The history of mountain building along the northern Tibetan margin since its initiation remains unclear. The exhumation evolutionary history of the Kunlun Belt, the first-order mountain range of northern Tibet, is resolved by using 40Ar/39Ar thermochronological analyses of Paleozoic and Mesozoic granitic intrusions. Four rapid exhumation events are identified from analyses employing multiple domain diffusion theories in the Carboniferous (~355-295 Ma), Triassic (~245-205 Ma), Cretaceous (~120-95 Ma), and Eocene (~40-35 Ma). The cooling rates and the therefrom deduced denudation rates are estimated for these stages. The events are interpreted to reflect the closure of the Prototethys Ocean in the early Paleozoic, closure of the Paleotethys ocean in the late Paleozoic, far-field effects from the closure of the Mesotethys Ocean, and far-field effects from the Paleogene convergence of India and Eurasia, respectively. These events collectively built up the present northern Tibetan margin.

The mechanisms that produced the evolution of the high topography along the northern Tibetan margin have long been a topic of debate [18]. Although the growth of the northern Tibetan margin in the Cenozoic has been well constrained to understand the far-field effects of Eurasia-Indian collision at ~47-65 Ma [911] and plateau rising, the high elevation is actually a final result of a long exhumation history since the closure of the Paleotethys Ocean in the late Paleozoic [2, 4, 6, 7, 12]. The evolutionary history during those times is crucial for understanding the dynamic setting of the northern Tibetan margin, which may control the evolution and final uplift of the high mountains in the region.

The Kunlun Belt is the first-order orogenic structure that defines the northern Tibetan margin, the exhumation history of which might reveal the underlying geodynamics and constrain the topographical evolution. A small number of research programs have concentrated on the uplift and exhumation evolution of the Kunlun Belt during the Cenozoic ([10]; Wang et al., 2004; [1, 8, 9, 11]), from which the earliest response to the Tibetan growth and the onset of mountain building have been resolved. A longer exhumation history since Paleozoic times is still lacking along the northern Tibetan margin and is needed for a better understanding of mechanism and to provide a panorama of the geological evolution since the closure of the Paleotethys Ocean.

Felsic intrusions are widespread in the basement of the Kunlun Belt [10] and provide indications of the magmatic events during the geological evolutionary history of the northern Tibetan margin [3, 5, 7, 1217]. These rocks have experienced cooling related to either exhumation or pure conduction. The argon isotopic system within slowly cooled rock is capable of recording the successive exhumation history of the intrusive rocks due to the nature of its temperature-dependent diffusion and could be taken as a proxy for uncovering the exhumation history of the basement rocks, which reflects the underlying dynamics.

In this article, we obtain results on the successive exhumation history of the batholith of the Kunlun Belt and resolve the different phases of the evolution since the Paleozoic by using 40Ar/39Ar thermochronology, based on which the exhumation and denudation history of the northern Tibet margin since the Paleozoic is reconstructed.

Located on the northern margin of the Tibetan plateau, western China, the Kunlun Belt is the first-order mountain belt of the northern Tibetan margin and separates the Qaidam Basin from the Tibetan plateau (Figure 1). Composed mainly of Devonian to Early Triassic marine sedimentary rocks, Jurassic to Cenozoic sedimentary rocks, and pre-Cenozoic granitoids, the Kunlun Belt is considered to be a result from opening and closing repeatedly of the same ocean and the uplift of the Tibetan plateau during the Paleozoic to Cenozoic [1820]. The granitoids rocks intruded into the pre-Cambrian metamorphosed basement in the lower crust and sedimentary rocks in the upper crust [1, 12, 18, 2023]. Radiometric dating shows that the granitoids are mainly two groups in age: 515-390 Ma (Cambrian-Devonian) ([12, 23, 24] and references therein) and 270-200 Ma (Permian–Triassic) ([1, 12, 2123] and references therein) since the Cambrian time. These rocks are considered to be a product of arc-type magmatism [7] and comprise parts of batholith of the Kunlun Belt, which was built by a long-lived south-facing subduction zone [3, 5]. These units also form the Qimantagh Mountains, surrounding the Kumukol Basin together with the Kunlun Mountains in the middle part of the northern Tibetan margin (Figure 1). The Qimantagh Mountains extend southeastward to and terminate at the Kunlun Belt. During the Mesozoic times, the Qimantagh and Kunlun Mountains shared similar uplift and denudation histories [11], suggesting that the two mountain ranges can be considered as one unit of the Kunlun Belt.

Seventeen samples were collected from major intrusive granitoid bodies with various U/Pb ages ranging from the Paleozoic to the Mesozoic [1, 2528] at a large number of localities to understand whether and when the studied area experienced the same exhumation history (Figure 1). Lithologically, the samples are undeformed granitoids (Table 1). Previous zircon U/Pb ages for some samples from the literature are also listed in Table 1 for comparison. Based on these U/Pb dates (Table 1), the samples can be divided into three groups: four from the Devonian (391±3394±13 Ma), twelve from the late Permian to early Triassic (253.1±5.4234.0±2.8 Ma), and one from the early Jurassic (194.0±1.0 Ma).

3.1. Analytical Technique

All samples were analyzed at the 40Ar/39Ar Laboratory at the Institute of Geology and Geophysics of the Chinese Academy of Sciences (IGGCAS), Beijing, China. The rock samples were crushed first, then, the K-feldspar and biotite crystals sized between 200 and 280 μm were handpicked under a microscope, and all grains with visible impurities were removed. The K-feldspars were carefully examined to meet the requirements of thermochronological analyses [29, 30]. Checked by using electron microprobe analyses, all those altered K-feldspars that the initial microstructures had been replaced by ultraporous adularia were removed. These K-feldspars likely suffered low-temperature (<450°C) recrystallization [31] and cannot be used for thermochronological study [29, 30].

All K-feldspar and biotite samples were wrapped in aluminum foil to form wafers that sized 4.0 mm in diameter and 2.0 mm in thickness. The wafers were stacked in quartz vials of 30 mm in length and 5.0 mm in inner diameter with the standard GA1550 biotite (98.8±0.5 Ma) [32] as a neutron fluence monitor. Then, the vials with samples were irradiated in vacuo within a cadmium-coated canister for 25 h in position H8 of 49-2 Nuclear Reactor (49-2 NR), Beijing, China. The H8 position lies in the core of the reactor and receives flux from all directions. Interference reactions were checked for irradiation by using pure CaF and K2SO4, which produced correction factors used in this study, which are 36Ar/37ArCa=0.000261±0.00014, 39Ar/37ArCa=0.000724±0.00028, and 40Ar/39ArK=0.000880±0.00023.

Samples were placed into a Ta tube resting in the Ta crucible of an automated double-vacuum resistance furnace to extract gas. These were incrementally heated in 12-17 steps of 10 min each from 550 or 600 or 650 to 1300 or 1350 or 1400°C for 40Ar/39Ar analyses of the biotite samples. To reveal the argon distribution in the K-feldspars as precisely as possible (for later thermal modeling), a high-resolution step-heating (36-40 steps from 450 to 1400°C) technique was employed. The extracted gas was introduced into the prep-line for purification on Al-Zr getters (5 min), and then, the purified gas was introduced into a MM5400 mass spectrometer for argon isotopic measurements with an electron multiplier, using 13 cycles of peak-hopping mode. The mass discrimination of the machine was monitored 2 to 3 times per day via measurements on atmospheric argon. The mean mass discrimination factor over this period was 1.0060±0.0002 per amu, and the uncertainty of this value is propagated into all age calculations. Hot blanks of the whole procedure were monitored every 3 measurements and were typically in the range 3.0×1016 mols for 40Ar and 9.0×1019 mols for 36Ar in nearly atmospheric ratios and 2-3 orders of magnitude smaller than sample signals. Although the mean blank errors were generally ~2% for 40Ar and ~5% for 36Ar, the large sample size (~10-15 mg in this study) of the samples relative to the blank minimized the impact of propagating these errors into the final age calculations.

Plateau ages for biotite were determined from 3 or more contiguous steps, comprising >50% of the 39Ar released revealing concordant ages at the 95% confidence level. The uncertainties in plateau ages were obtained by standard weighting of errors for individual steps according to the variance [33]. Thus, more precise determinations were given greater weight than those of lower precision, and the overall uncertainty about the mean value may be greatly reduced. Errors are reported at the 2σ confidence level.

Analytical results are listed in Table 1 and in the Supplementary Table (Table S1), and the biotite ages and the maximum and minimum ages of K-feldspars are also included. The age spectra are shown in Figure 2. To make comparison, zircon U/Pb ages of 10 samples (13kl-1, 13kl-5, 13kl-7-1, 13kl-11, 13kl-16, 13kl-17, 13kl-20-1, 13kl-27, 13kl-29, and 1335-2) from the same parental rock are also listed in Table 1 [1, 2528], which are consistent with the cooling sequence of their biotite ages and K-feldspars maximum ages of 40Ar/39Ar analyses.

3.2. 40Ar/39Ar Analytical Results

Apart from 13kl-17 (Figure 2(i)), all the biotites from samples of 13kl-5, 13kl-7-1, 13kl-11, 13kl-13, 13kl-14-1, 13kl-19, 13kl-20-2, 13kl-22, 1335-4, and 13312-1 yielded well-defined plateau ages (Figures 2(b)–2(f), 2(j)–2(l), 2(p), and 2(q)), ranging from 295.3±2.2 to 119.1±0.9 Ma (Table 1 and Figure 2). Biotite from 13kl-17 (Figure 2(i)) yielded a U-shaped age spectrum that is possibly related to excess 40Ar trapped both in nonvolume locations (grain boundaries) and in anion sites of mineral crystals [34]. The central portion of the age spectrum forms a short plateau consisting of 20% of the degassed 39Ar, interpreted as a geologically meaningful age (295.3±2.2 Ma).

The K-feldspar from samples of 13kl-1, 13kl-7-1, 13kl-11, 13kl-14-1, 13kl-15-1, 13kl-16, 13kl-17, 13kl-19, 13kl-20-2, 13kl-22, and 13kl-27 shows age spectra characterized by an age gradient in a staircase-shaped pattern (Figures 2(a), 2(c), 2(d), and 2(f)–2(m)), of which the two samples (13kl-16 and 13kl-17) are affected by a small amount of excess argon in the initial percentage of the total degassed 39Ar. The maximum ages of these age spectra range from 353.8±2.8 to 116.5±1.0 Ma, and the minimum ages range from 163.1±3.5 to 26.5±1.9 Ma.

On the contrary, the K-feldspar from samples of 13kl-5, 13kl-13, 13kl-29, 1335-2, 1335-4, and 13312-1 yields relatively flat K-feldspar age spectra (Figures 2(b), 2(e), and 2(n)–2(q)), with maximum ages ranging from 243.8±2.0 to 215.6.1±1.2 Ma. These age spectra displayed an age gradient for the first 10% of the released 39Ar, yielding minimum ages primarily ranging from 205.0±9.3 to 122.6±4.4 Ma.

The staircase-shaped age spectra of the K-feldspars are results of the progressive closing of the Ar domains within the K-feldspars as the parent rocks cooled from ~355 to 150°C at a cooling rate of 10°C/Ma [3436]. Therefore, the maximum ages of the age spectra reflect the closure times of the largest domains at ~350°C, whereas the minimum ages represent the closure times of the smallest domains when the temperatures reached 150°C. The intermediate-sized domains represent the closure times for temperature between ~350 and 150°C and result from the cooling between ~350 and 150°C caused by exhumation. The rough consistency between biotite ages with the K-feldspar maximum ages reflects the similar closure temperature of biotite (310°C [37]) to the largest domain of K-feldspar at a similar grain size and a cooling rate (10°C/Ma).

According to the maximum ages of K-feldspar, all samples can be divided into three groups. Group 1, 13kl-7-1, 13kl-15-1, 13kl-16, and 13kl-17, is in a range of 353.8-298.6 Ma (Table 1 and Figures 2(c) and 2(g)–2(i)); group 2, including the other samples except 13kl-11, ranges from 244.1 to 207.1 Ma (Table 1 and Figures 2(a), 2(b), 2(e), 2(f), and 2(j)–2(q)); and group 3, only one sample of 13kl-11, is 116.5 Ma (Table 1 and Figure 2(d)). These three groups of the maximum K-feldspar ages suggest three rapid cooling events most likely starting from the early Carboniferous (353.8 Ma), middle Triassic (244.1 Ma), and middle Cretaceous (116.5 Ma), respectively. Spatially, with the exception of the 13kl-11, all of the sample locations are distributed evenly along the Kunlun Belt, whereas the youngest generation of magmatism (13kl-11) appears to have been confined to the middle part of the Kunlun Belt, close to Golmud.

Although the temperature of crystallization for granite is much lower, the higher closure temperature (~900°C [38]) of the U-Pb system in zircon implies that the zircon U/Pb ages record times close to those of the rock (magma) emplacement, whereas the 40Ar/39Ar ages record times when the rocks cooled from ~350°C to 150°C [35, 36]. The same-rock discrepancies between the emplacement and 40Ar/39Ar ages could result from pure cooling from ~900 to 500°C [39], which may not have been correlated with tectonics. Conversely, cooling from ~500°C to the surface temperatures can be related to exhumation processes [10, 34]. These time differences also suggest that magma emplacement and exhumation or uplift were not coeval; usually, uplift occurs tens of millions of years after compared with magma emplacement in the process of orogeny. The minimum ages of K-feldspars imply that most of the batholiths of the Kunlun Belt cooled to 150°C during early Cretaceous.

In summary, all of the biotites and K-feldspars yield consistent ages following the thermochronology protocol, and the ages are also consistent with the U/Pb dates in the cooling sequence. The staircase-shaped spectra of the K-feldspars show slow cooling induced by exhumation. The 40Ar/39Ar geochronology based on the K-feldspars and biotites of the batholith rocks indicate that the outcropping rock should reside close to the ~350°C isotherm during the middle to late Carboniferous, middle to late Triassic, and middle Cretaceous and should have cooled to ~150°C by early Cretaceous. Considering that both the biotite and the largest domains in K-feldspar possess closure temperatures of approximately 350°C, only a lower limit of the temperatures at the initial time of exhumation is inferred. Thus, under an assumption of paleogeothermal gradient of 25°C/km [9], the 40Ar/39Ar geochronological results above set a lower limit on the total exhumation that is approximately 14.0 km.

5.1. Methodology

Over the past decades, mineral ages from emplaced rocks in the deep crust have been recognized as radiogenic thermochronometers. The radiometric ages recorded the time at which the decay products within a mineral cease to be lost via thermal diffusion at a specific temperature. Dodson [39] defined the closure temperature for mineral on which the thermal history of the emplaced rock can be constructed. By using the high-resolution step-heating strategy, 40Ar/39Ar geochronology has been at the forefront of efforts such as extracting information of both temporal and thermal evolution from a geochronometric system based on the nature of argon diffusion.

Usually characterized by staircase age spectrum patterns of 40Ar/39Ar, K-feldspar is capable of maintaining stable crystal structures at high temperatures up to 1200°C during laboratory step-heating [34, 40]. This provides a tool to reveal argon distribution patterns formed during the thermal event it has experienced. A series of closure temperatures rather than a bulk one is revealed by the staircase ages, suggesting that the staircase age spectrum of K-feldspar reflects a monotonic cooling event between 350 and 150°C as it cools (such as [34, 40, 41]). The multidomain diffusion model (MDD) was proposed to describe the case and to retrieve the thermal history using laboratory measurements [40]. The model predicts that the different argon diffusion domains close one after another during cooling, corresponding to the series closure temperatures between 350 and 150°C. However, the unmeasurable multiple domains defined in the model have faced unceasing questions since its debut. Wartho et al. [41] argued that only a single domain rather than multiple domains exists in a Madagascar gem K-feldspar. Moreover, a volume diffusion mechanism, supposed by MDD within K-feldspar, has also been debated. Several studies have illustrated that argon often obeys a “multipath” diffusion law [42] instead of a volume diffusion one alone [43], which predicts argon transport between lattices and in short-circuit pathways in minerals [4244]. However, it is difficult to extract thermal history information by using the multipath model due to the unspecified process that permits interaction between different diffusion mechanisms and the undeterminable exchange coefficients by laboratory measurement. Even so, the multipath model does predict several types of behavior of argon diffusion during degassing in laboratory, which can be tested using step-heating data of K-feldspar. Finally, the microstructure deformation or alteration at low temperature is another argument unfavorable to the multidomain diffusion model. These deformation or alteration of microstructure can change the argon domains at or below their closure temperatures. Perthitic lamellae, micropores, and deuterically recrystallized strips of albite and microperthite occurring in K-feldspar define diffusion domains [35, 45]. Therefore, it is important to select K-feldspars from undeformed felsic rocks whose alteration and recrystallization are characterized only by comagmatic microtextures or deuteric turbidity, exsolution, and myrmekite, which are favorable candidates for the multidomain model [35, 46]. In routine experiments, it is generally sufficient to assess the argon diffusion domains of K-feldspar using an electron microscope, petrographic microscope, or scanning electron microscope (SEM) for the suitability for thermal modeling.

Slab diffusion geometry is used for K-feldspar in the modeling because argon loss from it during in vacuo step-heating experiments can be best described by volume diffusion in slab geometry [30]. By adjusting the parameters, such as size (ρ), percentage of released gas (ϕ), and closure temperature (Tc) of different domains of K-feldspar, a number of age spectra are modeled. Then, the best-fit parameters for activation energy (E), frequency factor (D0), and the domain distribution (r/r0) from rim to core within the K-feldspar crystals can be obtained (Figures 3(a) and 3(b)) via matching these modeled age spectra closely up with the experimental ones (Figure 3(c)). The corresponding cooling histories are then calculated for each set of parameters, from which the best fitted cooling pattern is obtained finally (Figure 3(d)). A 90% confidence distribution from all the modeled cooling histories is given, from which a 90% confidence interval of median is shown (Figure 3(d)).

5.2. Modeling Results

The best-fit modeling parameters of the activation energy (E) and frequency factor (LogD0/r02) for all the samples are summarized in Table 2. The 90% confidence distributions of the modeled cooling histories and the 90% confidence interval of the median for all samples are shown graphically in Figure 4. Although they may not necessarily be unique, the same-sample modeled cooling histories are consistent with the biotite ages (Figure 4).

All seventeen cooling histories, which are compiled in Figure 5, form three distinct cooling trends corresponding to groups 1, 2, and 3 based on the maximum ages of K-feldspar, respectively. The four samples of group 1, 13kl-7-1, 13kl-15-1, 13kl-16, and 13kl-17, show similar cooling patterns: rapid cooling from ~355 to 295 Ma at a cooling rate of 10.0°C/Ma, followed by slow cooling from ~295 to 245 Ma at 0.17°C/Ma. Then, two of the samples, 13kl-7-1 and 13kl-17, return a rapid cooling rate during ~245-205 Ma, whereas the other two, 13kl-15-1 and 13kl-16, kept a slow cooling rate during this period (Figure 5).

The 12 samples of group 2 yielded very consistent cooling patterns: rapid cooling at a rate of 11.0°C/Ma from ~230 to 205 Ma and slowed cooling at a rate of 0.17°C/Ma from ~205 to 95 Ma (Figure 5). It is interesting that the two samples of group 1, 13kl-7-1 and 13kl-17, also revealed this rapid cooling event, but the other two samples of group 1 did not with unknown reasons.

The only one sample in group 3, 13kl-11, shows a distinct cooling pattern. It cooled rapidly at a rate of 13.5°C/Ma from ~120 to 95 Ma, followed by a slow cooling period from ~95 to 40 Ma, and then returns to a rapid cooling from ~40 to 35 Ma at a rate identical to a previous (U-Th)/He thermochronology result of 15.0°C/Ma [11] (Figure 5).

Based on the 90% confidence interval of median values of modeled cooling histories (Figure 5), the weighed average cooling rates for each period are calculated (Figure 5), from which the corresponding denudation rates are given under an assumption of a geothermogradient of 25°C/km [9, 11] (Table 3). Errors reported in Table 3 are at the 2σ level, which are propagated from each individual cooling history modeling and standard deviation of all the cooling histories used for calculation.

In summary, the modeling results reveal four rapid cooling events that correspond to four rapid exhumation events in the Kunlun Belt since the late Paleozoic. It is notable that the exhumation events recorded by the plutons occurred tens of millions of years later than their magmatic emplacement of granitoids during the orogeny. It is noted that variations of structure or thermal regime of expansion may occur in K-feldspar grains during step-heating experiments [35]. These variations result in inaccuracies in the cooling histories from the MDD model. The calculations indicate that the temperatures predicted using the MDD model exceed the “real” values by approximately 40°C [35]. Nonetheless, this is not concerned in this study because we are concerned about the cooling patterns, and the accuracies of the cooling rates are less important.

6.1. From Cooling to Uplift

The mineral radiometric ages may record the timing of cooling when the radiogenic products by the decay of isotopes are retained. Different types of cooling behavior of minerals are supposed [34], i.e., the pure conductive cooling of emplaced rocks in the absence of crustal denudation and the cooling related to uplift-induced denudation. Regarding denudation-related cooling, the recovered cooling histories can conversely constrain the denudation evolution of an uplifted terrane. The cooling histories presented above are inconsistent with pure conductive cooling characterized by constant cooling rates and thus instead reflect denudation processes [34]. Therefore, cooling histories could be related to denudation evolution and exhumation histories and even uplift of the studied region.

6.2. Slow Cooling and Reheating

Both slow cooling and reheating of emplaced rocks could yield staircase 40Ar/39Ar age spectra of K-feldspar observed in this study, and the reheating is most likely related to activity of the north Kunlun fault (Figure 1). Uncorrelated relationship between the distance of each sample to the north Kunlun fault and both the minimum K-feldspar cooling ages and the percentage of argon loss (Figure 6) indicates that thermal disturbances from the fault in the studied plutons can be ignored, because there is neither gradual decrease in the minimum age nor a gradual increase in argon loss observed toward the north Kunlun fault.

Another source of possible reheating may have derived from the plutons emplaced afterwards, as identified using a Ttemperaturettime solution plot ([34]; Cassata et al., 2009). Assuming plane slab geometry, the fraction argon loss (F, Table 1) is related to the dimensionless parameter Dt/a2 by the following equation:
where D is the diffusion coefficient, t is the reheating time, and r denotes the radius of the plane slab. In the Arrhenius relationship,
where D0 is the diffusion frequency factor, Eq is the activation energy, R denotes the gas constant, and T is temperature. Substituting equation (2) into (1) yields

Tt solutions for all samples are calculated using equation (3) (Figure 7). If the fractional loss observed in each samples resulted from an instantaneous heating event, the Tt curves in Figure 7 would be consistent, overlap each other, and be common to all the samples. We do not find that the Tt solutions are common to all the samples, suggesting that the studied plutons have not experienced reheating from rocks emplaced afterwards. In fact, the strategy we employed during sampling is to remain distant from the igneous rock bodies of a younger age.

6.3. Multiple Phases of Mountain Building along the Northern Tibetan Margin

Four rapid exhumation events are resolved from the modeling results (Figure 5). They occurred in the Carboniferous (355-295 Ma), Triassic (245-205 Ma), Cretaceous (120-95 Ma), and Eocene (40-35 Ma). These exhumation events mark significant tectonic movements since the Paleozoic, which collectively built up the present-day northern Tibetan margin in the Kunlun area (Figure 8).

In association with volcanic rocks, granitic intrusions of middle Paleozoic ages (420-390 Ma) extensively occur in the eastern Kunlun batholith [12, 23, 24, 26, 47]. This volcanism and magmatism are thought to be related to the closure of the Prototethys Ocean [1, 2, 7, 12], which existed between the Lhasa-Qiangtang and Kunlun-Qimantagh-Qaidam blocks from the Neoproterozoic to the Silurian [2, 7, 12] (Figure 8(a)). By late Silurian (~420 Ma), the basic framework of the northern Tibet margin had been shaped, with the Kumlun Belt and Qaidam Basin in place from south to north [12]. Although the suture zone of this closure has not been identified owing to strong deformation (Figure 8(b)), the occurrence of A-type granites indicates that the orogeny from the closure ended by ~390 Ma (early Devonian), and postcollisional extension began afterwards [12, 28, 48]. Geochemically, the A-type granites are products of the melting of an enriched refractory lithospheric mantle with high alkaline minerals (K2O and Na2O) [28, 48]. Therefore, the rapid cooling event recorded by these plutons during 355-295 Ma suggests rapid uplift in an extensional setting after convergent deformation in the Kunlun Belt during the early Paleozoic (Figure 8(c)), which possibly resulted from the convective removal of a thickened mantle root [49, 50] or the breakoff of an oceanic slab of the continental lithosphere during subduction [12, 51]. Both mechanisms could have caused the thinning of the thickened lithosphere following the orogeny and would have provided tremendous uplift. The uplift that initiated the Kunlun Belt is a result from the collision between the Qiangtang and Kunlun blocks, which is also recorded by a detrital zircon U/Pb age peak in the Jurassic to Lower Cretaceous sandstones in the Songpan-Ganzi and Qiangtang areas [2]. After the rapid exhumation, a period of slow exhumation at a rate of 0.17°C/Ma occurred between ~295 and 245 Ma (Figure 5), during which the Paleotethys Ocean opened.

Early Triassic granitoid intrusions formed the most extensive plutons in the eastern Kunlun and Qimantagh batholith [12, 23, 24, 26, 47] and are interpreted as the result of the closure of the Paleotethys, which formed during late Carboniferous to Permian times [2, 7, 12] (Figure 8(d)). Geochemically, these Triassic granites are I-type and mark the end of subduction and the beginning of collision. The rapid Triassic exhumation event (~245-205 Ma) resolved in this study reflects the uplift of the Kunlun Belt resulting from the collision of the Songpan-Ganzi and Kunlun blocks, which built the basic shape of the Kunlun Belt [1, 2, 6, 7]. A long period (~205-120 Ma) of slow exhumation then followed (0.17°C/Ma, Figure 5), which implies that a tectonically quiet period occurred during that time. This is consistent with the notion that the tectonic active regime moved to the south of Tibet, i.e., into the Lhasa area [2, 4, 6, 7].

The rapid uplift event in the Cretaceous (120-95 Ma) is seemingly confined to the middle part of the Kunlun Belt, close to Golmud (Figure 1). This event is possibly related to the far-field effects of the closure of the Mesotethys Ocean and the collision of the Lhasa and Qiangtang blocks to the south in the Cretaceous [2, 4], as marked by the Bangong suture zone (Figures 8(e) and 8(f)). The northward propagation of stress from the collision caused the uplift of the Kunlun Belt, although it is unknown why this far-field effect is limited to the middle portion of the Kunlun Belt.

A rapid uplift event in the Eocene (40-35 Ma) is also recorded by the Cretaceous granite (Figure 5) and is thought to have been related to the rejuvenation of the Kunlun Belt caused by the far-field effects of the convergence of the India and Eurasia plates and the growth of the Tibetan Plateau [9, 11], as marked by the Indus suture zone (Figure 8(g)). This event built the modern relief of the Kunlun Belt and has been observed in several studies of (U-Th)/He and 40Ar/39Ar thermochronologies [911]. More and more evidence suggests that extensive deformation went across the eastern and central parts of northern Tibet (Clark et al., 1999) within ~10 million years of the collision of India and Eurasia (41-52 Ma) (Rowley, 1996), although several recent studies have suggested that the collision may have occurred during the early Paleocene (~65-60 Ma) [52, 53]. Or to put it differently, the northern margin of Tibet was constructed during the early stage of the collisional process. Despite more than 2000 km of convergence since collision [9, 54], the northward propagation of stress was significant during the process of collision. Our previous study indicated that the highest relief of the Kunlun Belt throughout the entire Cenozoic was built by this uplift event and is almost twice that at present [11]. From ~40 Ma to the present, only minor uplift occurred, and erosion was the primary tectonic process, which lowered the relief of the Kunlun Belt to that of today. The sedimentation in the Qaidam Basin to the north and the Hoh Xil Basin to the south supports these observations. The deposition in the Qaidam Basin increased rapidly from 40 to 36.6 Ma [5557]. Growth strata from 50 to 40 Ma are extensively preserved in the outer zones of the Qaidam Basin [5860]. In the Hoh Xil Basin, next to the study area in the south, sedimentation rates increased rapidly at approximately 40 Ma [20]. All evidence cited above indicates increased denudation of the Kunlun Belt, which probably resulted from uplift at approximately 40 Ma.

Thermochronological 40Ar/39Ar mineral dating results on Paleozoic and Mesozoic granites along the Kunlun Belt, the first-order mountain of the northern Tibetan margin, are reported. Four rapid exhumation events in the Carboniferous (355-295 Ma), Triassic (245-205 Ma), Cretaceous (120-95 Ma), and Eocene (40-35 Ma) are resolved using thermal modeling employing the multidomain diffusion theory. The rapid exhumation event recognized in Carboniferous may reflect the thinning of the thickened lithosphere caused by the closure of the Prototethys Ocean, which initiated the Kunlun Belt. The rapid uplift event in the Triassic could have been a result of the closure of the Paleotethys Ocean, which built the modern shape of the Kunlun Belt. The rapid uplift event in the Cretaceous, which is confined to the middle part of the Kunlun Belt, may have been related to the far-field effects of the closure of the Mesotethys Ocean between the Lhasa and Qiangtang blocks to the south on the Kunlun Belt. Finally, the rapid event in the Eocene is a result of the far-field effect of the convergence of the India and Eurasia plates, which produced the highest relief in the Cenozoic and built the modern Kunlun Belt.

All the raw data are listed in the “Supplementary Table S1_Ar data.”

The authors declare that they have no conflicts of interest.

This study is supported by the Natural Science Foundations of China (41673015, 41521062, and 41930106) and “Strategic Priority Research Program” of the Chinese Academy of Science (XDB03020203). Supporting information in the form of raw data set (Table S1) is provided in the electronic supplement. The authors thank Profs. Franz Neubauer and Dawn Kellett for constructive suggestions and comments.

Exclusive Licensee GeoScienceWorld. Distributed under a Creative Commons Attribution License (CC BY 4.0)