Mineralogical and Geochemical Insights into Formation of the Muji Carbonic Springs, NW China

Abstract

The Muji carbonic springs on the northeastern margin of the Pamir Plateau provide a natural window into tectonically controlled CO₂ degassing within a continental collision zone. Through mineralogical and geochemical analyses, this study constrains the formation mechanisms and regional geological significance of carbonic spring systems. The formed deposits are dominated by calcite and aragonite, with minor dolomite, quartz, and gypsum. The compositions of major elements are consistent with the observed mineral assemblages, reflecting that the carbonate deposition was mainly governed by CO₂ degassing intensity and associated kinetic effects under cold-spring conditions. Carbon isotopes of the deposits are consistently enriched in heavy carbon with δ¹³C values of +3.5‰ to +9.1‰, indicating a persistent contribution of deep-sourced CO₂, most likely derived from metamorphic decarbonation of the crustal carbonates. Calcite exhibits moderate δ¹³C values due to rapid precipitation limiting isotope enrichment, whereas aragonite records higher δ¹³C signatures under subdued degassing and stable hydrodynamic regimes. The narrow δ¹⁸O range (−10.7‰ to −12.6‰), closely matching that of the spring waters, indicates that the tufas record the δ¹⁸O of the spring waters through DIC-water oxygen exchange. Trace element distributions (Sr–Ba–U) reveal systematic enrichment in deep-sourced fluids and progressive downstream geochemical alteration driven by spring–river mixing. The HD springs show high Sr and δ¹³C values, indicating minimal dilution of ascending CO₂-rich fluids, while MJX and MJXSP groups record variable degrees of shallow mixing. Collectively, the Muji system exemplifies a coupled process of “deep fluid input–shallow mixing–precipitation kinetics.” Its persistent heavy δ¹³C and trace-element enrichments demonstrate persistent metamorphic CO₂ release through fault conduits under ongoing compression. These findings establish the Muji springs as a key non-volcanic analogue for deep CO₂ degassing in continental collision zones and provides new insights into crustal carbon recycling and tectonic–hydrochemical coupling at plateau margins.

1. Introduction

Carbonic spring systems represent important natural windows for investigating the release of deep-seated fluids and their interactions with surface environments [1,2,3]. The precipitation of carbonate minerals such as travertine and tufa effectively reflects the interplay among deep CO₂ input, shallow water mixing, and precipitation kinetics [2,4]. Globally, typical sites such as the Tivoli springs, the Yellowstone hot springs, and the carbonic springs in the Tibetan Plateau [5,6,7] exhibit diverse mineral assemblages, carbon–oxygen isotopic compositions, and water–rock interaction features. These investigations highlight carbonic springs as not only key regulators of regional hydrochemistry and sedimentation but also as natural laboratories for exploring deep carbon cycling and tectonic activities.

The formation mechanisms and geochemical signatures of the carbonic springs vary significantly across regions due to differences in tectonic settings, fluid pathways, and near-surface environmental conditions. In active continental collision zones or plateau margins, deep fluids are often closely linked with major faults, detachments, or strike-slip structures [8,9,10]. These tectonic structures can enhance CO₂ release by concentrating stress and promoting episodic fluid ascent [11,12]. In contrast, springs in stable cratonic areas or intracontinental basins are much more strongly influenced by shallow groundwater circulation and local water–rock interactions. Moreover, the ascent of deep CO₂ and its reactions with shallow waters, sediments, or surrounding volcanic-metamorphic and sedimentary rocks introduce further variability. For instance, mantle-influenced systems are often identified by relatively high ³He/⁴He ratios, while δ¹³C values vary depending on the relative contributions of mantle, crustal carbonate degassing, or organic/soil-respired CO₂ [8,13]. Similarly, alkaline earth elements such as Sr and Ba can become enriched through deep fluid supply, CO₂-driven water–rock interactions, and carbonate precipitation/dissolution, though their concentrations remain strongly controlled by lithological sources, fluid residence times, and precipitation kinetics [6,14]. Therefore, carbonic springs in continental collision zones and plateau margins are not only critical conduits for deep CO₂ release but also key sites for unraveling tectonic–fluid–sediment interactions [15,16]. Comparative studies across regions provide valuable insights into their shared traits, differences, and dominant controls.

The Muji carbonic springs are located in the northeastern Pamir, a key zone of ongoing convergence between the Indian and Eurasian plates. Frequent fault activity and intense tectonic stress provide effective pathways for deep fluid ascent. The spring area hosts multiple dome-shaped tufa bodies with long-lived activity [17]. Deposits are dominated by calcite and aragonite, with minor dolomite, and are accompanied by vigorous CO₂ degassing [18,19]. These features suggest that the Muji springs may play an important role in deep CO₂ release and regional carbon cycling, yet they remain poorly constrained by integrated mineralogical, geochemical, and isotopic studies. In this study, we apply a combined approach using mineralogical analyses, major and trace elements, and stable isotopes to elucidate the genesis of Muji carbonic spring deposits. We focus on identifying the roles of CO₂ degassing, precipitation kinetics, and deep–shallow fluid mixing in shaping the deposits. Additionally, we integrate Sr-Ba-U systematics with δ¹³C_DIC to investigate spring–river coupling. Finally, by comparing the Muji system with well-studied carbonic spring systems such as Yellowstone and Tashkurgan, we highlight both the commonalities and distinctive features of Muji, thereby shedding light on fluid–tectonic–sedimentary coupling in the northeastern Pamir and its significance for the global deep carbon cycle.

2. Geologic Background

The Muji carbonic springs, located in the Muji Basin, Akto County, western Xinjiang, China (39°01′02″ N, 74°29′16″ E), are typical tectonic springs situated on the northeastern margin of the Pamir Plateau and the western edge of the Tarim Basin [17] (Figure 1). In terms of regional hydrology, the Muji Basin is drained mainly by the Muji River and its tributaries, which are fed by snowmelt and glacier meltwater from the eastern Pamir and surrounding high ranges [20]. Under the arid to semi-arid climate, most surface runoff is ephemeral, and a significant portion of meltwater and river flow infiltrates through highly permeable alluvial–proluvial fans. The Muji carbonic springs are preferentially located along these fan margins and fault zones, where deep-circulating waters re-emerge and locally mix with shallow groundwater and river water [19], establishing a direct hydraulic connection between the spring system and the surface drainage network. This region lies at the junction of the Tianshan, Kunlun, and Karakoram orogenic belts, representing a foreland transitional unit where the Pamir and the Tibetan plateau wedge northeastward into the Tarim foreland [21]. Intense plate convergence and crustal shortening have shaped the complex basin–range system and provided favorable conditions for the migration of deep-sourced fluids and their discharge on the surface.

The Muji Basin is bounded by three major active faults, including the Muji Fault, the northern segment of the Kongur Shan Fault, and the southwestern boundary fault of the Muji Basin. The major structure controlling spring discharge is the southwestern boundary fault of the Muji basin and its secondary faults. The Muji Fault (MJF) trends NW-SE, acting as the northern boundary of the Muji basin. It is an active fault characterized by dextral strike-slip with a normal component. Geological and geodetic studies indicate that the fault has remained active during the Late Quaternary, with a vertical slip rate of ~0.3 mm/yr and an extensional strain rate of 6–7 mm/yr [23]. The 2016 Mw 6.6 Aketao earthquake was associated with this fault. To the north, the Kongur Shan Extensional System (KES) and the Muztagh–Kongur Shan Fault (KSF) also bound the Muji–Tashkorgan Basin and significantly influence its geomorphic evolution [18,19]. However, the Late Quaternary activity of the northern Karakoram–Muji fault system remains debated: some studies suggest slip rates >4.5 mm/yr [24], whereas others argue that there is no significant late Quaternary strike-slip activity [25]. This discrepancy highlights the heterogeneous distribution of the regional stress field.

Stratigraphically, Paleozoic to Mesozoic marine carbonate successions are widespread in the surrounding region, interbedded with sandstones, conglomerates, and locally metamorphic rocks [26]. Notable outcrops of marine and metamorphosed carbonate rocks occur throughout the area, including marble of the Lower Paleozoic Bluntler Group and the Silurian strata, limestone of the Devonian Kiziltao Group, and marble and limestone within the Lower Permian strata [26,27]. The basin interior is filled with thick Quaternary alluvial–glacial and lacustrine deposits. In addition, the Triassic diorite–granite intrusions and gneissic domes are exposed in the northeastern Pamir. Fault zones act as major conduits for meteoric water infiltration; during deep circulation, water interacts with carbonate rocks, dissolving Ca²⁺ and HCO₃⁻. Upon ascent, CO₂ degassing induces carbonate supersaturation and precipitation, forming extensive tufa deposits. This “fault-controlled infiltration–deep circulation–spring discharge–tufa deposition” chain represents the fundamental mechanism of the Muji carbonic springs.

3. Sampling and Analytical Methods

3.1. Sample Collection

A total of 25 spring water samples, 3 river water samples, and 11 tufa (carbonate deposit) samples were collected from the Muji Basin. Based on geomorphic and hydrological characteristics, the Muji carbonic springs can be subdivided into three groups (Figure 1c). Group MJX springs are distributed along the riverbank zone and commonly occur on dome- or cone-shaped tufa mounds (“carbonate cones”) several meters above the active river channel. These features reflect long-term CO₂ degassing and carbonate accumulation influenced by lateral recharge from the river. HD springs discharge directly within or adjacent to the present river channel, showing vigorous effervescence, strong flow, and minimal tufa buildup due to rapid fluid ascent. MJXSP springs are located on the piedmont slopes or mid–upper alluvial fans, far from the main channel, and are characterized by weak or inactive discharge.

The spring water samples were preserved in two ways: one set was filtered through a 0.45 μm membrane and acidified to pH < 2 for trace element analysis; the other set was kept unacidified for dissolved inorganic carbon (DIC) isotope analysis. River water samples were taken for comparison. Tufa samples were collected from different surface positions near spring vents, including both surface and profile samples, to reflect spatial variations in mineralogical and geochemical characteristics.

3.2. Elemental and Mineralogical Analysis

Major elements of tufa samples were determined using an X-ray fluorescence spectrometer (Axios, PANalytical B.V., Almelo, The Netherlands). The instrument is a sequential wavelength-dispersive spectrometer equipped with a ceramic Rh-target X-ray tube (maximum power 4 kW, tube current 160 mA) and controlled by SuperQ Version 5.0 software. Powder pellets were prepared by drying <200 mesh (~75 μm) powders at 105 °C, mixing 4 g of the sample with boric acid, and pressing them into 32 mm diameter pellets under a load of 30 tons. The analytical precision of XRF was generally better than 2%, as monitored by repeated analyses. Analytical accuracy was assessed using certified reference materials (e.g., GBW series rock and soil standards), and measured values remained within ±2% of the certified compositions.

Mineralogical compositions of tufa samples were determined using an X-ray diffractometer (Rigaku Ultima IV, Rigaku Corporation, Tokyo, Japan). Samples were ground to <200 mesh powders and pressed into flat-surfaced mounts. Analytical conditions were Cu target, Kα radiation, 40 kV accelerating voltage, and 100 mA tube current. Continuous scanning was performed at a rate of 2°/min with a step size of 0.02°. The diffraction data were processed using MDI Jade 5 software, and mineral identification was based on the ICDD-JCPDS database. Mineral contents were estimated from peak area calculations.

Minor discrepancies between the XRF and XRD results (Tables 1 and 2) reflect differences in analytical scope and sample splitting. XRF provides bulk elemental compositions—including both crystalline and amorphous phases—whereas XRD detects only crystalline minerals. Because the two analyses were performed on different powder subsamples from the same hand specimen, micro-scale heterogeneity in detrital or accessory minerals (e.g., quartz, plagioclase, Fe oxides, Na-bearing phases) can result in slight mismatches between the chemical and mineralogical datasets.

Trace element concentrations in spring water and tufa samples were measured using an inductively coupled plasma mass spectrometer (ICP-MS, Thermo Fisher iCAP-TQ, Thermo Fisher Scientific Inc., Waltham, MA, USA). The instrument was installed in a class-1000 clean laboratory and equipped with an autosampler (ASX-560), water chiller (SH150-2100), 15 kVA UPS, Anton Paar microwave digestion system, acid evaporation system (model 09A24S), two CIF Scientific acid purification systems (for HNO₃ and HF), a Sartorius analytical balance (0.1 mg precision), drying oven (HHG-9079A), and a Milli-Q ultrapure water system. Liquid samples were diluted and corrected with internal standards before direct introduction. Solid samples were digested using microwave-assisted acid digestion followed by acid evaporation. Multi-point calibration curves were employed, with internal standards used to correct instrumental drift. Analytical reproducibility was better than ±5%.

3.3. Carbon Isotope Analysis

The stable carbon isotopic composition of dissolved inorganic carbon (δ¹³C_DIC) was determined using an isotope ratio mass spectrometer (IRMS, Delta V Advantage 253 Plus, Thermo Fisher Scientific, Waltham, MA, USA). Water samples were injected into pre-evacuated 12 mL Labco glass vials containing ~85% phosphoric acid (H₃PO₄) and equilibrated at 25 °C for 24 h to ensure complete reaction. The released CO₂ was purified using a vacuum extraction system and analyzed by IRMS. The analysis was calibrated against the international standard NBS-18, with an overall uncertainty of ±0.1‰.

Carbon and oxygen isotopes of tufa (δ¹³C_carb, δ¹⁸O_carb) were measured using a Thermo Fisher MAT-253 IRMS. Powdered samples reacted with anhydrous phosphoric acid at 70 °C to produce CO₂ for isotope analysis. Analytical reproducibility was better than ±0.1‰. All isotope measurements were calibrated against international and laboratory standards (NBS-19, IAEA-CH-6, IAEA-SLAP, VSMOW2), and analytical quality was monitored by replicate analyses and blanks.

4. Results and Discussion

4.1. Mineralogical Characteristics and Precipitation Mechanisms

In an active continental collision zone, understanding how deep-seated CO₂ migrates upward through fault-controlled conduits and becomes recorded in surface tufa deposits is critical for linking crustal degassing with near-surface carbonate precipitation. The deposits of Muji carbonic spring are mainly composed of calcite and aragonite, accompanied by minor dolomite and quartz (

Table 1. Mineralogical compositions of tufa deposits from the Muji basin.

No.	Sample ID	Quartz (%)	Plagioclase (%)	Calcite (%)	Aragonite (%)	Dolomite (%)	Gypsum (%)
1	MJX-01	1.9	–	89.9	8.2	–	–
2	MJX-02	1.6	–	96.6	1.2	–	0.6
3	MJX-22	1.6	0.4	40.8	57.3	–	–
4	MJX-24	1.7	–	98.3	–	–	–
5	MJX-25	1.7	–	31.0	67.3	–	–
6	MJX-26	2.6	0.6	88.8	7.5	0.6	–
7	MJX-31	1.2	–	98.8	–	–	–
8	MJX-41	0.9	–	89.1	9.5	–	0.5
9	MJX-43	2.1	–	88.8	9.1	–	–
10	MJX-44	1.7	–	94.6	2.6	–	1.1
11	MJX-66	0.9	3.7	94.5	1.0	–	–

Notes: “–” denotes values below detection or not determined. The corresponding spectra are shown in Figure S1.

). Trace gypsum occurs only in a few samples and is interpreted as a minor evaporative phase formed under the local arid climate, without affecting the primary carbonate precipitation pattern. These mineralogical features are consistent with both the macroscopic and microscopic characteristics of the tufa deposits. Hand-specimen observations show that the deposits are predominantly porous, laminated pisoidal and peloidal carbonates, forming thin surface crusts (approximately 0.2–1 mm) and mound- to cone-shaped buildups. Previous petrographic studies further revealed their microstructural origins: the pisoids and peloids contain micritic or organic-rich calcite clasts at the core, surrounded by 4–6 concentric laminae [17,18,21]. The outer shells display alternating micritic laminae (dark, commonly containing fine detrital quartz and microbial remnants)—potentially derived from wind-blown silt and/or weathering of local silicates—and clear sparry calcite laminae (white, composed of clean fibrous to granular calcite). SEM observations of thin extracellular polymeric films and radiating or dog-tooth calcite cements collectively support precipitation under low-temperature freshwater conditions. In most samples, calcite is the dominant phase, with abundances exceeding 90% in MJX-24 (98.3%), MJX-31 (98.8%), MJX-44 (94.6%), and MJX-66 (94.5%). Such calcite-dominated compositions are commonly interpreted as the result of precipitation under strong CO₂ degassing and high supersaturation. In contrast, aragonite-rich samples (MJX-22, 57.3%; MJX-25, 67.3%) are interpreted to have formed under relatively subdued CO₂ degassing and more stable hydrodynamic conditions. This pattern is consistent with the “degassing-rate–polymorph stability sequence” observed in global hot-spring systems such as Yellowstone and Tivoli [28].

Both calcite-rich and aragonite-rich samples show high CaO (46.5%–54.6%,

Table 2. Major element compositions and carbon–oxygen isotopic values of tufa deposits from the Muji basin.

No.	Sample ID	SiO2 (%)	Al2O3 (%)	TFe2O3 (%)	CaO (%)	MgO (%)	K2O (%)	Na2O (%)	δ13Ctufa (‰)	δ18Otufa (‰)
1	MJX-01	6.1	1.0	12.8	46.5	1.6	0.8	2.7	7.6	−11.8
2	MJX-02	5.5	1.4	3.1	49.5	1.3	0.4	0.5	6.8	−12.1
3	MJX-22	4.8	0.6	7.2	51.6	0.7	0.2	0.1	9.1	−10.7
4	MJX-24	3.8	0.7	2.7	53.5	1.0	0.1	0.1	7.4	−11.8
5	MJX-25	5.0	0.7	4.2	51.8	1.0	0.2	0.2	8.3	−10.7
6	MJX-26	4.0	0.8	0.4	51.8	2.4	0.2	0.2	6.3	−12.3
7	MJX-31	1.8	0.4	0.8	47.3	2.7	0.7	0.8	3.5	−11.9
8	MJX-41	1.2	0.2	0.5	54.6	1.1	0.1	0.1	7.9	−11.6
9	MJX-43	7.5	1.5	8.9	49.9	0.6	0.3	0.2	7.4	−12.5
10	MJX-44	2.1	0.4	0.4	53.2	1.0	0.1	0.3	7.2	−12.6
11	MJX-66	3.6	0.5	1.5	52.7	1.9	0.1	0.3	7.8	−11.5

), indicating that Ca availability was not a limiting factor for mineral precipitation. MgO contents are generally low ranging from 0.6% to 2.7%, and dolomite was not detected by XRD, suggesting that only a small proportion of the Mg²⁺ was incorporated into calcite or aragonite lattices to form high-Mg calcite.

Spring waters from the Muji system exhibit Mg/Ca molar ratios mostly in the range ~0.3–0.9, with an average of ~0.6 (

Table 3. Field temperature, major cations (Ca²⁺, Mg²⁺) and trace element concentrations (Sr, Ba, U) and δ¹³C_DIC values of Muji spring waters.

No.	Sample	T (℃) *	Ca2+ (mmol/L) *	Mg2+ (mmol/L) *	Sr (μg/L)	Ba (μg/L)	U (μg/L)	δ13CDIC (‰)
1	MJX-03	–	6.34	2.12	824.2	22.7	3.89	−2.2
2	MJX-06	11.6	3.54	3.24	677.4	12.4	3.68	1.1
3	MJX-10	12.1	6.24	3.90	1061.0	34.4	6.53	0.9
4	MJX-10-1	11.8	8.86	4.90	823.6	27.7	3.39	1.6
5	MJX-16	11.4	6.56	6.09	1818.4	62.9	1.87	−1.7
6	MJX-21	10.7	8.93	5.88	559.9	24.8	0.76	−0.5
7	MJX-26	10.7	7.31	3.96	1409.4	59.7	0.40	−2.0
8	MJX-31	10.8	7.98	3.18	27.3	2.8	0.03	0.2
9	MJX-36	10.9	5.09	4.05	223.2	7.6	0.05	−2.2
10	MJX-37	–	4.17	4.40	1955.1	154.1	3.50	1.0
11	HD-01	10.7	7.41	2.81	1298.5	55.4	1.40	−0.5
12	HD-02	10.4	8.21	2.59	929.5	25.8	1.18	−1.2
13	HD-03	10.8	9.16	3.68	1523.5	61.0	1.26	−0.3
14	HD-04	11.0	3.54	3.21	1202.7	50.4	1.52	−8.2
15	HD-05	11.2	5.16	2.86	1297.5	47.2	2.59	−1.2
16	HD-06	–	7.91	3.24	1556.9	73.3	1.98	−1.9
17	HD-07	–	2.72	1.81	1017.0	57.4	7.06	−4.3
18	HD-08	–	6.94	2.72	1168.3	26.8	1.23	−3.7
19	HD-09	–	7.34	1.87	977.5	42.4	0.94	−2.3
20	HD-10	–	9.13	5.14	1045.0	56.9	0.04	5.8
21	HD-11	10.8	8.26	4.16	1400.9	83.5	0.91	2.1
22	HD-12	–	7.71	4.98	78.3	9.6	–	−0.1
23	MJXSP-01	–	4.17	2.87	862.2	56.1	9.57	1.7
24	MJXSP-02	10.2	5.24	3.88	817.8	49.1	9.40	−0.5
25	MJXSP-03	10.6	5.49	3.95	563.4	80.4	5.90	−0.1
26	MJXR	10.6	1.76	1.20	1691.7	55.8	14.3	−3.3
27	MJXRD	–	1.83	1.52	1350.8	92.4	2.50	−4.2
28	MJXRU	–	1.27	0.56	186.6	9.5	0.02	−5.3

Notes: “–” denotes values below detection or not determined; No. correspond to sampling sites shown in Figure 1. “*”: data from [19].

), suggesting Mg enrichment but not Mg-dominated conditions. These values are lower than Mg/Ca ratios identified in experimental studies (>1), identified in experimental studies for strong Mg inhibition of calcite growth or for significant calcite growth and for promoting dolomite formation [29,30,31]. Under such low-temperature and low-Mg conditions, aragonite precipitation in the Muji system is unlikely to be governed by aqueous Mg or Sr concentrations. Instead, the occurrence of aragonite is better explained by degassing-controlled kinetic effects operating under relatively subdued CO₂ degassing, where lower supersaturation and more stable hydrodynamic conditions allow aragonite to nucleate and persist [6]. The relatively high δ¹³C values of the aragonite-rich samples further support a kinetic control. Therefore, the mineralogical differences between calcite- and aragonite-rich deposits at Muji are interpreted as the result of kinetic controls during rapid precipitation rather than direct chemical thresholds of Mg or Sr in the spring waters.

Some samples (e.g., MJX-43, MJX-01) exhibit anomalously high SiO₂ (up to 7.55%) and Fe₂O₃ (up to 12.6%), indicating detrital input or enhanced water–rock interaction during deposition. These “non-typical carbonate” end-members likely record mixing between deep fluids and host rocks along up-flow zones. Similar Fe–Si enrichments reported in the Latium–Tivoli graben [5] and the western Tibetan Plateau [14] suggest that such features mark transient enhancement of fluid–rock interaction during fault activity [1,27].

Macroscopic observations illustrate differences in color and fabric (Figure 2). Reddish to brown tones in MJX-01 and MJX-43 correspond to their high Fe₂O₃ contents, indicating significant ferruginous staining within these deposits. Aragonite enrichment samples MJX-22 and MJX-25 (57.3% and 67.3%) are light gray and relatively uniform and compact external textures. By contrast, calcite-rich samples such as MJX-24 show a brownish to yellow-brown color, whereas MJX-31 exhibits a dark gray to blackish gray appearance. Despite their contrasting colors, both samples display rough and granular external textures, consistent with rapid CO₂ degassing and high supersaturation during calcite precipitation. Overall, the Muji tufa are dominated by calcite with locally abundant aragonite, only trace to rare ordered dolomite, and occasional Fe–Si-enriched outliers. Together with spring-water Mg/Ca molar ratios mostly ~0.3–0.9 (Table 3), these features are consistent with precipitation regulated by CO₂ degassing and low-temperature kinetics, rather than a single controlling factor, and place the system within a fault-linked, cold-spring carbonate setting on the northeastern Pamir Plateau.

4.2. Stable Isotope Characteristics

Previous studies of the Muji carbonic springs have established several key geochemical characteristics that constrain the origin of CO₂ and the isotopic evolution of the spring waters [19]. The released CO₂ from springs exhibits δ¹³C values of −5.4‰ to −2.9‰ (average −3.9‰), indicating a mixture of deep-sourced CO₂ and carbon released from thermal decomposition of regional carbonate rocks, while soil or atmospheric CO₂ can be excluded due to high p_CO2 and abundant tufa precipitation. The spring waters have δ¹⁸O values ranging from −15.1‰ to −12.6‰, and δD around −106.8‰ to−89.6‰, reflecting meteoric recharge followed by water–rock interaction at depth and partial mixing with CO₂-rich deep fluids. Together, these isotopic features demonstrate that the Muji system involves deep CO₂ ascending along active fault conduits and interacting with meteoric waters before rapid degassing at the surface. These previously documented characteristics provide the necessary geochemical framework for interpreting the isotopic signatures of the tufa deposits described below.

Building upon this established geochemical context, the carbon and oxygen isotopic compositions of the Muji carbonic spring deposits exhibit a clear relationship with their mineralogical assemblages (Figure 3). Calcite-dominated samples (e.g., MJX-24, MJX-31) display relatively lower δ¹³C values (+3.5‰ to +7.4‰) (Table 1), consistent with intense CO₂ degassing during rapid precipitation. Under such conditions, fast CO₂ loss drives the fluid toward high supersaturation, promoting calcite nucleation and stabilization [6]. These samples exhibit rough and granular external textures, which are characteristic of carbonates formed under strong degassing where rapid precipitation generates irregular crystal aggregates [32,33]. In contrast, aragonite-rich samples (MJX-22, δ¹³C = +9.1‰; MJX-25, δ¹³C = +8.3‰) display higher δ¹³C values, which we interpret this enrichment as the result of weaker CO₂ degassing and limited re-equilibration with atmospheric CO₂ and partial preservation of deep-sourced isotopic signatures.

The δ¹⁸O values cluster narrowly between −10.7‰ and −12.6‰, distinctly lighter than marine carbonate standards and fall within the documented range of Muji spring waters (δ¹⁸O: −15.1‰ to −12.6‰) [19]. Because the Muji spring waters are characterized by meteoric δ¹⁸O–δD signatures, this overlap indicates that the tufa was influenced by meteoric-derived waters and that their oxygen isotopic compositions were largely controlled by water–carbonate equilibrium at low temperatures [19,34,35] and subsequently underwent rapid CO₂ degassing that produced a systematic negative δ¹⁸O shift [36]. Notably, the δ¹⁸O values do not vary significantly between calcite- and aragonite-rich deposits, implying that oxygen isotopes are less sensitive to mineral polymorphs and more strongly controlled by water source and temperature. Overall, the isotopic compositions are primarily governed by CO₂ degassing and water–gas exchange processes [29].

The isotopic behavior of the Muji springs broadly parallels Rayleigh-type degassing trends reported in hydrothermal travertine systems such as Mammoth Hot Springs [6]. In the Mammoth system, δ¹³C values increase from −1‰ to +5‰ and δ¹⁸O values from −20‰ to −14‰ along the vent-to-slope flow path, accompanied by decreasing temperature (70 → 30 °C) and rising pH (6.0 → 8.0) as CO₂ degasses. The Muji system, in contrast, maintains much higher δ¹³C values (+3.5‰ to +9.1‰) and a narrower δ¹⁸O range (−12.6‰ to −10.7‰) under nearly isothermal, cold-spring conditions (≈11 °C). These differences demonstrate that, whereas the Mammoth system is thermally driven by magmatic heat, the Muji springs represent a tectonically controlled cold-spring analog, where deep-sourced CO₂ ascends through active fault conduits under sustained crustal compression.

In this context, the consistently high δ¹³C values of the Muji tufa record the persistent contribution of deep CO₂ and limited isotopic re-equilibration during rapid ascent. The narrow δ¹⁸O range reflects oxygen-isotope exchange between the spring water and dissolved inorganic carbon (DIC) under semi-open degassing conditions. Consequently, the Muji system provides a non-volcanic analogue to the vent-to-slope isotopic evolution of Mammoth hot springs, capturing a fault-controlled degassing–precipitation coupling in a cold, collision-zone environment along the northeastern Pamir Plateau.

4.3. Geochemical Characteristics and Regional Implications

The Muji carbonic spring system records a coupled process of “deep fluid input–shallow mixing–precipitation kinetics,” governed fundamentally by tectonically controlled CO₂ release under cold-spring conditions. The δ¹³C values of deposits are consistently enriched (+3.5‰ to +9.1‰), while the δ¹³C of dissolved inorganic carbon (DIC) in spring waters ranges from −4.3‰ to +5.8‰ (Table 3), both significantly higher than soil-respired CO₂ or CO₂ derived from shallow organic matter degradation (−25‰ to −10‰) [37]. The distribution of δ¹³C–Sr further highlights differences among spring groups in terms of deep-fluid contributions and shallow-water mixing. The HD group is generally characterized by high Sr concentrations (1000–1600 μg/L) and mostly positive δ¹³C values (−2‰ to +6‰), with HD-10 and HD-11 showing the most enriched signatures (>+2‰), indicating that rapidly ascending deep CO₂ fluids can preserve their isotopic characteristics with limited dilution (Figure 4). In contrast, MJX samples display pronounced heterogeneity: some (e.g., MJX-37) resemble the deep endmember, whereas others (e.g., MJX-31 and MJX-36) exhibit low Sr and near-zero or slightly negative δ¹³C values, reflecting substantial dilution by shallow waters during ascent. Importantly, Muji spring waters exhibit systematic enrichment of strontium (Sr) relative to upstream river water (186 μg/L), with consistently intermediate-to-high Sr concentrations across all spring groups regardless of their δ¹³C values. This widespread enrichment suggests that Sr is primarily controlled by water–rock interaction. In contrast, δ¹³C is more sensitive to gas–water exchange, CO₂ degassing, and shallow water mixing—processes that can significantly modify the carbon isotopic signature.

The MJXSP group falls within relatively low Sr (500–800 μg/L) and intermediate δ¹³C (−0.5‰ to +2‰) ranges. Although its Sr concentrations are higher than those of river water, the δ¹³C values are clearly closer to shallow-water signatures, implying that deep-fluid contributions are relatively limited and shallow-water mixing plays a dominant role. River water samples also reveal a clear downstream trend: upstream waters (MJXRU) show the lowest Sr (186 μg/L) and most negative δ¹³C (−5.3‰), whereas downstream waters (MJXRD and MJXR) display significantly elevated Sr (>1300 μg/L) and enriched δ¹³C values (−4‰ to −3‰). These trends indicate that spring inflows progressively accumulate downstream, exerting significant modifications on both hydrochemistry and carbon isotopes. Overall, the distribution of Sr–δ¹³C not only distinguishes deep-source dominated (HD), mixed heterogeneous (MJX), and shallow-influenced (MJXSP) spring groups but also highlights the role of spring inputs in reshaping regional hydrochemistry and the carbon cycle, further supporting the coupled process of “deep input–shallow mixing–river modification.”

The distribution of Sr–Ba–U further elucidates the coupling between springs and river water. Muji spring waters contain Sr (563–1955 μg/L), Ba (7.6–154 μg/L), and U (0.03–7.06 μg/L), significantly higher than upstream river waters (Sr 186 μg/L; Ba 9.5 μg/L; U 0.02 μg/L) (Table 3; Figure 5). This pattern clearly indicates that the Muji spring water inflow substantially elevates Sr-Ba-U concentrations in the receiving river reach. Further downstream from the Muji river, however, the concentrations of Sr (1351 μg/L) and U (3 μg/L) are lower than those at site MJXR, demonstrating a notable decline. This trend reflects a dilution effect. However, the Ba concentrations remain relatively increased—likely due to stronger buffering by carbonate precipitation rather than dilution alone. Meanwhile, δ¹³C_DIC in downstream waters also increases from −5.3‰ to −3.3‰, consistent with elemental enrichment and confirming that spring inflow not only modifies the major and trace element composition but also exerts a measurable influence on riverine carbon isotopes.

From a regional perspective, the Muji spring waters exhibit distinctly higher Sr concentrations (27–1955 μg/L; median ≈ 1030 μg/L) than those from the Yellowstone (3.5–508 μg/L; median ≈ 54 μg/L). Although Ba concentrations in some Yellowstone springs reach higher maxima (1890 μg/L), the median Ba level in the Muji springs (≈48 μg/L) remains consistently higher, reflecting a more uniform deep-fluid contribution (

Table 4. Statistical summary of trace element concentrations (Sr, Ba, U) in carbonic spring systems of Muji, Yellowstone, and Tashkurgan.

Elements	N	Minimum	P10	Median	P90	Maximum	Mean ± SD
Muji
Sr (μg/L)	24	27.3	324.2	1031	154.7	1955	1023.1 ± 491.3
Ba (μg/L)	24	2.8	10.5	48.1	70.2	154	46.0 ± 31.7
U (μg/L)	23	0.03	0.12	1.50	6.95	9.57	2.75 ± 2.83
Yellowstone *
Sr (μg/L)	39	3.50	4.60	54.0	464	508	172.8 ± 190.2
Ba (μg/L)	15	1.0	2.8	18	1728	1890	442.5 ± 718.1
U (μg/L)	39	0.02	0.08	6.61	13.05	14.82	6.40 ± 5.59
Tashkurgan *
Sr (μg/L)	6	249	575	937	1735	2061	1082.2 ± 605.6
Ba (μg/L)	6	6.92	10.5	32.0	69.1	85.6	37.2 ± 28.6
U (μg/L)	6	0.01	0.03	0.12	0.39	0.54	0.18 ± 0.20

Note: *: data from ref [39,40,41] and See Supplementary Table S1 for details.; N–number of analyzed samples; P10 and P90—10th and 90th percentile values, respectively, representing the lower and upper bounds of the main data distribution; Mean ± SD—arithmetic mean and one standard deviation.

) [38,39]. In contrast, U concentrations in the Muji springs (0.03–9.57 μg/L; median ≈ 1.5 μg/L) are generally lower than those in the Yellowstone springs (0.02–14.82 μg/L; median ≈ 6.6 μg/L), although localized enrichment is observed at some sites (e.g., sample HD-07: 7.06 μg/L).

This difference likely arises from Yellowstone lies being located within a high-heat-flow volcanic–hydrothermal system characterized by oxidizing conditions that favor the enrichment of soluble U(VI). In contrast, the Muji system is a cold, CO₂-rich spring situated within an active strike-slip fault zone. Field measurements show that the Muji spring waters exhibit Eh values mostly between −98 and +46 mV, indicating overall relatively reducing conditions. Such an environment favors the reduction and immobilization of uranium, leading to generally lower U concentrations in the Muji waters.

The Tashkurgan springs lie within the same regional strike-slip boundary zone of the Pamir, approximately 150 km southwest of Muji springs, and therefore share a similar tectonic framework. However, unlike the cold CO₂-rich Muji springs, the Tashkurgan system is influenced by elevated heat flow associated with Mesozoic–Cenozoic granitic intrusions, giving rise to distinct hydrothermal conditions and geochemical signatures. Compared with the Tashkurgan hot springs (Sr = 249–2061 μg/L; Ba = 6.9–85.6 μg/L; U = 0.01–0.54 μg/L), the Muji waters occupy an intermediate-to-high compositional range, indicating significant but not extreme deep-fluid input. This Sr–Ba–U geochemical pattern is consistent with the combined influence of CO₂-rich deep-fluid inflow, water–rock interaction, and carbonate precipitation. Elevated Sr primarily results from the dissolution of carbonates and feldspars [42]; Ba variations are buffered by its preferential incorporation into calcite and aragonite [43]; and the spatial variability of U likely reflects local redox heterogeneity along the fluid flow paths.

Overall, this combination of enriched δ¹³C values, elevated Sr and Ba concentrations, and variable U content indicates that CO₂-rich deep fluids experienced significant upward migration, followed by dilution by surface waters and kinetic regulation during carbonate precipitation. Continuous tectonic activity rejuvenates the fault conduits, ensuring the persistent ascent of CO₂-charged fluids, while low-temperature kinetics and fluid–rock interactions govern the precipitation processes. Such sustained fluid ascent not only maintains CO₂ release but also promotes carbonate precipitation, forming an efficient near-surface CO₂ trapping mechanism. Therefore, the Muji system represents a typical non-volcanic analogue of deep CO₂ release within a continental collision zone, offering a new perspective on the interactions among tectonism, fluid flow, and sedimentation—and their broader implications for the global carbon cycle in the northeastern Pamir Plateau.

5. Conclusions

This study presents mineralogical and geochemical data for carbonic spring waters and tufa deposits from the Muji Basin. The deposits are mainly composed of calcite and aragonite, with those relative abundances reflecting altering CO₂ degassing intensities and associated precipitation kinetics under cold-spring conditions. The integrated geochemical and isotopic evidence reveals a coupled process of “deep fluid input–shallow mixing–precipitation evolution.” The persistently enriched δ¹³C values (+3.5‰ to +9.1‰) and Sr–Ba enrichment suggest that the CO₂ is primarily derived from metamorphic decarbonation of carbonate-bearing crustal rocks. Meanwhile, the progressive enrichment of Sr–Ba–U in downstream river waters confirms the significant influence of spring inflows on regional hydrochemistry and surface carbon budgets. These findings demonstrate that the Muji system represents a typical tectonically driven, non-volcanic example of deep CO₂ release within a continental collision zone. Continuous fault reactivation facilitates long-term crustal CO₂ discharge, while low-temperature kinetics regulate carbonate precipitation and partial carbon sequestration. Therefore, the Muji system provides an important natural analogue for studying deep CO₂ release and crustal carbon cycling in non-volcanic regions.