Molecularly Imprinted Polymers for Dispersive (Micro)Solid Phase Extraction: A Review

: The review describes the development of batch solid phase extraction procedures based on dispersive (micro)solid phase extraction with molecularly imprinted polymers (MIPs) and magnetic MIPs (MMIPs). Advantages and disadvantages of the various MIPs for dispersive solid phase extraction and dispersive (micro)solid phase extraction are discussed. In addition, an effort has also been made to condense the information regarding MMIPs since there are a great variety of supports (magnetite and magnetite composites with carbon nanotubes, graphene oxide, or organic metal framework) and magnetite surface functionalization mechanisms for enhancing MIP synthesis, in-cluding reversible addition-fragmentation chain-transfer (RAFT) polymerization. Finally, drawbacks and future prospects for improving molecularly imprinted (micro)solid phase extraction (MIMSPE) are also appraised.


Introduction
During the last two decades the large development of analytical instrumentation, mainly the introduction of mass spectrometry (MS) and tandem mass spectrometry (MS/MS), has facilitated the determination of analytes in biological, food, and environmental samples at trace concentrations. However, although the high sensitivity provided by the instrumentation and the direct injection/analysis of crude samples/extracts are not always possible, new sample preparation strategies are needed for potential interferences removal and analyte pre-concentration, for increasing the robustness and repeatability of measurements, for converting the analyte to a more suitable form for separation/detection, and also for avoiding conventional multiple-step pre-treatment methods [1]. Several extraction/preconcentration techniques have been therefore developed and among those techniques, solid phase extraction (SPE) and solid phase microextraction (SPME) are nowadays well established and commercially available methodologies. However, the main drawback associated with them is the moderate selectivity of sorbents, which can require further extract clean-up stages [2].
Molecularly imprinted polymers (MIPs) are versatile materials that mimic natural antigen-antibody mechanisms and allow molecules/analytes recognition [2,3]. These materials are prepared by polymerizing monomers and cross-linkers around the template molecules, leading to a highly cross-linked three-dimensional network polymer. After polymerization, the template molecules are removed, and the shape and size of the binding sites are established complementary to the target analyte. Synthesized MIPs are stable and show resistance to wide range of pH values, temperatures, and solvents and interact with target molecule in a selective way. Due to their practical features, MIPs have been used as selective sorbents for (micro)solid extraction (µ-SPE) procedures leading to

Dispersive (Micro)Solid Phase Extraction with Magnetic Molecularly Imprinted Polymers (MMIPs)
MMIP beads were first introduced by Ansell and Mosbach in 1998 as a core-shell structure (magnetic iron oxide, magnetite, Fe3O4) for performing drug radioligand binding assays [19]. Then, MMIPs (magnetic nickel hexacyanoferrate, NiHCF, nanoparticles coated with a molecularly imprinted polymer for the herbicide chlorotoluron) were proposed for preparing selective modified electrodes [20]. MMIPs as selective adsorbents for SPE procedures offer advantages such as avoidance of drawbacks associated with conventional batch SPE/µ-SPE procedures, which need filtration/centrifugation steps for separating the adsorbent from the bulk sample after the loading stage and from the extract after analyte elution. In addition, losses of adsorbent particles are minimized since adsorbent separation is easily and quickly achieved by applying a magnet [21]. As previously mentioned, MMIP nanoparticles can be stirred (dispersed) in the sample/extract (loading step) and in the eluting solution (elution step), taking advantage of their magnetic properties, but stirring can be also performed by vortexing and by ultrasound dispersion.
There are several strategies for preparing MMIPs, which lead to a great varietyof magnetic adsorbents. Moreover, despite free radical polymerization mechanism(s), which are mainly used to prepare MMIPs (and also MIPs), the heterogeneity caused by the fast chain propagation and irreversible termination reactions has led to the use of controlled radical polymerization strategies such as reversible addition fragmentation chain-transfer

Dispersive (Micro)Solid Phase Extraction with Magnetic Molecularly Imprinted Polymers (MMIPs)
MMIP beads were first introduced by Ansell and Mosbach in 1998 as a core-shell structure (magnetic iron oxide, magnetite, Fe 3 O 4 ) for performing drug radioligand binding assays [19]. Then, MMIPs (magnetic nickel hexacyanoferrate, NiHCF, nanoparticles coated with a molecularly imprinted polymer for the herbicide chlorotoluron) were proposed for preparing selective modified electrodes [20]. MMIPs as selective adsorbents for SPE procedures offer advantages such as avoidance of drawbacks associated with conventional batch SPE/µ-SPE procedures, which need filtration/centrifugation steps for separating the adsorbent from the bulk sample after the loading stage and from the extract after analyte elution. In addition, losses of adsorbent particles are minimized since adsorbent separation is easily and quickly achieved by applying a magnet [21]. As previously mentioned, MMIP nanoparticles can be stirred (dispersed) in the sample/extract (loading step) and in the eluting solution (elution step), taking advantage of their magnetic properties, but stirring can be also performed by vortexing and by ultrasound dispersion.
There are several strategies for preparing MMIPs, which lead to a great varietyof magnetic adsorbents. Moreover, despite free radical polymerization mechanism(s), which are mainly used to prepare MMIPs (and also MIPs), the heterogeneity caused by the fast chain propagation and irreversible termination reactions has led to the use of controlled radical polymerization strategies such as reversible addition fragmentation chain-transfer (RAFT) polymerization for preparing MIPs [22] and also MIP coatings over magnetic and non-magnetic supports [23][24][25][26]. RAFT polymerization provides more accessible sites for target adsorption and faster mass transfer because of the more homogenous polymeric network [27].

Classification of MMIPs
Based on MMIP structure, four types of MMIPs can be established: core-shell MMIPs, magnetic nanotube-supported MIPs, magnetic nanosheet-supported MIPs, and magnetic hollow porous MIPs [28].
A core-shell structure is the most widely used, and it consists of a core magnetic phase (typically magnetite) and a polymeric phase shell [29]. Magnetic nanoparticles (Fe 3 O 4 ) in the core-shell-based structures offer a high surface area for MIP anchorage, and the surface can be also modified (activated/functionalized) with hydroxyl groups and a SiO 2 layer to protect the core from oxidation or dissolution [30].
Magnetic nanotube-supported MIPs imply the presence of carbon nanotubes (CNTs) or multi-walled carbon nanotubes (MWCNTs) in the reaction medium during the co-precipitation and solvothermal synthesis of Fe 3 O 4 to prepare magnetic carbon nanotubes (MCNTs) in which the magnetic nanoparticles are linked onto the CNTs' surface [31]. The prepared MCNTs are then treated with a silica-based reagent to cover them with a SiO 2 layer (MCNTs-SiO 2 ) before MIP synthesis. Because of the large surface area of the CNTs, the prepared MCNTs-SiO 2 structures offer a higher specific surface area for MIP anchorage than that found in core-shell MMIPs, leading to larger binding/recognition sites for the target. On other occasions, MWCNTs, previously functionalized with carboxylic acid groups (COOH), are mixed with Fe 3 O 4 nanoparticles in the pre-polymerization solution for direct MIP synthesis [32].
Similar to CNTs, the use of graphene oxide (GO) during the solvothermal synthesis of Fe 3 O 4 leads to a GO-Fe 3 O 4 composite in which the magnetite nanoparticles are linked to the GO nanosheet [33,34]. MIP synthesis can be then performed after grafting the Fe 3 O 4 @GO surface with acrylic acid, and the resulting magnetic nanosheet-supported MIP provides high specific surface area and high affinity for the target molecule, as well as an extremely fast absorption rate [33]. On other occasions, the magnetic Fe 3 O 4 @GO composite can be functionalized with silica and vinyled reagents before MIP synthesis [34]. Other approaches are based on activated CNTs (presence of carboxyl groups absorbed onto the surface of GO through π-π attractions after acidic treatment) which, after hydrothermal treatment for synthesizing Fe 3 O 4 nanoparticles, lead to 3D magnetic GO-CNT composites [35].
Finally, well-designed magnetic hollow porous MIPs (magnetic HPMIPs) have been introduced as a sacrificial support in the molecular imprinting process. As reported, MIP synthesis is performed on the internal surface of mesoporous silica spheres (referred as MCM-48) followed by silica and template removal (typically hydrofluoric acid/ethanol mixtures) [36][37][38][39]. Previous to magnetite synthesis over the HPMIPs (co-precipitation method), a treatment with diluted perchloric acid was required to obtain 1,2-diol groups over the HPMIPs structures [37].
In addition to mesoporous silica, mesoporous carbon has also been found to be an excellent support for preparing hollow porous MIPs, with the advantage that carbon support is not removed (sacrificed) to obtain the required porosity of the material. Dglucose [40] and raw Pericarpium Granati (a medicinal plant) [41] have been used as sources of carbon for the synthesis of the magnetic mesoporous carbon (MMC) particles by hydrothermal methods (high temperatures as well as long synthesis times) in the presence of ferric and ferrous ions. Benefits of the hollow composites are the presence of high dense accessible recognition sites for molecular imprinting, and a high absorption capacity, which leads to higher enrichment factors when compared with traditional MIPs.
Hollow porous MIPs have been also designed by Fe 3 O 4 nanoparticle surface modification by a sol-gel route with silica-based reagents such as tetraethyl orthosilicate (TEOS), which promotes hydroxyl groups on the surface of the magnetic nanoparticles, followed by MIP synthesis, and template and silica layer removal [42][43][44].
Other magnetic HPMIPs have been synthesized by using hollow Fe 3 O 4 microspheres instead of conventional magnetite [45] (hollow Fe 3 O 4 microspheres are obtained by onepot hydrothermal methods [46]). In addition, other authors have prepared magnetic nanorings with abundant epoxy groups on the surface for imprinting purposes involving ring-opening reactions. The prepared material, named core-shell nanoring aminofunctionalized magnetic molecularly imprinted polymer (CS-NR-Mag-MIP), was found to offer high absorption capacities for bisphenol A [47] and sulfonamides [48].

Magnetite Surface Functionalization for Core-Shell MMIPs
There are some procedures for MMIP preparation in which Fe 3 O 4 nanoparticles are present in the polymerization mixture during MIP synthesis [49][50][51]. Although transmission electron microscopy (TEM) images show that the spherical Fe 3 O 4 nanoparticles are well enwrapped by the MIP shell [51], on other occasions the used of un-functionalized Fe 3 O 4 nanoparticles during MIP synthesis leads to uniform polymeric layer composites [50]. Dispersed spherical (nano)particles are preferred as adsorbents in SPE, and the spherical shape of a magnetite-based composite is guaranteed by performing MIP synthesis over surface functionalized magnetite. In addition, once Fe 3 O 4 nanoparticles are synthesized (magnetite is also commercially available), functionalization of the nanoparticles' surface makes it favorable for MIP adhesion, and also promotes a high specific surface area and improves polarity [52]. Although magnetite surface functionalization can be performed after Fe 3 O 4 synthesis or directly over commercial nanoparticles, one-step Fe 3 O 4 synthesis and surface functionalization procedures have been also described. After surface modification, polymerization can be performed by several polymerization methods using the adequate template molecule, monomer, cross-linker, initiator, and porogen. The resulting composite adsorbent will offer good selectivity/recognition for the target molecule as well as good magnetic properties [52].
Magnetite surface functionalization can be performed mainly by using silica-based, diol-based, and vinyled compounds (Table 1). However, there are other functionalization mechanisms as well as several combinations of surface modifier reagents for Fe 3 O 4 nanoparticle surface functionalization. Table 1. Functionalization reagents for magnetite core-shell magnetic molecularly imprinted polymers (MMIPs).
Polyethylene glycol (PEG) [53][54][55][56][57][58] Poly(vinyl alcohol) [59] Acrylic acid [60] Methacrylic acid (MAA) [61] Boronic acids: 2,4-Difluoro-3-formyl-phenylboronic acid (DFFPBA) a,b [ Diol-based reagents such as polyethylene glycol (PEG) [53][54][55][56][57] interact with the nanoparticle surface through one of the hydroxyl groups, allowing the remaining hydroxyl groups to be available to react with the components of the pre-polymerization mixture ( Figure 2). A similar mechanism is obtained for oleic acid [91][92][93][94][95][96][97][98][99][100][101][102], which interacts with the nanoparticle's surface through the hydroxyl groups but promotes the presence of vinyl groups in the modification layer. Similarly, poly (vinyl alcohol) [59] is also a source of hydroxyl and vinyl groups for reacting with the pre-polymerization components. A magnetic core surface rich in vinyl groups can be also obtained by treating the prepared Fe 3 O 4 nanoparticles with acrylic acid [60] or by one-step co-precipitation of Fe 3 O 4 in the presence of methacrylic acid [61]. The main advantage of using these reagents for magnetite surface modification is the simplicity of the procedure and the moderate operating conditions (room temperature or ice-bath). In addition, surface functionalization with these reagents avoids the electrostatic agglomeration of magnetite, which ensures the uniformity of magnetic nanoparticles in the pre-polymerization solution and a further MIP homogeneous embedding. However, drastic conditions, such as extreme pHs, when removing the template after MMIP synthesis can damage the link between the nanoparticle and the functionalization layer, which leads to a separation of the MIP layer (shell) from the magnetite nanoparticles (core).
The main advantage of using these reagents for magnetite surface modification is the simplicity of the procedure and the moderate operating conditions (room temperature or ice-bath). In addition, surface functionalization with these reagents avoids the electrostatic agglomeration of magnetite, which ensures the uniformity of magnetic nanoparticles in the pre-polymerization solution and a further MIP homogeneous embedding. However, drastic conditions, such as extreme pHs, when removing the template after MMIP synthesis can damage the link between the nanoparticle and the functionalization layer, which leads to a separation of the MIP layer (shell) from the magnetite nanoparticles (core).
Similarly, the presence of hydroxyl groups on the nanoparticle surface can also be achieved by using boronic acids that bind cis-diol-containing compounds and result as adequate for imprinting large biomolecules, such as proteins, and achieving oriented surface imprinting, depending on the affinity between the template molecule and the boronate residues [62][63][64][65][66][67] (Table 1). The boronic acid 2,4-difluoro-3-formyl-phenylboronic acid (DFFPBA) has been proposed for preparing MMIPs, which requires amino-functionalized magnetic nanoparticles before DFFPBA functionalization (easily achieved by Fe 3 O 4 synthesis in presence of 1,6-hexanediamine). Fe 3 O 4 @NH 2 can be directly treated with DFF-PBA (treatment at room temperature for 24 h) [62] or can be first silanized with TEOS and (3-Aminopropyl) triethoxysilane (APTES), leading to (Fe 3 O 4 @SiO 2 @DFFPB) [63]. In addition to DFFPBA, other boronic acids such as 4-formylphenylboronic acid (FPBA) in combination with sodium cyanoborohydride (NaBH 3 CN) have been found to be effective to prepare the surface of Fe 3 O 4 @NH 2 [64] or Fe 3 O 4 nanoparticles [65] for synthesizing MIPs for protein recognition. A similar strategy has been proposed using magnetite microspheres coated with porous TiO 2 (flower-like structure Fe 3 O 4 @pTiO 2 nanoparticles) prepared via a solvothermal method and further functionalized with γmercaptopropyltrimethoxysilane (γ-MPTS) to promote the presence of -SH before anchoring the boronic acid 4-vinylphenboronic acid (VPBA) [66]. More binding sites for templates (horseradish peroxidase), and thus, higher adsorption capacity, were found when using Fe 3 O 4 @pTiO 2 as a supporting material than when using Fe 3 O 4 @SiO 2 cores [65,66]. In addition, the strong electron-withdrawing effects of Ti(IV) endow the boronic acid with lowered pK a value that makes the Fe 3 O 4 @pTiO 2 @MIPs capture glycoproteins under moderate acidic conditions [66].
Finally, boronate-affinity magnetic hollow molecularly imprinted polymer sorbents for sialic acid (a compound exhibiting a cis-diol structure) have been also prepared by using mesoporous silica spheres (MCM-48) as a sacrificial support, glycidyl methacrylate (GMA) as a co-monomer to chemisorb Fe 3 O 4 nanoparticles, and 3-aminophenylboronic acid (APBA) as boronic acid [67]. After MCM-48@APBA preparation, the template, the cross-linker, and the initiator are added for performing MIP synthesis, followed by MCM-48 dissolution in a hydrofluoric/ethanol mixture (B-hMIP composite). Fe 3 O 4 nanoparticles are then synthesized (co-precipitation) in the presence of the prepared composite (B-hMIPs), leading to the magnetic hollow adsorbent [67].

Surface Functionalization with Silica-Based Reagents
Silica-based reagents (Table 1) are an alternative to vinyled and diol-based compounds in magnetite surface functionalization procedures for overcoming problems derived from core-shell breakdown as consequence of extreme pH and temperature operating conditions, since the resulting composites exhibit great stability [52]. TEOS is a typical silica-based compound used for Fe 3 O 4 modification at moderate operating conditions, resulting in Fe 3 O 4 @SiO 2 composites. The TEOS layer over the magnetite nanoparticles is a source of hydroxyl groups for further interactions with the pre-polymerization reagents . A typical diagram of a magnetic silica-based composite is illustrated in Figure 3.
The presence of amino (−NH 2 ) groups is guaranteed by using APTES (Table 1), and MIP synthesis can be performed directly by mixing the functionalized nanoparticles with the polymerization reagents [78,119]. On other occasions, the monomer, such as methacryloyl chloride, can be fixed to the functionalized silica layer after reaction with the immobilized amino groups [125]. In addition, previously modified Fe 3 O 4 with TEOS (Fe 3 O 4 @SiO 2 ) can be then covered with APTES for promoting the presence of −NH 2 groups [105,[120][121][122][123]. pounds in magnetite surface functionalization procedures for overcoming problems derived from core-shell breakdown as consequence of extreme pH and temperature operating conditions, since the resulting composites exhibit great stability [52]. TEOS is a typical silica-based compound used for Fe3O4 modification at moderate operating conditions, resulting in Fe3O4@SiO2 composites. The TEOS layer over the magnetite nanoparticles is a source of hydroxyl groups for further interactions with the pre-polymerization reagents . A typical diagram of a magnetic silica-based composite is illustrated in Figure 3. Introduction of carboxyl groups onto the Fe 3 O 4 @SiO 2 surface is performed by treating functionalized Fe 3 O 4 @-SiO 2 -NH 2 nanoparticles (TEOS and APTES covering) with poly(ethylene glycol)bis(carboxymethyl) ether before MIP synthesis [123]. Comparison between Fe 3 O 4 @SiO 2 -COOH@MIP and Fe 3 O 4 @-SiO 2 -NH 2 @MIP showed that the latter had high specific surface area and fast mass transfer rate toward the target (aminopyralid) [123].
These silanization procedures are time-consuming processes since they require a previous silanization stage (Fe 3 O 4 @SiO 2 ) followed by a treatment for incorporating the desired functional groups. In addition, the described procedures are reported to require reaction temperatures achieved by refluxing systems. Therefore, there have been described several one-step Fe 3 O 4 synthesis (solvothermal method) and surface functionalization procedures by incorporating into the reaction medium diol-based reagents such as PEG and ethylene glycol (EG) [133][134][135][136][137][138][139][140], and vinyled reagents (oleic acid [141] and hexanediamine [140]). Since the solvothermal method is required for Fe 3 O 4 nanoparticle synthesis, special laboratory devices such as Teflon-lined stainless steel reactors are needed. In addition, the synthesis/functionalization is performed at high temperature (200 • C) and for long times (up to 24 h).
One-step procedures for Fe 3 O 4 silanization (Fe 3 O 4 @SiO 2 ) and modification with functional groups such as −NH 2 have been also described for speeding-up the functionalization step. The use of APTES as a silica-based reagent (silanization) promotes the simultaneous functionalization with amino functional groups. In addition, if the modification is performed using APTES and mono acrylic acid at once, the resulting covering will also be rich in vinyl groups [142]. On other occasions, after Fe 3 O 4 functionalization with oleic acid, a further reaction with KH-570 ensures stability (silica covering) and abundant vinyl groups for further MIP synthesis [90].

Magnetite Surface Functionalization for Magnetic Nanotube-Supported and Magnetic Nanosheet-Supported MIPs
As summarized in Table 2, surface functionalization of mixed magnetic composites involving the presence of CNTs [31] and MWCNTs [143][144][145] has been efficiently achieved by using diol-based reagents such as EG and PEG [31,143,146,147], although some authors have described the convenience of a previous MWCNT@Fe 3 O 4 composite oxidation [144], reduction [145], or carboxylation [32,148] stage before functionalization/MIP synthesis. The activated surface of CNTs/MWCNTs improves the nanoparticles dispersion and the interaction of monomers with the CNTs/MWCNTs. However, some of these procedures require high temperatures and long times are also needed to complete the reactions [31,32,143,148]. After MWCNT@Fe 3 O 4 composite synthesis, silanization procedures have also been reported by using KH-570 under moderate reaction conditions (stirring/sonication, N 2 atmosphere, 70 • C, 10 min), which leads to a stable magnetic composite and also increases the reactive activity as a consequence of the anchored vinyl groups [149]. Functionalization with KH-570 can be also performed after a previous silanization of the prepared MCNTs with TEOS (MCNTs@SiO 2 ) [150]. Methacryloxypropyl trimethoxysilane (MAPTMS) has been also proposed as a silanizing agent and as a vinyled monomer for further IIP synthesis (Pb (II) ions as template and dithizone as a ligand) [148].
Regarding magnetic nanosheet-supported MIPs (Table 2), the GO@Fe 3 O 4 surface is usually functionalized by grafting with acrylic acid as shown in Figure 4 [33,151], which ensure the presence of vinyl groups for further polymerization. Acrylic acid was also used for surface modification of chitosan based GO@Fe 3 O 4 composites [152]. Silanization with TEOS to prepare MGO@mSiO 2 (mesoporous silica) has been also reported for direct MIP synthesis [153] and for a further functionalization with vinyltrimethoxysilane (VTTS) [34] and APTES [154] in order to facilitate the subsequent polymerization via vinyl or amino groups, respectively. However, prepared GO@Fe 3 O 4 nanoparticles [155,156], as well as 3D magnetic GO-CNT composites [35], were also directly used for MIP synthesis without functionalization.     usually functionalized by grafting with acrylic acid as shown in Figure 4 [33,151], which ensure the presence of vinyl groups for further polymerization. Acrylic acid was also used for surface modification of chitosan based GO@Fe3O4 composites [152]. Silanization with TEOS to prepare MGO@mSiO2 (mesoporous silica) has been also reported for direct MIP synthesis [153] and for a further functionalization with vinyltrimethoxysilane (VTTS) [34] and APTES [154] in order to facilitate the subsequent polymerization via vinyl or amino groups, respectively. However, prepared GO@Fe3O4 nanoparticles [155,156], as well as 3D magnetic GO-CNT composites [35], were also directly used for MIP synthesis without functionalization.

Other Mixed Composites for MMIPs
Various types of magnetic composites (Table 3) have been used as magnetic cores for MMIPs such as metal-organic frameworks (MOFs) and zeolite imidazolate frameworks (ZIFs). In some cases, the synthesis of Fe 3 O 4 nanoparticles following the hydrothermal process is performed in the presence of the framework (HKUST-1, a Cu-based porous MOF) and EG (diol groups), which also act as capping agents for avoiding aggregation [157]. Surface functionalization is further performed with VTMOS [157]. The procedure is timeconsuming, and a Teflon-lined stainless steel autoclave (synthesis at 200 • C) is required. On other occasions, the previously synthesized magnetite nanoparticles are allowed to react with the framework at moderate temperatures and short times. This is the case of Fe 3 O 4 @ZIF-8 composites, in which poly (styrenesulfonate sodium salt) is added to the reaction medium to allow the ZIF-8 shell growth (with the presence of 2-methylimidazolate as a precursor), and for which a further surface functionalization is not required [158]. Moreover, ZIF-L-based Co-based magnetic nanoporous carbon (Co-MNPC) is also directly mixed in the polymerization medium for preparing a magnetic selective Co-MNPC@MIP sorbent to aflatoxins [159].
Ni@MIL-100(Fe) MOF has also been used as a support for MIP synthesis [160]. In this case the framework exhibits magnetic properties, and after mixing with the template (hydroxychloroquine) and with the functional monomer (APTES) and the cross-linker (TEOS), MIP synthesis can be directly carried out.
Fast procedures for synthesizing magnetic composites have been also described for thiolated β-cyclodextrin assembled to gold nanoparticles (β-CD/Au), whose presence during the Fe 3 O 4 synthesis (co-precipitation method) leads to a β-CD/Au/Fe 3 O 4 composite functionalized for further MIP synthesis [161]. Other proposals suggested the previous synthesis of Fe 3 O 4 @mSiO 2 functionalized with APTES (presence of −NH 2 groups) magnetic core before surface grafting of β-CD and MIP synthesis (phthalic acid ester as a template) for preparing magnetic plasticizer MIPs [162]. In addition to the high selectivity inherent to the MIP layer, the prepared composite material was found to show large adsorption capacity and fast kinetic equilibrium.
The aminopolysaccharide nature of the biopolymer chitosan (CS) has also taken advantage of modifying magnetite for achieving a surface rich in functional groups for further polymerization. Preparation of Fe 3 O 4 @CS nanoparticles is easily performed following the hydrothermal synthesis of Fe 3 O 4 in the presence of CS [163]. Finally, one-step coprecipitation under alkaline conditions (Fe 3 O 4 synthesis) in the presence of the diazonium salt BF 4 ( + N 2 -C 6 H 4 -CH 2 -DEDTC) also generates a magnetic core that offers adequate functional groups for mixing with the pre-polymerization mixture and starting the MIP synthesis [165]. On other occasions, surface-modified Fe 3 O 4 with oleic acid was allowed to polymerize with arsenic (III)-2-acetyl benzofuran thiosemicarbazone complex as template, and methacrylic acid as a monomer (ionic imprinted polymer for As(III)) before a Pickering emulsion in the presence of nanoparticles of chitosan [164]. Extraction of As(III) from acid digests from rice and vegetable was achieved after pH adjustment, assisting the loading/elution process by ultrasounds [164].

Dispersive Solid Phase Extraction and Microsolid Phase Extraction with Non-Magnetic MIPs
As previous commented, dSPE/D-µ-SPE [9][10][11] can be performed by dispersing MMIP nanoparticles, and also non-magnetic MIP beads, by vortex and ultrasound stirring [10]. Table 4 summarizes the main features regarding dSPE/D-µ-SPE with non-magnetic MIPs. The adsorbents can be obtained by precipitation [166][167][168][169][170][171][172][173][174], and bulk [175][176][177] polymerization has been used for dSPE/D-µ-SPE by shaking the sample/extract-MIP bead mixtures for times varying from 5.0 min [166] to 3.0 h [169]. Absorption times can be reduced to 1 min when assisting the procedure by ultrasounds, enough time for isolating phenolic compounds in aqueous samples using 10 mg of MIP [170]. However, sonication times of 3.0 h have been proposed for fluoroquinolone pre-concentrations from waters using a dual-template MIP (dt-MIP) for norfloxacin and enrofloxacin as templates [172]. Authors, however, did not report insights regarding low adsorption rate for reaching the equilibrium between the targets and the dt-MIPs. The procedures were found to be effective and selective for isolating fungicides from cucumber [168], sulfonamides from milk [166], bioactive compounds (polydatin) from rat's plasma and urine [169], progesterone hormones from plasma, urine and waters [175], and also for purifying extracts (pre-concentrating targets) such as polydatin from Chinese medical medicines [169], antibiotics from pork [167], aflatoxins from cultured fish [171], pyraclostrobin from ginseng [174], folic acid from foodstuff [176], and herbicides from shellfish [177].     Ionic molecularly imprinted polymers (IIPs) have also been proposed for dSPE [173]. The bifunctional monomer 1-vinylimidazole was used for reacting with the template ((meta)arsenite, As(III)) and providing vinyl groups for polymerization [173]. The dSPE implied portions of 50 mg of IIP and vortexing at 1000 rpm for 1 min allowed the selective pre-concentration of inorganic arsenic species (As (III) plus As(V)) from rice extracts [173].
MIP synthesis around non-magnetic nanoparticles, such as silica nanoparticles, has been also performed to obtain stable adsorbents. Therefore, silica nanoparticles functionalized with KH-570 (SiO 2 @KH-570) by a base-catalyzed reaction of TEOS and KH-570 have been used as a core for preparing a selective MIP composite for the enantioseparation of racemic tryptophan (L-tryptophan recognition) in aqueous solutions [184]. Experiments were performed by oscillating the MIP-aqueous sample mixtures for 32 h at room temperature and using only 2 mg of adsorbent [184].
In addition, the excellent properties of MOFs have led to preparation of MOF-MIP composites based on UiO-66 MOF [178] and HKUST-1 MOF [180] by direct MIP polymerization on the MOF's surface. The dSPE was performed with 5 mg of UiO-66-MIP and shaking for 15 min to recover tetracyclines from chicken extracts [178]; whereas 2 mg of HKUST-1-MIP (vortexing for 2 min) proved adequate for nicotinamide pre-concentration [179].
Hollow non-magnetic composites based on silica [180] and carbon [181] have been also prepared for dSPE/D-µ-SPE. In both cases, after MIP synthesis over the nanoparticle, the supporting material was removed (hydrofluoric acid for silica [180], and calcination at 500 • C for carbon [181]) leading to a porous material with high surface area. Synthesis of hollow silica-based MIP composites required functionalization with KH-570 (source of vinyl groups) for an effective MIP synthesis and anchorage (estrogen recognition/pre-concentration from water [180]). However, TEOS and aluminum chloride were used for MIP synthesis (valsartan as a template) when preparing the hollow carbonbased aluminum-doped silica composite, promoting hydrolysis to generate silanol groups (Si−OH) followed by condensation of the silanols to obtain a polysiloxane (O−Si−O) [181]. Estrogen pre-concentration was designed by using 10 mg of the composite and shaking for 1.0 h [180]; whereas, valsartan and losartan isolation required 40 mg of adsorbent and sonication for 27 min [181].
Other composites such as MWCNT-MIPs have also been demonstrated to be effective adsorbents for dSPE of dioctyl phthalate in beverage samples [182]. Vinyl groups were incorporated on MWCNTs by reaction with sodium ethoxylate before MWCNT oxidation (presence of carboxyl groups), and MIP (dioctyl phthalate as a template) was further synthesized. The dSPE procedure was performed by mixing 60 mg of MWCNTs-MIPs with treated beverage samples (juice, dairy drinks, and carbonated drinks) and incubating at room temperature for 30 min on an oscillator [182]. Finally, graphene oxide-based MIPs (GO-MIPs) have also been used for dSPE when pre-concentrating bis(2-ethylhexyl) phthalate from waters by shaking 20 mg of adsorbent with the sample (water) at 600 rpm for 30 min [183].

Drawbacks and Future Prospects
MIMSPE procedures have been revealed as excellent approaches for miniaturization of SPE-based techniques in analytical chemistry, offering selective extraction/preconcentration when analyzing complex samples. Dispersive SPE/µ-SPE procedures based on MIPs (mainly MMIPs) have shown high potential of miniaturization, which implies the use of low amounts of adsorbents as well as low volumes of organic solvents for performing the elution stage.
However, MIPs and MMIPs face a number of challenges during the preparation (synthesis) stage and also during the application. MMIPs are synthesized in nonpolar solvents to avoid the disruption of the hydrogen bonding between monomer and templates. The generated hydrophobic surfaces lead to adsorption of interferences such as proteins. RAFT polymerization is a good alternative to overcome this problem since it allows the preparation of highly hydrophilic MIPs (or MIP external layers over nanoparticles), which can lead to efficient adsorbents for samples of a wide polarity range. Improvements have also been addressed to automate the techniques (similar to on-column/cartridges SPE) since batch MIMSPE procedures require several steps (conditioning, loading, washing, elution) and the procedures are not appealing processes when coping with hundreds of samples. In addition, the coupling (and also automation) of the MIMSPE devices directly with analytical instruments has not been explored yet.
In any case, MIMSPE procedures open a fascinating window to analyzing compounds from complex matrices, and continuous efforts in this research area should open more and more novel applications.

Conflicts of Interest:
The authors declare no conflict of interest.