Physical Properties of Polymers Under Soft and Hard Nanoconfinement: A Review

Polymeric materials under nanoconfinements have substantially deviated physical properties with respect to the bulk, especially glass transition temperature, physical aging, and crystallization behavior. Here we highlight the leading methods for creating various confinement systems. Upon these systems, recent advances on hard and soft confinement effect for glass transition, physical aging, mechanical properties and crystallization of polymers are reviewed in details. Furthermore, as nanoconfined systems in extreme conditions are experimentally inaccessible, simulation results describing confinement effect on such systems are also discussed.


INTRODUCTION
The states of polymeric materials are temperature dependent as shown in Fig. 1. The enthalpy, volume or other properties have different temperature dependence in two distinct states, i.e., glass state and liquid state. The intersection of the glass and liquid state lines corresponds to one crucial temperature for state transition which is referred to as glass transition temperature (T g ). Polymeric materials are in equilibrium state above T g . In contrast, the physical properties, including but not limited to volume, [1−3] enthalpy, [4−6] gas permeability [7−9] and elastic modulus, [10−12] gradually change with time if the polymers are annealed below T g . This evolution process is physical aging or structure relaxation. These time-dependent thermodynamic properties are attributed to macromolecular motion whose mode and intensity are relevant to the molecular mobility.
Polymeric materials under confinement originated from size effect, interfacial effect and spatial constraints have substantially different thermodynamic or mechanical properties from the bulk, as mobility of polymer chains is altered by confinement effect. For instance, the T g of polymers on silica substrate can be enhanced or decreased up to more than 30 °C from the bulk value. [13] And in particular, the aging rate of PS blocks in a PS-b-PMMA copolymer can be accelerated by a factor of three compared to that of neat PS with similar molecular weight. [14] Confinement can be classified into hard confinement and soft confinement, according to the relative modulus between polymer and interfacial materials.
In addition to the physical aging of amorphous materials, crystallization of polymers is greatly influenced by confinement. Together with mechanical properties, almost all concerned aspects of polymers are influenced by the effect of confinement. In the last several decades, the deviated physical properties under soft and hard confinement have attracted remarkable interests. A great deal of efforts have been taken to explore and rationalize the mechanism of the nanoscale confinement effect. Despite some advances, the underlying physics of confinement effect of polymers has not been well understood.
In this review, we highlight the methods for creating the model systems to investigate confinement effect under different geometries and dimensions, and summarize the major progress of the hard and soft confinement effect, on T g , structural relaxation, mechanical properties, and crystallization of polymers in recent works. Moreover, the applications of confined systems are emphasized, especially for which the better manipulation of properties of confined polymers will benefit the future design and applications. Although this work mainly focuses on the experimental results, the models or simulation methods for describing the confinement effect are also discussed.

METHODS FOR CREATING CONFINEMENT
The confinement systems can be created by various means, either from fundamental methods or using novel techniques. A major method of confinement creation is making thin films by spin coating or solution casting, as they have a very simple preparation process and create adjustable confinement provided by interface when the thickness is less than some thres-hold. [13,15−50] In general, hard substrates, including silica, quartz, mica, gold, aluminate, and polymers with relatively high Young's modulus, are used to provide the hard confinement. [13,15−33] On the other hand, soft confinement conditions are achieved by the free-standing surface, liquid interface, or the relatively soft polymer substrate. [13,15,16,19,24,27,29,31,32,36−50] Moreover, Murphy, Langhe and their coworkers reported that the complex alternating multilayer structure from stacked thin films can be produced by a layer-multiplying co-extrusion process, providing an efficient method for integrating multiple confinement from different interfaces. [51−53] The details of such co-extrusion process were introduced elsewhere. [54] Instead of the simple smooth interface, Yavari and coworkers used porous polyethersulfone to support thin films and thus providing special confinement effect. [34,35] Their work created a novel method for changing the 1D confinement condition for thin films and has great relevance in industrial application.
In addition to the 1D confinement in thin films, the hard or soft confinement can be achieved in 2D condition. Two main methods for creating 2D hard confinement have been introduced in recent works. [22,55−58] One of those is mixing the polymer solution with oriented carbon nanotubes to produce nanocomposites. [56] Instead of the nanotubes in nanocomposites, another 2D hard confinement environment was attained by placing polymers into nanoporous matrix like anodic aluminum oxide (AAO) template. [22,55,57,58] The nanoporous AAO template was filled with polymers, which is achieved by melting the polymer films on the top of AAO template to infiltrate at a suitable temperature. As such, the AAO matrix is able to provide 2D hard confinement to the polymer in the nano pores. On the other hand, a representative 2D soft confinement, i.e., freely standing polymer nanowires, was created by electrospinning. This notable method is extruding the polymer solution from plastic syringe with a stainless steel needle to a substrate under high voltage. [58] Subsequently, the electrospun nanowires are dispersed in solution. By this method, the polymer nanowires with free surface can be attained for the purpose of studying the soft 2D confinement effect.
Polymer nanoparticles can provide the 3D free-standing surface (soft confinement) [59−63] and hard confinement after capping a rigid shell, [59,63,64] making them useful to investig-ate 3D confinement effect. Polymer nanoparticles can be synthesized by emulsion polymerization with the details reported elsewhere. [65] The limitation of this polymerization method is that it is difficult to control molecular weight and particle size independently. Given this fact, another method called flash nanoprecipitation was developed and used to overcome this difficulty when the particle diameter is no more than 150 nm. The details of this method were reported in the literature. [66] Upon polymer nanoparticles, a notable 3D hard confinement model is the core-shell model. For instance, Guo and coworkers synthesized the silica-capped PS nanoparticles by the Stöber method. [59,63,64] The dispersed PS nanospheres were reacted with ammonia and tetraethoxysilane to coat a thin layer of silica, after pretreating with polyelectrolytes in water, and the details were reported in literature. [66,67] Silica-capped PS nanoparticles provide not only the hard confinement, but also an isochoric condition for the PS particles inside. Instead of hard shell, Zhu and coworkers used gold as hard core in the core-shell structure. [68] The thiolated PS with HAuCl 4 ·3H 2 O dissolved in distilled tetrahydrofuran (THF) reacted with NaBH 4 in THF-water solution to produce polystyrene-capped Au nanoparticles. The resulting material showed good core-shell structure and had substantially different properties from the bulk PS. In addition to nanoparticles, another typical method for creating 3D hard confinement is preparing nanocomposites by mixing the polymer with some hard nanoparticles typically made by silica or metallic oxides. [30,33,69−79] Sometimes, the surface of nanoparticles was functionalized to get better dispersibility. [69−76] On 3D confined materials, diblock copolymers were recently found to be an ideal model to study the 3D confinement effect, [14,80,81] as they can form various morphologies from selfassembled microphase separation, by controlling the molecular weight and molar ratio of the two components. These geometries formed by two polymer components with relatively soft and hard states provide the soft and hard confinements to each other simultaneously. This unprecedented method serves as a simpler and more efficient way to create 3D confinement with polymer-polymer interfaces.
With the confined polymers under various geometries and dimensions, the study of the hard and soft confinement effect on different properties becomes accessible. We review recent progress from such work in the following sections.

Effect on T g
The hard confinement effect on T g does not hold complete consistency throughout the related research activities in the last several decades. Some results showed that T g s of polymeric materials under hard confinement have substantial deviation from the T g s of bulk materials. The deviation direction depends on the segmental mobility, restricted mobility for T g enhancement [13,16,17,[24][25][26][27]30,33,56,58,68,[76][77][78][79]82,83] and enhanced mobility for T g depression. [13,17,18,21,25,33,55,79,83−85] However, Sun and coworkers reported reduced mobility of PS films due to adsorbed layer, resulting in suppressed T g , as opposed to the general correlation between mobility and deviation of T g . [86] In contrast to altered T g , other results showed invariant T g s of polymers under hard confinement, with respect to the bulk value. [30,51,55,56,59,63,69,76,79] Fortunately, the T g deviation showed consistent trend under large confinement intensity, regardless of the materials and confinement creation methods in many works. [16][17][18]21,25,30,33,68,73,75,77,87] Researchers dedicated to probing the underlying mechanism of hard confinement effect and a universal elucidation for these inconsistent of hard confinement effect on T g . In general, the existence of T g deviation depends on confinement intensity, [27,30,76] i.e., the confinement-interface-to-internal-polymer ratio. For example, Koerner and coworkers found that the T g of PS composited with SiO 2 remains the bulk value until the volume fraction of SiO 2 is greater than 40%. [76] Fakhraai and coworker indicated that the cooling rate dependence of T g reduction is much stronger than the thickness dependence of T g suppression, which gives an explanation to contradictory of hard confinement effect on T g with different cooling rates. [88] In addition, molecular weight is a crucial factor of the existence of T g deviation of polymers in the 2D hard nanoporous template. [55] When molecular weight of PS is larger than 175 kg/mol, the PS nanorods confined in AAO nanoporous template displays a lower T g than the bulk, as the radius of gyration derived by molecular weight plays an important role in the intrinsic size effect. [55] The intrinsic size effect suppresses T g when the chains packing are disturbed by the hard confinement. Moreover, adsorbed layer at interface is considered momentous in changing chains packing and mobility. [22,89,90] Several works interpreted altered T g under hard confinement in terms of the competition in adsorbed layer between chain pinning and packing frustration, in which the former impedes chain mobility but the latter acts as additional free volume in the proximity to interface accordingly playing significant role in T g reduction. [55,83,85] As such, the free volume holes diffusion (FVHD) model becomes a useful tool to describe the correlation between free volume in interface and the change of T g , concluding that hard confinement on T g does not merely depend on film thickness and interfacial interactions. [91,92] Correspondingly, Napolitano and coworkers indicated that adsorbed layer needs annealing time to form physiosorbed chains, and the ratio of adsorption time and the annealing time could be one explanation for the contradictive hard confinement effect on T g . [83,93] Another explanation for these inconsistent results is the attractive interaction between the polymer and the hard confinement. [13,17,25,26,30,33,79,87] As a representative result, enhanced T g s have been presented by PMMA or poly(2-vinylpyridine) (P2VP) thin films on silica substrate, as presented in Fig. 2, due to the hydrogen-bonding interaction between the polymers and silica. [13,17,26,30,33,79] Taking the opposite, for the interfaces lacking attractive interaction, T g was found to be lower than the bulk value. [13,17,33] Furthermore, PS presented restrained T g under confinement with interface of metallic oxide [55,73] or silica [17,21,75] , whereas T g of PS was improved with hard polymer confinement, such as PS thin films on PMMA [13,16] or polysulfone (PSF). [16] These results provide scientific insights for inconsistency of the hard confinement effect when the polymer or the hard confinement materials are different. Different interaction between polymer and confinement may induce reverse effect. However, some other results are in controversy, that is, deviation of T g had opposite tendencies from the bulk value, even though the confinement system was organized by same polymer and similar hard confinements. Askar and coworkers found that T g of PS in AAO was suppressed, [55,73] as opposed to the enhanced T g of PS under same confinement reported by Wei and coworkers. [58,68] Thus, the mechanism of the confinement effects on T g is remaining to be further revealed.

Effect on Physical Aging
Altered physical aging behavior caused from confinement is another major topic in last two decades. Although physical aging depends on the molecular mobility just as T g , the mechanisms of hard confinement effects on them are different. For instance, Flory and coworkers found that the PMMA nanocomposite with unmodified single wall carbon nanotube (SWNT/PMMA) and amino-functionalized SWNT (a-SWNT/ PMMA) showed different T g variations compared to neat PMMA, namely invariant T g for SWNT/PMMA and enhanced T g for a-SWNT/PMMA, while their aging responses were both decelerated. [56] Moreover, some researchers devoted to investigating the decoupling between molecular dynamics and physical aging. [69][70][71][72][73][74][75]78] Boucher, Cangialosi, and their coworkers were focused on different properties measured by various methods, including segmental dynamics from broadband dielectric spectroscopy (BDS), structural parameter from the Tool-Narayanaswamy-Moynihan (TNM) model, and segmental relaxation time from thermally stimulated depolarization current (TSDC), to verify that hard confinement changes physical aging of nanocomposites while keeping the same molecular mechanism such as Havriliak-Negami (HN) relaxation compared to neat polymer, as presented in Fig. 3. They concluded that the physical aging under hard confinement not only depends on the molecular mobility but also the phenomenon can be well described by the FVHD model.
While positive correlation between hard confinement intensity and its effects on physical aging is widely accepted, [23,27,51,[69][70][71][72][73][74][75]78] the unified theory of hard confinement effects on physical aging is not accomplished. Acce- lerated aging rate, [14,[69][70][71][72][73][74][75]78,80] decelerated aging rate, [27,28,30,33,51,56,77,79,94] and invariant aging rate [58,77] to the bulk were demonstrated in different works. It should be noticed that the definitions of aging rate are different in different works using specific measuring methods. The aging rate can be usually defined by the time to reach equilibrium state, or the slope of the changed properties with respect to the logarithm of aging time. The mismatch of the definition of aging rate might be one reason for this inconsistency. [95]

Effect on Mechanical Properties
In last several decades, researchers made great efforts to explore suitable methods for directly investigating mechanical properties of polymers under nanoscale. As various methods emerge, influenced mechanical properties under nanoscale hard confinement were exposed. The main mechanical property staying focused is stiffness, on which the consensus is that moduli of polymers increase with increasing hard confinement intensity, as depicted in Fig. 4. [15,96−99] Similarly, the correlated mechanical properties, compliance were found reduced due to the hard confinement. [100] Hu and coworkers indicated that enhanced entanglement interaction at interface may be a reason for restrained polymer mobility, resulting in higher modulus. [97] In addition, interchain chemical bonds drove another argument for suppressed mobility. [98,101] For instance, Zheng and coworkers announced that the rotation and translation of polymer segmental in poly(vinyl acetate) (PVAc) thin films were impeded due to the polar hydrogen bonds between polymer and substrate with hydroxyl group, which increase modulus of PVAc thin films. [101] Effect on Crystallization Confinement effect on crystallization became a research hotspot in last two decades. It is found that several specifications of crystalline or semi-crystalline polymers, including nucleation, [102−115] crystal orientation, [103,105,107,109,110,116] crystallization rate, [103,108,[112][113][114][115]117,118] and crystallization temperature (T c ) or melting temperature (T m ), [103,106,107,109,110,112,113,117] are altered due to hard confinement.
Instead of conventional heterogeneous nucleation in the crystallization of bulk polymers, the crystallization of polymers under hard confinement was reportedly dominated by homogeneous nucleation due to restricted mobility. [102][103][104][105][106][107]109,113] The homogenous nucleation is concomitant with compressed T c . [103,104,106,107,109,110,112] Suzuki and coworkers presented a "phase diagram" to describe two nucleation regimes depending on proposed temperature and AAO hole curvature. [103] Homogeneous regime is close to glass state, whereas heterogeneous nucleation emerges at high temperature. [103] Accordingly, restrained T c closed to T g is regarded as an indicator for homogenous nucleation. [104,106] Moreover, hard confinement, especially hard nanoparticles, increases nucleation sites or nucleation density, resulting in higher crystallization rate with respect to bulk materials. [108,111,112,114,115] For example, Zhao and coworkers found that the nucleation efficiency of poly(ethylene oxide) (PEO) depends on their grafting density on SiO 2 nanoparticles in the composites. [108] They speculated that the energy barrier for nucleation may be disturbed or the sites for nucleation may be activated due to interfacial structures like loops or tails. [108] In addition, Strawhecker and coworkers reported higher nucleation density in sodium montmorillonite filled PEO, which is attributed to interrupted spatial continuity and restrained heterogeneous nucleation. [112] The principal research of hard confinement effect on crystal orientation was carried upon AAO. In these works, crystal orientations parallel to [103,[105][106][107]109,116] and perpendicular to [105,110] the AAO open channel coexist, depending on pore size of AAO and cooling rate. Steinhart and coworkers proposed the "kinetics selective growth" crystallization mechanism and indicated that chain axis is oriented perpendicular to pore direction for poly(vinylidene fluoride) (PVDF) confined in AAO. [116] Any crystal orientation with <hkl>* direction (l ≠ 0) stops growing due to the confinement from channel walls. [116] After that, researchers systematically investigated crystal orientation of polymers in AAO and improved this mechanism. Su and coworkers reported that the fastest growth direction of PEO, i.e., <120>*, is inclined to parallel to the AAO channel direction. [107] Guan and coworkers set a criterion to determine crystal orientation under confinement, that is, crystal growth is oriented parallel to the axis of the AAO channels when the pore diameter is greater than contour length of polymer chains Nl, while the crystals are aligned perpendicular to the channel axis when the pore dia- meter is smaller than Nl. [105] Furthermore, Su and coworkers depicted the dependence of cooling rate on crystal orientation, indicating three zones of orientation behavior of polymer confined in AAO. Thus, this work provided evidence showing that the "kinetics selective growth" mechanism would be invalid under specific conditions. [119] The conclusions of hard confinement effect on crystallization rate are in controversy. That is, enhanced rate, [108,111,114,115,117] depressed rate or even completely suppressed crystallization, [103,112−114,117,118,120−122] are revealed in various works. In studies exploring the origins of these unmatched results, some publications showed that the interaction between polymer and hard confinement suppliers plays a crucial role in generating reduced chain mobility, which resulted in restrained crystallization. [118,121−124] However, Vanroy and coworkers suggested that suppressed crystallization kinetics of PET thin films confined between two adsorbing walls is ascribed to limited nuclei density instead of inhibited segmental mobility. [120] Kołodziejczyk and coworkers showed that hard confinement effect on crystallization rate depends on the pore size in AAO template. [117] Salol in AAO with pore size of 150 nm exhibited greater crystallization rate than the bulk. Nevertheless, with smaller pore size, crystallization is decelerated until being completely prohibited. Similar results were also presented by Dai et al. and Mijangos et al. [121,125] They found that the interfacial effect of AAO interrupts segmental mobility, while spatial heterogeneity enhances seg-mental mobility. [121,125,126] As such, interfacial effect increases but spatial heterogeneity reduces with decreasing pore size, resulting in inhibited crystallization of polymers in AAO compared to bulk materials. [121,125] Wu and coworker reported that temperature protocol is a noteworthy factor for the effect on crystallization rate. [114] In contrast to the restricted rate in cold crystallization (heating from room temperature to T c ), polylactide (PLA) composed with graphene displayed elevated rate in melt crystallization (cooling from high temperature to T c ) compared to neat PLA, as illustrated in Fig. 5.
It should be noticed that crystallization of polymer in AAO, including nucleation and growth process, may be substantially influenced by residual polymer on the surface of AAO template. [106,107,116] As such, future study of AAO confinement effect on crystallization should be extraordinarily careful to ensure experimental repeatability and reliability.

SOFT CONFINEMENT EFFECT Effect on T g
Different from inconsistent results for hard confinement effect on T g , an amount of research had compatible conclusions for T g under soft confinement, i.e., reduced T g with respect to the bulk value. [16,24,27,42−45,49,58,59,61−63,84,127−135] In these results, the relation between confinement intensity and magnitude of T g deviation is consistent, showing that T g decreases with increasing confinement intensity. [  described in Fig. 6, the soft poly(n-butyl methacrylate) (PnBMA) or poly(isobutyl methacrylate) (PiBMA) reduces T g of PS in vicinity of the interface. In addition to confinement intensity, it is found that the magnitude of T g reduction depends on materials' molecular weight. [42,133−135] The intrinsic size effect induces T g reduction in low molecular weight regime, whereas it is dominated by chain confinement effects in high molecular weight regime. [134,135] It is widely accepted that the T g reduction under soft confinement is attributed to the high mobility in the free surface. [63,129−132] For example, Zuo and coworkers found reduced T g in poly(ethylene terephthalate) (PET) thin films, while these films exhibited same T g as that of bulk materials after surface layer crystallizing. [132] Sharp and coworker showed suppressed T g and invariant T g in PS thin films with or without Al capping layer, respectively, compared to bulk T g . [130] They also showed that restrained T g could be recovered after removing Al capper, which gives a direct evidence to vital effect of free surface on T g reduction regardless of sample preparation history. [130] Accordingly, researchers put great effort into investigation on the thickness of high mobility surface, which is determined to be 2−4 nm. [131,136−138] Although large numbers of works presented constrained T g under soft confinement, a few opposite results still exist. For instance, T g enhancement in PS nanoparticles or poly(bisphenol A carbonate) (PBAC) nanoparticles was reported by Martínez-Tong and coworkers. [60] They speculated that these enhanced T g s should be attributed to the reduced number of conformational states in the nanoparticles constructed by few chains.

Effect on Physical Aging
A great number of works studying the soft confinement effect on physical aging showed that the aging of polymers was accelerated under soft confinement, compared to the bulk materials. [19,31,[36][37][38][39][40]46,48,52,53,59,61] However, different conclusions, restricted or invariant aging, were detected in a few works. [27,44,52,53,58,80] In some of those results, the invariant aging was observed when soft confinement situations were constructed by relatively soft rubbers or the stacked films, as opposed to rapid aging of polymer with free surface. [44,52,53] Additionally, these systems may lack sufficient confinement intensity, as the stiffness discrepancy between rubber and the investigated polymers was not large enough and the stacked structure lost vast free surfaces presented in the single freestanding film. In recent years, it was argued that polymers could have a second fast equilibration mechanism, especially for  polymers under soft confinement, in which the second mechanism emerges at short time scale. [36,37,61] Different time scales for the appearance of a fast equilibration mechanism might be a reason for incompatible soft confinement effect on physical aging. Though the conclusions of soft confinement effect on aging contain disagreement, the dependence of acceleration or deceleration on confinement intensity presented in these works is consistent, and the degree of aging rate deviation increases with enhanced confinement intensity. [27,36,[38][39][40][41]59]

Effect on Mechanical Properties
The results of soft confinement effect on mechanical properties are more comprehensive and systematic, compared to the works on hard confinement effect, as the measurement systems for soft confinement are remarkably diverse. Existing breakthroughs in measurement of mechanical properties rely upon development of novel techniques, among which one recent progress is the determination of Young's modules E, from a buckling method, [50,139−142] or an ultrathin film tensile test, [143] or by capillary wrinkling. [144] In all of these measurements, ultrathin film must be in contact with a liquid or a soft substrate to form a bilayer system. The induced interfaces may be nontrivial in assessment of confinement effect on mechanical behavior of polymer films. Hence, the sample setup requirements restrict the generality of these novel techniques to be applied in a variety of scenarios. However, the diverse methods showed the inconsistent stiffness results of the soft confinement effect. Stiffness increase [139,144−146] or decrease [15,50,140−143] with increasing confinement intensity is reported in different works. Moreover, the confinement effects on strength including failure stress and strain are not quantitively consistent. Together with the findings above, it may indicate that the deep analysis and understanding on confinement influence on mechanical properties depend on the specific materials systems and the associated confinement conditions. [140,143,146,147] For example, Lee and coworker found that the fracture stress of PS decreases in thinner films, while the fracture strain becomes larger. In contrast, Yoon and coworkers reported enhanced rupture stress and reduced rupture strain of polyisobutylene (PIB) under soft confinement as illustrated in Fig. 7. [146] Researchers devoted to expositing the physics of soft confinement on stiffness in last tens of years. It is found that the mechanical properties of polymers strongly depend on the entanglement of the polymer chains. [140,143,147] Thus, entanglement constraining polymer chain dynamics in different systems might be a crucial factor on the diverse soft confinement effects. Additionally, Torres and coworkers showed that the modulus of polymer with stiff backbones is independent of thickness, as opposed to constrained modulus with decreasing thickness in flexible polymer films, suggesting soft confinement effect on mechanical properties depending on main chain stiffness. [142] Another potential explanation is the competition between cohesive forces and the modulus of individual polymer chains. Page and coworkers suggested that the modulus of individual polymer chains plays a central role in the stiffness of thin films, as cohesive force is nonsignificant due to soft confinement. Thus, the modulus in thin films under soft confinement is enhanced with respect to bulk materials. [139]

Effect on Crystallization
The assorted morphologies formed by block copolymers are the most common situation to study soft confinement effect on crystallization, if the T c of crystallized block is greater than the T g of amorphous block. In last several decades, many efforts have been paid to studying such systems. [148−161] It is found that nucleation of polymer under soft confinement depends on confinement microdomain. [149,152,154,156,157,162] For lamella, heterogeneous nucleation is favored, [149,152,157,162] whereas the crystallization is controlled by homogeneous nucleation in cylindrical or spherical morphology. [152,156,157] Moreover, Ho and coworkers investigated crystallization of polystyrene-b-poly(Llactide) (PS-b-PLLA) copolymer and first reported undulated morphology due to the transition from homogeneous to heterogeneous nucleation. [154] As same as nucleation, crystallization kinetics of polymers under soft confinement relies on confinement morphology. The first order Avrami kinetics for cylindrical or spherical confinement and spherulitic crystallization for lamella were validated in many works. [149,152,156,157] Unfortunately, the results of soft confinement on crystallization rate did not show good consistency: reduced [148−150] or enhanced crystallization rate [162] was reported in different works.

DISCUSSION
Examining the results shown above, materials under soft confinement show more compatible conclusions than those under hard confinement. Moreover, several works have succeeded in predicting the soft confinement effect by different models. For instance, Tito, DeFelice and their coworkers offered a limited mobility model to simulate free volume and mobility. [163,164] They found that mobility of polymer films directly contacted with high mobility materials was clearly enhanced. The enhanced mobility reduces the T g of polymer films with free surface. Lang and coworkers utilized the dependence of interfacial adhesion energy to perform the T g suppression under soft confinement. [165] These models provide valuable tools to describe and understand the soft confinement effects. It has been indicated that stress has a great impact on aging rate in supported films or free-standing films with circular opening holder. [19,29] As such, the different results of confinement effect on physical aging might be attributed to mismatched thermal expansion between polymer and confinement provider, which introduces additional stress in aging process.
The hard confinement systems are complicated if the free surfaces are involved in the systems. In most cases, soft and hard confinements are co-existent, especially for a thin film supported on a hard substrate. The measured confinement effects reflect the overall interplay between the soft and hard confinement, relying on the ratio of two confinement intensities. The stronger confinement dominates the overall effect in these systems. For instance, the PMMA nanolines on silica showed reduced T g due to the high free-surface-to-interface ratio, in opposition to the enhanced T g of the PMMA thin films on silica, as displayed in Fig. 8. [24] The competition between hard and soft confinements may be one of the reasons for the inconsistency of confinement effect in some similar systems but having different confinement intensities. Similar systems, even constructed by same polymer and same confinement materials, can perform considerably different effects due to the discrepant fraction of free surface. Moreover, the compet-   ition can be used to explain the invariant effects results after changing the confinement condition. [13,31] For example, Roth and coworkers found that the T g of P2VP thin films on silica substrate remained elevated as ever after capped with PS layer, as the confinement induced by silica substrate was much stronger than the confinement provided by PS layer atop or the free surface. [13] Consequently, the inconsistency of hard confinement effect on T g or physical aging may be attributed to the complexity of confinement systems. In order to reduce the complexity of system and avoid the impact from free surface, fluorescence measurements are considered as an ideal technology in studies of thin films supported on substrate. A thick layer above the labeled thin films can impede the effects of free surface. The transformation between soft and hard confinement will bring complexity as well, especially in the investigation of crystallization in block copolymers. [166−171] For instance, during the crystallization of PEO blocks in PEO-PS block copolymer with microphase separation, [166][167][168][169]171] PS blocks form hard confinement to PEO blocks at the beginning of crystallization due to the lower stiffness of PEO blocks. However, with the crystallization progressing, the PEO crystals become harder than the amorphous PS. Consequently, the confinement provided by PS blocks shifts from hard to soft, with respect to PEO crystallization process.
In addition to the apparent confinement effects, the confinement systems are even more remarkable if utilizing the system as a tool to study some extreme or inaccessible states on experimental timescale. Upon the rapid aging induced by soft confinement, Boucher, Perez-De-Eulate and coworkers found evidence for existence of the two mechanisms of physical aging equilibrium that are impossible to achieve in such short time for bulk materials. [36,37,61] Moreover, Boucher and coworker found that the fictive temperature of PS reached the predicted Kauzmann temperature (T K ) in two days under soft confinement as demonstrated in Fig. 9. [37] The results are significant since the experimental evidence of the thermodynamic (ideal) glass transition is unprecedented to our knowledge, as more than millions of years are necessary to reach T K for bulk materials.

CONCLUSIONS
This review highlights classic and novel methods for making soft or hand confinements in 1D, 2D and 3D systems. The polymers have substantially different properties under confinement, compared to the bulk materials. The effect of soft and hard confinements on four major themes, glass transition temperature, physical aging, mechanical properties, and crystallization are reviewed in details. Most of the existing works show good consistency in the results of suppressed T g under soft confinement, while the hard confinement effect on T g needs further efforts to attain scientific insight and systematic conclusion due to the complexity induced by various interactions between the materials. The confinement effects on physical aging do not have consistent results, partially due to the fact that the definitions of aging rate are not the same in different works. Soft confinement effects on stiffness are in controversy just as on strength due to multiplex measure methods. In contrast, the results of effect of hard confinement are identical. Whereas it is indicated that hard confinement induces homogenous nucleation and reduced T c and T m , the effect of hard confinement on crystallization rate still causes dispute. Fortunately, crystal orientation of polymers in AAO has been well depicted by mature mechanism, depending on pore size of AAO and cooling rate. Soft confinement effect on crystallization, including nucleation and crystallization kinetics, principally relies on confinement microdomain. Although the mechanism of confinement effect on segmental motion has not been elucidated clearly, confinement systems are useful in studying the glass transition or physical aging. Comparing to the bulk, it is easier to make polymeric materials with confinement reaching some extreme conditions in short time scale due to the accelerated aging.