Cryogenic Performance and Modelling of Fibre- and Nano-Reinforced Composites: Failure Mechanisms, Toughening Strategies, and Constituent-Level Behaviour

Abstract

Composite materials are increasingly required to operate in cryogenic environments, including liquid hydrogen and oxygen storage, deep-space structures, and polar infrastructures, where long-term strength, toughness, and reliability are essential. This review provides a unique contribution by systematically integrating recent advances in understanding cryogenic behaviour into a unified multi-scale framework. This framework synthesises four critical and interconnected aspects: constituent response, composite performance, enhancement mechanisms, and modelling strategies. At the constituent level, fibres retain stiffness, polymer matrices stiffen but embrittle, and nanoparticles offer tunable thermal and mechanical functions, which collectively define the system-level performance where thermal expansion mismatch, matrix embrittlement, and interfacial degradation dominate failure. The review further details toughening strategies achieved through nano-addition, hybrid fibre architectures, and thin-ply laminates. Modelling strategies, from molecular dynamics to multiscale finite element analysis, are discussed as predictive tools that link these scales, supported by the critical need for in situ experimental validation. The primary objective of this synthesis is to establish a coherent perspective that bridges fundamental material behaviour to structural reliability. Despite these advances, remaining challenges include consistent property characterisation at low temperature, physics-informed interface and damage models, and standardised testing protocols. Future progress will depend on integrated frameworks linking high-fidelity data, cross-scale modelling, and validation to enable safe deployment of next-generation cryogenic composites.

1. Introduction

In extreme cryogenic environments, typically below 123 K, materials exhibit behaviour that differs fundamentally from their response at ambient temperatures. Such environments are found in applications including liquid hydrogen and oxygen propellant tanks, deep-space exploration, and polar infrastructure. In these settings, materials must maintain long-term strength, toughness, and reliability [1,2,3,4,5]. FRPs derive strength from continuous fibres and are widely used in aerospace and energy applications [6,7,8]. In contrast, NP composites enhance matrix toughness and thermal stability by leveraging the interfacial interactions and energy dissipation of nanoscale fillers [9,10,11]. This behaviour is underpinned by several key mechanisms: the high specific surface area of nanoparticles promotes strong interfacial bonding through van der Waals forces, hydrogen bonding, or covalent interactions, which enhances stress transfer and interfacial energy dissipation under cryogenic conditions. Additionally, nanofillers impose constraint effects by restricting polymer chain mobility and reducing free volume, thereby mitigating matrix embrittlement and improving dimensional stability. However, their cryogenic performance results from specific interactions such as thermal expansion mismatch between constituents, loss of molecular mobility leading to matrix embrittlement, and interfacial stress generation under combined thermal and mechanical influences, rather than from simple improvements in strength or toughness.

These interdependencies are illustrated in Figure 1, which presents the multi-scale framework used in this review. The diagram integrates four main aspects: (i) constituent materials including fibres, the matrix, and nanoparticles, (ii) key performance indicators such as thermal expansion coefficients (CTE), fracture toughness, and interfacial shear strength (ILSS), (iii) toughening mechanisms including nano-addition and synergistic effects, and (iv) modelling strategies ranging from molecular dynamics to finite element analysis. This framework provides a comprehensive view of how cryogenic conditions influence composite materials and helps guide the structure of the following sections.

At the material level, fibres, the matrix, and nanoparticles each respond differently to cryogenic temperatures, and these differences collectively shape the overall behaviour of the composite. Carbon fibres exhibit near-zero or even negative thermal expansion along the fibre direction, but expand in the transverse direction. In contrast, glass and aramid fibres contract in both directions [7,12,13]. This anisotropy contributes to dimensional stability but also leads to significant residual stress at the fibre-matrix interface upon cooling. For instance, the negative or near-zero axial CTE of carbon fibres combined with positive transverse CTE, versus the positive contraction of glass and aramid fibres in both directions, generates complex interfacial stress states (e.g., radial compression in carbon-fibre systems versus tensile debonding stresses in glass-fibre systems), which is a central driver of cryogenic damage such as microcracking and delamination. Most polymer matrix materials become stiffer and more brittle at low temperatures due to reduced chain mobility and shrinkage in free volume. Thermosetting polymers such as epoxy and polyimide lose significant toughness, while thermoplastics like polyether ether ketone (PEEK) retain better impact resistance and structural stability [14,15,16,17,18]. Nanoparticles offer additional avenues for material tuning. Ceramic fillers such as boron nitride (BN) and aluminium nitride (AlN) provide high thermal conductivity and low thermal expansion [2,19]. Zirconium tungstate (ZrW₂O₈) exhibits negative thermal expansion, which can offset matrix shrinkage [10,20]. Nanostructures like graphene, carbon nanotubes (CNTs), and MXenes (two-dimensional transition metal carbides/nitrides) enhance stiffness, improve interfacial bonding, and regulate heat transfer [21,22,23,24].

The combined behaviour of these components defines the system-level performance, where thermal mismatch, matrix embrittlement, and interfacial degradation become the dominant factors. Interfacial degradation primarily results from cyclic thermal stresses that induce microcracking and fatigue debonding, chemical incompatibility leading to adhesive failure under thermal cycling, and cryogenic contraction-induced loss of mechanical interlock. Thermal expansion mismatch introduces residual stresses that lead to microcracking and dimensional instability [7,25]. For example, between 4 and 293 K, glass-fibre composites contract by approximately 0.2% along the fibre axis and 0.7% transversely, indicating a matrix-dominated deformation response [13]. Matrix embrittlement reduces the threshold for crack initiation and promotes delamination under mechanical or thermal loading [26,27,28,29]. Repeated thermal cycling increases crack propagation and reduces the material’s ability to dissipate energy [30]. Interfaces are particularly sensitive. While cryogenic contraction may initially enhance interfacial shear strength due to radial compression [31,32], prolonged exposure or thermal cycling can lead to microcracking and adhesion loss. Studies show that interlaminar fracture resistance tends to peak near 77 K but declines significantly at 4 K [33,34,35,36]. Surface modifications such as graphene oxide (GO) deposition, CNT alignment, and nanosilica incorporation have been shown to improve interfacial shear strength and fracture toughness by 10–30% under cryogenic conditions [23,37,38,39].

To overcome these degradation mechanisms, researchers have developed multi-scale toughening strategies. Carbon nanomaterials improve energy dissipation by deflecting cracks [21,22,39]. Ceramic fillers help reduce thermal mismatch and mitigate internal stresses [10,19,20]. Hybrid systems, such as combinations of graphene and CNTs or GO and polyurethane, provide balanced improvements in stiffness, toughness, thermal cycling resistance, and compatibility with liquid oxygen environments [40,41]. Thin-ply laminates and hybrid-fibre architectures can further delay crack propagation and reduce delamination susceptibility [42,43], highlighting the benefits of structural tailoring.

Modelling strategies play a critical role in linking material behaviour across scales. Atomistic simulations, particularly molecular dynamics (MD), reveal mechanisms such as chain immobilisation and interfacial interactions [44]. Micromechanical representative volume element (RVE) models upscale these local effects to predict stress concentrations and microcrack initiation [34,45]. Finite element (FE) frameworks that incorporate progressive failure models simulate system-level degradation and help quantify structural reliability under cryogenic loading [46,47]. Experimental tools such as acoustic emission, in situ X-ray computed tomography (CT), and digital image correlation (DIC) enhance the validation of these models by providing detailed insight into failure mechanisms across multiple length scales [46,47,48].

In summary, the performance of FRPs and NP-reinforced composites in cryogenic environments is governed by complex, multi-scale interactions among fibres, the matrix, and nanofillers. Fibres offer dimensional stability, the matrix tends to stiffen and embrittle, and nanoparticles provide additional pathways for tuning material response. Key failure mechanisms include thermal mismatch, brittle matrix behaviour, and interfacial degradation. Specifically, interfacial degradation is caused by progressive loss of adhesion due to CTE mismatch stresses, polymer chain immobilisation reducing interfacial toughness, and cyclic thermal loading leading to fatigue-driven debonding. These can be mitigated through advanced nano-reinforcements and hybrid structural designs. Cross-scale modelling, from atomistic to continuum levels, offers a unified predictive framework that links fundamental mechanisms to macroscopic performance. Nonetheless, challenges remain in achieving accurate experimental-numerical integration and in translating detailed physical mechanisms into practical design rules. Addressing these gaps is essential to enable the reliable deployment of cryogenic composites in demanding applications such as liquid hydrogen storage, deep-space structures, and polar systems. Crucially, the performance of these materials is inextricably linked to their processing history. Manufacturing methods dictate defect populations (e.g., porosity) and initial residual stress states, which are significantly amplified at cryogenic temperatures. To provide a manufacturing context,

Table 1. Main manufacturing methods for fibre- and nano-reinforced composites in cryogenic applications.

Manufacturing Method	Key Process Characteristics	Typical Cryogenic Applications	Relevance to Cryogenic Performance (Pros/Cons)
Filament winding.	Continuous fibres wound under tension over a mandrel; high automation	Cryogenic storage tanks (Type III/IV COPVs for Liquid H2/O2) [46]	Pros: High fibre volume fraction; excellent hoop strength for pressure vessels. Cons: High residual stresses due to winding tension and thermal contraction; requires careful cure cycle control to prevent microcracking.
Autoclave moulding	Prepregs cured under high heat and pressure in a vacuum bag.	Aerospace Structures (Launch vehicle fairings, satellite components) [44]	Pros: Lowest void content and superior consolidation; consistent mechanical properties. Cons: High cost; process-induced thermal residual stresses can lead to warping at cryogenic temperatures
VARTM (Vacuum assisted resin transfer moulding)	Resin infused into dry fibre preforms under vacuum.	Polar infrastructure, Marine composites, Large structural panels [1]	Pros: Cost-effective for large parts; flexible fibre architecture. Cons: Slightly higher void content than autoclave; resin-rich areas may become brittle spots at low temperatures.
Ultrasonic/Shear mixing	High-energy dispersion of nanoparticles into resin prior to fibre impregnation	Nano-modified cryogenic composites (Toughened matrices for tanks/structures) [21]	Pros: Essential for breaking agglomerates of CNTs/Graphene to ensure barrier properties (H2 permeation) and toughening. Cons: Improper mixing introduces air bubbles; excessive heat during mixing can degrade polymer chains.

summarises the primary production methods employed for cryogenic fibre- and nano-reinforced composites, highlighting their specific relevance to low-temperature applications.

The primary objective of this review is to establish a comprehensive and coherent framework that integrates the current understanding of cryogenic composites across scales. It aims to systematically consolidate and analyse recent advances in four interconnected domains: (1) the cryogenic response of individual constituents (fibres, matrices, and nanofillers), (2) the resulting system-level performance and failure mechanisms, (3) strategies for performance enhancement and toughening, and (4) predictive multiscale modelling approaches. By synthesising these aspects, this work seeks to bridge the gap between fundamental material behaviour at the constituent level and the structural reliability of composite systems, thereby providing a clear roadmap for the development and application of next-generation fibre- and nano-reinforced composites in extreme cryogenic environments.

2. Effects of Cryogenic Temperatures on Composite Constituents

At deep cryogenic temperatures (below 77 K), polymeric matrix and reinforcement materials display distinctly different intrinsic material responses, driven by fundamental changes in molecular dynamics [1,5]. In terms of intrinsic material responses, molecular motions within the polymer matrix become frozen [49], leading to a sharp rise in storage modulus towards glassy limits [50,51]. While tensile strength often increases modestly before leveling off or diminishing [5,14,50], the matrix becomes inherently stiffer. By contrast, continuous fibres (e.g., carbon, glass, aramid) retain their high strength and stiffness while exhibiting very low or slightly negative axial thermal expansion [52,53,54,55]. For example, polyacrylonitrile (PAN)-based carbon fibre tows maintain about 230–240 GPa modulus and 4.9–5.0 GPa tensile strength from 300 K down to 77 K [52,53]. These intrinsic disparities precipitate specific composite-scale consequences. The matrix’s collapsed ductility causes yield and fracture strains to fall precipitously, resulting in a dramatic decline in overall toughness (impact strength, fracture toughness) [51,52,56]. Furthermore, the divergence between the contracting matrix and the dimensionally stable fibres creates a structural conflict, generating significant residual thermal stresses due to CTE mismatch [1]. Overall, the key trend is that matrices stiffen but lose ductility and toughness, while fibres remain strong and dimensionally stable—creating a property imbalance that drives thermal stress and limits composite toughness at cryogenic conditions.

The following sections briefly review the cryogenic performance of polymer matrix materials (thermosetting and thermoplastic resins) and reinforcement phases (continuous fibres and nano-fillers), highlighting key property trends and design implications. As depicted in Figure 2, epoxy-family systems exhibit pronounced increases in modulus and tunable thermal transport with boron nitride (BN) fillers, whilst Figure 2 illustrates the rate- and temperature-dependent tensile responses and thermal contraction of representative thermoplastics.

2.1. Effects of Cryogenic Temperatures on Polymer Matrix

2.1.1. Thermosets (Epoxy/Cyanate Ester/Bismaleimide)

At cryogenic temperatures, thermoset resins (epoxy, cyanate ester (CE), bismaleimide (BMI), etc.) exhibit behaviour defined by two distinct mechanisms: a viscoelastic-to-glassy transition and a fracture-driven embrittlement. Regarding the former, segmental relaxations cease as temperature drops, driving the material into a deeply glassy state. Dynamic-mechanical analyses (DMA) demonstrate that the storage modulus (E′) rises sharply, whilst the loss factor ($\tan \beta$) and loss modulus (E″) diminish accordingly [1,52,56]. As shown in Figure 2c, the DMA baselines capture this stiffening, which arises fundamentally from suppressed chain mobility and locked network conformations. Resins with rigid backbones and high crosslink densities, such as cyanate esters containing triazine rings, display particularly large increases in modulus due to this freezing effect [53,56]. Distinct from this elastic stiffening is the evolution of failure mechanics. While hardness and yield strength increase at low temperatures, they do so to a lesser degree than stiffness, leading to a “harder but not necessarily stronger” response [52,56]. For example, the tensile strengths of neat epoxy and cyanate ester systems increase modestly down to 77 K, but then plateau or slightly decline as the temperature approaches 4–20 K [53]. This strength limitation reflects suppressed plasticity and heightened defect sensitivity. At 77 K, although the threshold stress for yielding is higher, the fracture process zone effectively disappears at lower temperatures, allowing even small microscopic flaws to trigger catastrophic failure rather than ductile yielding [51,52]. In summary, thermosets at cryogenic temperatures are characterised by a strong increase in modulus and hardness, but only marginal gains in strength and a severe loss of ductility, consistent with a glassy, flaw-sensitive failure mode.

The ductility and toughness of thermosets collapse at cryogenic temperatures. Failure strains drop to single-digit percentages, and the stress–strain response becomes brittle (often producing mirror-like fracture surfaces). Plane-strain fracture toughness (K_IC) and energy release rate fall markedly with cooling [51,56]. Experimentally, many low-temperature epoxies show K_IC values at 77 K that are only a fraction of those at room temperature, with fracture behaviour approaching linear-elasticity [52,56]. CE and BMI resins likewise exhibit a pronounced loss of toughness. Certain formulations containing flexible chain segments may display a modest increase in fracture toughness at 77 K, although values remain well below those at ambient temperature [14]. The overall effect of deep cooling is a substantial reduction in crack resistance, whereby crack initiation occurs more readily, and crack propagation proceeds more rapidly under a given stress intensity. For cryogenic designs (e.g., superconducting magnet insulation), this loss of matrix toughness is a critical limitation, necessitating very careful defect control and, potentially, the inclusion of toughening agents [1,50,52].

Thermosetting resins also undergo notable changes in thermal properties. The thermal conductivity of neat epoxy resins is relatively low, approximately 0.10 to 0.15 W·m⁻¹·K⁻¹ at room temperature, and typically decreases further upon cooling [50]. For example, measurements on a representative epoxy formulation indicate a value of about 0.3 W·m⁻¹·K⁻¹ at 77 K, with some systems exhibiting even lower values [50]. This low conductivity arises from phonon scattering within the amorphous structure. As temperature decreases, disorder-induced scattering persists, preventing any significant rise in thermal conductivity (κ). As illustrated in Figure 2a,b, the incorporation of thermally conductive fillers, such as spherical hexagonal boron nitride (h-BN) and boron nitride nanosheet (BNNS) hybrids, enhances κ by several fold and helps maintain higher values at 77 K [57,58,59,60]. The coefficients of thermal expansion (CTE) of thermosetting resins decrease sharply upon cooling. Linear CTE values, typically in the range of 30 to 70 × 10⁻⁶ K⁻¹ at room temperature, are often greatly reduced by 77 K [14]. Rigid molecular systems, such as high-glass-transition-temperature (T_g) phenolics and cyanate esters, exhibit smaller coefficients of thermal expansion α(T). By contrast, flexible epoxy resins show higher α(T) values at room temperature, which decrease substantially in the cryogenic regime [14,49]. As summarised in Figure 2d–f, CE-cured epoxies show the downward trend of α(T) derived via thermo-mechanical dilatometry (TMA) or Dynamic-mechanical measurements (DMA) protocols alongside the viscoelastic baselines. Taken together, thermal property trends can be summarised as: (i) κ remains very low unless enhanced by conductive fillers, and (ii) α(T) decreases steeply with cooling, with rigid resins showing the lowest values—both of which have direct implications for matrix–fibre mismatch and thermal stress development.

Figure 2. Thermal transport and thermo-mechanical baselines of epoxy systems with/without boron nitride (BN) fillers, and CE-cured epoxies. (a) Thermal conductivity of neat epoxy and BN-sphere-filled epoxy at 77 K and 298 K [57]; (b) thermal conductivity of neat epoxy and BN-sphere-filled composites (77–298 K) [57]; (c) DMA E′/E″/tan δ baseline [61]; (d–f) CTE trends from TMA/DMA for CE-cured epoxies (*:anisotropic) [61].

Figure 2. Thermal transport and thermo-mechanical baselines of epoxy systems with/without boron nitride (BN) fillers, and CE-cured epoxies. (a) Thermal conductivity of neat epoxy and BN-sphere-filled epoxy at 77 K and 298 K [57]; (b) thermal conductivity of neat epoxy and BN-sphere-filled composites (77–298 K) [57]; (c) DMA E′/E″/tan δ baseline [61]; (d–f) CTE trends from TMA/DMA for CE-cured epoxies (*:anisotropic) [61].

2.1.2. Thermoplastics (PEEK/PEI/PI/PTFE/UHMWPE)

High-performance thermoplastics, including polyetheretherketone (PEEK), polyimide (PI), polyetherimide (PEI), ultra-high-molecular-weight polyethylene (UHMWPE), and polytetrafluoroethylene (PTFE), exhibit similar trends: stiffness increases markedly at low temperatures, while ductility is greatly reduced. For semi-crystalline thermoplastics, the effect is particularly pronounced. For example, multiple studies have shown that the Young’s modulus of PEEK increases as the temperature decreases from room temperature to 77 K and remains high even at 20 K. In contrast, UHMWPE demonstrates a several-fold increase in modulus by 77 K, primarily due to the freezing of amorphous chains and enhanced load sharing by crystalline fibrils. Oriented crystals continue to carry load effectively under these cryogenic conditions [62,63,64]. By comparison, amorphous systems and PTFE-rich materials also stiffen at low temperatures, but the magnitude of the increase is comparatively smaller [65]. As illustrated in Figure 3a–c, representative stress–strain curves at room temperature (RT), 77 K, and 20 K for PTFE, PEEK, and UHMWPE show an upward shift in ultimate strength and a contraction of fracture strain as temperature decreases.

The tensile strengths of unreinforced thermoplastics generally increase as the temperature decreases to 77 K, but then plateau or exhibit a slight decline below approximately 50 K, resulting in a characteristic maximum near the liquid-nitrogen temperature (77 K) [64,66]. For instance, controlled RT/77/20 K tests across PTFE, PEEK, UHMWPE show strength gains from room temperature to 77 K and a small decrease from 77 to 20 K, accompanied by strong strain-rate coupling [62,64]. High-T_g rigid thermoplastics (PI, PEI) start with higher strengths and show relatively smaller gains at 77 K [5]. Overall, engineering thermoplastics reach maximum strength around 77 K and do not significantly strengthen beyond that. Meanwhile, ductility is sharply reduced. Room-temperature elongations of several tens of percent decrease to single-digit values at 77 K, with fracture surfaces transitioning to mirror-like or multifaceted morphologies. Correspondingly, fracture toughness decreases progressively with lowering temperature [50,55,57,66]. These changes occur because molecular-level relaxation and crazing mechanisms are arrested, causing the polymer to fail in an almost purely elastic manner once the yield point is reached. Charpy and Izod impact strengths exhibit similar sharp reductions near and below 77 K [5]. Consistent with Figure 3, these behaviours reflect the suppression of molecular relaxation and crazing with decreasing temperature.

Thermoplastic matrices also exhibit reduced thermal conductivity and coefficients of thermal expansion (CTE) at cryogenic temperatures. Phonon transport is constrained by amorphous disorder, so thermal conductivity typically falls to a fraction of its room-temperature value, although increased crystallinity and molecular orientation can moderate this reduction. The incorporation of high-conductivity fillers, such as boron nitride and graphene, has been shown to enhance the low-temperature thermal conductivity of thermoplastic composites [27,65]. The linear CTE of thermoplastics, which is relatively high at room temperature (10 to 200 × 10⁻⁶ K⁻¹), decreases markedly upon cooling. Comprehensive measurements on PTFE, PEEK, and UHMWPE at room temperature, 77 K, and 20 K confirm a substantial reduction in α(T), tending towards a low plateau below 50 K [64,66]. This reduction in CTE partly alleviates thermally induced stresses in cryogenic structures; however, residual shrinkage and cure-induced stresses must still be carefully managed in the cooled matrix [1]. Compared with thermosets, thermoplastics generally retain slightly better toughness at cryogenic temperatures and exhibit lower CTE values, which help mitigate matrix–fibre mismatch. Nonetheless, their higher crystallinity sensitivity, processing-induced anisotropy, and interfacial adhesion challenges require careful control during manufacturing and structural integration.

Figure 3. Temperature-dependent tensile responses and thermal contraction of thermoplastics. (a) PTFE; (b) PEEK; (c) UHMWPE, with representative stress–strain curves at RT, 77 K, and 20 K under various crosshead speeds [67].

Figure 3. Temperature-dependent tensile responses and thermal contraction of thermoplastics. (a) PTFE; (b) PEEK; (c) UHMWPE, with representative stress–strain curves at RT, 77 K, and 20 K under various crosshead speeds [67].

Table 2. Intrinsic properties of matrices at RT/77 K/20 K.

Material	T (K)	$\mathit{E}$ (GPa)	$\mathit{\sigma }$ (MPa)	$\mathit{\kappa }$ (Wm−1K−1)	$\mathit{\alpha }\left(\mathit{T}\right)$ (10−6 K−1)	Refs.
Epoxy	300	2–4	60–90	0.2–0.3	40–70	[1,5,50]
	77	7–8	90	0.10–0.20	10–40
	20	8	90	0.03–0.10	5–20
CE	300	2.5–3.5	70–100	0.18–0.25	30	[1,61]
	77	4–8	100	0.06–0.12	15
	20	8	101	0.02–0.05	10–12
BMI	300	3.5	60	0.19–0.25	40–60	[1]
	77	6–7	70–80	/	20
	20	7	70–80	/	15
PEEK	300	3.7	100	0.29–0.32	47	[62,63,64]
	77	5.5–6.0	120	0.15–0.25	23.47
	20	6.0	110	0.12–0.20	6
PEI	300	2.8–3.3	110	0.22	55	[5,64]
	77	5.5–6.0	125	0.15–0.25	30
	20	6.0	120–130	0.10–0.20	20
PTFE	300	0.4–0.8	20–30	0.25–0.35	100–130	[64,65]
	77	0.8–1.0	20–35	0.2–0.3	30–60
	20	1.0	35	0.14	70
UHMWPE	300	0.8–1.5	20–40	0.40–0.5	100–200	[64]
	77	1.5–3.0	30–60	0.30–0.45	30–80
	20	2.5–3.8	45	0.25–0.40	10–40

summarises the intrinsic properties of common thermoset/thermoplastic matrices at room temperature/77 K/20 K, collating modulus E (GPa), tensile strength σ (MPa), thermal conductivity κ (W·m⁻¹·K⁻¹), and linear CTE α (×10⁻⁶ K⁻¹) with literature ranges and references. It serves as a quick-reference baseline for selecting resin systems and for estimating thermal-stress build-up during cooldown.

2.2. Effects of Cryogenic Temperatures on Reinforcement (Fibre and Nano) Properties

2.2.1. Continuous Fibres (Carbon, Glass, Aramid/Kevlar^®)

In deep cryogenic environments (around 77 K and below), the intrinsic properties of reinforcement materials remain overall quite stable, and their temperature sensitivity is significantly lower than that of polymer matrices. For continuous fibres, the axial stiffness of carbon fibres remains essentially unchanged or shows a slight increase as the temperature decreases from room temperature to liquid-nitrogen temperature (77 K). Glass fibres exhibit a modest increase in modulus or approach a plateau upon cooling, while aramid (Kevlar^®) fibres likewise display minimal changes in axial stiffness from room temperature to deep cryogenic conditions [52,66,68,69]. In terms of typical tensile strength, carbon fibres at room temperature have common average values on the order of several GPa, and at 77 K, they remain at similar levels, with only a significantly increased statistical scatter. The strength of glass fibres is generally insensitive to temperature, remaining in the range of 4–5 GPa. Aramid fibre strength has been reported to increase slightly when cooled to deep cryogenic temperatures (for example, from about 3.6 GPa to nearly 4 GPa), but the extent of improvement depends on the specific fibre grade and test protocol [68,69]. As shown in Figure 2 and Figure 3, continuous-fibre systems exhibit nearly constant axial stiffness together with pronounced anisotropy in thermal expansion and thermal conductivity at low temperatures.

It must be emphasised that all three types of continuous fibres already predominantly fail in a brittle manner at room temperature (extremely low elongation at break, no obvious yield region), and their ductility further decreases upon cooling, with almost no plastic energy dissipation occurring. This is not due to any weakening of bond energy or lattice strength at low temperature, but rather because reduced thermal activation diminishes microscopic relaxation via dislocations, chain segments, or disorder, amplifying the role of micro-defects in fracture. In the presence of cracks or flaws, the crack tip cannot be blunted by plasticity and thus fails suddenly in an almost linear-elastic manner, causing fracture toughness around 20 K to decrease further. Nominal strength becomes more dependent on flaw size and distribution, and strength scatter increases markedly at low temperature [29,70,71].

In terms of thermal properties, the three major types of continuous fibres exhibit distinctive yet predictable behaviours in thermal expansion and conductivity at deep cryogenic temperatures. Carbon fibres, owing to their high orientation and graphitized microcrystalline structure, possess an axial coefficient of thermal expansion (CTE) that is near zero or slightly negative at room temperature and remains near zero or negative at 77 K and below. By contrast, their transverse (radial) thermal expansion is positive and tends to converge upon cooling [53,72]. Glass fibres display positive and moderate axial thermal expansion at room temperature, which becomes significantly smaller upon cooling, indicating milder and more controllable dimensional change at low temperatures [73,74]. Aramid fibres (Kevlar^®), characterised by strongly oriented molecular chains and anisotropic hydrogen-bonded networks, exhibit extremely low or slightly negative axial CTE at room temperature, approaching zero in the 20–77 K range, while transverse expansion remains positive. This “near-zero or negative axial expansion” favours dimensional stability in slender cryogenic ties and straps [66,69]. As shown in Figure 2 and Figure 3a, a T700 tow demonstrates negative axial CTE with a narrow hysteresis between −150 °C and 150 °C, indicative of low contraction and high dimensional stability down to 77 K.

Regarding thermal conductivity, the axial heat conduction of carbon fibres is strongly influenced by molecular orientation and defect content. With decreasing temperature, the suppression of Umklapp scattering enhances axial thermal conductivity, which then approaches a low-temperature plateau governed by boundary and defect scattering. In contrast, the radial thermal conductivity remains very low and only weakly temperature-dependent on the order of about 0.5 to nearly 1 W·m⁻¹·K⁻¹ between 295 K and 77 K, leading to even stronger anisotropy at cryogenic temperatures [54,55]. As shown in Figure 2 and Figure 3, the axial channel exhibits a stronger temperature dependence than the radial channel, underscoring the anisotropy that directly informs thermal-diffusion and thermo-stress analyses under cryogenic conditions. Notably, the combination of anisotropic thermal expansion and low-temperature brittleness scatter in continuous fibres can amplify thermal mismatch and reliability issues in composite systems; thus, ply orientation, pre-stress, and fibre grade selection must be carefully optimised [29,70,71].

In recent years, cryogenic property databases and metrology have been significantly improved. Reviews and handbooks now consolidate thermo-mechanical–electrical data from ambient to cryogenic temperatures for use as design baselines. Consistent datasets confirm the near-zero axial CTE and low thermal conductivity of Kevlar^® 49 across the range of 0–300 K [1,5,66,69]. High-precision thermal characterisation at single-fibre and tow scales, using methods such as thermal-bridge, transient electro-thermal (TET), and time-domain thermoreflectance (TDTR), has quantified the anisotropy between axial and radial thermal conduction. Ultimately, these intrinsic constituent features—specifically the thermodynamic anisotropy and the suppression of plastic deformation—act as the physical precursors to composite-scale failure modes, such as interfacial debonding and hierarchical stress gradients, which are discussed in detail in Section 3. For example, T700 tows exhibit negative axial thermal expansion with narrow hysteresis between −150 °C and 150 °C, and the application of modest pre-load further decreases the apparent CTE, highlighting an “orientation-defect-stress” co-tuning strategy for enhancing dimensional stability at low temperatures [53]. Parallel studies on strength scatter and failure mechanisms emphasise that increasing molecular orientation and crystallinity while reducing surface and bulk flaws is critical to stabilising low-temperature strength. These insights extend to interfacial strength, thermal mismatch, and cyclic durability in laminates and sandwich structures used for hydrogen storage, superconducting magnets, and space cryogenic applications [29,71]. In summary, although the intrinsic stiffness and strength of continuous fibres remain stable at cryogenic temperatures, their brittle failure behaviour and amplified strength scatter make defect sensitivity, interfacial design, ply orientation strategies, and flaw control increasingly critical for ensuring reliability in demanding applications such as cryogenic fuel tanks, space structures, and superconducting magnet systems.

2.2.2. Nano-Reinforcements (CNT/Graphene/h-BN)

The intrinsic stiffness and strength of nano-reinforcements, including carbon nanotubes (CNTs), graphene, and hexagonal boron nitride (h-BN), remain extremely high at cryogenic temperatures, approaching their lattice limits with moduli near 1000 GPa and strengths of 100 to 130 GPa for graphene and tens to over 100 GPa for carbon nanotubes. These exceptional properties originate from the strong covalent bonding within the hexagonal lattice structure and are not diminished at deep cryogenic temperatures. However, macroscopic ductility remains negligible, and defect sensitivity increases as thermal activation is suppressed [70,71,75]. Graphene exhibits a negative in-plane CTE over a wide temperature range, with the magnitude decreasing toward zero as the temperature approaches the low-temperature limit [73,74,76]. Carbon nanotubes (CNTs) display near-zero axial CTE at room temperature that remains essentially constant between 20 and 77 K, enabling their use as highly dimensionally stable nano-rods [77,78]. Few-layer hexagonal boron nitride (h-BN) shows an extremely small in-plane CTE, reported as near zero or weakly negative in thin films at 300–400 K, and trends toward minimal expansion at lower temperatures, making it attractive for thermal matching in cryogenic assemblies [79,80,81,82]. As shown in Figure 4, Raman-derived α(T) for monolayer graphene is negative between 200 and 400 K and approaches zero as the temperature decreases.

Carbon nanotubes, graphene, and hexagonal boron nitride maintain very high intrinsic thermal conductivity at deep cryogenic temperatures, exhibiting a characteristic “increase followed by plateau or turnover” as the temperature decreases. The suppression of Umklapp scattering enhances thermal conductivity during cooling, whereas boundary, defect, and size effects impose strong limitations at very low temperatures. Classic measurements on single-walled carbon nanotube (SWCNT) ropes across 8–350 K show that thermal conductivity decreases with decreasing temperature and becomes linear below 30 K, reflecting phonon-dominated, boundary-limited transport [83]. Suspended few-layer graphene measured using a micro-bridge technique between 77 and 350 K exhibits higher in-plane thermal conductivity than supported counterparts and displays pronounced size dependence [84,85]. Similarly, suspended few-layer hexagonal boron nitride (h-BN) measured by a micro-bridge demonstrates high in-plane thermal conductivity (250 to 360 W·m⁻¹·K⁻¹ at 300 K) and a low-temperature increase with a kink near 60–80 K, consistent with reduced Umklapp scattering and boundary-limited transport [80,81,82]. As shown in Figure 5, these temperature-dependent signatures reflect the combined influence of reduced Umklapp scattering and boundary- or defect-induced limitations in one- and two-dimensional lattices. Importantly, while CNTs, graphene, and h-BN retain outstanding strength, dimensional stability, and thermal transport capacity at cryogenic temperatures, their effectiveness in composites is strongly governed by dispersion quality, interfacial bonding, and scale effects. These engineering factors often outweigh the intrinsic properties in determining the actual cryogenic performance of nano-reinforced composites.

Table 3. Intrinsic properties of reinforcement materials at RT/77 K/20 K (axial or in-plane).

Material	T (K)	$\mathit{E}$ (GPa)	$\mathit{\sigma }$ (GPa)	$\mathit{\alpha }\left(\mathit{T}\right)$ (10−6 K−1)	Refs.
Carbon fibre (T700)	300	230–240	4.9–5	−0.4–−0.38	[52,53,72]
	77	230–245	4.9–5.1	−0.3
Abo	20	230–245	4.8–5	−0.1–0
Carbon fibre (S-2)	300	86	4.7–4.8	2.8–3.0	[68]
	77	90		2.8
	20	92		2
Aramid (Kevlar®49)	300	113–130	3.6–4.0	−4.9–−2.9	[77,78]
	77		3.8–4.1	About 0
	20		3.7–4.0	About 0
CNT (SWCNT)	RT	1000	3.6		[73,76]
	77			−1–1
	20
Graphene (mono-/few-layer)	300	1000	100–130	−8.0	[73,76]
	77			−6–−2
	20			−1–0
h-BN (few-layer, suspended)	300	860	70	−1–1	[79,80,81,82]
	77			−1–0
	20			−0.5–0

describes the elastic modulus E, tensile strength σ, and coefficient of thermal expansion (CTE) α(T) of major reinforcements (carbon, glass, aramid/Kevlar^®, CNTs, graphene, h-BN) at room temperature (RT), 77 K, and 20 K.

3. Mechanical Performance at Low Temperatures

3.1. Thermal Contraction

Thermal expansion mismatch between fibres and the matrix is a primary driver of cryogenic behaviour in fibre-reinforced composites. At the phase level, constituent materials display divergent intrinsic responses: in the longitudinal direction, carbon and aramid fibres generally exhibit near-zero or even negative CTE, whereas glass and basalt fibres contract upon [86]. In contrast, the matrix undergoes significant positive contraction. Quantitatively, the matrix CTE typically remains one order of magnitude greater than that of the fibres [87], and this drastic mismatch generates strain incompatibilities that frequently exceed the reduced strain-to-failure threshold of the embrittled matrix. At the laminate level, these incompatibilities translate into complex residual stress states. In carbon-fibre systems, the significant matrix contraction combined with slight fibre transverse expansion produces interfacial compression, which can temporarily enhance fibre-matrix adhesion and increase tensile strength [88]. Conversely, in glass- and aramid-fibre composites, the simultaneous transverse contraction of both phases promotes tensile interfacial stresses, lowering the threshold for microcrack initiation [13,89].

At the molecular scale, carbon fibres themselves display hierarchical CTE behaviour, with a “core” region of negative axial CTE and an outer disordered shell of positive CTE. The contraction of the shell against the expanding core promotes surface cracking at cryogenic temperatures, thereby improving mechanical interlocking and interfacial bonding [7]. Although matrix CTE decreases with temperature, it remains one order of magnitude greater than that of fibres, making the matrix the dominant source of contraction [13]. Processing conditions and laminate architecture further modulate cryogenic CTE: lower cure temperatures reduce crosslink density and increase transverse CTE [87], while woven plies impose in-plane constraints on matrix contraction, leading to anomalously high out-of-plane expansion [25]. At the structural scale, near-zero CTE can be achieved by hybridising carbon and glass fibres, introducing off-axis plies, or optimising braid angle in textile composites [90,91,92].

These behaviours are illustrated in Figure 6, which compiles representative experimental data. Figure 6a illustrates the thermal expansion of several epoxies compared with that of a superconducting coil [26]. The matrices display sharp CTE increases above 200 K, while their values remain relatively low at cryogenic temperatures, consistent with suppressed chain mobility. The comparison highlights the dominant role of matrix contraction in the overall composite response. Figure 6b–d demonstrates the influence of creep loading on matrix CTE [7]. Untreated specimens follow a typical monotonic trend, whereas creep-conditioned samples exhibit systematically reduced or elevated CTE depending on loading direction, indicating that mechanical history modifies molecular packing and contraction behaviour. Collectively, Figure 6 highlights that fibre type, matrix chemistry, and thermo-mechanical history jointly govern the anisotropic and non-linear contraction of fibre-reinforced composites under cryogenic conditions.

In contrast to fibre reinforcement, which primarily governs anisotropy, nanoparticle modification provides a means of tuning the thermal contraction of composites by directly altering the matrix. Nanoparticles such as ZrW₂O₈ with negative CTE, or AlN and cupric oxide (CuO) with intrinsically low thermal expansion, have been shown to reduce the effective CTE of epoxies at cryogenic temperatures [9,20,93]. Their stiffening effect constrains matrix contraction, while their high thermal conductivity minimises thermal gradients and alleviates stress localisation during cooldown [15]. Functionalised nanofillers, including magnetite (Fe₃O₄)/GO hybrids, further enhance interfacial adhesion, redirect microcrack paths, and mitigate damage accumulation under thermal cycling [94].

As illustrated in Figure 7, these mechanisms are evident in experimental observations. Figure 7a shows that incorporation of AlN nanoparticles produces a monotonic decrease in epoxy CTE, reaching values below 28*10⁻⁶ K⁻¹ at 30 wt%. Figure 7b illustrates that ZrW₂O₈ reinforcement induces a pronounced reduction in matrix expansion, consistent with micromechanical predictions and confirming the efficiency of negative-thermal-expansion ceramics. Collectively, Figure 7 highlights how nanoparticle incorporation stabilises composite performance under extreme temperature variations, yielding reduced nominal CTE values and improved dimensional stability after repeated cryogenic exposure.

Ultimately, the core contradiction governing the thermal contraction of fibre-reinforced composites at cryogenic temperatures lies in the large disparity in the coefficients of thermal expansion between the matrix and the fibres. This intrinsic mismatch dictates residual stress evolution and interfacial stability throughout cooling cycles. By contrast, nanoparticle incorporation offers an effective pathway to tailor matrix contraction behaviour, enabling the fine-tuning of composite CTE and mitigating mismatch-induced damage. These findings underline the importance of integrated material design strategies that synergistically balance fibre anisotropy and matrix modification to achieve dimensional stability and structural reliability under extreme cryogenic environments.

3.2. Matrix Embrittlement

The embrittlement of the matrix at cryogenic temperatures primarily arises from intrinsic changes in molecular mobility, free volume, and chain morphology. Cooling below the glass transition temperature suppresses cooperative segmental motion and viscoelastic relaxation [17]. This transition into the glassy state is accompanied by a marked increase in storage modulus and tensile strength, while fracture strain and fracture energy decrease, signifying a ductile-to-brittle transition [17,95,96,97]. Such behaviour has been consistently observed in both thermoplastic and thermoset matrix systems, where cryogenic exposure enhances stiffness but sharply reduces elongation at break and impact resistance [17,96,98].

For semi-crystalline polymers, cryogenic conditions also induce morphological changes that promote brittle fracture. Reduced molecular mobility favours secondary crystallisation and densification of lamellae, yielding higher crystallinity but fewer tie molecules bridging crystalline domains [98,99]. This reduces lamellar spacing and plastic deformation capacity, thereby increasing brittleness. Studies on PA6, PA66, polyethylene, and polypropylene confirm that enhanced crystallinity at cryogenic temperatures correlates strongly with diminished extensibility and toughness [30,98,99,100].

At the macroscopic level, these mechanisms manifest as higher stiffness but lower fracture resistance. Fracture surfaces at room temperature typically display ductile features such as fibrillation, shear yielding, or microvoid coalescence. In contrast, under cryogenic conditions, they transform into smooth, mirror-like cleavage planes, characteristic of brittle fracture [17,96,101]. Mechanical testing consistently reveals rising modulus and tensile strength with decreasing temperature, accompanied by pronounced reductions in fracture strain, fracture toughness, and impact energy absorption [17,96,98].

This transition is clearly illustrated in fracture morphologies comparing cryogenic and room-temperature failures (Figure 8). Figure 8 shows that in epoxy- and PEEK-based systems, cryogenic fracture surfaces are significantly smoother and flatter than those at ambient conditions [15,93,94]. The absence of ductile deformation features directly reflects the suppression of energy-dissipating mechanisms in the matrix. This microstructural evidence reinforces the mechanical data: although stiffness increases at low temperature, the ability to absorb energy through plastic deformation is severely curtailed, leading to catastrophic brittle failure [15,17,93,94,96].

Furthermore, quantitative fracture energy data confirm that cryogenic exposure substantially reduces toughness relative to room temperature. Figure 9 illustrates that epoxy matrix systems experience a pronounced decrease in fracture toughness at −196 °C, with the trend consistently observed across multiple studies [102,103,104]. While the incorporation of nanoparticles such as polydopamine (PDA), block copolymers, or CuO nanorods can partially mitigate this loss, their effect eventually saturates. The dominant feature across all studies remains the intrinsic reduction in fracture toughness in the matrix under cryogenic conditions.

To quantify embrittlement effects, several performance metrics have been proposed. The most direct are fracture strain (ε_b), fracture toughness (G_IC), and impact resistance, all of which decrease markedly at cryogenic temperatures [17,96]. DMA further reveals a monotonic increase in E′ with decreasing temperature [97]. A combined descriptor, the brittleness index, defined as B = 1/(εb·E′), increases systematically with decreasing temperature and provides an effective measure of the brittle response of the matrix under cryogenic conditions [18]. Overall, matrix embrittlement represents the fundamental bottleneck governing the cryogenic failure of fibre-reinforced composites. Although interfacial toughening or nanoparticle modification can partially alleviate this issue, the root cause lies in the intrinsic molecular architecture of the matrix. Therefore, achieving reliable cryogenic performance ultimately requires innovations at both the molecular design and structural scales to develop matrices with inherent toughness and reduced temperature sensitivity.

3.3. Interfacial Strength

Interfacial performance is a critical factor governing the interlaminar load-bearing capacity and overall reliability of fibre-reinforced composites, and exhibits pronounced differences under cryogenic environments compared with room temperature. Numerous studies report that the ILSS of composites at cryogenic temperatures or lower is often higher than at room temperature. For example, microbond tests show that the ILSS of T700/modified epoxy increases by ~16% at −196 °C relative to room temperature [108], while unidirectional laminates exhibit an enhancement of nearly 70% in ILSS at 90 K [109]. However, under prolonged thermal cycling (e.g., 100–1000 cycles between −213 °C and 393 °C), this initial benefit may transition to interfacial degradation due to accumulated damage. Experimental data indicate that residual tensile and in-plane shear moduli decrease rapidly within the first 500 cycles, plateauing thereafter as microcrack saturation occurs. For instance, after 1000 cycles, residual tensile modulus reductions range from 2.5% to 7.5%, while in-plane shear modulus reductions vary from 4.8% to 26.6%, depending on the temperature span and upper limit [110]. This strengthening primarily arises from the mismatch in CTE between fibres and the epoxy matrix: during cooling, the matrix contracts and generates radial compressive stresses at the interface, producing a “clamping effect” that enhances interfacial friction and mechanical interlocking [108,111,112]. Reduced molecular mobility of epoxy chains at low temperature further stiffens the matrix, suppressing plastic yielding near the interface and thereby promoting more efficient stress transfer.

However, cryogenic exposure does not uniformly improve interfacial behaviour. Rapid cooling or repeated cryogenic-ambient cycles can introduce residual tensile stresses and microcracks, leading to partial debonding and strength degradation. Quantitatively, residual thermal stresses have been observed to increase by approximately 25–37% relative to room temperature baselines [46], eventually exceeding the local matrix strength. Consequently, a transition point is typically observed between 10 and 50 cycles (depending on resin toughness), where the accumulation of microcracks overrides the beneficial clamping pressure, driving a net reduction in interfacial shear strength. Short-term cryogenic shocks often deteriorate interfacial properties [105], whereas long-term exposure under stable cryogenic conditions may recover or even enhance strength due to sustained interfacial clamping [37,105]. Thus, interfacial behaviour at low temperature depends strongly on thermal history and cycling [37,106].

Systematic studies also demonstrate a strong dependence of ILSS on lay-up configuration under cryogenic conditions. Figure 10a–c show that unidirectional laminates exhibit a marked increase in ILSS with decreasing temperature, with values at 90 K being 67.5% higher than those at room temperature. In contrast, cross-ply and angle-ply laminates display only minor changes, indicating that ply orientation governs the extent of low-temperature strengthening [109]. The beneficial interfacial clamping effect is therefore most pronounced in unidirectional systems, while transverse or angled plies introduce competing stress fields that suppress the magnitude of improvement.

To mitigate cryogenic embrittlement and cyclic damage, the incorporation of nanofillers has emerged as an effective strategy for enhancing interfacial performance. GO, carboxylated graphene (G-COOH), functionalised carbon nanotubes (FCNTs), nano-silica (SiO₂), and graphene nanoplatelets (GNPs) have all shown significant toughening effects [38,107,111]. These nanofillers improve the interface through multiple mechanisms, crucially governed by functionalisation stability and geometric structure dependence.

Regarding functionalisation, surface groups such as carboxyl (-COOH), hydroxyl (-OH), and amino (-NH₂) form covalent bonds or strong hydrogen bonding networks with the epoxy matrix. Although chemical reactivity is suppressed at cryogenic temperatures, these pre-formed bonds remain highly stable. In particular, non-covalent interactions like hydrogen bonding often strengthen as thermal fluctuations diminish, thereby enhancing the static stability of the interface at low temperatures [40,113].

Furthermore, the reinforcement efficiency exhibits a clear structure dependence. One-dimensional (1D) nanofillers like CNTs primarily bridge microcracks and offer pull-out resistance, but their high aspect ratio makes them prone to agglomeration and viscosity increases. In contrast, two-dimensional (2D) sheets like GO and graphene provide a larger specific surface area for mechanical interlocking and can effectively deflect crack paths in multiple planes, acting as “shields” against crack propagation. However, their 2D geometry requires precise orientation control; randomly oriented sheets may act as defects, whereas aligned sheets maximise in-plane properties [31,114]. Their low CTEs and high thermal conductivity help to alleviate thermal mismatch stresses at the interface [107].

Figure 10. Temperature dependence of interlaminar shear strength (ILSS) for CFRP laminates with different lay-up configurations [115]. (a) Unidirectional; (b) cross-ply; and (c) angle-ply laminates tested from 90 K to 353 K. (d) Effect of GO content on ILSS at room temperature and 77 K [31].

Figure 10. Temperature dependence of interlaminar shear strength (ILSS) for CFRP laminates with different lay-up configurations [115]. (a) Unidirectional; (b) cross-ply; and (c) angle-ply laminates tested from 90 K to 353 K. (d) Effect of GO content on ILSS at room temperature and 77 K [31].

The role of GO addition has been quantitatively clarified. Figure 11d illustrates that the ILSS of both CFRP and GFRP composites increases with GO content up to an optimum (~0.2–0.3 wt%), beyond which aggregation reduces strength. At 77 K, the enhancement is more pronounced than at room temperature, confirming that GO not only strengthens fibre-matrix bonding but also synergises with the cryogenic clamping effect to deliver superior interfacial performance [39]. These findings emphasise that optimised GO loading can simultaneously maximise adhesion and minimise aggregation-induced defects, thereby improving interfacial reliability across both cryogenic and ambient conditions.

The microstructural evidence reinforces these macroscopic observations. Figure 11 shows SEM images of fracture surfaces that reveal clear differences in fibre-matrix adhesion with and without GO addition. At room temperature, neat epoxy composites exhibit smooth fibre pull-out surfaces, indicative of weak bonding. With 0.3 wt% GO, the fibres are more firmly embedded in the matrix and display increased surface roughness. At 77 K, the interface becomes even stronger, as evidenced by fewer pulled-out fibres and larger amounts of matrix adhering to the fibre surfaces. These features reflect the combined effects of GO functional groups, which form chemical bonds, and the wrinkled morphology, which enhances mechanical interlocking. Together, these mechanisms produce a robust fibre-matrix interface that is particularly effective under cryogenic conditions [39].

Beyond GO, comparative studies of different carbon-based nanofillers further clarify the balance between matrix- and interface-dominated effects. Figure 12 illustrates that CFRP laminates containing randomly dispersed nanofillers show limited or inconsistent ILSS improvements, particularly for CNTs, owing to agglomeration and the resulting embrittlement of the interlaminar region. In contrast, when CNTs are aligned through the thickness or chemically grafted onto the fibre surface, they deliver substantial gains in ILSS at both room and cryogenic temperatures. Alignment facilitates bridging across prospective delamination planes, while surface functionalisation improves dispersion and establishes chemical coupling. Electrophoretic deposition studies further demonstrate a clear concentration optimum, with ILSS peaking at intermediate CNT loadings but declining at higher levels due to aggregation.

The differences reported by Zotti et al. [114] highlight this duality. Their study compared CFRP systems incorporating GNPs, carbon nanofibres (CNFs), and CNTs under both ambient and cryogenic conditions. Some datasets showed negligible or even negative changes in ILSS for CNT-modified laminates, which were attributed to clustering, poor load transfer, and the tendency of CNTs to embrittle resin-rich interlaminar zones. Other datasets demonstrated clear ILSS improvements with CNTs, particularly when functionalisation and improved dispersion methods were employed. The apparent contradiction was explained as a competition between matrix embrittlement effects (caused by poorly dispersed nanofillers) and interfacial reinforcement effects (arising when nanofillers are well distributed, functionalised, or positioned at the fibre-matrix interface). A further series of tests confirmed the critical role of concentration: ILSS increased up to an optimal nanofiller content but declined at higher loadings due to aggregation. Collectively, these findings confirm that interfacial reinforcement dominates the overall response when dispersion and functionalisation are well controlled, explaining why ILSS trends initially appear inconsistent but converge once experimental design is accounted for.

Importantly, nanofiller modification not only strengthens the interface but also alters failure mechanisms across different thermal conditions. For unmodified composites, room-temperature failures are dominated by interfacial debonding and matrix shear yielding [106,116], whereas cryogenic failures are characterised by brittle delamination and rapid interfacial cracking [108,109]. With nanofiller incorporation, room-temperature failures tend to shift towards cohesive failure within the matrix adjacent to the interface, indicating stronger interfacial adhesion [39,115]. Under cryogenic conditions, cracks are more likely to deflect into the matrix or be bridged by nanofillers, with reduced fibre pull-out lengths and suppressed interfacial debonding [105,107,113]. This transformation from interfacial-dominated brittle fracture to matrix-dominated, energy-dissipating mechanisms highlights the superior resistance to delamination and cracking achieved in nanomodified composites under cryogenic environments.

A comparative summary of ILSS values for different fibre–nanoparticle reinforced resin systems at room and cryogenic temperatures is provided in

Table 4. Summary of ILSS values for different fibre–nanoparticle reinforced resin systems at different temperatures.

Resin	Fibre	ILSS at RT (MPa)	ILSS at CT (MPa)	Reference
Epoxy	Glass	27	20	[117]
Epoxy	Glass-Carbon hybrid	38	32	[117]
SiO2-modified epoxy	Carbon	97	158	[106]
Epoxy	Carbon	50	36	[108]
Modified epoxy	Carbon	32	36	[37]
GO-modified epoxy	Carbon	64	82	[39]
GO-modified epoxy	Carbon	65	110	[31]
Carboxylated graphene-modified epoxy	Carbon	32	37	[113]
FCNT-modified epoxy	Carbon	38	48	[112]
GNP -modified epoxy	Carbon	75	64	[109]
CNF -modified epoxy	Carbon	70	60	[109]
CNT-modified epoxy	Carbon	69	66	[109]

, which consolidates results across multiple studies and highlights the influence of fibre type, resin chemistry, and nanofiller modification.

Overall, interfacial enhancement under cryogenic conditions can markedly improve ILSS and mitigate delamination; however, its long-term reliability remains strongly dependent on thermal cycling history and the uniform dispersion of nanofillers. Future efforts should therefore focus on developing standardised testing protocols and scalable surface-modification strategies to ensure consistent and reproducible interfacial performance across diverse cryogenic applications.

4. Nano-Additives and Hybrid Architectures

4.1. Nano-Reinforcements

The strategic incorporation of nano-additives demonstrates significant potential for enhancing the cryogenic mechanical performance of fibre-reinforced composites, addressing critical challenges in extreme-temperature applications [118]. Multi-walled carbon nanotubes (MWCNTs) exhibit exceptional efficacy owing to their capacity for interfacial reinforcement through crack bridging and pinning mechanisms, which optimise stress transfer across fibre-matrix interfaces under cryogenic conditions. For instance, aligned carboxyl-functionalized MWCNTs (0.1 wt%) in epoxy matrices elevate flexural strength by 41% at −196 °C compared to unmodified systems [109]. This enhancement arises from synergistic mechanisms: matrix contraction induces radial compressive stresses that enhance interfacial adhesion, while MWCNTs suppress crack propagation through bridging effects. SEM validates these phenomena (Figure 13a,b), revealing reduced fibre debonding and distinct MWCNT pull-out traces at fracture surfaces [22,37], morphological evidence corroborated by prior cryogenic studies [119]. However, dispersion limitations impede scalability; agglomeration typically begins at low loadings, with viscosity increasing exponentially beyond approximately 2 wt% due to poor dispersion [120]. The dispersion limit (DL) for MWCNTs in polymer matrices is often below 2 wt% without functionalization, as higher loadings lead to significant viscosity rise and agglomeration. Specifically, agglomeration beyond 0.5 wt% increases viscosity by more than 47% [41], inducing stress concentration sites that compromise low-temperature impact resistance [22], a critical constraint for cryogenic structural applications.

While carbon-based nano-reinforcements excel in interfacial enhancement, SiO₂ addresses critical matrix-dominated failure modes by regulating thermal stress and suppressing cryogenic embrittlement. The incorporation of 1.5 wt% SiO₂ reduces the CTE of the epoxy matrix by 57.6% at 77 K, thereby alleviating interfacial thermal stresses resulting from cryogenic contraction. This stress mitigation serves as a key mechanism for suppressing microcrack initiation during thermal cycling [10]. The toughening effect synergistically combines plastic void growth and crack tip blunting: uniformly dispersed nanoparticles (~34 nm) trigger localised shear bands that dissipate fracture energy through microcrack bifurcation. SEM analysis validates this energy-absorption pathway (Figure 13c–f), showing river-like crack patterns transitioning to branched pathways around SiO₂ agglomerates [40], consistent with cryogenic toughening mechanisms reported for silica-epoxy systems [121].

Optimal performance requires precise process control: below 77 K, composites undergo a brittle-to-ductile transition reversal due to molecular chain freezing; at 4 K, “cleavage-style” fracture dominates, reducing fracture strain by 82% relative to room temperature [10]. This embrittlement aligns with cryogenic matrix behaviour described by Cotae et al. [122], underscoring the role of SiO₂ in delaying low-temperature failure mechanisms.

Table 5. Cryogenic performance of nano-additives in polymer composites.

Additive Type (Concentration)	Matrix	Temp. (K)	Key Performance Change	Key Mechanism
Vertically aligned MWCNTs (0.5 wt%)	Epoxy	77	IFSS ↑62.32%	Covalent bonding, fibre-matrix bridging [109]
MXene nanosheets (0.1 wt%)	Epoxy/CFRP	90	Crack propagation ↓3.2×; Peel stress ↓40%	Orthotropic alignment, CTE reduction [21]
GO nanosheets (0.25 wt%)	Epoxy/CFRP	77	ILSS ↑15.63% (RT); Recovery after cryo-cycling	Oxygen functional groups, hydrogen bonding [37]
G-COOH functionalized GN (1.5 g/L)	Epoxy/CFRP	77	ILSS ↑20.78%; FS ↑5.35%	Negative CTE, interfacial clamping [112]
CuO nanorods (1.5 wt%)	Epoxy	77	Tensile strength ↑44.0%	Negative thermal expansion (NTE) [10]
EFPS-grafted SiO2 (5 wt%)	Epoxy	90	Fracture toughness ↑48.82% (RT); TS ↑17.07%	Flexible chains, SiO2 debonding energy dissipation [123]

summarises key cryogenic performance metrics and mechanistic roles of nano-additives in polymer composites, including strength/toughness enhancements and interfacial reinforcement mechanisms across critical studies.

Micro-carbon additives (5–50 µm) serve as specialised impact modifiers rather than primary reinforcements in cryogenic composites, distinct from nanoscale counterparts like CNTs or nano-silica. While CNTs enhance interfacial shear strength through covalent bonding [109], micro-carbon particles preferentially absorb impact energy via microcrack deflection and particle pull-out mechanisms. At 77 K, epoxy composites with 15 vol% micro-carbon exhibit a 20.78% increase in impact strength compared to unmodified systems [112], attributed to superior crack-path tortuosity under dynamic loading, a phenomenon amplified by cryogenic matrix embrittlement [119]. However, synergistic conflicts arise in hybrid systems: GO composites containing micro-carbon show 12.3% lower flexural strength at 77 K than GO-only counterparts [40], due to competitive stress transfer pathways that diminish nanoscale reinforcement efficiency. This trade-off is exacerbated by inadequate dispersion control, where micro-carbon agglomerates larger than 50 µm act as stress concentrators during thermal cycling [30], accelerating fatigue-induced microcracking. Consequently, micro-carbon remains optimal for secondary impact-resistant layers in liquid hydrogen storage vessels [65]. Overall, the cryogenic enhancement effect of nano-fillers is highly dependent on dispersion quality and content optimisation. Inadequate control may readily induce a transition from toughening to embrittlement, which highlights the crucial importance of precise processing in translating mechanistic understanding into practical engineering applications.

4.2. Hybrid Composites and Synergetic Effects

Hybrid architectures leverage complementary reinforcement functionalities to overcome cryogenic limitations, where structural hierarchy dictates performance gains. In carbon/glass fibre interlaminar hybrids, carbon fibres provide compressive rigidity while glass fibres enhance strain tolerance, collectively increasing flexural strength by 35.37% at −196 °C compared to inverse ply sequences [124]. This synergy stems from CTE mismatch compensation: carbon fibres’ negative axial CTE counteracts epoxy shrinkage, generating interfacial compressive stresses that suppress delamination. For nanofiller-enhanced systems, the preparation process of GO and nano-Al(OH)₃ (NA) hybrid epoxy nanocomposites (Figure 14a) enables synergistic reinforcement. As quantified in Figure 14b, these hybrids exhibit a significant tensile strength increase at 90 K, while Figure 14c confirms a 21.11% elevation in K_IC versus single-additive systems. This cooperative toughening arises from crack-path bifurcation mechanisms detailed in Figure 14d, where GO nanosheets extend fracture trajectories and nano-Al(OH)₃ particles induce microcrack branching [40].

Beyond mechanistic advantages, hybrid architectures demonstrate quantifiable performance optimisation where conventional composites exhibit trade-offs between strength and fracture resistance at cryogenic temperatures. For nano-hybrids, the GO/Al(OH)₃ epoxy system achieves simultaneous 12.58% tensile strength and 30.27% fracture toughness gains at 90 K, eliminating the inverse relationship typically observed in mono-reinforced composites. SEM analysis of fracture surfaces confirms this failure inhibition, showing 3.2 times longer crack propagation paths in hybrid systems versus monoliths due to interfacial toughening mechanisms. The performance advantages of hybrid composites fundamentally stem from nanoscale interfacial engineering, where reinforcements actively mediate stress transfer across heterogeneous phases under cryogenic constraints. Specifically, MXene nanosheets (0.1 wt%) create continuous 2D networks that arrest microcracks at fibre/matrix interfaces. As illustrated in Figure 15a,b, which show SEM images of ILSS fracture surfaces and failure mode schematics at room temperature for CFRE and CFRE-MXene composites, the MXene modification results in a rough fracture surface with extensive matrix residue on fibres, indicating improved interfacial adhesion and a transition from debonding to matrix-dominated failure [21].

The performance advantages of hybrid composites fundamentally stem from nanoscale interfacial engineering, where reinforcements actively mediate stress transfer across heterogeneous phases under cryogenic constraints. Nanoscale bridging effects are paramount, as vertically aligned MWCNTs at 0.5 wt% form covalent bonds with epoxy matrices and simultaneously bridge carbon fibre interfaces mechanically. This results in a 62.32% increase in ILSS at 77 K compared to non-hybridised systems [109]. Similarly, MXene nanosheets (0.1 wt%) create continuous 2D networks that arrest microcracks at fibre/matrix interfaces (Figure 15c,d), suppressing crack propagation length by 3.2 times at 90 K [21]. Stress redistribution mechanisms further optimise load transfer: finite element modelling (FEM) verifies that MXene’s orthotropic alignment reduces interfacial peel stresses by 40% at −196 °C, redistributing loads toward high-stiffness fibres [21]. This prevents localised stress concentrations that trigger premature debonding. Crucially, hybrid systems maintain cryogenic interfacial integrity: modified epoxy/carbon fibre interfaces retain 89.2% higher frictional resistance at −196 °C versus room temperature [32]. SEM fractography consistently validates these mechanisms, showing MWCNT pull-out traces and matrix residue coatings on reinforcement surfaces even under extreme thermal contraction.

Table 6. Synergistic Effects in Hybrid Composites at Cryogenic Temperatures.

Hybrid System	Composition	Temp. (K)	Key Performance Advantage	Synergy Mechanism
CF/GF interlaminar hybrid	[C2G3]	77	Flexural strength ↑27.82% vs. pure GF	CTE mismatch-induced compressive stress [124]
Epoxy/GO/nano-Al(OH)3	0.1 wt% GO + 3 phr NA	90	Fracture toughness ↑30.27%; Liquid oxygen (LOX) compatibility ↑	GO crack-path tortuosity + NA crack refinement [40]
Carbon/Kevlar hybrid	[KCKCKC]s	77	ILSS: 35 MPa; Thermal stress ↓40%	Alternating plies mitigate CTE mismatch [125]
Flax/hemp epoxy	Flax-hemp-flax layup	77	Impact strength retention ↑21.6% (45 min exposure)	NaOH treatment, interfacial adhesion retention [126]

quantifies synergistic effects in hybrid composite architectures, highlighting performance gains (e.g., strength retention, damage suppression) and underlying synergy mechanisms under cryogenic conditions. In summary, hybrid composite designs offer value not only by enhancing individual properties but more significantly by overcoming the conventional strength-toughness trade-off. Achieving such synergy, however, relies heavily on controllable manufacturing factors, including ply sequencing and interfacial engineering, which are essential for transforming mechanistic insights into reliable cryogenic applications.

5. Modelling Strategies for Low Temperature

5.1. Molecular Dynamics

Molecular dynamics (MD) simulations have emerged as a powerful tool for elucidating the interfacial interactions and failure mechanisms of nanoparticle- and fibre-reinforced composites at the atomistic scale. The modelling procedure generally follows a standardised workflow. First, polymer matrices are constructed via cross-linking algorithms or reactive force fields, such as diglycidyl ether of bisphenol A (DGEBA) cured with amine hardeners, or thermoplastic systems including polymethyl methacrylate (PMMA) and polyethylene (PE) [127,128,129,130,131,132]. Reinforcing agents such as CNTs, graphene nanoplatelets (GNPs/GO), polyhedral oligomeric silsesquioxane (POSS), or inorganic oxide nanoparticles are subsequently introduced [44,129,130,133,134,135]. The geometry of the reinforcements (length, diameter, aspect ratio, orientation) and surface modifications (e.g., carboxylated graphene or γ-aminopropyltriethoxysilane, γ-APS) are typically considered as key parameters [136,137,138,139].

As illustrated in Figure 16, distinct strategies are adopted for different types of reinforcements. For fibre-reinforced systems, amorphous SiO₂ is commonly employed to represent the glass fibre surface, with silane coupling agents grafted to establish covalent linkages between the fibre and epoxy [136,140]. Epoxy monomers such as DGEBA and curing agents (e.g., triethylenetetramine, or m-phenylenediamine, mPDA) are polymerized through cross-linking to form a three-dimensional network chemically bonded to the fibre surface. Both van der Waals and hydrogen bonding interactions, as well as explicit covalent bonds at the interface, are considered [136,140]. In typical simulations, the system is partitioned into fixed, interfacial, and matrix regions, enabling atomistic tracking of debonding, crack initiation, and propagation under applied tensile or shear loads [136]. For nanoparticle-reinforced systems, a fully cross-linked epoxy network is generated, into which a SWCNT is embedded and equilibrated under canonical ensemble (NVT) and isothermal-isobaric ensemble (NPT) conditions to obtain a stable interface [141]. Reactive force fields (ReaxFF) are often employed to capture bond scission, radical formation, and interfacial reconstruction during high-temperature ablation, revealing the role of CNT-induced interphase densification in suppressing matrix degradation [141].

Mechanistic insights from these simulations have demonstrated that CNTs significantly improve tensile strength and fracture toughness of epoxy matrices through chain adsorption, crack deflection, and energy dissipation [128]. The degree of reinforcement is strongly dependent on CNT length, diameter, and alignment, with results showing good agreement with single-fibre pull-out tests [137]. Bone-shaped CNTs further enhance interfacial strength via mechanical interlocking at their enlarged ends [133]. Graphene-based reinforcements exhibit sensitivity to sheet waviness, orientation, and volume fraction, with multilayer graphene models revealing a competition between chain confinement and interlayer sliding [130,132,134]. POSS and oxide nanoparticles create interphases with distinct elastic properties compared to the bulk matrix, thereby serving as effective parameters for mesoscale models [127,129]. In fibre-reinforced systems, ReaxFF simulations of epoxy/glass interfaces have revealed bond rupture and hydrogen bond breakage during debonding, while silane coupling agents substantially improve fracture energy though peak strength remains relatively unaffected [136]. The introduction of CNTs on carbon fibre surfaces generates a tri-phase interphase that significantly enhances stress transfer, in agreement with experimental single-fibre shear tests [137]. As shown in Figure 17, the failure morphology of glass fibre/epoxy interfaces highlights the importance of interfacial chemistry: uncoated fibres rely mainly on van der Waals and hydrogen bonding, which are susceptible to water adsorption and nanovoid coalescence, leading to interfacial debonding underload. In contrast, silane-coated fibres form Si-O-Si and Si-O-C covalent bridges, and the combined effect of covalent bonds, hydrogen bonds, and van der Waals interactions markedly improves adhesion, shifting the failure mode from interfacial debonding to mixed interface-matrix or matrix cohesive failure [142]. Quantitative descriptors derived from molecular dynamics, encompassing adhesion energy, hydrogen bond statistics, interfacial contact area, and water radial distribution peaks, facilitate the parameterization of cohesive zone models within representative volume elements. This calibration couples atomistic failure initiation with the prediction of macroscopic interlaminar shear strength [131].

To extend predictive capability to structural scales, several studies have integrated MD with continuum-level models. A multiscale framework combining MD with the finite volume direct averaging micromechanics (FVDAM) method was developed to transfer cyclic traction-separation laws and damage accumulation rules from MD to continuum cohesive layers, enabling macroscale cyclic debonding predictions [143]. Graphene/epoxy interfacial properties obtained from MD have also been embedded into cohesive zone models and extended finite element method (XFEM) simulations to predict fracture toughness as a function of platelet aspect ratio and distribution [130]. A sequential coupling of MD with the scaled boundary finite element method (SBFEM) has further enabled efficient predictions of interfacial failure using MD-calibrated traction-separation and shear lag models [138]. More recently, a three-dimensional Lennard-Jones cohesive zone model (LJCZM3D) was parameterised by MD and integrated with phase-field formulations to capture coupled in-plane and out-of-plane fracture in graphene nanocomposites [135]. In addition, a multiscale framework was proposed for basalt fibre (BF)-reinforced Nylon (PA) 66 composites (Figure 18), in which non-equilibrium MD (NEMD) was used to evaluate interfacial thermal conductivity and molecular mechanics of fibre/graphene/coupling agent interactions, mesoscopic RVEs were constructed to investigate stress distribution in BF/PA66 systems, and macroscale layerwise finite element models were combined with infrared thermography to predict both mechanical and thermal performance of composite laminates [144]. This hierarchical approach provides a full-chain link from atomistic interactions to structural response.

However, these multiscale approaches entail key limitations and simplifying assumptions when bridging MD-derived interfacial laws to continuum models. Critical limitations include the scale separation assumption, which often neglects mesoscale phenomena such as defect evolution or dynamic interface changes; the idealisation of complex atomic interactions into simplified traction-separation laws, which may overlook rate-dependent or temperature-sensitive bonding behaviour; and the computational expense of MD simulations, restricting parametric studies and model validation. Transfer assumptions typically involve homogenising discrete interface properties into continuum parameters, presuming linear elasticity or perfect bonding outside calibrated ranges, and ignoring size effects when upscaling nanoscale findings. Addressing these aspects is essential for enhancing predictive accuracy in cryogenic applications.

Temperature effects have also been systematically investigated. At cryogenic temperatures, epoxy matrices exhibit a substantial reduction in hydrogen permeability due to increased cross-link density and reduced free volume, which hinder molecular diffusion pathways [145]. Graphene/GO-C-S-H nanocomposites demonstrate enhanced stiffness and modulus at low temperature owing to stronger hydrogen bonding, though defects can compromise this effect [146]. In graphene/copper systems, dislocation motion is more effectively suppressed at low temperatures and high strain rates, resulting in improved strength and modulus, while the reinforcing effect diminishes at elevated temperatures [147]. For PMMA/CNT composites, reduced chain mobility at cryogenic conditions intensifies confinement, thereby increasing modulus and strength [148]. MD simulations of CNT-PMMA under tensile loading further reveal that stronger van der Waals interactions and chain entanglement at low temperatures enhance load transfer, leading to increased yield stress and fracture strain [149]. Low-temperature toughening strategies such as POSS nanoparticles and hyperbranched polymers have also been shown to improve modulus and fracture resistance of epoxy matrices [44,129]. Fibre-reinforced systems demonstrate similar sensitivity: MD simulations of glass fibre/polypropylene interfaces show that interfacial van der Waals and hydrogen bonding interactions strengthen at low temperatures, thereby increasing interfacial stiffness, while elevated temperatures weaken adhesion and promote cohesive matrix failure [139]. Strain rate also plays a critical role, as insufficient relaxation at high rates leads to stronger interfaces and altered failure modes [139]. In carbon fibre/epoxy systems, cryogenic conditions promote brittle debonding, although surface treatments such as CNT grafting or silane coupling agents can preserve high interfacial strength [136,137].

Taken together, MD simulations have provided standardised atomistic methodologies to probe interfacial structure-property relationships in both nanoparticle- and fibre-reinforced composites. Multiscale frameworks enable the systematic transfer of atomistically derived parameters into mesoscale RVEs and macroscale structural models, bridging the gap between nanoscale physics and engineering-scale predictions. Cryogenic simulations further reveal that reduced free volume and constrained chain mobility enhance brittleness of matrices [145,146,147,148,149], that nanoparticle reinforcements exert even stronger confinement and hydrogen bonding effects at low temperature [44,129,146,147,148,149], and that fibre/matrix interfaces, while stiffer, exhibit more brittle failure modes under cryogenic conditions unless appropriately modified [136,137,139]. These findings collectively deepen our understanding of composite behaviour across multiple scales and provide critical insights for the design of advanced composites with reliable performance in extreme environments. Nevertheless, the intrinsic spatial and temporal limitations of MD simulations still constrain their direct applicability to structural design. Future progress will therefore rely on tighter integration with experimental validation, machine-learning-based surrogate models, and engineering-scale finite element frameworks to enable data-driven, predictive design of cryogenic composite structures.

5.2. Micromechanics Models

The investigation of composite materials under cryogenic conditions necessitates the application of micromechanics models due to the inherent limitations of conventional macroscale approaches in capturing complex microstructural interactions and failure mechanisms at low temperatures [34]. Micromechanical models provide a fundamental understanding of the deformation and damage processes at the fibre-matrix level, which is critical for predicting the overall performance of composites in cryogenic applications such as liquid hydrogen tanks and superconducting magnets [45,150]. These models are particularly essential for elucidating the effects of thermal residual stresses induced by CTE mismatches between fibres and matrix, which are exacerbated at cryogenic temperatures [151].

However, the development and validation of micromechanics models for cryogenic composites face several core challenges. Firstly, the significant temperature dependence of constituent material properties, such as the increased brittleness of polymer matrices and altered interfacial behaviour, complicates accurate constitutive modelling [152]. Secondly, the generation of substantial thermal residual stresses during cooling from curing temperatures to cryogenic service conditions can lead to premature microcracking and interfacial debonding, which are difficult to quantify without detailed micromechanical analysis [153]. Additionally, experimental validation under cryogenic conditions is inherently challenging and costly, limiting the availability of empirical data for model calibration [154]. These challenges underscore the need for advanced micromechanical frameworks that can effectively address multiphysics coupling and microstructural complexities unique to cryogenic environments.

Recent breakthroughs in micromechanical modelling have successfully addressed critical challenges in cryogenic applications through sophisticated integration of three key aspects: temperature-dependent material behaviour, thermally induced residual stresses, and progressive damage mechanisms. Among various modelling approaches, RVE models have emerged as particularly effective tools for quantifying micro-scale stress fields under complex thermo-mechanical loading conditions [152]. The fundamental Mechanics of Materials (MOM) approach utilises two-phase RVEs with uniformly distributed fibres(Figure 19a), employing iso-field assumptions and rule-of-mixtures formulations to derive effective elastic properties [155]. For advanced composite systems requiring higher fidelity, Kundalwal and Ray [156] present an enhanced three-phase RVE that explicitly incorporates a distinct fibre-matrix interphase region (Figure 19b). This three-phase MOM extension introduces interphase-specific constitutive equations and advanced homogenization rules, enabling accurate prediction of how interphase properties, including stiffness and thermal expansion coefficients, govern bulk composite behaviour under cryogenic conditions [152]. The Composite Cylindrical Assemblage (CCA) model (Figure 19c) provides specialised solutions for composites with complex fibre geometries through concentric cylindrical RVEs representing hollow circular fibres embedded in a matrix medium. Originally developed by Hashin and Rosen [157], this variational method delivers exact solutions for hexagonal fibre arrays (Figure 19d) alongside approximate solutions for random fibre distributions. Building upon these foundational approaches, Ren et al. [45] pioneered a trans-scale framework for Composite Overwrapped Pressure Vessels (COPVs), modelling thermal mismatch stresses during cooldown from 455 K to 50 K. Their hexagonal RVE formulation demonstrated thermal stresses alone induce matrix cracking at 50 K, with base stresses increasing 25% versus room temperature.

Analytical micromechanics evolved through homogenization techniques accounting for cryogenic property variations. Shindo et al. [34] developed temperature-dependent Halpin-Tsai equations modified with Mori-Tanaka approximations, capturing the 37% increase in critical energy release rate observed in glass/epoxy composites at 77 K compared to room temperature. Their model revealed that matrix microcracks paradoxically enhance fracture toughness by 15–25% through crack deflection and fibre bridging. Interface modelling has seen transformative advances through generalised cohesive zone formulations. Yuan et al. [158] established temperature-dependent traction-separation laws incorporating cryogenic interfacial properties, predicting within 10% error the 83.3 MPa interlaminar shear strength at 4 K.

Micromechanical model predictive capabilities under cryogenic conditions have been validated through experimental error quantification. Ren et al. [45] compared trans-scale RVE predictions against carbon fibre/epoxy testing at 50 K and reported that the error in matrix cracking initiation temperatures was less than 8%. Their model captured the 37% thermal residual stress increase during cooldown from 455 K to 50 K, validated through neutron diffraction. For interface modelling, Yuan et al. [158] quantified temperature-dependent cohesive zone model accuracy through cryogenic end-notched flexure (ENF) tests at 4 K. Their results demonstrated that the error in interfacial shear strength predictions was less than 10%, with maximum discrepancies reaching 9.8% at liquid helium temperature. Shindo et al. [34] conducted a comprehensive error analysis on their virtual crack closure technique (VCCT) implementation, revealing that neglecting fibre/matrix interface friction caused 22.7% overprediction of Mode II energy release rates at 77 K in woven composites. Their experimental validation through four-point bend end-notched flexure (4ENF) tests demonstrated that incorporating a friction coefficient of μ = 0.3 reduced prediction errors from 18.4% to 6.1% at liquid nitrogen temperature.

The most extensive validation efforts have focused on damage progression modelling. Mishnaevsky [159] reported 12–15% underprediction of delamination growth rates in wind turbine blade composites at −40 °C when using conventional progressive damage models, attributing this to unaccounted matrix embrittlement effects. His modified framework, incorporating cryogenic brittle transition physics, reduced errors to 7% across the 293 K to 77 K range. For nanoparticle-reinforced systems, Zheng et al. [153] quantified machine-learning accelerated SCA accuracy through synchrotron X-ray tomography at 50 K, showing 92.3% matrix crack density prediction accuracy but 18.7% delamination area underestimation in high thermal gradient regions.

Validation metrics consistently show superior model performance at lower temperatures. Shindo et al. [34] documented average prediction errors decreasing from 14.2% at 293 K to 6.8% at 77 K and 4.9% at 4 K, attributed to reduced viscoelastic complexity and more linear material responses. Similarly, Mishnaevsky [159] observed a 40% improvement in fatigue life predictions at −40 °C compared to room temperature, owing to diminished time-dependent deformation mechanisms. These findings collectively demonstrate that while modern micromechanical frameworks achieve remarkable accuracy (typically <10% error) for static cryogenic predictions, challenges persist in modelling cyclic damage accumulation and localised thermal-gradient effects where errors can exceed 15%. Future validation efforts should prioritise in situ cryogenic microscopy techniques to capture real-time damage evolution and improve interface characterisation methodologies.

Critical limitations persist in cryogenic micromechanical modelling despite significant advances. Current frameworks inadequately address temperature-dependent interfaces, with cohesive zone models failing to capture ductile-to-brittle debonding transitions below 50 K. Multiscale transition inaccuracies constitute another limitation, as homogenization-induced errors reach 22% at 20 K in COPV dome regions [45], and machine learning approximations introduce 15–18% errors in localised thermal stress predictions [160]. Computational frameworks face crippling efficiency-accuracy tradeoffs: high-fidelity analysis requires prohibitively expensive computational costs [159], while accelerated methods sacrifice sub-micron damage resolution [160]. Future priorities include developing in situ cryo-TEM/SEM techniques, creating damage-aware scale-bridging functions [45], integrating quantum computing optimisation [160], and establishing digital twin frameworks [159]. In summary, while current micromechanical models have achieved a prediction error of less than 10% for static cryogenic properties, significant uncertainties remain under service conditions such as thermal cycling, hydrogen permeation, and localised thermal gradients. Enhancing the engineering applicability of these models will require future integration of in situ cryogenic experimental techniques and digital twin technologies to better capture complex multiphysics interactions and damage evolution under realistic operational environments.

5.3. Finite Element Analysis

Under cryogenic conditions, the mechanical response and damage evolution mechanisms of fibre-reinforced and nanoparticle-reinforced composites differ significantly from those at room temperature. Epoxy matrices generally exhibit increased stiffness but reduced fracture strain, showing pronounced brittleness, while residual thermal stresses arising from the CTE mismatch between fibre and matrix promote stress concentration and microcrack initiation during cooling; subsequent thermal cycling further accelerates crack propagation and cumulative damage [25,27,29,47,48,161]. These complex temperature effects limit the ability of experimental methods alone to fully capture the underlying mechanisms, making FEA and multiscale modelling indispensable tools in this field.

For fibre-reinforced composites, numerical modelling has primarily focused on incorporating temperature-dependent constitutive models for the epoxy matrix together with progressive damage criteria. Epoxy has typically been described using elastic-plastic or brittle damage laws, exhibiting stiffening but reduced ductility at low temperatures, while fibres are usually modelled as orthotropic elastic phases. At the laminate level, failure initiation has been captured using three-dimensional Hashin or Puck criteria [27,43,47]. Interfacial behaviour has often been addressed with cohesive zone models (CZMs), which employ traction-separation laws to simulate fibre-matrix debonding and interlaminar delamination, successfully reproducing crack initiation and propagation under cryogenic conditions [25,48]. Figure 20a illustrates a typical multiscale modelling framework for fibre-reinforced composites: the macroscopic laminate is discretised into a seven-layer RVE, further decomposed into yarns, matrix, and interlayer interfaces; yarns are then resolved into individual fibres embedded in the matrix, enabling the transfer of damage and failure information across micro-, meso-, and laminate scales.

In contrast, nanoparticle-reinforced epoxy composites place greater emphasis on particle-matrix interfacial behaviour and energy dissipation mechanisms. Multiscale finite element frameworks often employ micro-RVEs with explicitly embedded nanoparticles, where particle-matrix interfaces are represented by cohesive elements to capture debonding processes. Simulations have identified three major toughening mechanisms: particle debonding, plastic void growth in the surrounding matrix, and shear-band formation between particles [163]. Particle size and interfacial strength are critical parameters, and the concept of an optimum particle size effect has been proposed: when particle radius approaches the critical separation length of the interface, partial debonding maximises energy dissipation, thereby overturning the conventional assumption that smaller particles are always more effective [163]. Parametric studies further reveal that the volume fraction and aspect ratio of GO nanoplatelets dominate the reinforcement effect, with the contribution of volume fraction reaching 46.1% [164]. Hybrid toughening strategies, such as GO combined with polyurethane (PU), have also been investigated through modelling, showing synergistic rigid-soft interactions and predicting fracture toughness improvements exceeding 200% at 90 K, while maintaining compatibility in liquid oxygen environments [162]. Similarly, MXene nanosheets have been shown to bridge fibre-matrix interfaces and mitigate residual thermal stresses, enhancing transverse and interlaminar properties under cryogenic conditions [21]. In addition, inorganic fillers such as nano-silica have been represented by elastic-particle/brittle-matrix interface models, highlighting their ability to mitigate matrix embrittlement at low temperatures [163,165]. Figure 20b demonstrates a modelling approach for nanoparticle-reinforced composites under impact loading: PU particles and GO nanosheets are explicitly embedded in the epoxy matrix with refined meshing to capture local interfacial failure and stress concentration, while three-dimensional solid and cohesive elements are employed to simulate damage progression under an impact energy of 98 J.

Beyond modelling frameworks, different reinforcement strategies also lead to distinct failure modes under cryogenic conditions. Figure 21 compares the typical failure characteristics of fibre-reinforced and nanoparticle-reinforced composites through combined numerical and experimental observations. In fibre-reinforced systems (Figure 21a), matrix cracking and interlaminar delamination dominate the failure process, and as the temperature decreases from 193 K to 93 K, damage zones expand and fracture surfaces evolve from relatively rough to flat and brittle, indicating a substantial reduction in matrix toughness. By contrast, nanoparticle-reinforced systems (Figure 21b) exhibit more complex, multi-mechanism failure processes: pure epoxy shows rapid propagation of a main crack; the addition of PU particles induces particle bridging and tearing; GO nanoplatelets generate microcracks that interact with the main crack; and the combined PU/GO system produces significant crack deflection accompanied by particle bridging, tearing, and distributed microcracking, which markedly enhance energy dissipation. These findings indicate that fibre-reinforced composites in cryogenic environments predominantly fail by matrix embrittlement and delamination, while nanoparticle-reinforced systems achieve effective toughening via crack deflection, particle bridging, and microcrack dispersion.

These findings indicate that fibre-reinforced composites in cryogenic environments predominantly fail by matrix embrittlement and delamination, while nanoparticle-reinforced systems achieve effective toughening via crack deflection, particle bridging, and microcrack dispersion. Overall, finite element and multiscale modelling approaches have successfully captured the key deformation and failure mechanisms of both fibre- and nanoparticle-reinforced composites under cryogenic conditions. However, their predictive accuracy remains highly dependent on the fidelity of interfacial constitutive laws and the precision of temperature-dependent material parameters. Future research should therefore prioritise experimental data inversion, interfacial parameter calibration, and the simulation of cyclic cryogenic environments to establish robust, quantitatively reliable predictive frameworks for engineering design.

6. Challenges and Future Directions

Despite notable progress in understanding the cryogenic behaviour of nano- and fibre-reinforced composites, critical challenges remain unresolved. Closing these gaps is essential if laboratory results are to evolve into dependable engineering solutions. Several research directions deserve particular attention.

The first difficulty lies in characterising material properties across the cryogenic regime. Predictive modelling depends on reliable datasets, yet available measurements of fibres, matrix, and nanoparticles are scattered and sometimes contradictory. Reported coefficients of thermal expansion, fracture toughness, or interfacial shear strength often diverge by an order of magnitude. Moreover, the time-temperature dependence of viscoelasticity and plasticity is seldom examined below 77 K, leaving the ductile-brittle transition poorly constrained. Establishing consistent baselines will require more systematic use of advanced techniques. In situ cryogenic dynamic mechanical analysis, nano-indentation, and high-resolution thermal analysis provide particularly valuable tools for this purpose. Beyond single tests, questions of cryogenic ageing, thermal cycling, and chemical compatibility with liquid hydrogen or liquid oxygen must also be addressed, since they define the realistic boundary conditions that ultimately govern service life.

A second, equally pressing issue concerns the fibre-matrix interface. This region remains the most sensitive and least predictable element under cryogenic conditions. Current models often employ simplified cohesive-zone formulations that miss key aspects: temperature-dependent adhesion, stress fields generated by mismatched coefficients of thermal expansion, or the toughening role of nanoparticles at the interface. More promising strategies would link atomistic insights from molecular dynamics with micromechanical and continuum approaches to create physics-informed multiscale models. At the same time, surface functionalisation, nanoparticle distribution, and fibre sizing deserve closer scrutiny. These variables strongly affect cryogenic interfacial strength yet are rarely parameterised in numerical simulations, partly because of the experimental difficulty in isolating them. Furthermore, while carbon-based nanofillers currently dominate the literature, there is a pressing need to expand research into non-carbon nanomaterials and hybrid filler systems. These materials offer distinct advantages, such as electrical insulation and tunable negative thermal expansion, which are essential for specific cryogenic applications like superconducting magnet insulation, yet they remain underrepresented in current structural performance studies.

The third challenge is the modelling of damage mechanisms. Failure is inherently multi-scale, involving matrix microcracking, fibre-matrix debonding, delamination, and eventual global fracture. Traditional criteria, such as Hashin or Puck, provide useful starting points but are insensitive to temperature and therefore cannot reproduce the transition from ductile to brittle behaviour at low temperatures. Recent advances in modelling approaches offer new possibilities. Phase-field fracture, peridynamics, and stochastic multiscale frameworks can incorporate temperature-dependent fracture energies, embrittlement kinetics, and evolving crack networks. Yet bridging molecular-scale phenomena, such as chain immobilisation or nanoparticle toughening, with system-level predictions of delamination growth remains a daunting open problem. Progress here will determine whether simulations can genuinely predict reliability rather than merely post-rationalise experiments.

Finally, model validation requires experimental strategies beyond conventional tensile or fracture tests. Standard tests at low temperature deliver only averaged data and are plagued by reproducibility issues arising from thermal gradients and handling difficulties. Newer techniques are beginning to address these limitations. Digital image correlation, acoustic emission, and X-ray computed tomography have provided in situ views of damage initiation and propagation under cryogenic conditions. When combined with simulations, these approaches offer a path towards robust calibration and a deeper understanding of multi-scale fracture processes. However, to enable cross-comparison between laboratories and eventual certification for aerospace or energy use, the field urgently needs standardised cryogenic testing protocols.

Taken together, these directions suggest that future work should move towards an integrated research framework. Accurate characterisation, physics-informed interface and damage models, and comprehensive experimental validation must be linked, rather than pursued in isolation. Such integration could bridge atomistic mechanisms and structural reliability, paving the way for the safe deployment of advanced composites in hydrogen storage, deep-space missions, and other extreme environments.

7. Conclusions

This review has systematically examined the performance and modelling of nano- and fibre-reinforced composites under cryogenic conditions, spanning constituent behaviour, system-level responses, enhancement mechanisms, and predictive strategies. At the constituent level, fibres generally preserve their stiffness (e.g., T700 carbon fibres maintain ~230–245 GPa modulus from 300 K to 77 K), whereas the polymer matrices stiffen but suffer severe embrittlement (e.g., epoxy fracture strain drops to ~1% at 77 K). Nanoparticles, in contrast, introduce powerful opportunities to tailor both thermal and mechanical properties, as demonstrated by the 15.63% enhancement in interlaminar shear strength (ILSS) with 0.3 wt% graphene oxide (GO) addition at 77 K. At the composite scale, performance is governed by the interplay of thermal expansion mismatch, matrix embrittlement, and interfacial degradation, with mitigation strategies such as nano-addition (e.g., 4 wt% SiO₂ retaining 98% interfacial adhesion after 25 thermal cycles), hybrid architectures (e.g., [C2G3] carbon/glass configuration increasing flexural strength by 35.37% at −196 °C), and thin-ply designs offering notable improvements in toughness and reliability. Based on the reviewed evidence, carbon fibre/epoxy systems modified with well-dispersed nanoparticles (e.g., SiO₂, MXene) or hybrid carbon/glass fibre architectures emerge as particularly promising for applications demanding high-dimensional stability and thermal cycling resistance, such as liquid hydrogen storage tanks and deep-space structures. For instance, the inclusion of 0.1 wt% MXene nanosheets reduces crack propagation by 3.2 times at 90 K, highlighting their potential for cryogenic applications where damage tolerance is critical.

Methodologically, progress has been advances across multiple scales. Molecular dynamics has clarified atomistic interactions, micromechanics has linked those interactions to mesoscale responses, and finite element modelling has extended the insights to laminate and structural levels. In parallel, state-of-the-art in situ experimental techniques, such as digital image correlation, acoustic emission, and synchrotron-based tomography, have captured real-time fracture and damage evolution, enabling increasingly robust model validation.

Despite this progress, major challenges remain. Reliable deployment of cryogenic composites in aerospace, hydrogen storage, and deep-space infrastructures demands more comprehensive characterisation across the full cryogenic regime. Interfaces remain the most unpredictable regions, with their behaviour controlled by temperature-dependent adhesion, residual stress fields, and nanoparticle toughening effects—factors rarely parameterised in existing models. Furthermore, current damage models struggle to reproduce ductile-to-brittle transitions or capture the stochastic nature of microcracking under thermal cycling. Standardised cryogenic testing protocols, coupled with integrated computational–experimental validation strategies, are urgently needed to reduce uncertainty and accelerate certification.

Looking forward, the future of this field lies in integrated, physics-informed research frameworks that seamlessly combine high-fidelity data, multiscale simulation, and advanced experimental validation. Such convergence will bridge atomistic mechanisms with structural reliability, enabling predictive rather than retrospective modelling. Beyond consolidating existing knowledge, this review highlights the imperative of cross-scale integration and standardised validation, offering a distinctive framework to guide both fundamental discovery and engineering translation of cryogenic composites.