**Alternative Green Extraction Phases Applied to Microextraction Techniques for Organic Compound Determination**

#### **Eduardo Carasek \*, Gabrieli Bernardi, Sângela N. do Carmo and Camila M.S. Vieira**

Department of Chemistry, Federal University of Santa Catarina, Santa Catarina, SC 88040-900, Brazil **\*** Correspondence: eduardo.carasek@ufsc.br

Received: 29 May 2019; Accepted: 9 July 2019; Published: 16 July 2019

**Abstract:** The use of green extraction phases has gained much attention in different fields of study, including in sample preparation for the determination of organic compounds by chromatography techniques. Green extraction phases are considered as an alternative to conventional phases due to several advantages such as non-toxicity, biodegradability, low cost and ease of preparation. In addition, the use of greener extraction phases reinforces the environmentally-friendly features of microextraction techniques. Thus, this work presents a review about new materials that have been used in extraction phases applied to liquid and sorbent-based microextractions of organic compounds in different matrices.

**Keywords:** biosorbents; microextraction; organic compounds; green extraction phases

#### **1. Introduction**

Sample preparation is a crucial step in analytical methods for determining organic compounds. The isolation of the analytes from the matrix is a major task to ensure the quantification and unambiguous identification of such compounds [1]. Classical sample preparation techniques, such as liquid–liquid extraction (LLE) and solid-phase extraction (SPE), are usually time-consuming and labor-intensive. These techniques usually use large volumes of organic solvents, which are expensive and generate a considerable amount of waste that is harmful for human health and the environment [2].

Microextraction techniques such as those based on sorbent microextraction and liquid-phase microextraction are considered of great importance, since they represent an environmentally friendly alternative to classical extraction methods [3]. There are different microextraction configurations and modes of use. Sorbent microextraction may be considered as an advanced and miniaturized solid phase extraction (SPE) technique. Solid phase microextraction (SPME) [4] and thin film microextraction (TFME) [5] belong to this category. Similarly, liquid phase microextraction (LPME) can be considered as miniaturized liquid–liquid extraction procedures [6]. Most LPME techniques used include dispersive liquid–liquid microextraction (DLLME) [7] and single drop microextraction (SDME) [8,9].

In general, microextractions are carried out using an appropriate extraction phase, which can be a liquid [6] or a solid material [10], depending on the technique chosen. There is a large variety of extraction phases commercially available. However, in the last decade, efforts have been devoted to the development of new materials to be used as "greener" extraction phases. The green aspects of these alternative materials contribute to a less harmful and lower-cost analysis [3]. Furthermore, their usage reinforces the environmentally friendly character of microextraction techniques. In some specific cases, it increases selectivity and hence applicability for treating complex samples. For example, ionic liquids (ILs) and their tunable properties meet the criteria for extracting some compounds.

Based on that, the aim of this work is to review the new materials used as green extraction phases for the determination of organic compounds by microextraction and chromatographic techniques. Furthermore, some recent applications of these materials in various matrices are presented.

#### **2. Biosorbent-Based Extraction Phases**

Natural, renewable and biodegradable sorbents are denominated biosorbents and have attracted a great deal of attention in the sample preparation area, due to their low cost, non-toxicity and high availability [11]. There are several materials from different sources that can be used as biosorbents, such as agricultural waste products, industrial by-products and biomass derived from usually discarded materials [12]. Some materials such as cork, bamboo charcoal, bract, and recycled diatomaceous earth have already been applied in extraction phases to a large number of microextraction techniques based on solid phase extraction. Thus, in the following topics, the characteristics and some relevant recent applications of these materials in the biosorbent-based extraction phase will be discussed. More information about the applications and validation parameters of the reported methods are summarized in Table 1.

#### *2.1. Cork as a Biosorbent*

Cork is the bark of the cork oak tree (*Quercus suber* L.) and as a lignocellulosic material, it is composed of 40% suberin, 24% lignin, 20% cellulose and hemicellulose and 15% of other extractives [13]. In 2013, Dias et al. [14] proposed, for the first time, the use of a cork-based biosorbent as a coating for the solid phase microextraction technique (SPME) introduced by Pawliszyn et al. in 1990 as a miniaturized technique [4]. The procedure to obtain SPME cork fiber involves immobilization of the cork powder (approx 200 mech) on a nitinol wire of 0.2 mm thickness and approximately 2 cm length. After this, the wires with biosorbent are heated at a temperature of 180 ◦C for 90 min. Before use, the cork fibers produced are conditioned at 260 ◦C for 60 min in a gas chromatograph (GC) injection port [14].

The fiber characterization conducted using Fourier transform infrared spectroscopy (FTIR) and scanning electron microscopy (SEM) showed a heterogeneous chemical composition. Lignin presents several aromatic rings that may allow π-π interactions between sorbent phase and analytes, mainly the non-polar compounds. On the other hand, cellulose and hemicellulose exhibit a number of O–H groups in their structure, allowing for hydrogen-bonding and dipole–dipole interactions with the compounds presenting intermediate polarity. Furthermore, a homogeneously distributed coating and a porous structure are reported for the surface of the fiber. The coating thickness obtained for the proposed fiber was about 55 μm [14].

The SPME biosorbent-based fiber has already been successfully applied for the determination of polycyclic aromatic hydrocarbons (PAH) [14], organochloride pesticides (OCPs) [15] and UV filters such as 3-(4-methylbenzylidene) camphor (4-MBC) and 2-ethylhexyl 4-(dimethylamino) benzoate (OD-PABA) [16] (Table 1). In the work proposed by Dias et al. [14], the cork fiber extracted the PAHs by adsorption through π-π interactions and suberin was reported to play a more important role than lignin, in this case. The cork fiber was compared to commercially available fibers such as polydimethrylsiloxane/divinylbenzene (PMDS/DVB), divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) and polydimethylsiloxane (PDMS), presenting similar or better extraction efficiency for most compounds. An advantage reported by the authors was the lifetime of the coating layer, which was higher than the commercial ones, 50–100 times against 40 times, respectively.


*Separations* **2019**, *6*, 35

When used for OCP determination [15] the extraction efficiency of the cork fiber was attributed mostly to dipole–dipole interactions with the analytes. The authors also reported the occurrence of hydrogen bonds with the compounds containing oxygen atoms. In the work of Silva et al. [16], the cork fiber extraction efficiency for 4-MCB and OD-PABA was compared with PDMS/DVB and PDMS fibers, and the results showed a better extraction efficiency when the cork fiber was used, for both analytes.

Cork has also been used with thin film microextraction (TFME) [17]. TFME comprises a new geometry for SPME, aiming to provide more sensitivity for this technique. The device used in TFME consists of a support coated with a thin layer of a sorbent phase that can be used in headspace or immersion mode. Moreover, this technique has been designed to fit a commercially available 96-well plate system providing high-throughput analyses. To date, there has been only one study published using cork with TFME, by Morés et al. in 2017 [17]. The TFME cork coating coupled with 96-well plate system was used as a high-throughput method for the extraction of emerging contaminants in a water sample by high-performance liquid chromatography-diode-array detector HPLC-DAD (Table 1). In this work, analytes with the log Kow ranging from 2.49 to 5.92 were successfully extracted by the cork sorbent phase.

Another microextraction technique that used cork as the sorbent phase is called bar-adsorptive microextraction (BAμE) [18]. The BAμE device consists of a finely divided powder (up to 5 mg) fixed with suitable adhesives in polypropylene supports with cylindrical bar format. In the experimental procedure, the adsorbing bars are placed in direct contact with the sample, under constant stirring. Due to the low density of the polypropylene support, this floats just below the vortex formed by agitation, preventing direct contact of the bar with the flask's walls containing the sample, thus increasing the useful life of the device. After extraction, the desorption step consists of completely inserting the bar into vials containing a few microliters of a suitable extraction solvent.

Cork biosorbent has been used twice with BAμE. It was first used in 2015 for determination of polar and intermediate polarity compounds (parabens, benzophenone and triclocarban) in water samples by HPLC-DAD [18]. In this study, bars of 7.5 and 15 mm in length were used. Hollow cylindrical polypropylene tubes (15 mm length and 3 mm diameter) were coated with adhesive films followed by a layer of the cork powder (200 mesh). Before use, the bars were conditioned under ultrasound agitation with 250 μL of acetonitrile (ACN) for 15 min. The half bars (length of 7.5 mm) were obtained by cutting the 15 mm length bar in half. As shown in Table 1, the quantification limits ranged from 1.6 to 20 μg L−<sup>1</sup> using a bar of 15 mm and 0.64 to 8 μg L−<sup>1</sup> using a bar of 7.5 mm.

The second study was published in 2017. At this time, the use of cork BAμE bars was extended to biological samples for determination of two potential lung cancer biomarkers (hexanal and heptanal) in human urine by HPLC-DAD [19]. In this study, the adsorptive bar surface was impregnated with 2,4-dinitrophenylhydrazine (DNPH) so that derivatization and extraction were accomplished simultaneously on the surface of the bar under acidic conditions. Relative recoveries in urine samples varied from 88 to 111% (Table 1). Figure 1 illustrates a scheme of the biobased BAμE procedure used. According to the authors, one of the main advantages was the low cost of the method, since polypropylene tubes, adhesive and cork obtained from cork stoppers were used to produce the devices.

**Figure 1.** Scheme of the biobased bar-adsorptive microextraction (BAμE) procedure to determine hexanal and heptanal in human urine by HPLC-DAD. Reproduced with permission from [19], Copyright Elsevier, 2017.

#### *2.2. Bract as a Biosorbent*

Another lignocellulosic material was reported in 2017 as a green extraction phase for SPME [20]. The material, called bract, is the non-developed seeds obtained from the tree *Araucaria angustifolia* (Bert) O. Kuntze, a conifer found in the south and southeast of Brazil and in eastern Argentina. This material is composed of 45% lignin, 46% holocellulose (cellulose and hemicellulose) and 15% total extractives. The process for obtaining bract-based fibers is similar to those already described for cork fibers. Both materials are similar; however, the cork powder presents a better attachment to the nitinol wire, so it is easier to handle. Bract has been used as an environmentally friendly and low-cost biosorbent coating for SPME for the determination of OCPs in river and lake water [20] and PAH's in river water [21] (Table 1). The characterization of bract fiber carried out by thermogravimetric analysis (TGA), SEM and FTIR showed that the fiber offers satisfactory thermal stability with no decomposition observed up to 260 ◦C. SEM micrographs presented a highly porous and rough morphology and a film thickness of approximately 60 μm [20]. Like cork, bract is also a lignocellulosic material. The FTIR spectrum revealed peaks related to O–H bond and C–H stretching assigned to polysaccharides and lignin. C=C stretching from the aromatic rings and a peak related to C–O–C bond were also identified. Figure 2 shows a scheme of the preparation (2A) and SEM micrographs obtained for bract SPME fiber (2B).

**Figure 2.** (**2A**) Scheme of the preparation of SPME fibers and (**2B**) SEM micrographs obtained for bract fiber (**a**) magnification of 1500× (**b**) magnification of 3000× and (**c**) a cross-section of the proposed fiber at magnification of 300×. Reproduced with permission from [20], Copyright Elsevier, 2017.

When applied to the determination of OCPs in water samples by gas chromatography–electron capture detection (GC-ECD), a satisfactory analytical performance was reported with limits of detection (LODs) ranging from 0.19 to 0.71 ng L−1. In addition, the biosorbent-based fiber provided efficient extractions when compared with the commercial mixed coating fiber DVB/Car/PDMS. In 2018, bract fiber was used for the determination of PAH's in water samples by gas chromatography–mass spectrometry (GC-MS) [21]. In this study, the LODs varied from 0.003 to 0.03 μg L−<sup>1</sup> (Table 1).

Bract has also been used with TFME combined with a 96-well plate for the determination of steroid estrogens in human urine by liquid chromatography fluorescence detector (HPLC-FLD) [22]. At this time, the target compounds presented intermediate polarity with log P ≤ 4.12. The extraction efficiency of the bract layer was explained by the π-π interactions between lignin and the analytes. The LODs of the method varied between 0.3 μg L−<sup>1</sup> for 17-β-estradiol and 3 μg L−<sup>1</sup> for estrone (Table 1). As an advantage, in this study, the authors reported the use of the 96-well plate system, allowing for 1.7 min/sample turnaround times for the proposed method.

#### *2.3. Recycled Diatomaceous Earth as a Biosorbent Material*

Diatomaceous earth is an amorphous silicate sediment originating from fossilized unicellular microorganisms on algae of the class Bacillariophyceae centricae. This material is composed mainly of silica dioxide and small amounts of aluminum, iron, calcium, magnesium, sodium and potassium. After being used for the filtration and clarification of beer in a brewery, the diatomaceous earth was subject to thermal treatment and then used as the extraction phase for SPME [23]. The FTIR characterization of the material revealed O–H bonds from silanol groups. Moreover, asymmetric stretching was reported assigned to Si–O–Si bonds, frequently found in silicate materials. This biosorbent has been used along with SPME for the determination of PAH's in river water samples by GC–MS [23]. In the comparison with the commercial fibers PDMS/DVB and PDMS, the biosorbent showed better extraction efficiency for most compounds, except for acenaphthylene (C12H8), fluorene (C13H10) phenanthrene (C14H10) and pyrene (C16H10), for which PDMS/DVB was better. In this work, the authors did not provide a possible explanation for the interactions between analytes and the biosorbent. However, a fiber limitation was reported regarding the use of salt in the optimization step. According to the authors, salt particles added to the samples can remain adsorbed in the surface of the extraction phase, causing fiber damage. In this case, if the salt addition is necessary to improve extraction efficiency, a cleaning step with water may be done before the fiber insertion into the GC injection port.

Other applications of this sorbent include TFME with 96-well plate system for the determination of endocrine disruptors in water samples by HPLC-DAD [24] and with BAμE in the determination of methyl and ethyl paraben, benzophenone and triclocarban in water by HPLC-DAD [25] (Table 1). In the work of Kirschner et al. [24], bisphenol A (BPA), benzophenone (BzP), triclocarban (TCC), 4-methylbenzylidene camphor (4-MBC) and 2-ethylhexyl-p-methoxycinnamate (EHMC) were successfully determined from environmental water samples. Considering the analyte structure, the authors attribute the extraction efficiency of diatomaceous earth to the O–H moieties presented in the sorbent and the O–H and N–H groups in the target compounds. In this work, information about the extraction phase stability in the presence of organic solvents was provided. After successive extractions, the biosorbent blades were able to be used without expressive loss in the extraction efficiency for at least 20 extraction/desorption cycles. The proposed method exhibited satisfactory analytical performance, with LODs varying between 1 and 8 μg L−<sup>1</sup> and determination coefficient ranging from 0.9926 to 0.9988.

#### *2.4. Other Materials Used as Biosorbents*

A range of other materials characterized as biosorbents have also been used in combination with microextraction techniques for organic compound determination. Although there are still few applications involving these materials, a brief description is provided, along with the existent applications.

Bamboo plants are characterized by rapid growth and are widely distributed in China. Bamboo charcoal is obtained by submitting the bamboo to high temperatures (over 800 ◦C), producing a material with high density, porous structure and a large surface area. Bamboo charcoal was proposed as a novel and inexpensive SPME coating material for determination of 11 phthalate esters (PAE) in water samples by GC-MS [26].

Another material used as a biosorbent was obtained from powdered seeds of the *Moringa oleifera* tree. This material is considered to possess a highly fibrous and naturally functionalized surface. The characterization of the moringa-based biosorbent using SEM and FITR showed a porous framework of interconnected fibers, and various functional moieties were identified such as O–H, N–H and C–H and CH2 groups. The first application for organic compound determination was in 2016, as a sorbent for the determination of 13 phthalate esters (PE) in a milk sample by micro-solid phase extraction (μ-SPE) and GC-MS [27]. The relative polar PEs interacted with the sorbent through the functional moieties identified. The more the alkyl chain of PEs increased, the lower the extraction efficiency became.

An eco-material denominated montmorillonite (MMT) clay, modified through the intercalation of ionic liquids (IL), has also been applied in the extraction phase [28,29]. MMT is a clay mineral composed of structural layers consisting of an octahedral alumina sheet sandwiched between two tetrahedral silica sheets. MMT is found in sediments, soils or rock and has been modified to adsorb organic compounds of low polarity from aqueous solutions. In 2016, Fiscal-Ladino et al. [28] used rotating-disk sorptive extraction (RDSE) and MMT in the extraction phase for the determination of polychlorinated biphenyl (PCB) compounds in water samples with GC-ECD (Table 1). The RDSE device consists of a rotating Teflon disk containing an embedded miniature magnetic stirring bar. In this study, SEM was employed to characterize the novel sorbent, and the results showed clusters of particles with a narrow size distribution of approximately 25 mm. The extraction efficiency achieved for the MMT modified with 1-hexadecyl-3-methylimidazolium bromide (HDMIM-Br) phase was compared with commercial phases and showed the highest response for all the studied analytes.

Very recently, the viability of MMT-HDMIM-Br as a green sorbent for RDSE was again investigated [29]. In this study, cork and montmorillonite clay modified with ionic liquid were explored for the determination of parabens in water samples by high-performance liquid chromatography—tandem mass spectrometry (LC-MS/MS). The proposed method presented limits of detection of 0.24 μg L−<sup>1</sup> for the cork and 0.90 μg L−<sup>1</sup> for the MMT-HDMIM-Br with correlation coefficients higher than 0.9939 for both biosorbents.

#### *2.5. Concluding Remarks about Biosorbents*

In general, biosorbents demonstrated great versatility for the extraction of the different classes of compounds. Lignocellulosic biosorbents, such as cork, bract and *Moringa oleifera* seeds, are mainly composed of lignin, cellulose and hemicellulose. These macromolecules have a number of chemical groups that are capable of interacting with a wide range of analytes with different polarities. The works reported in this review showed studies in which cork was able to satisfactorily extract non-polar compounds, such as PAH's, and compounds with intermediate polarity, such as parabens, benzophenone and triclocarban. Bract biosorbent demonstrated similar behavior, presenting satisfactory extraction for compounds with low polarity, such as OCPs, and for those with intermediate polarity, such as steroid estrogens.

The extraction efficiency of these lignocellulosic materials is mostly explained by the π-π interactions between lignin and the analytes, or through hydrogen-bonding and dipole-dipole interactions between O–H groups presented in the cellulose and hemicellulose with the O–H, N–H bonds and Cl present in the analytes. When biosorbents were used with SPME for PAH extraction in water samples, bract fiber showed lower LODs than cork and diatomaceous earth fiber. The same was observed for OCP determination in water samples. Bract has a higher percentage of lignin in its structure than cork, which could explain the higher extraction efficiency for the non-polar compounds.

By using BAμE as the microextraction technique, the diatomaceous earth bar provided lower LODs than the cork bar for the extraction of parabens, benzophenone and triclocarban in water samples. Although diatomaceous earth has been used for PAHs determination, it has shown good extraction

efficiency for compounds with intermediate polarity, which was mainly due to interactions through O–H groups. It is also worth mentioning that the porous structure of these biosorbents plays an important role in the extraction through physical interaction with the analytes.

As a final remark, cork has been the material most used with different microextraction techniques and for a large variety of compounds. This fact can be related to the ease with which it is obtained through the reuse of wine bottle corks. The other biosorbents are more limited, such as bract, which is obtained from trees in southern Brazil and in eastern Argentina. Similarly, diatomaceous earth is a sub-product from the beer filtration and clarification process. Most of the works report the comparison with commercial extraction phases. In general, the results are similar or even better, in some cases. However, the procedures employed in the preparation of the devices, in particular for SPME, may be a limitation for the widespread use of these bio-based extraction phases.

#### **3. Ionic Liquids (ILs) as Green Extraction Phase**

Ionic liquids (ILs) are non-molecular solvents with melting points below 100 ◦C, negligible vapor pressure at room temperature, high thermal stability and variable viscosity. The ILs' miscibility in water and organic solvents can be controlled by selecting the cation or anion combination or by the addition of certain functional groups in the IL molecule. Most often, ILs are composed of large asymmetric organic cations and inorganic or organic anions. The most usually employed IL anions are polyatomic inorganic species, such as PF6− and BF4−, and the most relevant cations are a pyridinium and imidazolium ring with one or more alkyl groups attached to the nitrogen or carbon atoms [30]. ILs have been successfully applied to the liquid phase microextraction technique (LPME) as a less toxic alternative to conventional organic solvents. Considering these most notable properties, the potential usage of ILs as the extraction phase for LPME and applications has already been extensively reviewed by different authors [31–35]. The successful use of ILs in extraction phases is related to their structure. In addition to the common interactions existing in conventional solvents, ILs also have ionic interactions which confer miscibility when dissolved in polar substances. At the same time, the presence of alkyl chains in the cation determines the solubility in less polar substances. A review by Han and coworkers in 2012 presents the physical properties of some of the most commonly used ILs [32].

In 2003, Liu et al. [36] reported the first application of ILs in the extraction phase in single drop microextraction (SDME). SDME is a simple, easy-to-operate and reliable LPME-based method developed in the 1990s [8]. In this report, IL-based SDME coupled with HPLC was applied for the preconcentration and analysis of polycyclic aromatic hydrocarbons (PAHs) using the IL 1-octyl-3-methylimidazolium PF6 [C8C1IM-PF6] as the extraction solvent. Compared with 1-octanol, ILs provided higher enrichment factors (EFs), enabling the use of extended extraction times and larger drop volumes. In 2015, Marcinkowski et al. reviewed the analytical potential of ILs in SDME [37].

One of the most relevant applications of ionic liquids concerns their use as the extraction phase in dispersive liquid-liquid microextraction (DLLME). DLLME is a powerful extraction technique in which microliter volumes of an extraction solvent are dispersed in the sample to extract and preconcentrate the analytes [7]. The tunable properties of ILs have made these solvents particularly attractive for DLLME applications. Trujillo-Rodríguez et al. [38] reviewed in 2013 the use of ILs in the different types of DLLME and Rykowska et al. [39] recently reviewed modern approaches for IL-DLLME. In a recent application, IL-DLLME was used for the first time in the extraction phase for cortisone and cortisol determination from human saliva samples by HPLC-UV. The method provided high selectivity and EFs to achieve biological levels [40].

#### *3.1. Magnetic Ionic Liquids (MILs) as Green Extraction Phase*

A subclass of the ILs, denominated magnetic ionic liquids (MILs), has also been used as a green alternative to conventional organic solvents in LPME applications. Their physicochemical properties are similar to conventional ILs; nonetheless, MILs exhibit a strong response to external magnetic fields. MILs are obtained by the introduction of a paramagnetic component into the cation or anion of the IL

structure. Often the paramagnetic component is comprised of a transition or lanthanide metal ions [41]. Synthesis, properties and analytical applications of MILs, including micro extractions, have been already reviewed [42].

MILs have been applied to many LPME techniques. Table 2 shows the most recent applications (since 2017). However, most of the MIL-based extraction approaches are performed using DLLME. In this case, a mixture containing the MIL, dissolved in a small amount of an organic solvent, is dispersed in the sample and then recovered with a magnetic rod. The first application of MIL-DLLME was described in 2014 for the extraction of triazine herbicides in vegetable oils using 1-hexyl-3-methylimidazolium tetrachloroferrate ([C6mim] [FeCl4]) as the extraction phase [43]. Very recently, Sajid et al. [44] reviewed significant milestones of employing MILs for analytical extraction application and the main drawbacks of using MILs with DLLME.

Among the most recent applications, one in particular has attracted significant attention, since a new generation of MILs suitable for in situ DLLME were presented. MILs comprising paramagnetic cations containing Ni(II) metal centers coordinated with four N-alkylimidazole ligands and chloride anions were used for in situ DLLME and extraction of both polar and non-polar pollutants in aqueous samples. In this work, a metathesis reaction was originated by mixing a water-soluble MIL into the aqueous sample followed by the addition of bis [(trifluoromethyl) sulfonyl] imide ([NTf2-]) anion. This reaction produced a water-immiscible extraction solvent containing the preconcentrated analytes. The MIL was then isolated by magnetic separation and subjected to analysis using reversed-phase HPLC-DAD. The proposed methodology achieved higher extraction efficiency when compared to the conventional MIL-dispersive liquid-liquid microextraction. Extraction efficiencies ranging from 46.8 to 88.6% and 65.4 to 97.0% for the [Ni(C4IM)4 <sup>2</sup>+]2[Cl<sup>−</sup>] and the [Ni(BeIM)4 <sup>2</sup>+]2[Cl−] MILs were obtained [45].

MILs have also been successfully applied to the SDME technique. In a recent study, a high-throughput parallel-single-drop microextraction (Pa-SDME) was developed [46]. According to the authors, Pa-SDME combines some advantageous features of trihexyl (tetradecyl) phosphonium tetrachloro manganite (II) ([P6, 6, 6, 14<sup>+</sup>]2[MnCl4 <sup>2</sup>−]) MIL such as drop stability and extraction capacity with the 96-well plate advantages for obtaining high-throughput analysis. In this study, the determination of parabens, bisphenol A, benzophenone and triclocarban was conducted from environmental aqueous samples by HPLC-DAD. The method validation was carried out after the optimization step, and LODs ranging from 1.5 to 3 μg L−<sup>1</sup> were achieved. Coefficients of determination were higher than 0.994, and intraday and interday precision ranged from 0.6 to 21.3% (*n* = 3) and 10.4–20.2% (*n* = 9), respectively. Relative recovery ranged between 63% and 126%. Figure 3 shows the Pa-SDME lab-made extraction apparatus used for the extractions.


**Table 2.** Recent applications of magnetic ionic liquids (MILs) for extraction of different analytes from various matrices. DLLME—liquid–liquid dispersive microextraction.


**Table 2.** *Cont.*

**Figure 3.** An overview of the extraction procedure using the novel parallel-single-drop microextraction (Pa-SDME)/MIL-based approach. Reproduced with permission from [46], Copyright Elsevier, 2017.

#### *3.2. Deep Eutectic Solvent (DES) and Natural Deep Eutectic Solvents (NADES) as a Green Extraction Solvent*

The concept of deep eutectic solvents (DESs) was first introduced by Abbot et al. in 2003 [56]. DESs consist of two solid compounds interacting via hydrogen bonds to form a liquid phase with a lower melting point compared to each individual component [57]. The most popular DES involves the combination of choline chloride (ChCl) with urea, carboxylic acids (e.g., citric, succinic, and oxalic acids) and glycerol acting as hydrogen bond donors (HBDs). The use of ChCl has been related to some advantages for DES production, including ease of preparation, biocompatibility, non-toxicity and biodegradability. Although DESs are considered a subclass of IL, they are cheaper and easier to prepare due to the lower cost of the raw materials. Also, they present less toxicity and are often biodegradable, which makes them valuable alternative solvents. One of the most attractive features of these solvents is that, like ILs, their chemical properties can be tuned through the manipulation of their chemical structures (HBA and HBD) to interact more effectively with the target analytes. Florindo et al. (2018) [58] provide a closer look into DES intermolecular interactions; however, there is still a lack of knowledge regarding this topic.

In 2017, Shishov et al. published a review of the applications of DES in analytical chemistry, including their use in the extraction phase in microextraction techniques [59]. Nowadays, these solvents represent a very promising alternative in the sample preparation area, mainly due to their easy acquisition and versatility for extract different classes of compounds. In the work of Farajzadeh et al., a gas-assisted DLLME method using a mixture of ChCl and 4-chlorophenol (1:2 molar ratio) as the extraction solvent was developed for pesticide residue determination in vegetable and fruit by GC-FID [60]. The proposed method was optimized, and enrichment factors and extraction recoveries were achieved in the range of 247–355 and 49–71%, respectively.

Two hydrophobic deep eutectic solvents were synthetized and used as the extraction solvent with air-assisted DLLME (AA-DLLME) for pre-concentration and extraction of benzophenone-type UV filters from aqueous samples and determination by HPLC-DAD [61]. DESs were obtained by mixing DL-menthol and quaternary ammonium salts with a straight-chain monobasic acid. After optimization, a DES consisting of DL-menthol and decanoic acid mixture (1:1) was chosen for UV-filter extraction. Analytical parameters of merit were evaluated, and the developed method exhibited low limits of detection (0.5 to 0.02 ng mL−1) and repeatability in the range of 1.5–4.9 and 0.6–5.6% for intraday (*n* = 6) and interday (*n* = 6) determinations, respectively. The method was applied to determine the benzophenone-type filters in environmental water samples, and relative recoveries ranged from 88.8 to 105.9%.

Recently, a novel approach for effective liquid-liquid microextraction based on DES decomposition was reported [62]. In this work, DESs were synthesized from tetrabutylammonium bromide and long-chain alcohols. Afterwards, they were decomposed in the aqueous phase, resulting in an in-situ dispersion of organic phase and extraction of hydrophobic analytes. The method was applied to 17b-estradiol microextraction from transdermal gel samples. Efficient extraction of 95 ± 5% and reproducibility of 6% were obtained. Figure 4 shows a scheme of liquid-liquid microextraction based on in-situ decomposition of deep eutectic solvent.

**Figure 4.** Schema of liquid-liquid microextraction based on in situ decomposition of deep eutectic solvent. Reproduced with permission from [62], Copyright Elsevier, 2019.

Natural deep eutectic solvents (NADESs) are considered a sub-class of DESs, and they consist of a mixture of cheap and natural compounds such as sugars, alcohols, organic acids, and amino acids [63]. The most significant features of NADESs include adjustable viscosity, since they are liquid at temperature below 0 ◦C, sustainability, and the capability of dissolving a diverse range of analytes with different polarities. In a review by Hashemi et al., (2018) [63] the authors show the most common compounds for preparation of DES and the chemical structure of the most used NADES components. Cunha et al., 2018 [64] presented the main LPME techniques using DES or NADES as extraction solvents to determine several polar volatile and non-volatile compounds from food and water matrices. Table 3 shows some recent applications of DES/NADES (since 2017) with microextraction techniques for the determination of organic compounds in various matrices by chromatographic methods.


**Table 3.** Recent applications of deep eutectic solvents (DES)/natural deep eutectic solvents (NADES) for extraction of different analytes from various matrices.

1—5, 6, 7, 8-Tetrahydro-5, 5, 8, 8-tetramethylnaphthalen-2-ol; 2—tetrabutylammonium chloride ([N4444] Cl, TBA).

#### **4. Supramolecular Solvent (SUPRAS) as Extraction Phase**

Supramolecular solvents (SUPRASs) are nano-structured liquids in which the spontaneous association of different molecules self-organizes in a biphasic system formed by a continuous and a dispersed phase. SUPRASs' amphiphilic nature highlights one of their main advantages, providing excellent solvation for a wide range of organic and inorganic compounds. This characteristic is due to the presence of supramolecular aggregates, promoting a solvent with different degrees of polarity [77–79]. Supramolecular solvents have already been used in the determination of several classes of compounds such as parabens, pesticides, polycyclic aromatic hydrocarbons and bisphenols [80–83].

A very recent SUPRAS paper showed SUPRAS applicability with LPME as the extraction solvent. In this work, cationic surfactants didodecyldimethylammonium bromide (DDAB) and dodecyltrimethylammonium bromide (DTAB) were used in the extraction solvent mixture. SUPRAS-based liquid phase microextraction (SUPRAS-LPME) was also used in the preconcentration of five TCs (tetracycline, oxytetracycline, chlortetracycline, methacycline and doxycycline) in milk, egg and honey samples. An alkaline solution of the analytes was preconcentrated via electrostatic and hydrophobic interactions in the presence of the SUPRAS extraction solvent. The extraction mechanism was confirmed by the exploration of SUPRAS' Zeta potential and particle size. According to the authors, the results showed an excellent quantification method using SUPRAS and LPME for the determination of TCs in various matrices [84].

Recently, an innovative study proposed the application of a novel hexafluoroisopropanol (HFIP)/Brij-35 based SUPRAS in the determination of some organic compounds in water samples also using LPME. Brij-35 is a budget-friendly and non-toxic anionic surfactant that has a high cloud point (>100 ◦C). Presenting characteristics of a strong hydrogen-bond donor, elevated density and high hydrophobicity, HFIP was used in Brij-35 s density regulation and cloud-point reduction to below room temperature. The HFIP/Brij-35 SUPRAS-based LPME procedure allowed its preparation at room temperature with centrifugation only, making it very simple. Quantification of parabens with HFIP/Brij-35 showed good linearity and correlation coefficients higher than 0.9990. Spiked samples provided recoveries from 90.2% to 112.4% and relative standard deviation of lower than 9% [85].

#### **5. Bio-Based Solvents**

Bio-based solvents are a group of green solvents that have several advantageous characteristics such as low toxicity and non-flammability, besides being biodegradable and renewable, as they are produced from biomass and agricultural materials [86]. One example is ethanol, which has been used for decades in classical liquid-liquid extraction. Among others, glycerol, 2-methyl tetrahydrofuran (meTHF), ethyl lactate, p-cymene, and terpenes are part of the bio-based solvent groups that have attracted interest for applications in separation methodologies due to their characteristics [87]. A very interesting bio-based solvent is d-Limonene. It is derived from citrus peel and, like other bio-solvents, is low-cost and biodegradable and exhibits low toxicity. Its main attraction has been in the substitution of traditional solvents such as acetone, toluene, and chlorinated and fluorinated solvents in several applications. Due to its characteristics as a degreaser, this solvent has been applied to the removal of oils and fats [88]. In addition, d-Limonene may be a substitute for toxic organic solvents in Soxhlet extraction procedures [89].

Recently, a liquid–liquid dispersive microextraction (DLLME) method applying d-Limonene and B-carotene for the determination of b-cyclodextrin (b-CD) was developed [90]. These two mixed bio-solvents show a strong adsorption characteristic. When placed in the presence of b-CD, B-carotene forms a complex that increases the absorbance of the extracted phase, generating an excellent analytical signal for the determination of the target compound. The validation of the method presented an excellent limit of detection (0.00004 mol L−1) with a linear range from 0.0004 to 0.006 mol L−1. The method was applied for the determination of b-CD in water and pharmaceutical samples, obtaining recovery values between 94.2 and 108.0%, confirming the efficiency of the method.

Despite the advantages, these solvents are still poorly explored as the extraction phase for microextraction techniques, mainly with chromatography techniques. Some drawbacks may have to be overcome, such as the high viscosity that causes poor analyte mass transfer and also the incompatibility with analytical instrumentation [91]. Nevertheless, an alternative would be their combination with other green solvents or their use as modifiers of solid sorbents, as already proposed by Hashemi et al. (2018) [63].

#### **6. Conclusions**

The development of alternative green extraction phases represents an important research field in chemical analysis for the determination of different analytes from various matrices. This approach has been exploited in several recent publications and highlighted in this article. The use of biosorbents in analytical chemistry, mainly applied to microextraction techniques, is a very promising eco-friendly and cheap alternative. Cork, bract and diatomaceous earth have shown tremendous potential as alternative sorbents to commercial phases (PDMS/DVB, DVB/Car/PDMS and PDMS). However, preparation of the fibers may limit their use. Additional efforts need to be made in order to expand the applicability of the existing biosorbents to different groups of analytes. The use of green extraction phases as an alternative to conventional organic solvents has led to remarkable improvements with regard to environmentally friendly aspects. DES/NADES and SUPRAS are a very promising alternative in sample preparation. Additional research is needed to exploit their interactions with the analytes and also to expand their applicability. The application of bio-based solvents in the extraction phase should be further investigated, since there are only a few reports regarding these subjects.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors are grateful to the Brazilian Government Agency Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), for the financial support which made this research possible.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **Hunting Molecules in Complex Matrices with SPME Arrows: A Review**

#### **Jason S. Herrington \*,**†**, German A. Gómez-Ríos \*,**†**, Colton Myers, Gary Stidsen and David S. Bell**

Restek Corporation, Bellefonte, PA 16823, USA; colton.myers@restek.com (C.M.); gary.stidsen@restek.com (G.S.); david.bell@restek.com (D.S.B.)

**\*** Correspondence: jason.herrington@restek.com (J.S.H.); german.gomez@restek.com (G.A.G.-R.)

† Those authors contributed equally to this work.

Received: 1 August 2019; Accepted: 28 November 2019; Published: 15 February 2020

**Abstract:** Thirty years since the invention and public disclosure of solid phase microextraction (SPME), the technology continues evolving and inspiring several other green extraction technologies amenable for the collection of small molecules present in complex matrices. In this manuscript, we review the fundamental and operational aspects of a novel SPME geometry that can be used to "hunt" target molecules in complex matrices: the SPME Arrow. In addition, a series of applications in environmental, food, cannabis and forensic analysis are succinctly covered. Finally, special emphasis is placed on novel interfaces to analytical instrumentation, as well as recent developments in coating materials for the SPME Arrow.

**Keywords:** SPME; green chemistry; air sampling; complex matrices; mass spectrometry

#### **1. Introduction**

Solid phase microextraction (SPME) is a concept that embraces an array of technologies, or devices, with several common features:


SPME, originally conceived and patented by researchers at the University of Waterloo (UW) in the late 1980s, was licensed and commercialized at the beginning of the 1990s by Supelco Inc. (now Millipore-Sigma, Bellefonte, PA, USA). The first peer-reviewed manuscript, published in 1989 [7], brought to light the most know configuration of SPME: "the fiber" [8]. As shown in Figure 1, a traditional SPME device is composed of the following parts: a color coded screw hub (A), a sealing septum (B), a septum piercing needle (C), a fiber attachment needle (D), and a coated fused silica fiber (E). Though multifarious SPME devices have been developed since the mid-1990s for thermal, liquid, and inclusive laser desorption [1,4,9], the thin cylindrical geometry described by Bellardi et al. [7] is the most well-known and the leader in sales worldwide. Indeed, GC coupled to several detection systems (e.g., ECD, FID, and MS) is the most commonly used instrument to interface SPME devices.

**Figure 1.** Solid phase microextraction (SPME) Arrows and a traditional SPME fiber; (**1**) 1.5 mm SPME Arrow; (**2**) 1.1 mm SPME Arrow; and (**3**) 23 gauge traditional SPME fiber. d1: Support tubing; d2: septum piercing needle; d3: phase diameter; d4: phase support tubing diameter; l3: phase length; a3: phase area; and v3: phase volume.

Even though the SPME patent did not thwart academia/industry from conducting research on SPME devices and extraction phases [10–12], it categorically prevented corporations from commercializing improved versions of the "traditional" fiber (e.g., enhancements to the substrate to make a more robust technology). The expiration of intellectual property a few years ago enabled commercial vendors to not only offer the "traditional" SPME fibers, but to also mechanically and chemically enhance versions of this technology. An example of these enhancements includes the first large volume SPME fiber developed by CTC Analytics AG for GC applications, known as the SPME Arrow (see Figures 1 and 2) [13].

In parallel with the development of the SPME Arrow, other green chemistry technologies aiming to overcome the drawbacks of the traditional SPME fiber have also appeared [14]. Among them, one can highlight the thin film microextraction (TFME), the stir bar sorptive extraction (SBSE), and the in-tube extraction (ITEX). As recently reviewed by Dugheri and Olcer [9,14], most of these technologies offer better analytical features over the traditional SPME fiber. However, some of them are harder to automate [15] or are not compatible for direct immersion experiments [16]. Thus, the focus of the current review article is to summarize the fundamental and operational aspects of SPME Arrows, as well as recent developments and future directions [17,18].

**Figure 2.** Restek PAL traditional SPME fiber and SPME Arrow manual extraction and injection kit. (**1**) Extraction guide; (**2**) injection guide; (**3**) Arrow/fiber syringe; (**4**) large inner diameter (ID) locking screw; and (**5**) small ID locking screw.

#### **2. SPME Arrow Design**

As portrayed in Figure 3, the SPME device workflow comprises several steps including the following: 1. piercing the septum of the vial with the outer needle; 2. exposing the extraction phase to the sample and collecting the analytes of interest for a fixed period of time; 3. withdrawing the extraction phase into the needle of the SPME device; 4. transporting the SPME device to the instrument station, and 5. transferring the SPME device into the injection port of the instrument, so the analytes are eluted from the extraction phase via thermal desorption.

**Figure 3.** SPME (Arrow and traditional fiber) workflow for gas chromatography (GC) analysis. Manual SPME extraction (**1**–**3**) and injection (**4**–**6**). (**1**) Load SPME in manual syringe and extraction holder and position over sample vial. (**2**) Penetrate sample vial septum. (**3**) Press down on manual holder plunger to expose SPME extraction phase and begin sampling. (**4**) Load SPME in manual injection holder and position over GC inlet. (**5**) Penetrate GC inlet septum. (**6**) Press down on manual holder plunger to expose SPME extraction phase and begin desorption.

SPME Arrows were designed to overcome short-comings associated with the workflow of traditional SPME fibers, such as limited mechanical robustness ([19]), poor inter-device reproducibility, and small extraction phase volumes ([9]). The body of an SPME fiber is made of stainless steel, and the extraction phase is typically coated on fused silica. Consequently, one of the most common complaints of traditional SPME fiber end-users is a lack of physical durability of the fused silica ([20]). Though this problem can be partially alleviated by coating the extraction phase on a metal core (e.g., nitinol) [20,21], the body of a traditional SPME fiber is also commonly reported as fragile (see Figure 4). Consequently, commercial vendors have begun offering pre-drilled GC inlet septa and thin-walled vial septa to help mitigate issues with the damage of the core during the extraction and desorption steps. It is not uncommon for a SPME device to become injured, even within the first use. Failure rates appear to be highest amongst new end users; however, experienced end users are not immune to these issues either. Due to the delicate nature of traditional SPME fiber devices, most of them cannot be repaired. Manual extractions and desorptions appear to increase traditional SPME fiber failure rates, so the adoption of robotic autosamplers (e.g., CTC PAL or more commonly referred to as "rail" systems) has helped alleviate this problem to some degree. However, regardless of manual or automated injection, traditional SPME fiber devices typically fail to reach their true potential lifetime due to some sort of physical damage. "Potential" lifetime is stated because the majority of these mechanical failures appear to take place well before the fiber phase has been exhausted (i.e., it has not reached the life time of the coating).

**Figure 4.** Mechanical failures common to traditional SPME fibers.

#### *2.1. Physical Attributes*

SPME Arrows only share the following physical attributes with traditional SPME fibers: 1. the dimensions and thread types on the color-coded hubs; 2. the dimensions of the needle ferrule; and 3. a stainless steel composition of the support tubing and septum piercing needle. Beyond the aforementioned, SPME Arrows diverge from traditional SPME fibers in an attempt to increase mechanical durability. Traditional SPME fibers have 23 or 24 gauge (i.e., 0.573 or 0.511 mm, respectively) external diameters on their septum piercing needle. The SPME Arrow was initially developed with 1.1 and 1.5 mm (i.e., ~17 and 15 gauge) external diameters on their septum piercing needle, which is approximately 2 and 3 times the diameter of traditional SPME fibers, respectively. Figure 1 provides a scale image for visual comparison of the physical attributes of SPME Arrows and a 23 gauge traditional SPME fiber. Furthermore, Table 1 breaks down the divergent physical attributes of the SPME Arrows and a 23 gauge traditional SPME fiber. Most notably, the support tubing (i.e., plunger), septum piercing needle, and phase support tubing of the 1.1 and 1.5 mm SPME Arrows have increased

external diameters by 264–332%, 174–237%, and 583–449%, respectively over a 23 gauge traditional SPME fiber. These increased external diameters are largely responsible for the increased mechanical robustness of SPME Arrows. In particular, the increase in the support tubing appears to contribute the most improvement in durability, as this seems to be the most common failure point associated with traditional SPME fibers [13].


**Table 1.** Dimension breakdown of 23 gauge traditional SPME fiber (100 μm polydimethylsiloxane (PDMS), 1.1 mm SPME Arrow (100 μm PDMS), and 1.5 mm SPME Arrow (250 μm PDMS).

It is important to note that the SPME Arrows have "arrow" shaped tips from which they garner their name. These arrow tips have the same external diameter as the SPME Arrows' septum piercing needle (i.e., 1.1 or 1.5 mm). The arrow tip helps increase the mechanical robustness of the SPME Arrow, as the force required to penetrate the vial and/or GC inlet septa is less on an SPME Arrow when compared to a traditional SPME fiber, despite the increase in diameter. It has been demonstrated that a 1.1 mm SPME Arrow only requires 799 g of force to penetrate and headspace vial septum, whereas, a 0.63 mm traditional SPME fiber requires 1188 g (~50% more than the SPME Arrow) of force to penetrate the same configuration [22]. Furthermore, the SPME Arrow tip has been demonstrated to cut slits in vial and GC inlet septa, as opposed to coring septa. Therefore, septa lifetime appears to be as good, if not better, than traditional SPME fibers [23]. Furthermore, when retracted (i.e., not extracting/desorbing), the Arrow tip serves the purpose of a protective cap, thereby minimizing the diffusion of compounds into the septum piercing needle and ultimately reaching the phase. This helps minimize background contamination in between analyses or losses of analytes while the fibers are store for analysis on the tray of the GC system [21].

More recently, the SPME Arrow has been advanced with a 0.804 mm support tubing housed inside a 1.5 mm septum piercing needle (denoted as 1.5\* mm in Table 2). This design was released to overcome problems associated with using SPME Arrows for direct immersion (DI) extractions, such as phase swelling and the subsequently sloughing off and/or damaged when the SPME Arrow support tubing is retracted inside a 1.1 mm septum piercing needle. The wider 1.5 mm septum piercing needle provides enough clearance for the swollen phase on the 0.804 mm support tubing, thereby mitigating the sloughing issues.



\* 0.804 mm support tubing housed inside of a 1.5 mm septum piercing needle to allow for phase swelling during direction immersion (DI) extractions.

#### *2.2. Phase*

SPME Arrows were also designed to overcome the small phase areas and volumes associated with traditional SPME fibers. As shown in the Table 1, a 23 gauge traditional SPME fiber with 100 μm of polydimethylsiloxane (PDMS) has a 9.40 mm2 and 0.600 μL phase. The 1.1 mm Arrow with 100 μm of PDMS, which represents the most direct comparison to the aforementioned, has a 44.0 mm2 and 3.80 μL phase; which is a 468% and 633% increase in area and volume, respectively. This increase in phase area and volume has resulted in an increase in sensitivity and/or capacity, which is addressed later on in the performance and applications sections of this manuscript.

As shown in Table 2, SPME Arrows have been developed with most of the traditional SPME fiber phase offerings. The carbowax-polyethylene glycol (PEG) and carbowax/templated resin (CW/TPR) phase are the only phases not currently available on the SPME Arrow platform. Though the SPME Arrow has most of the phases, it is important to note that not all phase thickness/configurations have been replicated in the SPME Arrow. For example, although the SPME arrow is available in PDMS, it is only available in 100 and 250 μm PDMS configurations. The 7 and 30 μm PDMS configurations found in the traditional SPME fiber offerings are not available in the SPME Arrow. Additional phase thickness deviations may be observed when looking at the other phases (e.g., 75/85 μm of carbon on the traditional SPME fiber compared to 120 μm of carbon on the SPME Arrow).

#### **3. SPME Arrow Accommodations**

End users may not purchase a SPME Arrow and begin using it as direct replacement for the traditional SPME fiber. There are several things an analyst will need to modify/replace in order to accommodate the SPME Arrows' larger external diameters, including: 1. the injection tool (both manual and robotic); 2. the GC inlet; and 3. the GC inlet liner.

First, the manual tool may need modification or replacement (see Figure 2). For example, the traditional SPME fiber holder [19] does not accommodate SPME Arrows. Despite the SPME Arrows sharing a similar length and the same dimensions on the color-coded hubs, the Supelco manual holder tip diameter is too small. Some users have worked around this problem by drilling out the manual holder with a 3/64" metal compatible drill bit. Alternatively, other users have purchased the Restek PAL manual injection kit (Restek Corporation, Bellefonte, PA, USA), which accommodates both traditional SPME fibers and SPME Arrows without modification.

In terms of robotic platforms, CTC Analytics AG does not support the use of SPME Arrows on their second generation PAL systems (e.g., PAL/PAL-xt). Therefore, users with these systems will need to modify their existing SPME holder or acquire an appropriate SPME Arrow holder from Chromtech (Bad Camberg, Germany). Newer generations of robotic arms, such as PAL3 systems (e.g., RTC and RSI), fully support the SPME Arrow as long as users acquire the appropriate tool from CTC Analytics AG. In the case of rail systems commercialized by other vendors such as Agilent (Santa Clara, CA, USA), Gerstel (Linthicum Heights, MD, USA), Shimadzu (Kyoto, Kyoto Prefecture, Japan), and Thermo (Waltham, MA, USA), users need to contact the manufacturer to acquire the appropriate tool for their robotic system. Furthermore, the fiber conditioner found on second and third generation PAL systems does not accommodate the SPME Arrow diameters. Therefore, users need to directly condition the SPME devices in the GC inlet or acquire an SPME Arrow-specific conditioner module from the appropriate manufacturer.

In terms of the GC system, the inlets that are currently installed on the instruments made by major manufacturers (e.g., Agilent, Shimadzu, and Thermo) do not accommodate SPME Arrows. However, factory-modified GC inlets are commercially available for those manufacturers. Alternatively, users can modify their existing GC inlets by drilling out the excess of metal with a 3/64" metal drill bit. Likewise, traditional SPME fibers require the use of 0.75 mm inlet liners or greater. With the smallest outer needle diameter of 1.1 mm, SPME Arrows require the use of lager inlet liners. Several commercial vendors provide straight-walled inlet liners capable of accommodating SPME Arrows. For example, a 1.8 mm ID straight/inlet liner can accommodate both the 1.1 and 1.5 mm SPME Arrows. It is important to note, that in lieu of SPME Arrow-specific liners, end users may be able to use a standard 2.0 mm straight-walled liner and not suffer any significant performance losses [24].

#### **4. SPME Arrow Performance**

#### *4.1. Benchmarking*

In the previous sections, several advantages and disadvantages of SPME Arrows when compared to traditional SPME fibers were reviewed. However, those were mostly focused around the physical attributes associated with SPME Arrows. The current section focuses on how the aforementioned physical differences in phase area and volume translate into analytical performance. Several benchmarking applications have been chosen to compare SPME Arrows with traditional SPME fibers. It is important to note that the 1.1 mm SPME Arrow is the focus for the remainder of this section and for most of the manuscript, as this configuration is available in the most popular SPME phases (e.g., polydimethylsiloxane (PDMS) and divinylbenzene (DVB)), as shown in Table 2, whereas the 1.5 mm SPME Arrow is only available in a 250 μm polydimethylsiloxane (PDMS) configuration, which is anticipated to have less applications. It is important to point out that the 1.5\* mm SPME Arrows commercially launched a few weeks prior to the writing of this manuscript, so there are little data to warrant a discussion beyond what was mentioned previously.

When comparing a 100 μm PDMS 1.1 mm SPME Arrow against a 100 μm PDMS 23 gauge SPME fiber after the headspace (HS) extraction of volatile organic compounds (VOCs) in drinking water [spiked at 2.5 ppb as per International Organization for Standardization (ISO) method 17943] via GC coupled to mass spectrometry, it was found that equilibration times, extraction times, and desorption temperatures were equivalent among the two devices. Figure 5 presents a chromatogram overlaying instrumental response of the SPME Arrow and SPME fiber for the 92 VOCs. The SPME Arrow's response is higher than the traditional SPME fiber. On average, the SPME Arrow demonstrated a ~4× increase in response over traditional SPME fibers. This observation is consistent with the SPME Arrow's larger phase volume, which correlates to a greater volume/mass of target analyte collected and thereby an increased analytical response. It is important to note that "on average" was stated in the previous point, because the increase in response for very volatile compounds like vinyl chloride was ~10× on the SPME Arrow vs. the traditional SPME fiber. However, a semi-volatile compound like naphthalene only saw ~2× increase in sensitivity on the SPME Arrow compared to the traditional SPME fiber. Such differences in extraction recoveries are correlated to different vapor pressures (i.e., Henry's law constants) of said analytes and, expectedly, the availability in the headspace being the rate-limiting step. Likewise, it was observed that the SPME Arrow generated linear results (0.998 median R2) over a wide calibration range (0.0025–166 μg/L); with excellent precision (3.24% median RSD); and low sensitivity (30.0 ng/L median MDL) for all 92 ISO 17943 VOCs [25].

**Figure 5.** SPME Arrow overlay (TIC) on traditional SPME fiber for ISO 71943 headspace- volatile organic compounds (HS-VOCs) in drinking water. Acquired on Agilent 7890B/5977B GC-MS.

#### *4.2. Method Development*

Initial studies with SPME Arrows indicate that method development should follow the same logic and approach already demonstrated to be optimum for traditional SPME fibers [26]. For example, Herrington et al. evaluated SPME Arrows to see if there was a deviation in the extraction times required for SPME Arrows given the increase in phase volume. For this study, 100 μm PDMS 1.1 mm SPME Arrows and 100 μm PDMS 23 gauge traditional SPME fibers were evaluated for 92 HS VOCs, which had been spiked in drinking water at 2.5 ppb per method ISO 17943 [25]. Everything was equivalent (e.g., equilibration times and desorption temperatures), except for the extraction times. Extraction times of 15, 30, 60, 120, 240, 480, 960, and 1920 s were evaluated for each SPME (n = 3 for each SPME and each extraction time). Figure S1 shows the results from which the following two observations were made: 1. the SPME Arrow continued to demonstrate an increase in response (i.e., amount collected per unit of time) [9], which was again attributed to the increase in phase area over traditional SPME fibers; 2. both SPME types equilibrated at the same time (~120 s) for most of the 92 VOCs evaluated. This observation was consistent with the fact that both SPME types were 100 μm PDMS. Since the phase thickness was the same, it was deemed reasonable that the gas phase kinetics was the same; therefore, the equilibrium times were equivalent.

#### *4.3. Troubleshooting*

An extensive literature search only produced one reported issue associated with the SPME Arrow. Hartonen et al. reported chromatographic issues of double amine peaks when using SPME Arrows for extracting HS amines from water [27]. The root cause was determined to be an increased amount of water vapor extracted from the samples, due to the SPME Arrows' increased surface area and volume. However, it is important to note that other work on HS volatiles from drinking water [27] did not report any chromatographic issues and/or water issues. However, this work was conducted on thermally conditioned fibers with the use of sodium chloride in samples and split mode during desorption: the latter two were important conclusions Hartonen et al. arrived at [27]. In addition, Helin et al. did not report any issues when using SPME Arrow for HS amines in water with an SPME Arrow [17]. Finally, Gionfriddo et al. demonstrated short-chain aliphatic amines are best analyzed with the use of derivatization [28].

Beyond the aforementioned works, it is important to note that other work on HS volatiles from drinking water investigated the use of the 1.5 mm SPME Arrow with 250 μm of PDMS over a 1.1 mm SPME Arrow with 100 μm of PDMS [25]. As shown in Figure S2, light gases like chloroethane had ~2.5 times the response on the 1.5 mm SPME Arrow compared to the 1.1 mm SPME Arrow. However, the chromatography began to tail and split. A split injection could overcome this poor chromatography; however, then any sensitivity gains would be lost. It is believed that this tailing was a product of the fact that the phase was so thick on the 1.5 mm SPME Arrow that these lighter gases deeply penetrated into the thick phase, which then caused tailing up desorption. This work, Hartonen et al.'s work, and work not shown here have tended to suggest that SPME Arrows perform best with the use of a small (e.g., 2:1 or 5:1) split during thermal desorption. However, more extensive work will have to investigate this in the future. For instance, forthcoming work should consider the use of programmed temperature vaporization (PTV)-type inlets for SPME Arrows [14].

#### **5. SPME Arrow Applications**

#### *5.1. Environmental Analysis*

SPME technology has applicability for environmental pollutants in air, water, soil and sediment, and it can be used in the field or in the laboratory for sample preparation [17,18]. An advantage of the SPME Arrow is the larger phase volume that allows for the collection of a higher amount of the target compounds, so reporting requirements for environmental data users can be met.

One drawback of the traditional SPME fibers for headspace analysis is that the amount of analyte enriched on the coating is not sufficient to meet the reporting limits established by environmental agencies. The SPME Arrow design allows for four-to-five times higher analytical responses of compounds than a traditional SPME fiber [29]. For instance, the work performed by Kremser et al. presented a comparison of several techniques for determining volatile organic compounds including: purge and trap (P and T), ITEX, sample loop, traditional SPME fiber, gas-tight syringe, and SPME Arrow. As can be seen in Table 3, the detection limits for the SPME Arrow are in the same range as P and T and ITEX techniques, which are representative of exhaustive extraction techniques. However, the automation of workflow for the SPME Arrow is not only simpler than for P and T but also compatible with direct immersion experiments, which are not doable by ITEX [14].

In another study, Kaziur and collaborators developed a method capable of detecting picogram-per-liter (0.05 and 0.6 ng L−1) levels of water taste and odor compounds (i.e., isopropyl-3-methoxypyrazine, 2-isobutyl-3-methoxypyrazine, geosmin, 2-methylisoborneol, 2,4,6-trichloroanisole, 2,4,6-bromoanisole, and beta-ionone) in water samples [18]. As a matter of fact, this fully automated workflow was more sensitive than existing methodologies (see Table 4).


SPME PAL SPME Arrow

BIN

IPMP, IBMP, GSM, MIB,

TCA, BIN, TBA

40

10

 1 (2.6)–1000 (2600)

 0.05–0.6

 5–100

 0.2–0.5

<7

<11

 -

 [46]

Headspace work with SPME Arrow has been performed for the analysis of volatile amines in waste water. For example, after careful optimization of the extraction conditions, Helin et al. observed that the SPME Arrow produced lower detection limits, and higher recoveries of dimethyl amine (DMA) over the traditional SPME fiber (88% vs. 57%, respectively) [17].

The SPME Arrow has also been used for the analysis of semi-volatile compounds using immersion extraction. Typically, semi-volatile sample preparation for water samples has used liquid-liquid extraction (LLE) or solid phase extraction (SPE). Boyaci et al. did a comparison of SPME technology versus LLE and SPE, and it was found that the reduction of unwanted matrix interferences, the solventless extraction, the reusability, and the feasibility for high throughput sample analysis made SPME a more attractive technique [47]. Kremser et al. did extensive work with freely dissolved polycyclic aromatic hydrocarbons (PAHs) in water [48]. Extraction times and stirring rates were optimized to determine the method detection limits, and these were compared to previously published data for traditional SPME and SBSE (see Table 5). Direct immersion experiments using the SPME Arrow for freely dissolved semi-organic compounds showed detection limits five times lower than the traditional SPME fiber and similar results to stir bar sorptive extraction (SBSE). When comparing the SPME Arrow to SBSE, the authors highlighted the easiness of automation and utilization of shorter extraction times.

The SPME Arrow has also been used for environmental air sampling with the in-field analysis of biogenic volatile organic compounds (BVOCs). Monoterpenes (gamma-pinene and d3-carene) and aliphatic aldehydes (octanal and decanal) were selected by Barreira et al. to represent expected compounds to be found in field testing [49]. In-laboratory extraction efficiencies showed that the SPME Arrow had two-to-three times more area count than the traditional SPME fiber. Sampling was then performed in a boreal forest in Hyytiälä, Finland. Barreira et al. determined that the extraction efficiency of the SPME Arrow was two-times higher or greater, depending on compound, than the traditional SPME fiber, allowing for more sensitive testing (see Figure 6) [49].

**Figure 6.** Comparison between the average mass of identified monoterpenes (α-pinene, 13-carene, and limonene) and aldehydes (octanal, nonanal, and decanal) collected with different polydimethylsiloxane-divinylbenzene (PDMS-DVB) SPME devices (fiber and Arrow) from ambient air and measured by GC-MS. Figure reprinted with permission of Elsevier from Reference [48], 2019.


#### *5.2. Food Analysis*

There are not as many publications on SPME Arrow in food analysis, as compared to traditional SPME fibers [8]. However, in the past couple of years, several applications have spanned the analysis of diverse matrices including fish and rice [52,53]. For instance, Song and coworkers compared carboxen/polydimethylsiloxane (CAR/PDMS) SPME sorbents in the fiber format to the arrow format for the HS extraction of volatiles present in salt-fermented sand lance fish sauce [53]. The researchers reported that alcohols, aldehydes and pyrazines, with the exception of 1-pentenol, were more effectively extracted using the SPME Arrow. Some compounds that are believed to be important to the flavor profile were only observed using the arrow device. Lan and coworkers described a modified zeolitic imidizolate framework (ZIF-8) as a solid phase microextraction support on the arrow construct [54]. The group compared the novel adsorbent to a commercially available carboxen/polydimethylsiloxane device for the sampling of volatile amines in wastewater and food samples (salmon and mushrooms). The researchers found that the commercial ZIF-8 material exhibited a small pore size (5.6 Å), which likely excluded the model amine compounds, resulting in a low extraction efficiency. The acidification of the material significantly increased the extraction of small volatile amines, presumably due to an increase in pore size. The modified ZIF-8 arrow design provided comparable extraction efficiencies for small, volatile amines from several matrices as compared to commercial carboxen/polydimethylsiloxane SPME Arrow devices. Yuan and coworkers investigated metal organic framework (MOF) sorbents applied to arrow SPME devices for the determination of PAH contaminants in fish samples [55]. The authors employed a zirconia based UiO-66-molybdenum disulfide composite and compared the recovery of PAH contaminants from seafood samples to both commercially-available arrow and fibers coated with PDMS/carboxen/DVB coatings. The combination of the MOF composite sorbent and the arrow format resulted in and increased number of PAH species detected. Lan et. al. described the analysis of low molecular weight aliphatic amines from various matrices using several different modifications of silica sorbent applied to the arrow format [56]. The scientists demonstrated that dimethylamine and trimethylamine could be effectively detected and quantitated in mushroom samples using these devices.

#### *5.3. Terpenes in Cannabis*

With the rapid growth of the cannabis market taking place, the need for analytical testing has become more critical. An area of interest in this market is being able to identify and place cannabis flowers into the correct chemical variety, otherwise known as chemovar. To properly classify cannabis chemovars, a comprehensive chemical profile examining compounds, such as terpenes and cannabinoids, is collected [57]. An analysis of terpenes is typically done via headspace—gas chromatography—mass spectrometry (HS-GC-MS). Herein, a 1.1 mm 120 μm divinylbenzene (DVB)/PDMS SPME Arrow was used to analyze the terpene content in *Humulus lupulus* (hops), and individual terpene responses were compared to that of a HS-syringe method typically used for this application and a 65 μm DVB/PDMS traditional SPME fiber. Hops were used in place of cannabis, as cannabis could not be legally obtained. Hops were ground using drying ice and then stored in the freezer until needed. Ten-to-fifteen milligrams of ground hops were added to a 20 mL crimp top HS vial.

Current methodologies recommend extraction times of 10 min for the traditional SPME fiber analysis of terpenes [58]. Improvements to this parameter alone have the potential to decrease instrument runtimes and increasing the number of samples that contract laboratories are able to test. As can be seen in Figure 7, average responses for a 10 min extraction time using a traditional SPME fiber are equivalent to that of a 2 min extraction time with the SPME Arrow. This can be done without sacrificing reproducibility, as both techniques showed % RSDs under ≤10%. Lighter terpenes (monoterpenes) gave better responses on the SPME Arrow, while the heavier terpenes (sesquiterpenes) were more comparable. However, the traditional fiber did provide slightly better responses for the sesquiterpenes.

**Figure 7.** Extraction of VOCs in cannabis-related applications. Comparison of the traditional SPME fiber vs. SPME Arrow. Acquired on Agilent 7890B/5977B GC-MS.

#### **6. Future Directions**

The future of the SPME Arrow is promising, and among the multiple research and application avenues, this new geometry of SPME may grow rapidly. The following three areas are anticipated to expand: the development of new extraction phases [26,55,56], the Arrow's direct interface with MS instrumentation [59], and the Arrow's evolution towards smarter devices.

The first premise beyond the development of novel extraction phases is that the performance of these materials is better than those already commercially available by improving either robustness, extraction capabilities or device ruggedness [6,14,60]. For instance, the lack of inter-device reproducibility of one of the most popular extraction phases in SPME for GC analysis, the triple phase (DVB/Car/PDMS), has been its Achilles' heel [61]. Therefore, the recently launched SPME Arrow with a blended triple phase is expected to make a greater impact in the food and fragrance industry [62]. The main issue with the "original" tri-phasic SPME fibers relies on the process used for its manufacturing given that, if the two extraction phases (DVB-PDMS and Car-PDMS) are not properly aligned, this can lead to differences in the amount of extraction phase per coating length, thus leading to significant differences in the amount of analyte extracted—particularly volatile compounds [61]. In tri-phasic SPME Arrows, the particles of DVB and Car are embedded on the PDMS, resolving the issue of multiphase irregularities and consequently leading to better inter-device reproducibility.

As recently reviewed by Gomez-Rios and Mirabelli [4], the direct interface of SPME devices to MS has been increasing in the last five years, and SPME Arrows have not been an exception. For instance, research from the Zenobi's group demonstrated the applicability of an automated SPME

Arrow workflow, hyphened with a dielectric barrier desorption ionization (DBDI) coupled to an LTQ Orbitrap, to analyze common organic contaminants in treated wastewater (see Figure 8) [59]. Limits of detection as low as 3 ng/L were attained with a total analysis time per sample of less than 10 min (see Figure S3). Given that the ionization mechanism by DBDI, as well as by other direct-to-MS technologies, primarily relies on proton transfer for analyte ionization, a good response is typically attained for polar compounds with high proton affinity, whereas low ionization efficiencies are observed for nonpolar compounds. To overcome this limitation, a dopant-assisted DBDI ionization approach has been proposed to more efficiently ionize PAHs extracted from water. Though this strategy allowed for an up-to-an-order-of-magnitude signal enhancement for the analytes monitored, others go in detriment, and, consequently, the use (or not) of dopant must be driven by the target analytes. Certainly, DBDI is only one of the possible means of interfacing SPME Arrows to MS instrumentation. Other options include but are not limited to direct analysis in real time (DART) [63–65], atmospheric pressure photon ionization (APPI) [66], and thermal desorption-electrospray ionization (TD-ESI) [67]. It is anticipated that SPME Arrow-like geometries [11] compatible with liquid desorption and complex biological matrices will be developed in the near future and potentially coupled to direct-to-MS technologies such nano-electrospray ionization (nano-ESI) [68], desorption electrospray ionization [69], and the microfluidic open interface (also known as Open Probe Sampling Interface, OPSI) [70–72], as means to enhance the speed of analysis and the sensitivity [73]. Areas where such devices can make a great impact include but are not limited to, food fraud [74], volatomics [64], and rapid screening in the clinical chemistry realm [75].

**Figure 8.** Schematic drawing (**a**) and image (**b**) of the sample introduction system (with and without dopant), source, and dielectric barrier desorption ionization (DBDI) interface with the mass spectrometer. Figure reprinted with permission of Elsevier from Reference [59], 2020.

Finally, as part of the advances on analytical chemistry towards the Internet of Things and other intelligent platforms [76], the development of "smart" SPME devices has begun. The so-called "smart" SPME devices, currently commercialized by PAL, have been exclusively designed for their rail systems and comprise a chip that automatically informs instruments about the phase coated on the device and its usage history. Though the smartness of the device is strictly related to physical/mechanical aspects of the device, the development of Arrows with "smarter" coatings [6] (e.g., ionic liquids [77], MOF [78], and carbon nanotubes (CNT) [79]) and "smarter" geometries and substrates components [80] is expected.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2297-8739/7/1/12/s1, Figure S1: Extraction times for SPME Arrow and traditional SPME fiber for ISO 17943 HS-VOCs. Acquired on Agilent 7890B/5977B GC-MS, Figure S2: 1.5 mm SPME Arrow (red trace) versus 1.1 mm SPME Arrow (black trace) for analysis of chloroethane in water samples according to method ISO 17943 HS-VOCs. Acquired on Agilent 7890B/5977B GC-MS., Figure S3: SPME-Arrow and DBDI for determination of ppt of contaminants in waste water. Quantitation of four compounds of varying polarities and contaminant classes: (a) DEET, (b) Tamoxifen, (c) Pyrene, (d) Metolachlor. Figure reprinted with permission of Elsevier from Reference [59], 2020.

**Author Contributions:** C.M., G.S., D.S.B., J.S.H. and G.A.G.-R. wrote the review. J.S.H. and G.A.G.-R. planned and revised the document. All authors have read and agree to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** All the authors work at Restek Corporation and Restek Corporation commercializes SPME Arrows.

#### **References**


gesellschaft.de/dev/validierungsdokumente?download=32:f41-din-38407-41-2011-06&lang=de (accessed on 16 January 2020).


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*
