Solid-State Electromechanical Smart Material Actuators for Pumps—A Review

Abstract

Solid-state electromechanical smart material actuators are versatile as they permit diverse shapes and designs and can exhibit different actuation modes. An important advantage of these actuators compared to conventional ones is that they can be easily miniaturized to a sub-millimeter scale. In recent years, there has been a great surge in novel liquid pumps operated by these smart material actuators. These devices create opportunities for applications in fields ranging from aerospace and robotics to the biomedical and drug delivery industries. Although these have mainly been prototypes, a few products have already entered the market. To assist in the further development of this research track, we provide a taxonomy of the electromechanical smart material actuators available, and subsequently focus on the ones that have been utilized for operating pumps. The latter includes unidirectional shape memory alloy-, piezoelectric ceramic-, ferroelectric polymer-, dielectric elastomer-, ionic polymer metal composite- and conducting polymer-based actuators. Their properties are reviewed in the context of engineering pumps and summarized in comprehensive tables. Given the diverse requirements of pumps, these varied smart materials and their actuators offer exciting possibilities for designing and constructing devices for a wide array of applications.

1. Introduction

Mechanical—conventional and smart material—actuators are power transducers that transform input energy into output mechanical work and, therefore, can also move and/or control the movement of structures. Conventional mechanical actuators include fluidic, hydraulic, magnetic, as well as electromagnetic actuators and electric motors [1]. Smart material mechanical actuators are emerging actuators that are either based on pure smart materials or composites. These smart materials, also termed as ‘functional’, ‘intelligent’, ‘adaptive’, or ‘active’, respond to environmental stimuli with specific shape alteration modes within certain time frames [2]. Smart material mechanical actuators can be solids, liquids, or gels, and can be excited electrically, optically, thermally, chemically, magnetically and so on [3,4]. In a plethora of areas in which these actuators have been applied, there has been a significant amount of research conducted into liquid pumps operated by solid-state electroactive mechanical smart material actuators since the first such pump was published in 1975 by Thomas and Bessman [5], creating opportunities for applications ranging from aerospace and robotics to biomedical and drug delivery technologies. Electrically stimulated transducers are attractive due to the accessibility of electrical power and the advances in control systems that operate on electrical signals. In this paper, we focus on solid-state electromechanical actuators that have been incorporated into liquid pumps as driving mechanisms [2].

To the best of our knowledge, this is the first review to focus on solid-state electromechanical smart material actuators for driving liquid pumps, which focuses on describing the motion mechanisms of these materials for this application. Considering the breadth of the field, our focus was on the actuator side of the topic, especially since there are comprehensive reviews already available in the literature that discuss the properties of pumps created from electroactive polymers [6,7,8] as well as their biomedical applications [9]. This work is also a continuation of our previous work [8], which concentrated on the characteristics and performances of the pumps themselves. The pumps that have been operated with these smart material actuators have mainly either been diaphragm or linear peristaltic pumps (Figure 1), where the stroke volume ΔV is the difference between the maximum volume Vmax and the minimum volume Vmin of the pumping chamber, and its dead volume V₀ is the minimum volume of fluid contained between the inlet and outlet during a pumping cycle.

Solid-state electroactive mechanical smart material actuators have several advantages over conventional ones. First, they can be easily upscaled or downscaled, with the latter being particularly beneficial since miniaturizing conventional actuators has limitations. Secondly, they can also provide different actuation modes depending on their geometries and designs. These actuation modes can either have one, two or multiple degrees-of-freedom (DoF) and can exhibit unidirectional or bi-directional bending, twisting, buckling, expansion, contraction, elongation, shortening or even a combination of these form alterations in an operating cycle. When actuators respond unidirectionally to an applied voltage, deflection occurs only in one direction. However, when actuators respond bi-directionally, deflections in different directions are achieved by changing the polarity of the applied voltage. Finally, smart material actuators can be conformed in numerous geometries, such as sheets or membranes, plates, rods, tubes, rings, wires, springs, etc., while designs can range from single-layered to multi-layered structures. Multi-layered structures include stacked single-layered actuators, which can be either uni-morphs or bimorphs. Uni-morphs constitute bilayers where a smart material layer is combined with two passive material membranes, while bimorphs are tri-layers that consist of an inert layer and two smart material layers. Both are cantilever actuators that induce single-DoF bending upon actuation [10]. Pumps, which are comprised of distinct chambers that have diaphragms, typically require actuators conformed into sheets, uni-morphs or bimorphs. Plates, rods, wires and springs can be particularly beneficial when applying forces to deform pump chambers and create peristalsis, whereas actuators conformed to tubes and rings, can potentially act as both, the pump chamber as well as the actuator. Some of the principal designs for applying actuators with unidirectional expansion or bi-directional expansion/contraction, bending and buckling, either integrated into the pump diaphragm or as external transducers, are depicted in Figure 2.

The aim of the present paper is twofold. Firstly, to apprehend the smart materials that have been considered for operating mechanisms in liquid pumps and to comprehend the several opportunities that such actuators can provide. Secondly, to set the basis on which further research into smart material pumps can be designed and built.

The current paper is structured as follows: Section 2 contains a taxonomy of the available solid-state electromechanical smart material actuators and discusses which have been utilized for pumping. These include unidirectional shape memory alloy (U-SMA)-, piezoelectric ceramic (PEC)-, ferroelectric polymer (FEP)-, dielectric elastomer (DE)-, ionic polymer metal composite (IPMC)- and conducting polymer (CP)-based actuators. Section 3 presents a literature review of the smart materials on which the aforementioned actuators are based on, focusing on their working principles, their properties and limitations, as well as current efforts for material improvements. Typical actuators are also presented. Representations of smart material operating principles and illustrations of their most typical actuator configurations and actuation modes, considering their (potential) applications in membrane and/or linear peristaltic pumps, are included. All actuation modes in this paper are single- or two-DoF unless specified otherwise. Properties like their work density, typical and maximum achievable actuation strain and stress, the required driving voltage or field, bandwidth and cycle life are important when applying smart material actuators in applications, and are thus included in tables for each smart material. Key material properties, such as elastic moduli and, depending on the smart material, e.g., electrical resistivities, phase transition temperatures, dielectric constants, etc. are also given. As the variability of IPMCs and CPs, which are soft electroactive polymers (EAPs), is high, these are presented as general categories [11]. Next, in Section 4, a discussion of the actuating technologies in the context of engineering pumps that are operated by smart material actuators is included. Finally, the paper closes with conclusions in Section 5.

2. Taxonomy of Solid-State Electromechanical Smart Material Actuators

A taxonomy of the solid-state electromechanical smart material actuators that have been reported in the literature is presented in Figure 3.

On a first level (Figure 3, Level 1), solid-state electromechanical smart materials can initially be broadly categorized according to their characteristic actuation mechanisms into those that are based on shape memory alloys (SMAs), piezoelectrics and on two types of EAPs, namely, electronic EAPs and ionic EAPs. A substantial amount of research has recently been conducted on all these smart materials since they can potentially mimic the properties and/or functions of human muscles and, hence, are also termed as ‘artificial muscles’ [12,13,14,15].

The aforementioned smart materials can either be heat, i.e., SMAs, charge, i.e., ionic EAPs or electric field driven, i.e., piezoelectrics and electronic EAPs. The magnitude of the voltage applied across these electric field-driven actuators can be tweaked either by altering the thickness of the actuators and/or by adjusting their dielectric constants [16,17] while actuation fields are limited by the dielectric breakdown strength of their dielectric constituents.

The actuation mechanism of SMAs entails a phase transition [18]. They are actuated due to thermal energy diffusion [19], but as they are conductive and have a high electric resistance, they can be heated by applying an electric current (joule heating) [20] with a voltage typically < 10 V. SMA actuators can further be classified into U-SMAs and bi-directional SMAs in accordance with their actuation modes [21,22,23,24,25,26].

Actuators based on piezoelectrics change shape from the orientation of electric dipoles within the smart materials in response to the applied electric fields that are in the order of 100 MV/m. They can be segregated into inorganic piezoelectrics, PECs, organic ones, piezoelectric polymers (PEPs) and PEP composites [27]. According to their structure, PEPs can be divided into amorphous PEPs and semi-crystalline PEPs (Figure 3, Level 2). The latter are comprised of electroactive FEPs (some types of FEPs can also be actuated by light or heat [28]) and liquid crystal elastomers (LCEs) (most types of LCEs are photo- or thermo-actuated) [29].

Electronic EAPs actuate due to coulombic forces that are generated from the creation of electric fields. These actuators typically necessitate electric fields of 100 MV/m for excitation [16,17]. Electronic EAPs are subsequently segregated into DEs [30] and voided-charged polymers (also known as ferroelectrets or piezoelectrets) [27].

Actuation in ionic EAPs results from volume changes that occur due to the transportation/insertion of ions within the actuators during excitation. These ions can either be employed by liquid or solid-state electrolytes [16]. Since ions from the latter are relatively immobile compared to the solvated ions, and therefore, limit the actuation of ionic EAPs, we only focus on actuators that use liquid electrolytes. These are usually either aqueous or organic electrolytes (ions solvated with water or organic solvents) or room-temperature-ionic-liquid electrolytes. Liquid electrolytes enhance actuation responses mainly because solvated ions are more mobile and larger (ions plus solvation shell), resulting in accentuated volume changes and therefore electromechanical properties. Aqueous electrolytes are typically more ionically conductive (e.g., 73 S/m for 1 molar sulfuric acid) compared to room-temperature-ionic-liquids (e.g., 1.15 S/m for 1-ethyl-3-methylimidazolium tetrafluoroborate), but the large ionic radii of room-temperature-ionic-liquids can produce large peak strains and stresses [12]. Ionic EAP transducers require the application of low voltages ≤ 10 V [31] for excitation. The magnitude of the voltage required depends on the conductivity of the components employed and the actuators’ thickness. Values are limited by the electrochemical stability potential window of the electrolytes, beyond which they degrade [13]. For example, the potential window for water is 1.23 V, above which electrolysis transpires. The electrochemical stability potential window of room-temperature-ionic-liquids is wider than that of water, and therefore, these electrolytes provide a more stable operation compared to when using aqueous ones, but only when dehydrated and encapsulated, as charged systems tend to always be hygroscopic. Ionic EAPs can be divided into IPMCs, CPs and high-specific surface-area carbon actuators. The latter are composites with electrodes that can either be based on carbon nanotubes (CNTs) [32], graphene [33], carbide-derived carbon (C) (also known as curved graphene) [34], C aerogel [35], or C black [36]. When operating within the electrochemical stability potential window of their electrolytes, IPMCs and high-specific surface area carbon actuators exhibit double-layer capacitive actuation, whereas CPs provide actuation due to redox reactions. Finally, the electroactive properties of ionic EAP can typically be modulated by the choice of membrane, electrode thickness, electrolyte/s and, if tri-layers, by whether or not their ion-conducting polymer layer contains a reinforcement.

On a last note, it should be mentioned that there has also been research conducted on combinations of smart materials from Figure 3 to create actuators. Moreover, it is also worth noting that although there are more smart material categories, e.g., shape memory polymers, shape memory ceramics [18], etc., these are not included in the taxonomy as their applicability in/as actuators has not been demonstrated yet.

As U-SMA-, PEC-, FEP-, DE-, IPMC-, as well as CP-based actuators are the only solid-state electromechanical smart material actuators that have been studied for applications in operating pumps [8], they therefore constitute the scope of our paper. From these, U-SMAs are pure materials, whereas PECs, FEPs, DEs and IPMCs are composites. In principle, CPs can either be used as pure materials or smart material composites.

3. Electromechanical Smart Material Actuators

3.1. Actuators Based on Unidirectional Shape Memory Alloys

3.1.1. Materials, Actuators and Working Principles

Unidirectional shape memory alloys (U-SMAs) were discovered in the 1950s [37]. Typical U-SMAs for actuators have been nickel (Ni)-titanium (Ti) alloys (also known as nitinols), as they exhibit material as well as electromechanical properties superior to other U-SMAs. Less expensive, but with additionally lower stability than NiTis are copper (Cu)-based (mainly Cu-aluminium (Al)-Ni, Cu-zinc (Zn)-Al) and silver (Ag)-based (e.g., Ag-manganese (Mn)-silicon (Si)) alloys [18].

Upon actuation, U-SMAs exhibit unidirectional contraction, where they present a catch-state. At temperatures below their austenite phase transformation temperature, U-SMAs constitute twinned martensites that can be easily deformed mechanically through de-twinning. When voltage is applied and the U-SMAs are heated above their austenite phase transformation temperature, they transition to a high-temperature austenite phase before returning to their memorized, fully contracted, shapes [19,20]. When voltage is removed and U-SMAs subsequently cool down, their initial low-temperature twinned martensitic phase is recovered (Figure 4).

U-SMA actuators can be manufactured as bulks, typically forming plates, sheets, rods of circular cross-sections and wires (Figure 5a). Other geometries can also be configured by deforming and then annealing the U-SMAs at certain high temperatures to reset their memorized shapes, for example, springs [19]. NiTi spring actuators are common [38,39,40,41] as they can be designed according to force and displacement requirements [19]. What is more, external biasing elements can be coupled to U-SMA bulk actuators to induce bi-directional motion. Common examples are spring actuators that can be combined with dead-weights, passive biasing springs or additional U-SMAs in an antagonistic design, which is the most common configuration [42,43] to acquire bi-directional contractions [41] (Figure 5b). In the last two configurations, mechanical work output is obtained from in-between the two springs [27]. It has also been shown that a bistable piston pump can be constructed using U-SMA wires and restoring springs [44,45]. Finally, U-SMA thin films can be bonded to a passive metal or polymer layer to obtain uni-morphs that exhibit bi-directional one-way bending [46,47] (Figure 5c).

3.1.2. Properties of Unidirectional Shape Memory Alloys

Key material and electromechanical properties of TiNi, CuZnAl and CuAlNi are shown in

Table 1. U-SMA (NiTi, CuZnAl and CuAlNi) material and electromechanical properties.

	Smart	NiTi	CuZnAl	CuAlNi
Materials Properties
Young’s modulus (GPa)		28–83 [56]	70–100 [56]	80–100 [56]
Resistivity (106 Ωm)		0.5–1.1 [56]	0.07–0.12 [56]	0.1–0.14 [56]
Phase transformation temperature (austenite finish temperature Af) (°C)		200 [56]	200 [56]	200 [56]
Typical actuation strain, ε (%)		4% [31]
Maximum actuation strain, εmax (%) (1st actuation cycle)		6–8 [56]	4–6 [56]	5–6 [56]
Typical actuation stress, σ (MPa)		100–130 [56]	40 [56]	70 [56]
Maximum actuation stress, σmax (MPa)		500–900 [56]	400–700 [56]	300–600 [56]
Driving voltage (V)		5 [31]
Bandwidth (Hz)		milliseconds (active cooling)–minutes (passive cooling) [51,52,53,57]
Cycle life		300 @ ε = 5%–>105 @ ε = 0.5% [58]	>104 [56]	>5 × 103 [56]

. U-SMAs exhibit extremely large energy densities per cycle (10 MJ/m³) [48]. U-SMA wires, in a straight form, can contract up to ε = 8% in length. When wires are conformed into springs, strains of ε > 100% can be achieved, albeit with lower stresses [12]. The stresses of U-SMAs are large, with, e.g., NiTis generating σ = 130 MPa. Low voltages, in the range of 5 V [49] are required for U-SMA thermal actuation [20]. By using very large brief current pulses, >10⁹ A/m², contraction times can be reduced from minutes to several milliseconds since greater electrical power is delivered for the phase transformation [31,50,51]. By employing rapid, larger current pulses, heat dissipation is also minimized, leading to higher actuation efficiencies. Frequency bandwidths of U-SMAs can be limited when high-temperature austenite phases are restored to low-temperature martensite phases during complete operating cycles through passive cooling [38,52,53]. Cooling times can be reduced to the range of milliseconds through active cooling by using heat transfer mediums such as gases, liquids or Al heat sinks [51,54]. Actuators based on U-SMA films can dissipate heat relatively quickly due to their high surface area and have been reported to operate at frequencies up to 100 Hz [46]. A limitation of U-SMAs is the potential fatigue upon cycling, which results in diminishing strains and stresses with increasing cycle count [55]. Moreover, the cycle life of NiTi actuators at high strains can be limited, from 10⁵ cycles at strains of ε = 0.5% to a few hundred cycles at ε = 5% as their shape memory effect degrades significantly [48].

3.2. Actuators Based on Piezoelectric Ceramics

3.2.1. Materials, Actuators and Working Principles

Piezoelectric ceramics (PECs) were discovered in the 1940s [59], although the direct piezoelectric effect was detected in the 1800s by Pierre and Paul Jacques Curie [58], according to which an electric field is induced in a piezoelectric material in response to the applied stress. The piezoelectric strain coefficient d describes the relationship between the charge induced and the applied stress [60]. PECs are non-centrosymmetric crystals with unit cells that contain dipole moments due to the arrangement of the ions. These unit cells form distinct regions that have similar dipole moments, but as these domains are arranged randomly, there is no net polarization within the material. By applying a polarizing field to the material, all the dipoles can be aligned, and a net polarization can thus be achieved [61]. Aluminum nitride (AlN) and zinc oxide (ZnO) crystals are common smart materials, with lead zirconate titanates (PZTs) and the recently developed lead magnesium niobate-lead titanates (PMN-PTs) characterized by exceptional piezoelectric properties [62]. The most common electrodes applied to PECs are Ag, Ni-gold (Au), chromium (Cr)-Au and Ni- vanadium (V) [60].

When an electric field is formed across a PEC parallel to the electric dipole moments, actuation (termed the indirect piezoelectric effect [1]) transpires, which will remain after the voltage is removed. PECs exhibit bi-directional actuation. Specifically, material expansion occurs when the direction of the electric field is opposite to the electric dipole moments and contraction takes place when the reverse field is created [63] (Figure 6).

PEC actuators can be electroded with single crystals formed as rods (rectangular or circular cross-sectioned), disks, rings, plates or sheets that can either exhibit bi-directional expansion or contraction, depending on the polarity of the electric field and the dipole moments [60,64] (Figure 7a). To significantly decrease actuation voltages [1], PEC actuators can be designed as multi-layered structures comprised of electroded PEC films (thinner than 100 μm) stacked with alternate strips polarized in opposite directions. These bi-directionally contract or expand upon actuation [64,65] (Figure 7b). Actuation displacements of PECs can be amplified in bimorph conformations [12]. PZT bimorphs with both membranes oriented to have the same polarity can either bend bi-directionally due to the expansion of one strip and contraction of the other, or elongate or shorten bi-directionally due to the same deformation mode of both strips, depending on how actuation voltages are applied [64] (Figure 7c). As indicated in Figure 7d, uni-morphs comprised of a PEC and a passive layer can bend in both directions when excited [66]. THUNDER (thin layer composite uni-morph ferroelectric driver and sensor) [67,68] and RAINBOW (reduced and internally biased oxide wafer) [69] actuators are both initially bent uni-morphs that either further buckle or flatten out depending on the polarity of the field applied. THUNDERs are fabricated by bonding a pre-stressed PEC sheet, electroded on one side, to a stainless-steel substrate on the other. RAINBOW actuators are comprised of a PEC layer that is reduced on one side and electroded on both sides. As RAINBOW actuators are monoliths, they present superior properties to THUNDERs. Improvements of the THUNDER actuators such as the LIPCA (or lightweight piezo-composite curved actuator) [70] have been reported, where the metallic membranes have been replaced by fiber-reinforced epoxy layers, which reduce the actuator’s weight by up to 40% and increase its induced displacements by up to 60%.

3.2.2. Properties of Piezoelectric Ceramics

Material and electromechanical properties of PMN-PT and PZT-5H are presented in Table 2.

PEC single crystals achieve high power densities in the range of 1 kJ/m³ [71]. The induced stresses are large, in the range of σ = 100 MPa [1], while the strains are typically extremely small, in the magnitude of ε = 0.1% (generally lower than 100 μm) [1]. PECs typically require large excitation fields in the range of MV/m [12]. Moreover, they typically have very large bandwidth (up to 10 MHz resonance frequency) [13] and long cycle lifetimes over 10⁶ cycles [72].

Table 2. PEC (PMN-PT, PZT-5H) material and electromechanical properties.

	Smart Material (Single Crystal)	PMN-PT	PZT-5H
Properties
Young’s modulus, Y (GPa)		27 [72]	37 [72]
Dielectric constant		4700 [72]	3300 [72]
Normal longitudinal piezoelectric strain coefficient, d33 (pC/N)		2000–3000 [61]	593 [61]
Typical actuation strain, ε (%)		0.6 [12]	0.2 [73]
Typical actuation stress, σ (MPa)		100 [12]	108 [73]
Driving electric field (MV/m)		100 [12]
Bandwidth (Hz)		≤10 MHz (resonance frequency) [13]
Cycle life		106 [72]

Table 2. PEC (PMN-PT, PZT-5H) material and electromechanical properties.

	Smart Material (Single Crystal)	PMN-PT	PZT-5H
Properties
Young’s modulus, Y (GPa)		27 [72]	37 [72]
Dielectric constant		4700 [72]	3300 [72]
Normal longitudinal piezoelectric strain coefficient, d33 (pC/N)		2000–3000 [61]	593 [61]
Typical actuation strain, ε (%)		0.6 [12]	0.2 [73]
Typical actuation stress, σ (MPa)		100 [12]	108 [73]
Driving electric field (MV/m)		100 [12]
Bandwidth (Hz)		≤10 MHz (resonance frequency) [13]
Cycle life		106 [72]

3.3. Actuators Based on Ferroelectric Polymers

3.3.1. Materials, Actuators and Working Principles

Research on ferroelectric polymers (FEPs) began in 1969 when Kawai started investigating polyvinylidene fluoride (PVDF or PVF₂) [74]. Ferroelectricity is a type of piezoelectricity [62]. The most common FEPs are β-phase PVDF homopolymers and copolymers [75,76,77], which have the largest piezoelectric coefficients d of all bulk PEPs [62] due to the high electronegativity of the fluorine (F) atoms on their polymer backbones [77,78]. These polymers are semi-crystalline, containing of -CH₂CF₂- repeating units on their polymer backbones that are arranged in crystals, hence, further organizing the electric dipole moments in parallel [62,78].

There are many types of β-phase PVDFs that have been fabricated using different methods [79]. The most typical include pre-strained PVDF (also known as relaxor PVDF), irradiated PVDF and copolymers of PVDF with either small mass fractions of larger monomers such as chlorofluoroethylene (<10%) [77] or trifluoroethylene (TrFE) [80]. FEPs fabricated by combining these techniques have also been reported in the literature, with irradiated P(VDF-TrFE) possessing the best electromechanical properties of all FEPs [81,82,83]. Currently, research is being conducted into PVDF—C black [84], CNTs [85] or piezoelectric inorganic particles, such as barium titanate (BaTiO₃) [86], PZT [87] or ZnO [88], to enhance piezoelectricity [87]. Typical electrodes are made of Al, Cu or Au [75,76,89] and CP electrodes have also recently been studied. Although the latter are more compliant, they are however less conductive than traditional electrodes, which has resulted in actuators with lower-induced deformations [90,91].

β-phase PVDF polymers exhibit bi-directional actuation. They contract when the direction of the electric field formed is opposite to the electric dipole moments and expand when the reverse occurs. This is probably due to the alignment of their polarized domains with the electric field; however, the mechanism is still under discussion [28,89] (Figure 8). β-phase PVDF polymers do not present catch-states [48].

β-phase PVDFs are usually manufactured as membranes [76,89]. They can be single layer actuators that expand or contract bi-directionally, depending on the polarity of the voltage applied and the dipole moments of the polarized domains (Figure 9a). They have also been rolled into tubular actuators of circular cross-section to elongate bi-directionally [92] (Figure 9b). Most typically, PVDF membranes have been incorporated into uni-morphs (Figure 9c) or bi-morphs (Figure 9d) [90]. The FEPs in these uni-morphs and bi-morphs have either constituted single or multiple stacked electroded membranes that are bonded to a metal or polymer passive part [73,82,83,93,94,95]. Finally, FEPs have also been conformed to bellows actuators by folding β-phase PVDF-based bimorphs in two, resulting in an elliptical conformation that either further buckled or flattened upon actuation [96] (Figure 9e).

3.3.2. Properties of Ferroelectric Polymers

Material and electromechanical properties of irradiated P(VDF-TrFE) are exhibited in

Table 3. FEP (irradiated P(VDF-TrFE)) material and electromechanical properties.

	Smart Material	Irradiated P (VDF-TrFE)
Properties
Young’s modulus, Y (GPa) (depending on composition and level of irradiation)		0.3–1.2 [48]
Dielectric constant		55 [48]
Dielectric breakdown (MV/m)		200 [2,13]
Normal longitudinal piezoelectric strain coefficient, d33 (pC/N)		−33 [73]
Typical actuation strain, ε (%)		2 [73]
Maximum actuation strain, εmax (%)		7 [73]
Typical actuation stress, σ (MPa)		20 [13]
Maximum actuation stress, σmax (MPa)		40 [83,93,97]
Driving electric field (MV/m)		100 [13]
Bandwidth (Hz)		100 [13]
Cycle life		≤5 × 107 [48]

. β-phase PVDFs have extremely large energy densities (~1 MJ/m³) [13]. Moreover, they can exhibit moderate strains up to ε = 7% actuated at DC fields of approximately 150 MV/m and ε = 2% at low-frequency AC fields of 200 MV/m [73]. Stresses attained can range σ = 20 [13]–40 MPa [83,93,97]. Typically, β-phase PVDFs can be actuated at fields of ~100 MV/m, as they present dielectric breakdown at fields of 200 MV/m [2,13]. Bandwidths are high, reaching up to 100 Hz [13] and cycle lives can reach up to 5 × 10⁷ [48]. The main limitation is that the process of transforming PVDF into β-phase PVDFs is complex, resulting in reproducibility issues [76].

3.4. Actuators Based on Dielectric Elastomers

3.4.1. Materials, Actuators and Working Principles

Dielectric elastomers (DEs), smart material tri-layers that constitute electroded elastomeric membranes, were invented in the 1990s. Polydimethylsiloxane (PDMS) silicones [98] and very high bond (VHB) acrylic [99,100,101,102,103,104,105] have been the most investigated elastomers. Dielectric membranes are commonly pre-strained (100%) to align the elastomers’ dielectric dipoles [106], and thus, obtain optimal actuation properties [107]. Alternatively, dielectric membranes can be loaded with piezoelectric particles, e.g., titanium dioxide (TiO₂), to increase their dielectric constant [108]. Conventional electrodes for DE actuators are thin metal layers, e.g., Au, Ag grease, graphite or carbon (C)-based films (e.g., C powder in oil or grease, or C—elastomeric matrix composites). Novel electrodes in research are metal salts—elastomer composites that are also intrinsically patternable, CNT electrodes that are additionally self-clearing, and electrodes based on implanting metallic nano-clusters in the elastomeric membrane [109].

When a potential is applied across the electrodes of a DE actuator, coulombic forces between the two oppositely charged electrodes generate Maxwell compressive stresses within the elastomer, resulting in uni-directional in-plane expansion of the smart material [3,13,110]. Once the voltage difference is removed, DEs return to their initial conformation (Figure 10).

DE actuators are typically based on un-strained or pre-strained films. When using the latter, the membranes should be fixed with axial constraints. DEs can be comprised of a single film or multiple films that are stacked or rolled together [111] to increase actuation stresses, since the actuation forces are in parallel [13]. Single-layered actuators or thin stacked actuators comprised of a small number of DE layers in parallel, can either initially be flat or molded into domes. Both can induce either out-of-plane buckling when supported perimetrically by rigid frames [112] or provide in-plane elongation when clamped at one end [105] (Figure 11a). Thick stacked actuators that consist of many unstrained DEs have been reported and used for the contraction they exhibit upon excitation in the direction perpendicular to their expansion [113,114] (Figure 11b). DEs can also be molded into tubes or one or more DE films can be rolled around an axis to form tubular actuators of circular cross-sections, which can either elongate or expand when actuated depending on whether they are fixed on one or both ends [108,115,116,117,118,119,120] (Figure 11c). Finally, DEs can be incorporated into uni-morphs [121,122,123] and bimorphs [124]. The former is comprised of a DE tri-layer connected to a flexible passive substrate, and the latter consists of two DEs bonded to a flexible layer. DE uni-morphs exhibit bi-directional one-way bending, and DE bimorphs display two-way bending upon actuation (Figure 11d).

3.4.2. Properties of Dielectric Elastomers

Table 4. DE (silicone-based and VHB acrylic-based) material and electromechanical properties.

	Smart Material	Silicone-Based	VHB Acrylic-Based
Properties
Young’s modulus, Y (MPa) (depending on fabrication process and strain)		0.1–1 [48]	1–3 [48]
Dielectric constant		∼3 @ 1 kHz [48]	∼4.8 @ 1 kHz [48]
Dielectric breakdown (MV/m)		200 [2,13]
Typical actuation strain, ε (%)		10–30% [125]
Maximum actuation strain, εmax (%)		120 [48]	380 [48]
Typical actuation stress, σ (MPa)		0.3 [48]	1.6 [48]
Maximum actuation stress, σmax (MPa)		3.2 [48]	7.7 [48]
Driving electric field (MV/m)		110–350 [48]	125–440 [48]
Bandwidth (Hz)		1400 [126]	10 [126]
Cycle life		107 @ ε = 5% (no failures observed)–106 @ ε = 10% [48]	107 @ ε = 5% (no failures observed)–106 @ ε = 50% [48]

presents the material and electromechanical properties of silicone-based and VHB acrylic-based DE actuators. The energy densities of DEs are extremely high (3.4 MJ/m³) due to the large strains they produce [13]. These are typically ε = 10–30% [125], however, strains up to ε = 120% have been reported for silicone-based DEs and ε = 380% for VHB acrylic-based DEs [126,127]. Special designs of DEs have exhibited extremely high reversible strains, in the order of ε = 1165%, through the ‘snap through’ phenomenon, in which they are subject to constant pressures during actuation [102,128,129,130,131,132,133,134]. The DE actuation stresses are small. Silicone-based DEs typically attain stresses of σ = 0.3 MPa and maximum stresses of σ = 3.2 MPa, whereas VHB acrylic-based DEs produce σ = 1.6 MPa and σ = 7.7 MPa, respectively [48]. Large fields in the range of 100 MV/m are required for actuation [48]. These are close to the dielectric breakdown strength of the polymers [2].

Extremely high bandwidths can be attained with silicone (1400 Hz is typical), whereas 10 Hz is typical for VHB-acrylic DEs [126]. The main limitation of DEs is their failure, either due to the dielectric breakdown of the polymer caused by the large driving fields [135] or due to the electrode cracking. Humidity also affects actuation [136]. By pre-straining the polymer films prior to fabrication, DE thicknesses are decreased, which reduces the actuation voltages required to attain the necessary electrical fields for excitation [13]. This strategy, however, can also result in elastomer relaxation and, subsequently, material failure [13].

3.5. Actuators Based on Ionic Polymer Metal Composites

3.5.1. Materials, Actuators and Working Principles

Ionic polymer metal composites (IPMCs), tri-layers that consist of an ionically conductive polymer containing mobile solvated counter-ions (co-ions), and two metal electrodes on either side, were developed in the 1930s and have been researched thoroughly since 1992 [137,138,139]. These ionically conductive polymers are usually ionomers; polymeric membranes with ionic (cationic or anionic) groups covalently linked to their polymer backbones. Ionomers most commonly used for fabricating IPMCs are Nafion and Flemion, which contain side chains with sulfonate (RSO₃⁻) and carboxylate (RCOO⁻) ionic groups attached to the main chain, respectively [140]. Co-ions employed by liquid electrolytes balance the charge of the ionic groups within the IPMCs’ ionomer and are required for actuation [141,142,143,144]. Co-ions can either be mobile cations or anions, depending on the type of the ionomer employed. Finally, since IPMCs most commonly contain liquid electrolytes, noble metal electrodes are typical, e.g., platinum (Pt), palladium (Pd), Au, etc., as they are non-redox active under these conditions [137,145]. Less stable metal electrodes like Cu, Ag, etc., have also been researched, to a lesser extent, however [145].

When voltage is applied to Nafion or Flemion-based IPMCs with noble metal electrodes operated with aqueous electrolytes, the most common medium for operating IPMCs, mobile cations migrate together with the solvating water molecules towards the cathode. Increased volumes at the negative electrode induce ionomer swelling, and therefore, IPMC bending can be reversed [146] (Figure 12). IPMCs do not present a catch-state.

IPMCs have typically been based on thin, up to 0.25 mm-thick single membranes of ionomers or on thicker layers produced by hot-pressing two or more thin layers together to increase actuator force capabilities [147]. Thicker or different shapes have also been fabricated by 3D-printing [148] or by fusing precursor granules in molds of the desired thickness or shape [142]. Cantilevers have been the most common type of actuator, inducing bi-directional, two-way bending (Figure 13a) [149,150]. Either flat or as domes, they have also been fixed perimetrically to induce bi-directional, two-way buckling [151] (Figure 13b). Rings that have been created from two attached IPMC actuators have exhibited two-way expansion and contraction when both actuators are actuated simultaneously [148] (Figure 13c). Finally, rectangular rods have also been reported to induce multi-DoF bending when voltages are applied to two or four external faces [152] (Figure 13d).

3.5.2. Properties of Ionic Polymer Metal Composites

Table 5. IPMC material and electromechanical properties.

	Smart Material	IPMC with Noble Metal Electrodes
Electro-Mechanical Properties
Young’s modulus, Y (MPa)		0.1 [48]
Typical actuation strain, ε (%)		0.5 [48]
Maximum actuation strain, εmax (%)		>10 [153,154]
Typical actuation stress, σ (MPa)		3 [155]
Maximum actuation stress, σmax (MPa)		30 [82,156]
Driving voltage (V)		1–5 [153,157]
Bandwidth (Hz)		100 [153,157]
Cycle life		106 @ ε = extremely small [157]

presents the material and electromechanical properties of IPMCs with noble metal electrodes. IPMCs can exhibit maximum energy densities of 5.5 kJ/m³ per cycle. Actuation strains are small, ε = 0.5% [48] and can reach up to ε > 10% [153,154] depending on the ionomer, electrolyte, as well as the design/quality of the electrodes [48]. They also exhibit small-induced stresses of σ = 3 MPa [155] that can reach up to σ = 30–40 MPa [82,156].

IPMCs require low voltages of 1–5 V for actuation. The voltage limits of IPMCs employing aqueous electrolytes must not exceed the stability window of water. To avoid evaporation of the solvent and ensure stable response, IPMCs should either be encapsulated or employ room-temperature ionic liquids as the electrolytes. IPMCs can reach high 100 Hz operating bandwidths at low strains [153,157]. Finally, IPMC actuators have the potential to consecutively vibrate more than 10⁶ times at extremely low strains [157]. IPMCs operating in these regimes have, however, limited practicality.

The main limitation of IPMCs is that they exhibit non-linear, time and history-variant properties due to liquid back-flux with time [141]. Moreover, their metallic electrodes have been investigated as they crack, especially at larger strains, from the deformation accompanying electromechanical actuation, resulting in individual metallic islands on the surface of the membrane [158].

3.6. Actuators Based on Conducting Polymers

3.6.1. Materials, Actuators and Working Principles

Research on intrinsically conducting polymers (CPs), conjugated polymers with alternating single and double bonds on their polymer backbones [77], started in 1834 [159]. They are semi-conductors that can become conductors through chemical or electrochemical doping (extrinsically CPs), in which the ions also act as charge carriers along with the electrons. These smart materials are, therefore, electronically as well as ionically conductive. Examples include poly(3,4-ethylenedioxythiophene) (PEDOT) [160], polyacetylene (PA), polypyrrole (PPy), polythiophene (PTh), polyaniline (PANI), poly(p-phenylene) (PPP), polyfluorene (PF), etc., and derivates made from their monomers with added functional groups to alter their properties [10,161,162,163,164]. PEDOT is used as/in CP actuators due to its high conductivity and chemical stability, whereas PPy and PANI are common due to their chemical stability and the induced strains [10]. Solvated mobile co-ions are usually provided from aqueous electrolytes or room-temperature-ionic-liquids when oxidizing or reducing the CPs to balance the charge of the polymeric chains, resulting in actuation [77]. Conjugated polymers, such as PPy, can be p-doped. As the polymer backbones are oxidized, the CPs are combined with the anions from an electrolyte (anion-exchanging CPs). Other conjugated polymers, such as PTh, can be n-doped, with their polymer chains being reduced. These CPs are thus intercalated by the cations of a liquid electrolyte (cation-exchanging CPs) [10,12].

When voltages are applied to CPs (oxidized or reduced conjugated polymer—electrolyte systems) conformed to free-standing electrodes (as represented in Figure 14), they are oxidized or reduced and the solvated co-ions from the electrolyte are inserted/ejected to/from the CPs’ structure to neutralize the charge, thus creating volumetric changes. CPs therefore exhibit bi-directional operation in which they expand and contract [12,165]. Bending in CP-based tri-layers has been explained in [166] as a result of differential expansion of both electrodes in every cycle. CPs present continuous catch-states, as along the (redox) charge, deformation is also maintained.

CP actuators are typically film- or fiber-based [161] (Figure 15). Historically, the first types of CP-based actuators were free-standing electrodes that are, however, typically not used today. The most common conformations reported in the literature are uni-morphs (Figure 15a) and bimorphs (Figure 15b). These have either been a single or two anion- or cation-exchanging CP membranes combined with an inert ionically conductive polymer membrane saturated with a liquid electrolyte [167,168,169]. Both of these types of actuators exhibit bi-directional two-way bending [10]. Other examples of bimorphs have been anion- and cation-exchanging CP layers on either side of a porous membrane containing mobile solvated co-ions that induce bi-directional elongation and shortening upon actuation [170] (Figure 15b). Finally, tubular actuators have been comprised of a passive material with embedded CP-based fiber actuators to provide bending (Figure 15c). These fibers constituted CPs surrounded by an ionically conductive membrane—electrolyte and were further wrapped within a Cu sheath [171,172,173]. When axially fixed, these CP tubular actuators have been reported to bi-directionally expand and contract [174]. Finally, composite laminates have also been comprised of CPs with electrodes. These can either actuate linearly or they can bend, depending on the design [175,176,177].

3.6.2. Properties of Conducting Polymers

Key material and electromechanical properties of CP actuators are shown in

Table 6. Material and electromechanical properties of CPs.

	Smart Material	CP
Electro-Mechanical Properties
Young’s modulus, Y (GPa)		0.8–3 [48]
Conductivity (S/m)		10,000–45,000 [47]
Typical actuation strain, ε (%)		2–10 [77]
Maximum actuation strain, εmax (%)		40 [77]
Typical actuation stress, σ (MPa)		1 [77,178]
Maximum actuation stress, σmax (MPa)		35 [77,178]
Driving voltage (V)		≤1 [10]
Bandwidth (Hz)		1 Hz [10]
Cycle life		106 [48]

. CPs present typical energy densities of 0.1 MJ/m³ per cycle, which can reach up to 1 MJ/m³ [48]. Although large maximum strains of ε = 40% have recently been reported [77], the induced strains are typically low, in the range of ε = 2–10% [77]. CPs commonly have operating stresses of σ = 1–35 MPa [77,178], with induced stresses dependent on the thickness of the CP electrode layers of the actuators. Thinner layers produce lower stresses, and although thicker membranes bend slower, they produce higher stresses [161]. CPs require low voltages for actuation, typically ≤1 V [10]. Frequency bandwidths have been limited to the order of 1 Hz for films and fibers up to several μm thick since, after CPs switch between their oxidized and reduced states, volume changes depend on the mass transport of the co-ions of the used electrolyte. Higher frequencies up to 100 Hz or even kHz at low strains can be obtained by using thinner CPs, higher voltages, and incomplete electrochemical cycling to keep the polymer always in an electrically conducting state [10]. CPs have been shown to reach up to 10⁶ cycles [48]. CP actuators should be encapsulated to avoid solvent or ion exchange, redox reactions with some components of their environment [77,179]. Moreover, overoxidation and overreduction must be avoided to prevent irreversible processes leading to loss of conductivity [161].

4. Actuator Considerations in Pumps

As discussed in [6], most studies on pumps driven by smart material actuators have been conducted on PECs. Some have utilized U-SMA and DE actuators and a few have also used IPMC actuators. Limited work has been based on FEP and CP actuators. From these, some PEC- and SMA-operated pumps have entered the market, whereas the rest remain prototypes.

When engineering pumps, it is important to initially consider their operating environments (liquid, air or vacuum for space applications), although encapsulation can aid operation in varying conditions. Furthermore, if interface with living tissues and organisms is required, biocompatibility characteristics should also be examined. Regarding a smart material’s biocompatibility, first, the toxicity of its materials should be evaluated (whether they trigger an immunological response or not), which can be resolved via encapsulation. Secondly, the required excitation voltages should be reviewed. Electrolysis of water occurs in biological fluids not lower than 1.23 V that can result in dangerous by-products for organisms, and thus, actuators operated above 1.23 V should be encapsulated. It goes without saying that the higher the actuation voltages, the riskier the systems are due to the risk entailed if an encapsulation were to fail. Finally, when using high voltages/currents, thermal insulation should also be considered as the heat generated can affect the interfacing living tissues as well as the liquids pumped.

If pumps for drug delivery or other biomedical applications are considered, the TiNi U-SMAs are highly biocompatible and are therefore widely used. U-SMAs can work in dry as well as in wet environments and in vacuum [180,181,182]. U-SMA transition temperatures should either be engineered considering the pump’s operating environment and/or the fluids handled or should otherwise be thermally insulated to avoid un-intentional actuation. PECs, FEPs and DEs typically require high electric fields for actuation, translating into high voltages that limit their compatibility with living tissues and organisms. Electric fields can either be decreased by reducing thicknesses or these actuators must be shielded from their environments. PECs can typically operate in air as well as in liquids and vacuums. Some, such as AlN, ZnO and PMN-PT are non-toxic, while, for example, PZT is toxic due to its Pb content. Some researchers have been experimenting, however, with the latter by means of encapsulation as well as surface treatment with highly biocompatible Ti [61,183].

Ferroelectric β-phase PVDFs can operate in liquids, air and vacuum [89] and are non-toxic [61]. The toxicity of DE actuators depends on the electrodes used [184]. While encapsulated in protective layers, they can operate in dry as well as in wet environments. Finally, regarding ionic EAPs actuators, their toxicity is determined by the materials and the electrolytes utilized. Even though Nafion for IPMCs and conjugated polymers for CPs are harmless, most room-temperature-ionic-liquids are harmful [185]. Encapsulation of these actuators is required to additionally extend their lifetime in air and vacuum by preventing the liquid electrolytes from escaping the actuators and to allow sustained operation in foreign liquids by preventing ion exchange with biological systems [179]. From the actuators reviewed, U-SMA-, PEC- and CP-based exhibit catch-states, establishing them as particularly attractive for inhibiting potential back-flow in peristaltic pumps. Key material and electromechanical properties of U-SMA, PEC, FEP, DE, IPMC and CP smart materials and their actuators are presented in Table 1, Table 2, Table 3, Table 4, Table 5 and Table 6. An advantage of smart materials and their actuators, especially for biomedical applications, is their compliance. FEPs, DEs, IPMCs and CPs that are polymer-based trilayer actuators with metallic electrodes, are soft as their metal layers are very thin. Even SMAs and PECs, which can possess higher Young’s moduli compared to the organic smart materials exhibited in the review, are compliant conformed in thin structures.

Operation stability is another factor to consider as it affects the pump’s operating life. A disposable microfluidic pump might, for example, be required to operate for one cycle, whereas a cardiac pump to work perpetually for years. U-SMAs can operate up to 10⁵ cycles (ε = 0.5%), PECs and FEPs can actuate 10⁶ cycles and up to 5 × 10⁷ cycles, respectively, while DEs can work for 10⁵ cycles with no failure observed (ε = 5%). IPMCs and CPs have been reported to actuate up to 10⁶ cycles (ε = extremely small for IPMCs).

Generated actuator strains ε and stresses σ can affect pump performance in terms of the volumes pumped per cycle and the ability to create a pressure head, which regulate flow rates, respectively. Resonance frequencies can also influence the liquid volumes pumped. Actuating frequencies can furthermore affect fluid pumping frequencies, which can be important when, for example, administering drugs or even blood so, that in the latter, the volume of blood pumped by the heart is emulated. High stresses usually translate into low strains and similarly, high frequencies translates into low strains and sometimes stresses. In general, low strains of smart materials that induce planar displacements can be mechanically amplified by bonding with passive layers to create bending uni-morph or bimorph actuators with greater displacements. Low stresses can be increased by using actuators in parallel and by utilizing multi-layered stacked or rolled smart materials to increase force capabilities. Moreover, when utilizing ionic actuators such as IPMCs and CPs, properties can be altered by the electrolyte/s used.

U-SMA materials typically generate small to moderate strain outputs (~5%) and large stresses (~100 MPa). Considerations when employing their actuators are their biasing to induce bi-directional motions, and to therefore, obtain full operational cycles and accelerating their restoration phases, which are based on cooling, to attain higher operational frequencies. PECs are robust materials that produce large stresses (~100 MPa) at extremely small strains (~0.5%) while they can actuate at high operational frequencies (up to 10 MHz). FEPs typically generate small strains and moderate stresses (~2% and ~20 MPa, respectively) and can operate at high bandwidths (up to ~100 Hz). Material reproducibility issues with FEPs should be considered. DE actuators can be easily conformed to various geometries. High to extremely high reversible strains (for silicone-based DEs, 120%, acrylic VHB, 380% up to ~1100% through the ‘snap-through’ phenomenon, respectively) and small to moderate actuation stresses (~0.5 MPa and ~1.5 MPa, respectively) can be generated at small to high bandwidths (~1500 Hz and 10 Hz), depending on the employed materials. Lateral constraints for pre-straining the actuators’ polymers should be considered. High frequency is obtained at the expense of strain and sometimes stress, all three typically cannot be high at the same time. IPMCs demonstrate typically small strains (~0.5%), small-induced stresses (~3 MPa) and can work at high operating frequencies (~100 Hz). CPs induce low to moderate stresses and strains (~1 MPa and 2–10%). When utilizing both materials and their actuators in applications, encapsulation should be considered to avoid the exchange of the electrolyte components with the environment.

Finally, a risk to be considered when utilizing composite tri-layers such as PECs, FEPs, DEs, IPMCs and, in some cases, CPs is delamination, as there is often a mismatch of the constituents’ Young’s moduli coefficients, and the composites therefore gradually degrade with incremental cycles. Furthermore, FEP and DE actuators that present high strains may be limited due to the fatigue of the electrodes that are applied to the surface of their polymers as well as due to the dielectric breakdown of the polymers.

5. Conclusions

The global pump market is projected to grow until 2024 with a 5% compound annual growth rate, and that of micro-pumps is estimated to increase by 20% as these devices will be integral components in micro-fluidic systems such as labs-on-a-chip and micro-dosage, point-of-care testing, and micro-total analysis systems.

Solid-state electromechanical smart material actuators can provide interesting solutions for driving these devices, as they can be easily miniaturized to a sub-millimeter scale compared to conventional ones. Moreover, due to the material and electromechanical properties of the materials and their actuators, as well as their common actuation modes, these technologies are extremely diverse.

A review of the solid-state electromechanical smart material actuators for operating liquid pumps has been presented. From a variety of smart material actuators, the current manuscript has identified the application of U-SMAs, PECs, FEPs, DEs, IPMCs and CPs in liquid pumps. Smart material working principles, their most common actuator configurations, as well as their displacement modes, have been detailed. The focus has also been on smart material characteristics, properties, and limitations, as they are important when employing smart material actuators to move and/or control structures. While commercial application of these actuators is at an early stage due to remaining challenges in material properties, there is intensive R&D within the field, and the demand for such devices will likely increase the interest in the coming years.