Subduction-Induced Fractionated Highly Siderophile Element Patterns in Forearc Mantle

Abstract

Compositions of highly siderophile elements (HSEs) in forearc mantle have been little studied and effects of slab dehydration on their abundances in forearc mantle remains unclear. This study reports two different kinds of HSE patterns for peridotites from a New Caledonia forearc ophiolite. The Group-I samples show relatively flat patterns of Ir-group-platinum-group elements (IPGEs) and enrichment of Pt over Pd. Such patterns imply that interstitial sulfides were significantly removed through melt extraction, whereas sulfides enclosed within silicates were mostly unaffected. Meanwhile, Pt-Fe alloys were generated, resulting in suprachondritic Pt/Pd ratios. In contrast, the Group-II samples display convex HSE patterns and are depleted in all HSEs except for Ru, yielding strongly positive Ru anomalies. This indicates that both enclosed and interstitial sulfides were substantially consumed, whereas chromite was generated to stabilize Ru. Compared to abyssal peridotites, subduction-related peridotites commonly have stronger fractionation in the HSEs. Therefore, the HSE data of mantle peridotites are potentially able to discriminate the tectonic settings of ophiolites.

1. Introduction

Highly siderophile elements (HSEs), also known as iron-loving elements, include platinum-group elements (PGEs; Os, Ir, Ru, Rh, Pt, Pd), Re, and Au. Core separation is expected to strip almost all HSEs away from the mantle, due to their extremely high metal/silicate partition coefficients (D^{metal/silicate} > 10⁴) [1]. Nevertheless, they are present in the Earth’s mantle in chondritic relative abundances and at higher levels than expected, which has been explained by the late accretion hypothesis [2,3,4]. Such a model has been challenged by the discovery of non-chondritic HSE patterns of mantle samples [5,6,7], which have been explained by variable processes, e.g., addition of non-chondritic extraterrestrial material [8] or differentiated outer-core material to the primitive mantle [7]. Moreover, abundant evidences suggest that non-chondritic HSE patterns can result from partial melting and metasomatism within lithospheric mantle [6,9,10,11].

Subduction zones are likely to be important locations for generating non-chondritic HSE compositions in the subarc mantle (including mantle wedge and forearc mantle), which is evidenced by the highly fractionated HSE patterns of mantle wedge xenoliths, e.g., high Pt/Pd ratios [12,13]. The fractionation might be caused by the mobility of HSEs in slab-released fluids/melts [14], and/or the formation of PGE alloys during hydrous melting [12]. In comparison, whether subduction also produces fractionated HSE patterns of the forearc mantle remains unclear, as HSE compositions of forearc mantle have been rarely reported due to the lack of xenolith samples. This study presents HSE data of mantle peridotites from a New Caledonia ophiolite that originated in forearc settings [15,16,17]. The data support the existence of two different types of fractionated HSE patterns in forearc mantle.

2. Geological Setting and Sample Description

2.1. Geological Setting

The main island of New Caledonia is located in the Southwest Pacific and represents the emerged part of the Norfolk Ridge (Figure 1a), which is considered to be a continental crust fragment rifted from the east margin of Gondwana in the late Cretaceous [18]. The islands of Quvéa, Lifou, and Maré are the original island arc. The main island of New Caledonia has been traditionally divided into four major tectonic units: The continental basement (including “Montagnes Blanches” Nappe, late Cretaceous Paleogene cover, Teremba Terrane, Koh-Central Terrane, and Boghen Terrane), the magmatic Poya terrane, a high-pressure metamorphic belt, and ophiolitic massifs [19].

The New Caledonia ophiolites are exposed in a large massif (the Massif du Sud) occurring in the south of the main island and smaller isolated massifs along the west coast, such as Me Maoya, Kopeto, Koniambo, and Tiebaghi (Figure 1a). They are connected with the oceanic lithosphere of the Loyalty Basin [20,21] and have been regarded as the archetype of forearc ophiolites [22]. The New Caledonia ophiolites are allochthonous and were thrust from NNE to SSW over the continental basement of the main island. The New Caledonia ophiolites consist almost exclusively of mantle rocks (predominantly refractory harzburgites) and crustal rocks are commonly missing. Only a lower crustal cumulate sequence has been observed in the Massif du Sud. It has been suggested that the New Caledonia ophiolites were generated during the Paleocene, and were emplaced onto the main island during the early Eocene [23,24]. Mantle peridotites selected in this study were collected from the Me Maoya klippe, which overlies the mafic Poya terrane (Figure 1b).

2.2. Sample Description

Twenty spinel harzburgites from the Me Maoya massif were selected in this study. They were collected from big boulders in the creek, and no field relation between different samples exists. These harzburgites consist mainly of olivine (67–83% in modal contents), orthopyroxene (16–32%), and spinel (1–2%). Primary clinopyroxene is completely absent in all hazburgites. More than half of the studied samples are fresh, with loss on ignition (LOI) less than 1 wt% (

Table 1. Highly siderophile elements (HSE) data (in ppb) of the Me Maoya harzburgites.

	Al2O3	MgO	CaO	LOI	Ol Fo	Sp Cr#	Os	Ir	Ru	Pt	Pd	Re	(Os/Ir)N	(Ru/Ir)N	(Ru/Os)N	(Pt/Pd)N
Standards
WPR-1							16.9	16.7	23.7	308	240	11.2
UB-N							3.54	3.20	6.33	6.95	5.83	0.22
BHVO-2							0.08	0.07	0.11	5.52	2.73	0.57
Group-I harzburgites
15NC11	0.39	43.41	0.36	6.17	0.92	0.55	1.76	0.62	1.09	7.46	0.60	0.011	2.59	1.18	0.46	8.10
15NC12	0.32	42.37	0.40	8.15	0.91	0.55	1.09	0.75	0.46	3.01	0.60	0.029	1.32	0.41	0.31	3.27
15NC17	0.19	45.53	0.45	0.64	0.91	0.68	3.45	2.32	4.33	4.07	0.05	0.013	1.36	1.25	0.92	53.03
15NC21	0.43	45.68	0.48	0.74	0.92	0.56	0.39	0.32	1.33	0.56	0.06	0.012	1.11	2.79	2.51	6.08
15NC23	0.32	46.53	0.40	0.30	0.92	0.56	0.30	0.49	0.76	1.47	0.27	0.019	0.56	1.04	1.86	3.55
15NC25	0.74	45.09	0.51	0.24	0.93	0.57	1.62	1.23	2.78	3.19	0.09	0.197	1.20	1.52	1.26	23.09
15NC27	0.47	43.40	0.56	3.58	0.91	0.55	1.07	0.92	2.40	3.23	0.16	0.006	1.06	1.75	1.65	13.15
15NC28	0.28	45.09	0.42	1.36	0.91	0.66	1.73	0.79	1.87	4.90	0.13	0.016	2.00	1.59	0.80	24.56
15NC29	0.36	45.29	0.56	0.28	0.91	0.59	0.84	0.58	0.72	3.29	0.12	0.001	1.32	0.83	0.63	17.86
15NC30	0.78	44.45	0.50	0.10	0.92	0.59	0.24	0.43	1.86	3.26	0.04	0.002	0.51	2.90	5.70	53.10
15NC31	0.10	45.24	0.22	2.46	0.91	0.72	0.70	0.52	2.06	1.33	0.04	0.010	1.23	2.66	2.17	21.66
15NC31-R							0.46	0.66	2.68	1.20	0.15	0.020	0.64	2.72	4.29	5.21
15NC33	0.50	44.84	0.42	0.82	0.92	0.56	0.53	0.74	3.40	0.28	0.04	0.026	0.65	3.08	4.72	4.56
15NC34	0.37	45.42	0.49	−0.02	0.91	0.57	0.79	0.67	3.64	4.55	0.23	0.008	1.07	3.64	3.39	12.89
Group-II harzburgites
15NC13	0.41	43.33	0.42	5.96	0.92	0.59	0.02	0.10	1.43	0.12	0.10	0.005	0.18	9.59	52.63	0.78
15NC15	0.24	43.72	0.36	6.18	0.92	0.58	0.05	0.06	1.21	0.18	0.03	0.002	0.76	13.53	17.81	3.91
15NC20	0.42	45.99	0.40	0.04	0.92	0.60	0.02	0.20	0.72	0.08	0.06	0.001	0.09	2.42	26.50	0.87
15NC22	0.41	45.50	0.37	−0.06	0.92	0.63	0.03	0.28	0.54	0.03	0.04	0.002	0.10	1.29	13.25	0.49
15NC24	0.29	45.75	0.36	0.06	0.92	0.64	0.03	0.11	0.54	0.04	0.01	0.001	0.25	3.29	13.25	2.61
15NC24-R							0.01	0.10	0.60	0.04	0.04	0.002	0.09	4.03	44.17	0.65
15NC26	0.31	45.46	0.32	1.88	0.92	0.59	0.06	0.07	1.01	0.06	0.06	0.001	0.78	9.68	12.39	0.65
15NC26-R							0.02	0.04	1.07	0.05	0.06	0.002	0.46	17.95	39.38	0.54
15NC32	0.33	45.87	0.34	0.02	0.92	0.62	0.02	0.04	0.92	0.07	0.03	0.003	0.46	15.43	33.86	1.52

1: Major element contents are in wt%; 2: HSE contents are in ppb; 3: Chondrite normalized values are from Fischer-Godde et al. [33]; 4: The data of major elements and the contents of Re and Os in the studied samples are from Liu et al. [25].

). In comparison, the remaining eight samples have experienced variable degrees of serpentinization with higher LOI values of 1.36–8.15 wt%. They display porphyroclastic to slightly deformed textures (Figure 2), in which orthopyroxenes commonly occur as porphyroclasts and olivines occur as neoblasts (Figure 2a). Orthopyroxene generally displays irregular forms and shows abundant exsolution lamellae of clinopyroxene (Figure 2a). Olivine neoblasts are relatively small (~0.1–1 mm) and locally display kink bands (Figure 2a). Large spinel grains usually display a lobate texture, whereas small subhedral spinels are interstitial among silicate minerals (Figure 2c,d). Quite small (<0.06 mm) clinopyroxene can be observed to be interstitial along the boundary of olivine, othopyroxene, and spinel (Figure 2c,d). In the serpentinized harzburgites, olivine commonly develops a mesh texture and has been variably replaced by serpentine (Figure 2b). Interstitial sulfide is almost absent in studied samples; only tiny sulfide droplets and ferro-alloys with sizes <15 μm locally occur as inclusions within olivine, orthopyroxene, and spinel (Figure 2e,f). Some sulfides have been partially replaced by Fe hydroxide phases in the serpentinized samples.

3. Analytical Methods

Whole-rock HSE contents were analyzed using isotope dilution. Detailed analytical procedures have been previously described [27]. About 2 g of sample powders were mixed with appropriate amounts of a mixed ¹⁸⁷Re-¹⁹⁰Os spike and a mixed ¹⁹¹Ir-⁹⁹Ru-¹⁹⁴Pt-¹⁰⁵Pd spike into Carius tubes, which were digested by a mixed acid of 3 mL 12 N HCl and 6 mL 16 N HNO₃ at ~240 °C for 48–72 h in an oven. Osmium was extracted from the aqua regia solution by solvent extraction into CCl₄ and further purified by micro-distillation [28]. Rhenium and remaining PGE (Ir, Ru, Pt, Pd) were first separated from the solution into subgroups (Re–Ru, Ir–Pt, and Pd) by a 2 mL anion exchange resin (AG-1×8, 100–200 mesh). Subsequently, the Re–Ru was further purified by a 0.25 mL anion exchange resin; the Pd and Ir–Pt were further purified by an Eichrom LN spec resin to completely remove Zr and Hf.

Osmium concentrations and isotopes were measured by negative thermal ionization mass spectrometer (N-TIMS) on a GV Isoprobe-T instrument. Details are given in Liu et al. [25]. The concentrations of other HSEs were measured on a Thermo Fisher Scientific Neptune multi-collector ICP-MS (MC-ICPMS) with an electron multiplier in peak-jumping mode or using Faraday cups in static mode, according to the measured signal intensity. In-run precisions for ¹⁸⁵Re/¹⁸⁷Re, ¹⁹¹Ir/¹⁹³Ir, ⁹⁹Ru/¹⁰¹Ru, ¹⁹⁴Pt/¹⁹⁶Pt, and ¹⁰⁵Pd/¹⁰⁶Pd were 0.1%–0.3% (2 s). The Re, Ir, Ru, Pt, and Pd standards were used to correct mass fractionations. The correction rate is less than 0.3%. The total procedural blanks were 3 pg for Os, 4 pg for Re, 3 pg for Ir, 9 pg for Ru, 10 pg for Pt, and 6 pg for Pd. Three standards, i.e., WPR-1, UB-N, and BHVO-2, were used to monitor the accuracy of the analytical procedures. The ¹⁸⁷Os/¹⁸⁸Os ratios for these standards were previously reported [25] and their HSEs contents are given in Table 1. These values are consistent with the published values within errors [29,30,31].

4. Results

The HSE contents are listed in Table 1. The reproducibility of the concentrations, based on three duplicate analyses of samples (i.e., 15NC24, 15NC26, and 15NC31), are more than 10% for some HSEs, indicating that the “nuggest effect” (inhomogeneous distribution of PGE-bearing phases) is significant for the Me Maoya harzburgites. However, the HSE patterns themselves are generally reproducible. Compared to the primitive mantle (PM; Reference [32]), the Me Maoya harzburgites have variably lower contents of HSE (Figure 3), i.e., ~0.002–0.98 times the PM. The studied samples can be divided into two groups according to their HSE patterns. The Group-I harzburgites display relatively flat patterns from Os to Pt and are depleted in Pd, whereas Re is variably enriched relative to Pd (Figure 3a). They have high Pt/Pd ratios, giving maximum (Pt/Pd) _N (N: Chondrite-normalized; Reference [33]) ratios of 53.1 (cf. 1.03 of the PM). Compared to the Group-I harzburgites, the Group-II samples are strongly depleted in all HSEs except for Ru (Figure 3b). They display convex patterns, with strong fractionation in IPGEs (Os, Ir, and Ru). They have low (Os/Ir)_N ratios of 0.09–0.78, which are much lower than the inferred values (~1.01) of the primitive mantle. They also display strongly positive Ru anomalies (Figure 3b) and have supra-chondritic (Ru/Ir) _N ratios of 1.29–17.95 (cf. 1.46 of the PM). Compared to the Group-I samples, the Group-II samples have higher Ru/Os but relatively lower Pt/Pd ratios (Figure 4), i.e., 0.4–7.8 and 5.0–81.5 versus 16.8–71.5 and 0.8–6.0, respectively. There is no significant difference in the Os isotope ratios between the two groups of the Me Maoya harzburgites. The ¹⁸⁷Os/¹⁸⁸Os ratios do not show any correlation with Ru/Ir or Pt/Pd ratios (Figure 5).

5. Discussion

Both major elements and Os isotope data of the studied samples have been previously reported [25]. They have refractory compositions, with high MgO (42.37–46.53 wt%) but low Al₂O₃ (0.1–0.78 wt%) and CaO (0.22–0.56 wt%). They have high olivine fosterite contents (Fo) of 0.91–0.93 and spinel Cr# values of 0.55–0.72. The refractory features of the Me Maoya peridotites are also supported by their low HSE contents (i.e., ~0.002–0.98 times the PM). The refractory compositions imply that the Me Maoya peridotites represent residues after high degrees of partial melting and have affinities to forearc peridotites, which is consistent with previous inference that the New Caledonia ophiolite are relics of ocean lithosphere formed in a forearc setting [15,16,17].

Mantle peridotites of the New Caledonia ophiolites have commonly experienced strong serpentinization and supergene weathering, which produced the world-famous nickel laterite ore deposits [34]. The studied Me Maoya harzburgites have also been subjected to variable degrees of alteration, as evidenced by their variable LOI values of 0–8.15 wt%. Therefore, the effects of alteration processes on their HSEs should be evaluated before scrutinizing the HSE data. Serpentinization might have little effects on HSEs of mantle peridotites, as HSEs are unreactive in a reducing environment [7,8,35,36]. In contrast, supergene weathering occurred under an oxidizing environment, and thus, could potentially affect HSEs of the studied samples [7,37,38]. Nevertheless, no correlation between the LOI values and HSE contents (Figure 6) suggests that alteration (neither serpentinization nor supergene weathering) is not the culprit for non-chondritic HSE patterns of the Me Maoya harzburgites.

The Group-I harzburgites have HSE contents lower overall than the values of the PM (Figure 3a). They display relatively flat IPGEs (i.e., Os, Ir, Ru) patterns and strong depletion in Pd (Figure 3a). Such fractionated HSE patterns can be produced by partial melting, as melt extraction can result in depletion of PPGEs (i.e., Rh, Pt, Pd) and Re relative to IPGEs in mantle peridotites [6]. The fractionation has been commonly attributed to incongruent melting of base metal sulfides (BMS) during partial melting, which produces Cu-Ni-rich sulfide melts and leaves refractory monosulfide solid solutions (Mss) in the residues [39]. The refractory Mss concentrates IPGEs, whereas the Cu-Ni-rich sulfides are enriched in PPGEs and Re [9,39]. The Cu-Ni-rich sulfide melts can be physically removed along with magmas, resulting in the depletion of PPGE and Re relative to IPGEs in the residual mantle [40].

Although two samples are depleted in Pt over Pd, most Group-I harzburgites do not show any Pt depletion at all or are even slightly enriched in Pt, with (Pt/Pd) _N ratios of 3.27–53.10. Enrichment of Pt over Pd in mantle peridotites cannot be simply explained by incongruent melting of sulfides, as they have similar partition coefficients in sulfides [41]. Thus, other mechanisms should be sought to account for their high Pt/Pd ratios. As shown in Figure 7a, similarly high Pt/Pd ratios have been also previously reported for mantle wedge xenoliths [12,13], which have been explained by formation of Pt–Fe alloys during multi-stage melt depletion [12]. Awaruite, i.e., Pt-rich Fe-Ni alloy (Ni₃Fe-Ni₂Fe), has been reported for depleted harzburgites from the New Caledonia ophiolites [42], although it has not been discovered in the studied samples. Formation of such Pt-rich alloys during hydrous melting can effectively fractionate Pt from Pd in mantle residues. It should also be noted that Re is variably enriched over Pd in Group-I samples. This might be due to subduction-related metasomatic processes, during which Re was added by the metasomatic agents.

Compared to the Group-I harzburgites, the Group-II samples are extremely depleted in all HSEs except for Ru (Figure 7b). Their HSE display highly fractionated convex patterns, with remarkably positive Ru anomalies, and also show strong fractionation in IPGEs, with sub-chondritic Os/Ir ratios (i.e., 0.09–0.78). Such features indicate that BMS in their protoliths, not only interstitial sulfides, but also sulfide inclusions within silicates, should have been substantially removed during hydrous melting.

Although their sub-chondritic Os/Ir can be explained by sulfide breakdown in the presence of hydrous fluids, as has been proposed for mantle xenoliths [46], removal of sulfides alone is hard to explain their remarkably positive Ru anomalies (i.e., supra-chondritic Ru/Ir ratios). Supra-chondritic Ru/Ir ratios have been reported for peridotites from different tectonic settings and might be an indigenous feature of the Earth’s mantle [8]. It has led to speculation that fractionated meteoritic materials were added as late-veneer after core segregation [8], or a portion of Ru was not stripped to the core and preserved in sulfides within the mantle [47]. However, elevated Ru/Ir ratios >10 are not common for mantle samples, and have been rarely reported for some mantle xenoliths [35]. Therefore, extremely high Ru/Ir ratios up to 26 of the Group-II samples, coupled with their convex HSE patterns, are unlikely indigenous features of Earth’s mantle.

Besides, it has been suggested that strongly supra-chondritic Ru/Ir ratios of mantle peridotites might reflect addition of Ru from the percolating melts [35]. However, the New Caledonia peridotites with Os concentrations of <1 ppb have ¹⁸⁷Os/¹⁸⁸Os ratios comparable to those of the ambient oceanic mantle (i.e., ¹⁸⁷Os/¹⁸⁸Os < 0.13). This implies that the HSE-rich phase, e.g., sulfide, in the subducting slab can remain stable during dehydration at shallow depths [25]. Thus, minimal HSEs were transported to the forearc mantle via slab dehydration. The lack of interstitial sulfide in the studied samples also supports that the effect of the percolating melt on Me Maoya peridotites is extremely limited.

A positive Ru anomaly reflects the stability of Ru-bearing phases during melt extraction. Both laurite (RuS₂) and Os-Ir-Ru alloys are two major platinum-group metals (PGM) that stabilize Ru in the mantle, both of which have been discovered in mantle peridotites [48,49]. Hydrous melting at high fugacity of oxygen (f_O2) in forearc mantle is conducive to PGE alloy nucleation, as it can efficiently reduce the fugacity of sulfur (f_S2). During partial melting, Mss in mantle peridotites are totally substituted by laurite, and then by Os–Ir–Ru alloys with a further decrease in f_S2, whereas Pt-rich alloys start to form at higher f_S2 than do Ru-rich alloys/sulfides [50]. Although both laurite and Os-Ir-Ru alloys can stabilize Ru in the mantle, the enrichment of Ru in Group-II samples requires a phase that does so preferentially over Os, Ir, and the other HSEs. Chromite is, plausibly, such a candidate. Based on a positive relationship between whole-rock Cr and Ru contents, it has been inferred that chromite is an important host mineral of Ru in komatiites [51], which is supported by high Ru contents in chromites obtained by laser ablation [52,53]. Strong compatibility of Ru in chromites has been explained by two different mechanisms. The first possibility is that Ru-bearing minerals (e.g., laurite) occur as micro-scale inclusions within chromites [54]. Alternatively, Ru could substitute into the crystal lattice of chromites at high f_O2 and low f_S2, as the ionic radii of Ru³⁺ is theoretically similar to Fe³⁺ and Cr³⁺ within the spinel structure [55]. Both mechanisms can stabilize Ru in forearc mantle, whereas other HSEs are scavenged. This explains the strong positive Ru anomalies observed for the Group-II harzburgites.

Two different types of HSE patterns observed for New Caledonia harzburgites indicate that HSE in forearc mantle have been variably fractionated. This implies that the forearc mantle has been heterogeneously affected by hydrous melting (Figure 8). In some portions of the forearc mantle, the interstitial sulfides were significantly removed, whereas sulfides enclosed within silicates and spinels remained mainly untouched. Meanwhile, Pt-rich alloys were produced in these regions during high degrees of hydrous melting. In contrast, some other portions of the forearc mantle might have higher f_O2 or contain less sulfur, in which almost all sulfides have been stripped along with melt extraction. In this case, breakdown of sulfides will release all HSEs to the hydrous magmas, whereas Ru was partitioned to chromites.

Combined with HSE compositions of mantle wedge xenoliths [12,13], our data support that subduction can result in strong fractionation in HSEs of the subarc mantle at different depths. Subduction continuously fluxes fluids from the slab to the overlying mantle. The fluids transport fluid-mobile elements (such as Cs, Rb, and Sr) and potential oxidants (e.g., H₂O, S, and C), which oxidize the subarc mantle [56]. Sulfides in the subarc mantle are quickly consumed during hydrous melting under oxidizing conditions, as water can increase the sulfur solubility in melts [57]. Hydrous melting should result in a quick decrease of sulfur fugacity in the subarc mantle, driving the formation of PGM [50]. Thus, subduction-modified mantle is more likely to obtain fractionated HSE patterns compared to mid-ocean ridge mantle, where anhydrous melting is expected. Compared to abyssal peridotites, some mantle wedge xenoliths and forearc peridotites have remarkably higher Ru/Ir and Pt/Pd ratios (Figure 9). Thus, HSE compositions of mantle peridotites can be applied to discriminate tectonic settings of ophiolite formation (Figure 9). For example, high Ru/Ir ratios have also been reported for mantle peridotites from the Oman ophiolites [58], consistent with their affinity to supra-subduction zone (SSZ) settings [59]. The HSE compositions of peridotites from worldwide ophiolites are compiled in Figure 9. The Ru/Ir ratios in peridotites from Oman and Scotland ophiolites display increasing tendencies at low Al₂O₃ contents, which are similar to those in New Caledonia ophiolites and some mantle wedge xenoliths. This implies that Oman ophiolite and Scotland ophiolite may have undergone evolution in the subduction zone setting, which are consistent with previous studies on those ophiolites [58,60].

6. Conclusions

Refractory mantle peridotites from the New Caledonia ophiolites provide a unique opportunity to study the HSE signatures of forearc mantle and evaluate effect of slab dehydration on their abundances. The peridotites from the Me Maoya klippe of the New Caledonia ophiolites show non-chondritic, strongly fractionated HSE patterns. There are no correlations between HSE concentrations and loss on ignition among the samples, which suggests that HSEs are not affected by secondary alteration processes. Based on the characteristics of HSE, the studied peridotites can be classified into two distinct types. The Group-I samples have relatively flat patterns of IPGEs, coupled with supra-chondrite Pt/Pd ratios, resembling the features of some mantle wedge xenoliths. This reflects that interstitial sulfides were substantially consumed but sulfides enclosed were retained during partial melting of peridotites, and Pt-Fe alloys were generated to produce Pt enrichment over Pd. The Group-II samples are characterized by convex HSE patterns with a strong positive Ru anomaly, which reflects that both interstitial and enclosed sulfides were significantly removed, and Ru was partitioned to chromites. The HSE data of the New Caledonia ophiolite peridotites and wedge xenoliths support that subduction-related peridotites commonly display much stronger fractionation in HSE patterns than abyssal peridotites. Therefore, tectonic settings of ophiolites can be effectively discriminated by using the HSE compositions of the constrained mantle peridotites.