pb dates, respectively, indicating that the breccia formed after 73.9 Ma. They also demonstrated 71.6 ± 7.3 Ma and 85.9 ± 8.2–65.5 ± 6.5 Ma fission-track dates, implying that the fault breccia samples likely formed before ~70 Ma. Their thermal histories were highly consistent: both showed rapid cooling during 70–65 Ma and slow cooling during 65–0 Ma, implying that the fault was likely still active during 70–65 Ma, resulting in the rapid exhumation.

1. Introduction

The timing of fault activity is usually determined according to the stratigraphic units cut by the fault, and there are several methods that can be used to date the fault rocks directly, such as Ar-Ar dating of mylonites (e.g., [1,2]) and electron spin resonance dating of fault gouges (e.g., [3,4]). Since the end of the last century, low-temperature thermochronology has created new possibilities for obtaining the timing of fault activity; e.g., using apatite fission-track (AFT) dating to date fault gouges (e.g., [5,6,7,8,9]), cataclasite (e.g., [10,11]) and mylonite (e.g., [9,12]) or using horizontal or vertical sections to estimate fault history (e.g., [13,14,15]). Although these methods can be used to directly date fault rocks or to constrain the surrounding rock evolution on both sides of the fault, they also have certain disadvantages; for example, it is difficult to determine the time span of fault activity using a single dating method, and a large number of samples are needed to compare the date variations across the fault.

This paper proposes to constrain the age of fault activity by using both zircon U-Pb and apatite fission-tracking to date the formation of fault breccia. This approach is suitable for studies of the faults with few outcrops; e.g., hidden faults passing through a city. We studied the Guangsan fault to illustrate this approach. The Guangsan fault is a deep and hidden fault [16] across from the urban centre of Guangzhou, a large city of more than 10 million people, that has influences on urban construction and earthquake hazards [17]. Dating the fault activity helps us to understand the evolution of the fault.

2. Geological Setting and Sampling

The Guangsan fault, with a depth of 30 km [18,19], is the middle section of the Gaoyao–Huilai fault, which is one of the longest EW-trending faults in the South China Block. The Gaoyao–Huilai fault zone (Figure 1a) is located in southern China, near the South China Sea. It is about 200 km long and 10–60 km wide [16]. It determines the flow direction of the rivers and the topography of the Pearl River drainage, which is one of the biggest drainage systems supplying the northern South China Sea (Figure 1a). The fault formed during the Trassic and was again active during the Jurassic–Cretaceous with granite intrusion and volcanic overflow along the fault; e.g., basalt and trachyte (64–57 Ma) [16].

The Guangsan fault and Shougouling faults are both located in the middle section of the Gaoyao–Huilai fault zone and are the two branches of the fault zone (Figure 1a). Studies have mainly focused on the Shougouling fault. Zou et al. obtained a ⁴⁰Ar/³⁹Ar plateau date of 172.27 Ma from the mylonite [20]. Zhu et al. obtained AFT dates of 36.2 ± 0.9 Ma, 29.9 ± 4.4 Ma and 27.1 ± 4.3 Ma for fault gouges [6]. Quaternary fault activity has also been documented [6,20,21]. Unlike the Shougouling fault, which is exposed between the Cretaceous strata and intrusive rocks, the Guangsan fault is mostly hidden under the Quaternary strata. Therefore, study of the Guangsan fault is still limited to the Quaternary strata [21,22,23]. This study provides new evidence of the pre-Quaternary history of Guangsan fault.

The strata exposed in the study area include Cretaceous and Quaternary strata (Figure 1b). The Cretaceous strata are composed of the Baihedong formation, Sanshui formation and Dalangshan formation, from bottom to up [27]. The temporal boundaries between the formations are 93 and 80 Ma, respectively [28]. The Baihedong formation comprises siltstone and mudstone with thin interlayers of limestone and gypsum, the Sanshui formation comprises red sandstone and conglomerate with thin interlayers of gypsum and the Dalangshan formation comprises sandstone and mudstone [27,28,29]. Both the Sanshui formation and Dalangshan formation contain granitic conglomerates, and those of the Dalangshan formation are coarser than those of the Sanshui formation [29]. The Baihedong formation, Sanshui formation and Shuzhugang rock body (mainly composed of rhyolite porphyry) are cut by the Guangsan fault and form a sinistral strike-slip fault.

Seven samples were collected in Guangzhou city along the fault zone: four samples (#1, #2, #3, #4) from the deep subway excavation (samples #2 and #4 were located perpendicular to the direction of the fault and were up to tens of meters wide, while samples #1 and #3 were located on the fault plane; see Appendix A), two samples from drill rock cores (MKZ2-A90, TTL-23) at Jiangtai Road subway station and the final sample (#7) from drill rock core nearby Chishajiao subway station. Samples #2, #4 and TTL-23 were rhyolite porphyry (country rock), samples #1, #3 and MKZ2-A90 were fault breccia (fault rock) and sample #7 was conglomerate (country rock) from the Sanshui formation. Photos of the samples are shown in Appendix B. Zircon U-Pb dating was conducted on all samples and AFT dating on four samples (#1, #3, MKZ2-A90 and TTL-23).

3. Analytical Methods

Zircon and apatite were separated by crushing, sieving and magnetic and heavy-liquid separation and then handpicked under a binocular microscope. Then, the zircon and apatite grains were mounted in epoxy resin and polished to expose internal surfaces.

The zircon grains were photographed using a JSM-IT100 scanning electron microscope (JEOL Ltd., Tokyo, Japan) coupled to a Gatan MiniCL cathodoluminescence (CL) spectroscope (Gatan, Inc., Pleasanton, CA, USA). Zircon U-Pb dating was performed using an Agilent 7900 inductively coupled plasma–mass spectrometer coupled to a GeoLas HD 193 nm laser ablation system at SampleSolution Analytical Technology Co., Ltd in Wuhan, China. Argon was used as the make-up gas and mixed with the carrier gas Helium before use in the plasma–mass spectrometer. The spot size and frequency of the laser were set to 32 µm and 5 Hz, respectively. Zircon 91500, GJ-1, Plesovice and glass NIST610 were used as external standards. The software packages ICPMSDataCal [30] and Isoplot v3.06 [31] were used to perform the quantitative calibration for U-Pb dating and to plot concordia diagrams, respectively.

Apatite grains were etched with 5.5 M HNO₃ for 20 s at 21 °C. Then, direct U determination was performed using an Agilent 7800 inductively coupled plasma–mass spectrometer coupled (Agilent, Yokogawa, Japan) to a New Wave UP 213 nm laser ablation system at Isotope Geology Laboratory of ChronusCamp in São Paulo, Brazil. A mixture of argon (950 mL/min), helium (440 mL/min) and hydrogen (4.5 mL/ min) was used as the carrier gas for the plasma–mass spectrometer. The laser spot size was set to 40 µm and the frequency to 5 Hz. The Mud Tank and Durango apatites were used as the standard samples [32,33]. The AFT dates of the grains were calculated from the spontaneous fission-track density and the U concentration. Then, the confined fission-track lengths were measured under an optical microscope. The software Low-T Thermo (V5.0) [34] was used to perform the thermal history modelling.

4. Analytical Results

4.1. Zircon U-Pb Dating

A total of 25 spots on the zircon grains were dated for each sample. CL images of the analysed zircon grains are illustrated in Appendix C, and all spots were located at the edges of the zircons, which meant that the dates were representative of the latest crystallographic regrowths. All the zircons were euhedral or subhedral, except for a few zircons from sample #7, which were rounded. The grain sizes mainly ranged from 70 to 200 μm. They exhibited clear oscillatory zoning. All the analyses (Appendix D) indicated Th/U ratios > 0.1, most of them > 0.4, indicating that all the zircons were of magmatic origin. The zircon dating U-Pb concordia diagrams are illustrated in Figure 2. The ²⁰⁷Pb/²⁰⁶Pb ages of the zircons (>1000 Ma) and ²⁰⁶Pb/²³⁸U ages of the remaining zircons were used for the discussion.

The zircon grains of sample #2 were 70–150 μm long. Among the 20 spot analyses with concordances ≥ 90%, the dates of 18 data points were concentrated in the late Cretaceous (91.8–78.2 Ma). The dates of two were early Cretaceous (113 ± 2.7 Ma and 150 ± 2.1 Ma).

Most zircon grains of sample #4 were 140–200 μm long. Among the 22 spot analyses with concordances ≥ 90%, the dates of 20 data points were concentrated in the late Cretaceous (91.5–75.9 Ma). The dates of two were early Cretaceous (120 ± 1.2 Ma) and one was early Paleozoic (437 ± 4.1 Ma).

The zircon grains of sample TTL-23 were mostly 130–200 μm long. The dates of 20 data points with concordances ≥ 90% were concentrated in the late Cretaceous (91.7–74.9 Ma).

The zircon grains of sample #1 were mostly 100–200 μm. Among the 25 spot analyses with concordances ≥ 90%, the dates of two data points were late Cretaceous (88.2 ± 1.2 Ma and 82.1 ± 1.0 Ma), five were early Cretaceous (169–135 Ma) and two were Triassic (218 ± 1.9 Ma and 243 ± 2.8 Ma), while 15 of the data points were concentrated in the early Paleozoic (422–462 Ma) with a weighted mean of 449.8 ± 3.8 Ma (N = 14, MSWD = 2.7). Only one data point was dated to the Proterozoic (1050 ± 46 Ma).

The zircon grains of sample #3 were mostly 150–200 μm. Among the 22 spot analyses with concordances ≥ 90%, 20 points were concentrated in the late Cretaceous (93.5–79.4 Ma). The other two were Jurassic (161 ± 2.7 Ma) and early Paleozoic (415 ± 3.8 Ma).

The zircon grains of sample MKZ2-A90 were mostly 90~200 μm. Among the 19 spot analyses with concordances ≥ 90%, 11 data points were concentrated in the late Cretaceous (87.7–73.9 Ma). The date of one was Jurassic (157 ± 1.4 Ma), the dates of three points were Triassic (238–235 Ma) and the dates of four points were early Paleozoic (475–437 Ma).

The zircon grains of sample #7 were mostly 90–200 μm. Among the 23 spot analyses with concordances ≥ 90%, the date of one was early Cretaceous (103 ± 1.5 Ma). The dates of 2 were Triassic (233 ± 2.7 Ma and 235 ± 2.4 Ma) and the dates of 12 were concentrated in the early Paleozoic (405–470 Ma) with a weighted mean of 438.5 ± 6.2 Ma (N = 7, MSWD = 1.5). The dates of seven were Proterozoic (747–2637 Ma).

4.2. AFT Analysis Results

The AFT results for four samples (TTL-23, #1, #3 and MKZ2-A90) are summarized in

Table 1. The AFT results for this study.

Sample	Depth (m)	N (Grains)	Ns (Tracks)	Rho (s) ×106	Rho (Zeta) ×105	Zeta (mMS) ×103	UUNK (µg/g)	Pooled Age (Ma)	Central Age (Ma)	Minimum Age (Ma)	p (χ2)	Mean Length (μm)	n (Tracks)
TTL-23	41.8–41.9	40	169	0.26	1.65	1.772	9.06	51.5 ± 4.2	71.6 ± 7.3	52.1 ± 13.1	0.01	14.92 ± 0.84	98
#1	~24	37	106	0.82	1.65	1.772	2.25	57.9 ± 5.8	65.5 ± 6.5	65.4 ± 6.5	0.93	14.93 ± 0.93	35
#3	~26	40	127	0.19	1.65	1.772	5.65	61.9 ± 5.8	69.3 ± 6.3	69.5 ± 6.8	0.95	14.77 ± 0.97	108
MKZ2-A90	28	40	145	0.23	1.65	1.822	5.26	70.0 ± 6.8	85.9 ± 8.2	73.1 ± 8.0	0.49	14.82 ± 0.87	47

Note: All the errors are 1σ. n: number of confined tracks.

, and the detailed results are listed in Appendix E. The AFT date of the rhyolite porphyry sample TTL-23 was 71.6 ± 7.3 Ma and the AFT dates of the fault breccia samples (#1, #3 and MKZ2-A90) were 65.5 ± 6.5 Ma, 69.3 ± 6.3 Ma and 85.9 ± 8.2 Ma, respectively.

Although sample TTL-23 had the largest dispersion (Figure 3) and the lowest p (χ²)-value (Table 1), the overdispersion was ignored because the sample was rhyolite porphyry. All the other samples had lower dispersions (Figure 3) and passed the χ²-test (p (χ²) > 5%, Table 1), implying that these three samples most likely had only one single date population. The mean track lengths of all samples ranged from 14.77 ± 0.97μm to 14.93 ± 0.93 μm (Table 1), and the length distributions of all samples were unimodal, with a peak at ~15 µm (Figure 4), indicating fast cooling histories since the length standards were comparable to most age standards and the standard deviations were <1 µm.

The thermal history modelling was conducted for samples TTL-23 and #3 as they had sufficient confined tracks (n = 98 and 108, respectively). D_par was not used for compositional correction. For the thermal history modelling, we used the following parameters and models: the surface temperatures at sea level were set to 20 °C; the fanning curvilinear fit annealing model [36] was used as the AFT annealing model for c-axis projected track lengths; the c-axis projected confined track-length distribution was used for the Kolmogorov-Smirnov test; and the initial c-axis projected track length was set to 16.3 μm. The overall fit was determined by the minimum of two probabilities (age and length). The results are shown in Figure 5. The mean thermal histories (MTHs) within the 50% confidence intervals (i.e., goodness-of-fit (GOF) ≥ 0.5) were used as the final thermal history modelling result. The results indicated that the thermal histories of samples TTL-23 and #3 showed relatively rapid cooling from 70 to 65 Ma and slow cooling from 65 to 0 Ma. Their results are remarkably consistent.

5. Discussion

The zircon U-Pb dates of the rhyolite porphyry samples from country rock were concentrated in the late Cretaceous. In contrast, the zircon U-Pb dates of the sedimentary samples from country rock ranged from early Cretaceous, Triassic and early Paleozoic to Proterozoic. The zircon U-Pb dates of the fault-zone samples were also dispersed, ranging from late Cretaceous, late Jurassic–early Cretaceous and early Paleozoic to Proterozoic. This implies that the source rock of the fault breccia samples was most likely rhyolite porphyry and sedimentary rock.

The late Cretaceous zircon U-Pb dates of the three rhyolite porphyry samples were concentrated in the range from 91.8 to 74.9 Ma, and the late Cretaceous zircon U-Pb dates of the three fault breccias were concentrated in the range from 93.5 to 73.9 Ma. Combining these two groups, it can be inferred that there was multiple-stage magmatic activity or an extended period of volcanic activity in the late Cretaceous from 93.5 to 73.9 Ma. Therefore, the entire zircon U-Pb dataset was used to obtain a maximum age for the faulting since the faulting occurred after the deposition of the rhyolite porphyry; i.e., the faulting occurred after 73.9 Ma.

The AFT results, which indicated similar AFT dates (around 70 Ma), unimodal distribution of AFT confined lengths (Figure 4) and high χ² values for AFT grain dates (Table 1), implied that the three fault breccia samples likely underwent complete annealing with the rhyolite porphyry around 70 Ma. This also implied that the fault breccia samples likely formed before ~70 Ma. Both friction and hydrothermal infilling are possible reason for this reset, but further research is needed.

Based on the reheating model, the thermal history modelling results of samples TTL-23 and #3 showed rapid cooling during 70–65 Ma and slow cooling during 65–0 Ma. The cooling trends of both samples were highly consistent. One possible reason is that the rock from samples TTL-23 and #3 underwent the same cooling history after the forming of the fault breccia. Another possible reason is that the rock from sample #3 was cut during periods of uncertainty, as most of the zircon U-Pb dates from sample #3 were late Cretaceous, and rhyolite porphyry was most likely the main material. In either case, the implication is that the fault was likely still active during 70–65 Ma and causing rapid exhumation.

The Dalangshan formation around the Guangsan fault includes granitic gravels coarser than those of the Sanshui formation [29]. The Dalangshan formation is roughly contemporaneous with the faulting, which implies that the fault zone possibly provided coarse material for the Dalangshan formation. Rapid exhumation also occurred during ~60–80 Ma along the continental margin of the South China Block [37,38,39], suggesting that both the faulting and exhumation were probably driven by extensional collapse of the continental margin of the South China Block after the Yanshan orogeny [16,20].

6. Conclusions

We present a method for dating faulting. It involves comparing the zircon U-Pb dating of both surrounding rock and fault breccia to determine whether the breccia originates from the surrounding rock, providing an upper limit for the time of faulting. The apatite fission-track dates of the surrounding rock and fault breccia provide a lower limit for the time of faulting.

When applied to the Guangsan fault in South China, the results revealed that the surrounding rock had zircon U-Pb dates of 91.8~74.9 Ma and the fault breccia dates of 93.5~73.9 Ma. The two kinds of rocks had similar AFT dates of around 70 Ma and underwent rapid cooling during 70–65 Ma. This implies that faulting along Guangsan fault likely occurred during ~73–65 Ma.