, Volume 82, Issue 1, pp 197–209 | Cite as

Reversed-Phase Functionalised Multi-lumen Capillary as Combined Concentrator, Separation Column, and ESI Emitter in Capillary-LC–MS

  • Estrella Sanz Rodriguez
  • Shing Chung Lam
  • Paul R. Haddad
  • Brett PaullEmail author
Part of the following topical collections:
  1. 50th Anniversary Commemorative Issue


For the first time, a multi-lumen capillary (MLC) (126 parallel channels of 4.2 µm i.d) has been modified to produce a C18-functionalised silica porous layer open tubular (PLOT) capillary column for both on-capillary preconcentration and separation. The modified multi-lumen capillary used in this dual mode provided significant advantages over typical nano/capillary-LC–MS systems, in that it facilitated both higher sample loading capacity, the use of elevated flow rates, and simplified equipment requirements. Following modification, 100% of the channels displayed a homogenous porous silica layer, 257 ± 36 nm thick. The PLOT-MLC was first evaluated for on-capillary solid-phase extraction. Extraction of caffeine, ofloxacin, atrazine, and diuron was carried out offline using an 8-cm-long PLOT-MLC, with quantification achieved using HPLC coupled to a quadrupole-time of flight (QTOF) mass spectrometer. The results confirmed reversed-phase selectivity and average recoveries obtained were around 70%. Subsequently, a 65-cm-long PLOT-MLC was evaluated as a separator column using a capillary liquid chromatography (Cap-LC) system equipped with a nano-injector and coupled to the mass spectrometer. The short PLOT-MLC provided a baseline separation in isocratic mode [water:acetonitrile (each with 0.1% formic acid) = 70:30, v/v] of ofloxacin, atrazine, and diuron. Finally, direct coupling of the PLOT-MLC with the QTOF via a capillary electrosprayer facilitated the simultaneous use of the modified capillary as a solid-phase concentrator, separator column (carrying out concentration-focusing-separation on the PLOT-MLC) and electrospray emitter. This configuration greatly simplifies the traditional capillary-LC–MS equipment requirements, via the removal of all connectors and additional capillary between injector and MS inlet, and is demonstrated herein with large volume sample loading and step-gradient elution/separation with sensitive MS detection.


Liquid chromatography Mass spectrometry Multi-lumen capillary C18-funtionalised fused silica Porous layer open tubular columns 


Open tubular liquid chromatography (OT-LC) has failed to emulate the success of open tubular gas chromatography (OT-GC), primarily due to the low micrometre scale i.d. columns required in LC to compensate for the 10,000 times lower diffusion coefficients in liquids compared to gases [1]. Theoretically, the OT-LC column i.d. should be in the order of a few micrometers (less than 6 µm) to obtain satisfactory efficiency [2]. In addition, there exists another consideration, namely an extremely low column capacity (thus complex sub-µL injector requirements), demanding the production of longer columns (> 1 m), which can be challenging in terms of uniform stationary phase layer formation and increased backpressures. For such OT columns reported to date (which have been typically ~ 10–20 µm i.d.), both these factors have limited potential wider application. Added to these rather fundamental restrictions, in the case of micro-bore capillary OT columns (e.g., sub-10 µm i.d.), practical and instrumental difficulties also exist, related to the low operational flow rates (due to increasing column backpressures), which necessitates specialist nano-flow pumps, and zero-dead volume nano-connectors, all of which can present complications, especially for gradient elution. Add to this reduced detection options, and the result is often compromised repeatability, reproducibility, and sensitivity [3]. These various issues have also acted to limit the use of such open tubular capillaries for on-capillary and on-line micro-extraction [4].

However, despite the above difficulties, many approaches have been successfully demonstrated to produce porous layer OT (PLOT) capillary columns, including roughening of the capillary internal surface [5, 6] and various ways to immobilise or grow either polymer-based [7, 8, 9, 10, 11, 12, 13] or silica-based porous stationary phase layers on the capillary wall [14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25]. Silica is the most widely used stationary phase medium in both packed and monolithic chromatographic columns for reversed and normal-phase LC, as it possesses many advantages over polymer-based substrates, including higher mechanical stability and greater resistance to swelling or shrinkage in non-aqueous solvents [7]. Conversely, the fabrication of silica-based PLOT columns is often considered to be complex and tedious, with long processing times compared to the preparation of similar polymer-based columns [26]. However, in 2016, the development of silica-based OT-PLOT capillary columns took a considerable step forward with the work of Hara et al. [1], who applied a sol–gel approach to successfully produce homogeneous porous silica layers (from 300 to 570 nm thickness) in 5.7 µm i.d. capillary columns, of lengths up to 2.5 m. The porous layer was subsequently functionalised using an octadecyldimethyl-N,N-dimethylaminosilane/toluene (20:80, v/v) solution. Separation efficiencies of up to N = 600,000 were demonstrated, realising the true potential of OT columns for LC, and undoubtedly reigniting interest in silica-based OT-LC. However, to obtain such efficiencies, which required limiting the volume of the solute band to below 1 nL, on-column fluorescence detection was required, together with nL min−1 flow rates, requiring a split ratio of 1/250,000, thus illustrating perfectly the practical difficulties mentioned above. Extremely small injection volumes, in the order of picoliters, were injected, and only highly fluorescent coumarins were shown as test solutes, injected at high concentrations, namely from 350 to 500 mg L−1.

A potential route to overcome issues related to high backpressures and limited sample loading capacity, as seen with the conventional PLOT capillary formats, is to use multiple parallel channels within a single capillary. Such capillary formats are now widely available in numerous designs and dimensions as the so-called fused-silica photonic crystal fibers (PCFs), also known as micro-structured fibers (MSFs) or multi-lumen capillaries (MLCs). The potential advantage of these fibers relates to the large number of precisely spaced, homogenous, and parallel micro-channels, which collectively provide a higher overall column capacity as a consequence of increased surface area, and thus capacity for larger sample loading, together with a reduced flow resistance and thus compatibility with higher operational flow rates. These capillary formats are available with the individual channel dimensions in the single digit micron range (e.g., 4 µm), and, therefore, have the potential not only for the development of selective micro-extraction capillaries, but also as multi-lumen PLOT capillary columns for liquid chromatographic separations [4]. To date, there has been very limited work reported using such MLCs in separation science, and their potential applications are still being explored. These recent applications include capillary electrophoresis (CE) methods for the separation of dye-labeled peptides and DNA [27, 28, 29], an application of wall-modified MLCs for the extraction of polyaromatic hydrocarbons (PAHs) [4] with subsequent separation using GC–MS [30], the OT-LC separation of 2–3 fluorinated species on fluorosilane-modified MLCs [31], and most recently, as an open tubular enzyme reactor [32] and in on-line solid-phase extraction [33].

Although the use of micro-bore MLCs in capillary LC and micro-extraction is relatively new, the concept of an array of capillary columns for LC was already being considered over 35 years ago. In their work, Meyer et al. [34] undertook theoretical calculations which predicted that multi-capillary columns could exceed the performance of packed beds over a wide range of efficiencies if variations in linear velocity between capillaries could be maintained to approx. 1% or less, which corresponds to diameter variation of 0.5%. Today, this fabrication challenge still remains difficult to overcome and is aggravated by the introduction of the stationary phase. However, continually improved manufacturing processes and specific strategies to form a monolayer without cross-linking and/or to increase surface coverage of the stationary phase are helping to improve upon the limited separation efficiencies described first by Meyer et al. (~ N = 100) in 1983 [34], Belov et al. (~ N = 2000) in 2005 [35], and Daley et al. (~ N = 700) in 2011 [31].

In addition to the above studies on MLCs for LC separations, Oleschuk et al. have recently been developing a series of multi-channel emitters for electrospray ionisation (ESI) based on MLCs [36, 37, 38]. The ESI emitter, through which the fluidic sample is delivered and sprayed into the mass spectrometer (MS) across an applied potential difference, is an essential element in the performance of ESI. When working with nano-scale LC, where flow rates under 500 nL min−1 are typical, small (< 20 µm i.d.) tapered emitters are used widely. However, such emitters exhibit a susceptibility to clogging, and thus alternative emitter formats employing multiple fluidic flow paths (i.e., multi-channel emitters) have been recently studied by different groups. For example, Oleschuk et al. [36] have explored the use of MLCs as multi-channel ESI emitters, modifying the exit surface of the silica-based MLC using chlorotrimethylsilane to allow the effective electrospray of up to 99.9% aqueous solutions, demonstrating excellent signal stability and sensitivity. Later, the same group fabricated a polymer nozzle array using a micro-structured fiber as a template for the generation of a 5-cm nanoelectrospray emitter [36, 38].

Herein, following on from the work of both Hara et al. [1] and Oleschuk et al. [36, 37, 38], we report upon an investigation into the modification of 4.2 µm i.d. parallel micro-channels within MLCs, with a uniform porous C18-silica layer. The reversed-phase PLOT-MLC was then evaluated for off-line in-capillary micro-extraction by LC–MS and also as a separator column in a standard capillary-LC–MS system. Finally, realising our main objective, we demonstrate, for the first time, the triple application of the modified MLC as a single platform for in-capillary micro-extraction, chromatographic separation, and multi-channel ESI emitter.


Chemicals, Standards, and Materials

De-ionised water was obtained from a Millipore Milli-Q water purification system (Bedford, MA, USA). Acetonitrile (ACN) was purchased from Honeywell Burdick & Jackson (Muskegon, MI, USA). Methanol, acetone, toluene, sodium hydroxide, polyethylene glycol (PEG, MW = 10,000 g mol−1), tetramethoxysilane (TMOS), and urea were obtained from Sigma-Aldrich (St. Louis, MO, USA). Acetic acid, formic acid (FA), hydrochloric acid (HCl, 37%), and tetrahydrofuran (THF) were sourced from Merck (Darmstadt, Germany). Octadecyldimethyl-N,N-dimethylaminosilane (ODS-DMA) was purchased from Fluorochem (Hadfield, UK). PTFE filters (0.22 µm × 25 mm), tubing, unions, adapters, sleeves, ferrules, and fittings were acquired from Kinesis Inc. (Vernon Hills, Ill, USA). Multi-lumen capillary (MLC) (LMA-15) was provided by NKT Photonics (Birkerød, Denmark). A scanning electron micrograph image of the unmodified MLC can be found at Electronic Supplementary Material Fig. S1.

Four model compounds of potential environmental significance, including one polar small molecule (caffeine), one pharmaceutical (ofloxacin), and two pesticides (atrazine and diuron), all from Sigma-Aldrich (St. Louis, MO, USA), were used as the test mixture for the various studies carried out herein.

Preparation of Porous Layer Open Tubular Multi-lumen Capillary (PLOT-MLC)

The desired porous silica layers within the MLCs were produced by following the basic protocol proposed by Hara et al. [1, 20, 21, 22] to modify their 5.7 µm i.d. fused-silica capillaries. Briefly, the layer was formed in-capillary via a “wetting transition” effect that can be induced through sol–gel processing. In their latest article, Hara et al. [1] reported three different recipes, varying the amount of TMOS and PEG to obtain different layer thicknesses. In this study, we tested two of the three recipes, specifically to obtain a thick and thinner stationary phase layer, naming them A and B, respectively. A homemade pressure bomb driven by nitrogen (Electronic Supplementary Material Fig. S2) and based on a published design [12] was used to pump the different solutions and solvents specified within each recipe through the MLC. A double reservoir allowed the preparation of two MLC capillaries at the same time. Nitrogen was delivered from a cylinder controlled by a high-pressure regulator operating at 100 bar. If smaller flow rates were required, pressure was reduced as necessary. As an example of the capillary preparation process, the following recipe was used to obtain the thinner layer (B): ~ 1-m-long MLCs were first treated with 0.1-N NaOH at 40 °C for 2 h, and washed consecutively with 0.1-N HCl, water, and acetone. A mixed feed solution was then prepared as follows: 0.215 g of PEG and 0.45 g urea were dissolved in 5 mL of 0.01 M acetic acid. Next, 2.5 mL of TMOS were added and the solution was stirred at 0 °C for 30 min. Following this, further stirring at 25 °C for 10 min was carried out. The feed solution was then filtered through a PTFE filter and subsequently pumped through the MLCs for 10 min after the first drop was observed at the exit of the fibers. The gel was formed and subsequently aged by leaving the MLC at 25 °C for 24 h. Next, a hydrothermal treatment was performed in an oven to produce mesopores from the ammonium carbonate generated by the hydrolysis of urea. The temperature was raised from 40 to 95 °C for 10 h and then kept at 95 °C for 15 h, before the temperature was cooled to 40 °C for 5–6 h. The MLCs were then washed with methanol for 5 days. After the washing process, the modified capillary columns were dried at 120 °C in an oven for 24 h. Once the porous silica layer was formed, in-situ functionalisation was carried out with a 20:80 (v/v) ODS-DMA/toluene mixture solution. Prior to charging with the functionalisation mixture, the MLCs were washed with tetrahydrofuran and toluene, both for 1 h. The ODS-DMA/toluene mixture was charged into the MLC at 35 bar for 24 h at 60 °C. Finally, a toluene wash and equilibration step with acetonitrile was carried out. Three different 1-m-long MLCs were modified following recipe B. A 1 m long MLC was also modified following recipe A, where the only difference with the preceding protocol was the amount of TMOS and PEG used, namely, 0.365 g of PEG and 3.6 mL of TMOS.

Scanning Electron Microscope (SEM)

Scanning electron microscopy images of modified MLCs were recorded using an Hitachi SU-70 field emission scanning electron microscope (FESEM). C18-silica layer thickness measurements and flow-through diameter (diameter of the open channel after the layer formation) within the modified MLC channels were undertaken by employing the measurement tool within the microscope software using a TEM grid for calibration.

Mass Spectrometry

A Bruker micrOTOF-Q II mass spectrometer (MS) from Bruker Daltonik GmbH (Bremen, Germany) was used throughout. Instrumental parameters are detailed at Electronic Supplementary Material Table S1. The MS was equipped with an Apollo II electrospray ionisation (ESI) source. Three different instrumental setups (A, B, and C, see Fig. 1) together with three different ESI sprayers were used in this study: a standard ESI sprayer, an ESI nano-sprayer, both from Bruker Daltonics (Bremen, Germany), and a capillary electrophoresis (CE) ESI–MS sprayer from Agilent Technologies (Waldbronn, Germany). Optimal ESI source parameters differed according to each ESI sprayer. Therefore, Table S1 shows the optimal values for each. Data acquisition was carried out in full MS positive ionisation mode. Extract ion chromatograms (EICs), corresponding to [M+H]+ ions for caffeine, ofloxacin, atrazine, and diuron, namely, m/z 195.08, 362.15, 216.10, and 233.02, respectively, were used to record the data obtained. MicrOTOF-Q II instrument control, data acquisition, and processing were via Bruker Daltonics Compass Software (version 1.3 SR1).

Fig. 1

Three instrumental setups using for the different tests carried out in the present study: setup A: LC–MS setup used for the evaluation of off-line PLOT-MLC micro-extraction; setup B: capillary-LC–MS configuration used for the evaluation of PLOT-MLC as a reversed-phase LC separator column; setup C: capillary-LC–MS configuration used for the evaluation of the PLOT-MLC as combined concentrator, separator column, and ESI emitter. On the left side setup C: a schematic shows the CE–ESI sprayer. On the right side of setup C: two positions for the divert valve used to control the flow through the PLOT-MLC. Chromatographic and instrumental parameters for the three setups are detailed at “Experimental” section. MS instrumental parameters for each configuration are shown at Electronic Supplementary Material Table S1

Off-Line PLOT-MLC Micro-extraction Procedure

To evaluate the applicability of the modified MLCs for in-capillary micro-extraction, an 8-cm-long modified PLOT-MLC was used. Eluate collected from the PLOT-MLC was analysed using LC–MS (setup A, Fig. 1). A standard ESI electrosprayer was used at the ESI source. The LC system included an Agilent 1200 series Binary Pump and Autosampler (Waldbronn, Germany). Chromatography was performed on a C18 column (150 × 4.6 mm, 5-µm particle size) from Thermo Scientific (Waltham, MA, USA), using a water:ACN with 0.1% FA, mobile phase gradient, from 90:10 to 5:95 in 15 min. A flow rate of 1 mL min−1 was split after the column with a T splitter at a split ratio of 4:1, providing 0.250 mL min−1 to the ESI source. The C18 column was held at 25 °C. The sample injection volume was 20 µL. Workflow was as follows: de-ionised water was spiked with caffeine, ofloxacin, atrazine, and diuron, each at concentrations of 20 µg L−1. 100 µL of the mixed standard solution was passed through the modified MLC using a syringe pump at a flow rate of 2 µL min−1, for in-capillary micro-extraction. The capillary eluate during loading was collected and 20 µL analysed by LC–MS to assess % extraction efficiency for each test compound. Following extraction, 20 µL of ACN was pumped through the MLC to elute the retained compounds. The eluate was collected and again analysed by LC–MS. This step was carried out in triplicate (eluate 1, 2, and 3) to determine the elution volume necessary to obtain maximum % recoveries. Between each extraction experiment, the modified MLC was preconditioned by flushing with 100 µL of ACN and 100 µL of de-ionised water. Before loading of the sample, 20 µL of ACN was passed through the MLC, and collected and analysed to verify no sample carryover. A schematic presenting this workflow can be found at Electronic Supplementary Material Fig. S3. Each extraction experiment was carried out in triplicate and calculation of % retention and recoveries based upon a five-point calibration curve.

Evaluation of PLOT-MLC as a Separator Column

As shown in Fig. 1, in setup B, the mobile phase was delivered isocratically (water:ACN with 0.1% FA) using an Agilent 1200 Series capillary pump (Waldbronn, Germany) equipped with electronic flow control (EFC) at a flow rate of 0.5 µL min−1. The pump was connected to a manual Cheminert nano-volume injector from Valco Instruments Co. Inc. (Houston, TX, USA) equipped with an internal loop of 20 nL. The PLOT-MLC (inlet) was connected directly into the injector and kept at 25 °C. The PLOT-MLC outlet was connected to a 10-cm × 20-µm i.d fused-silica capillary, itself connected to the ESI nano-sprayer at the MS. The ESI nano-sprayer includes a narrow bore fused-silica (22 cm x 20 µm i.d.) capillary as the emitter tip. *Different flow rates and mobile phase compositions were studied to assess chromatographic performances of the PLOT-MLC separator column.

Evaluation of PLOT-MLC as Combined Device for Extraction, Separator Column, and ESI Emitter

Figure 1 shows the setup C employed in this study. The CE–ESI–MS sprayer is typically used as the interface between a CE system and an MS detector. It is also called a triple-tube sprayer and a schematic can be found at Fig. 1. The PLOT-MLC was installed through the spray needle (first tube) up to the tip and is used as the ESI emitter. This interface utilises an additional sheath liquid channel (second tube), compared to a standard ESI sprayer, which flows around the first tube. Both concentric tubes are surrounded by the third tube transporting the nebulising gas (N2). The three flows mix at the stainless-steel needle tip, providing electrical contact at the outlet end and generating the spray and the subsequent ionisation process. In this study, ACN with 0.1% of formic acid was used as sheath liquid. This was delivered at 0.650 mL min−1 using an Agilent 1100 Series Pump (Waldbronn, Germany) equipped with a splitter (ratio 1:100). The sheath liquid flow rate at the CE sprayer was 6.5 µL min−1. The nebulising gas was delivered to the ESI source at 6 mL min−1. Mobile phase (e.g., water:ACN, 70:30, v/v, with 0.1% FA) was delivered using the capillary pump at 0.5 µL min−1. A divert valve installed on the MS was used to manage the different flows through the PLOT-MLC. The valve was a standard 6-port valve (see Fig. 1, setup C). A Chemyx Fusion 100 syringe pump (Indianapolis, IN, USA), was used to deliver the sample at a flow rate of 0.5 µL min−1 and was connected into the valve at position 3. The capillary-LC pump was connected to position 5. The PLOT-MLC inlet was connected to position 4 and then fed directly into the CE–ESI sprayer as shown. Positions 6 and 2 were directed to waste. Position 1 was not used. The precise workflow was as follows: Initially the divert valve was in the ‘inject’ position, to be sure that the sample delivered by the syringe pump filled all connecting tubing up to the divert valve. Next, when the divert valve was switched to the ‘load’ position, ports 3 and 4 were connected, allowing the sample to be loaded onto the PLOT-MLC (extraction/loading step). Different extraction times, corresponding to different injection volumes, were investigated. Ports 5 and 6 were connected, sending mobile phase from the capillary pump to waste. The ‘inject’ position switched the valve (ports 4 and 5 connected) such that the mobile phase flowed directly into the PLOT-MLC (elution step) and out into the MS. In this position, sample from syringe pump at port 3 was directed to waste at port 2. The Agilent 1200 Series LC Binary Pump, Capillary Pump, and Agilent 1100 Series Pump were controlled by HyStar 3.2, Chemstation B.04.01, and Chemstation B.01.03 software, respectively.

Results and Discussion

Modified PLOT-MLC

Figure 2 shows SEMs of the obtained porous layers within 1-m-long MLCs applying the two recipes (A and B) to obtain a thicker (Fig. 2a1–a2) and a thinner stationary phase layer (Fig. 2b1–b6), respectively. Although recipe A provided a homogenous layer in many channels, most of them presented a non-homogenous layer and, in some channels, a cross-linked layer or apparently empty channels. The fact that the channels within the MLC used are ~ 30% smaller than the 5.7 µm i.d. capillary used by Hara et al. [1], could explain the cross-linking or ‘bridging’ seen in Fig. 2a2 using the recipe for the thicker layer. However, with recipe B, SEM imaging confirmed that 100% of the channels were modified with a homogenous porous layer. Figure 2b3–b6 shows the porous layers formed applying recipe B within the 4.2 µm i.d. channels, in three separately modified MLCs. Taking a measurement of the layer thickness across all the channels within a single modified MLC revealed a 257 ± 36-nm layer thickness (14% RSD, n = 126). The measurements of the flow-through diameter (3.84 ± 0.09 µm) showed a 2.5% difference in diameter which will result, according with theory and models proposed by Meyer et al. [34], in a 5% difference in linear velocity. In these circumstances, the prediction is that fewer than 6400 plates can be expected from these MLCs when applied to LC separations.

Fig. 2

Modified PLOT-MLC: a1a2 SEM pictures of exampled channels of modified MLC following the recipe A to obtain the thicker C18-silica layer proposed by Hara et al. [1]. b1b6 SEM pictures of exampled channels of the modified MLC following the recipe B to obtain the thinner C18-silica layer proposed by Hara et al. [1]. b3b6 SEM pictures showed the thickness of the C18-silica layer obtained in different channels and different modified MLCs cuts

It is worth noting that, as Hara et al. [1] observed in their study, the volume of the obtained layer is directly proportional to the mass of TMOS fed into the capillary: the mass fed to the 5.7 µm i.d. capillary used by Hara et al. [1] was ~ 1.8 times greater than that for the 4.2 µm i.d. MLC channels used in this study, and the volume of the porous layer volume produced was ~ 1.7 times larger in their study. Thus, the MLC volumetric phase ratio (m), i.e., the ratio of the porous layer volume to that of the flow-through volume was found to be m ≈ 0.30, which is very similar to the phase ratio obtained in the 5.7 µm i.d. capillary (m = 0.26) by Hara et al. [1] when applying recipe A. This confirms the observation of Hara et al. [1] that the deposited silica forms a layer with a predetermined density, independent of the capillary diameter.

Application of Modified PLOT-MLC for Micro-extraction

Using an 8-cm-long length of the modified PLOT-MLC, application to on-capillary micro-extraction was investigated using a test mixture of caffeine, ofloxacin, atrazine, and diuron at low µg L−1 concentrations. Quantification of the test compounds following extraction and elution was achieved using reversed-phase LC–MS (Fig. 1, setup A). The workflow followed is detailed within the “Experimental” section and Electronic Supplementary Material Fig. S3. Figure 3 shows the extracted ion chromatograms (EICs) for a standard mixture of the test compounds before extraction [Fig. 3, chromatograms (a)], and the PLOT-MLC eluate collected during on-going extraction [Fig. 3, chromatograms (b)]. As can be seen from the chromatograms, no significant peaks for ofloxacin, atrazine, and diuron were detected in the capillary eluate. Quantification of each compound revealed loadings of 98.65 ± 0.64, 95.71 ± 1.64, and 99.21 ± 0.46%, for ofloxacin, atrazine, and diuron, respectively. However, as Fig. 3b for caffeine clearly shows, this more polar compound was mostly unretained (9.08 ± 3.62%). Calibration curves used to obtain those results were built using aqueous standard solutions of the test compounds at concentrations of 5, 10, 15, 20, and 30 µg L−1 for each compound. Regression coefficients were in the range of r2 = 0.980–0.999. In Fig. 3, chromatograms (c) show the extracted ion chromatograms (EICs) for the first ACN eluate (ACN eluate 1) for the four test compounds. The improvement in signal response is clear as a concentration step was achieved. As can be seen in Fig. 4a, average total recovery was ~ 70%, the majority of which was recovered in the first ACN elution (Fig. 4b). Quantification of the ACN eluate was carried out using the standard mixture solutions at 5, 10, 20, 50, and 75 µg L−1 of each compound in ACN. Regression coefficients for the calibrations curves obtained were from r2 = 0.990 to 0.996. For a 100% recovery, the theoretical maximum concentration factor (CF) under the current test conditions would be 5, based upon the 100:20 load and eluate ratio. Experimental CFs were 0.5, 4, 3.4, and 2.9 for caffeine, ofloxacin, atrazine and diuron, respectively (Fig. 4a). Although ofloxacin, atrazine, and diuron were almost 100% retained, further optimisation of elution conditions would appear to be necessary to improve % recovery. Recoveries for the retained compounds, ofloxacin, atrazine, and diuron, decreased proportionally as their affinity for the C18-silica layer increased, which would suggest that the low recoveries were an elution volume issue. Greater elution volumes (times) and optimisation of elution solvents would likely provide higher recoveries, although this extraction method optimisation was not the aim of the present study. Regardless, these results further confirmed the reversed-phase selectivity of the modified MLCs which was the main objective of these experiments. Despite the above recovery data, no carryover was observed between extractions after applying the cleaning protocol specified in the “Experimental” section.

Fig. 3

Application of modified PLOT-MLC for micro-solid-phase extraction: extract ion chromatograms (EICs) for ofloxacin, caffeine, atrazine, and diuron of a 20 µg L−1 aqueous standard solution, b aqueous eluate collected during extraction/loading step, and c first ACN eluate. Analysis were carried out using setup A showed in Fig. 1. Chromatographic conditions and instrumental parameters are detailed at “Experimental” section. Workflow and MS instrumental parameters are shown at Electronic Supplementary Material Fig. S3 and Table S1, respectively

Fig. 4

a Average (%) recovery and standard deviation (SD) for the four test compounds using modified PLOT-MLC for micro-solid-phase extraction. b % eluted in each of the three ACN eluates collected. Analysis were carried out using setup A shown in Fig. 1. Chromatographic conditions and instrumental parameters are detailed at “Experimental” section. Workflow and MS instrumental parameters are shown in Electronic Supplementary Material Fig. S3 and Table S1, respectively

Modified PLOT-MLC as a C18 Separation Column

Different flow rates (0.3, 0.4 and 0.5 µL min−1) and mobile phase compositions (water:ACN, with 0.1% FA, at the ratio of 90:10, 80:20, 70:30 and 50:50, v/v) were tested using instrumental setup B (Fig. 1). A standard mixture of caffeine and diuron at 1 mg L−1 and ofloxacin and atrazine at 0.25 mg L−1 concentrations was used as a test sample mixture. Isocratic elution using water: ACN, 70:30, v/v, with 0.1% FA as mobile phase at 0.5 µL min−1 was found to be suitable for the separation of these compounds. As can be seen in Fig. 5, even with a relatively short PLOT-MLC column (65 cm) a baseline separation for ofloxacin (co-eluted with caffeine), atrazine, and diuron was achieved. However, although promising, peak efficiencies were rather low (N = 312, 874, and 1000 for ofloxacin, atrazine, and diuron, respectively), here reduced to the complexity of coupling the MLC to standard single bore capillaries, wherein the uniform distribution of flow into the multiple channels is extremely difficult to achieve. Limitations in maximum efficiency from variations in flow-through diameters were unavoidable with the current MLCs; however, the use of the 10 cm × 10 µm i.d. fused-silica capillary to connect the PLOT-MLC to the nano ESI sprayer, in addition to the fused-silica capillary within the nano ESI sprayer (22 cm × 20 µm i.d.) (Fig. 1, setup B), was contributing a significant extra-column band broadening of the peaks which could be eliminated. Clearly then, to reduce the extra-column volume of the system, a system, whereby such connections are removed altogether (setup C), would be preferable. To achieve this, the system was modified to incorporate a CE–MS–ESI sprayer, which provides the capability of using the PLOT-MLC itself as both separator column and ESI emitter.

Fig. 5

Modified PLOT-MLC as C18 separation column coupled to MS: overlaid extracted ion chromatograms (EIC) for ofloxacin (1), caffeine (2), atrazine (3), and diuron (4) obtained using setup B (Fig. 1). Injection of 20 nL of a mixture at 1 mg L−1 for caffeine and diuron and 0.25 mg L−1 for ofloxacin and atrazine. Isocratic mode, water:ACN (70:30, v/v) with 0.1% FA at 0.5 µL min−1 flow rate. Other chromatographic conditions and instrumental parameters are detailed in the “Experimental” section. MS instrumental parameters are shown in Electronic Supplementary Material Table S1

Modified PLOT-MLC as Combined Concentrator, Separator Column, and ESI Emitter

Figure 1 shows the setup created for this study and, within the “Experimental” section, the full workflow, and experimental conditions are specified. The PLOT-MLC was installed within the co-axial CE–MS sprayer, such that the MLC projected out of the nebuliser tip at an optimal distance of 1 mm, to avoid any dead volume and enable the production of a stable spray. As the dimensions of the nebuliser are standard to couple with the ESI source, the tubing reserved for the sheath liquid is relatively large and a relatively high flow rate for the sheath liquid (usually 1–10 µL min−1) is required to maintain stable electrospray ionisation. However, although the sheath liquid serves to improve ionisation efficiency, the sheath liquid does introduce significant dilution of the PLOT-MLC effluent, which flows at 0.5 µL min−1. Accordingly, an optimisation of the sheath liquid flow rate was carried out. Different flow rates, namely, 0, 1.5, 4.5, 6.5, and 9.5 µL min−1, were evaluated using a standard solution of 0.15 mg L−1 of ofloxacin and atrazine and 1 mg min−1 of diuron. A 3.5-fold increase in peak area for ofloxacin and atrazine and twofold increase for diuron was observed when the sheath liquid flow rate was varied from 0 to 6.5 µL min−1. Higher flow rates decreased the response slightly. Consequently, 6.5 µL min−1 was chosen as the optimal sheath liquid flow rate. The responses obtained during these optimisation studies confirmed that the PLOT-MLC emitter was indeed able to generate an efficient electrospray, even with a high percentage of water in the mobile phase (70%). Highly aqueous solutions are particularly difficult to spray with hydrophilic silica-based emitters. The previous studies have modified the silica exit surface with a silylation reagent to reduce its wettability [34]. Using the modified C18-silica PLOT-MLC this issue was eliminated, as the silica surface was already fully hydrophobic, which combined with the optimal flow of sheath liquid at the ESI source, resulted in a stable and reliable ES ionisation. The spray generated a similar signal to a commercial tapered capillary emitter. This was the main objective here, as the aim was not try and produce a multi-tip emitter with improved ionisation.

The analytical performance of the modified PLOT-MLC in a combined concentrator/separator/emitter format was evaluated by constructing a calibration curve of five standard solutions of a mixture of ofloxacin and atrazine. The concentrations of each in the standard solutions were 25, 50, 75, 100, and 150 µg L−1. The injection volume (loaded on the PLOT-MLC by the syringe pump, see Fig. 1, setup C) was kept constant at 2.5 µL, loading the standard solutions over 5 min at a flow rate of 0.5 µL min−1. To achieve on-capillary focusing, standards were prepared in 100% water. During the loading process, MS data acquisition was operational to evaluate if each compound was fully retained. This was, indeed, the case and over the concentration range investigated, the calibration curves revealed a regression coefficient (r2) of 0.999 for both compounds. Under these conditions, retention time relative standard deviations (%RSD) for ofloxacin and atrazine was 5.55% and 8.65%, respectively, demonstrating that the combined extraction and elution procedure was reproducible and independent of concentration within the range and injection volume studied. Similarly, to evaluate the effect of different sample loading volumes with a fixed standard concentration, the following loading volumes were tested, 1.5, 2.5, 5, 7.5, and 12.5 µL, for a 150 µg L−1 standard of ofloxacin and atrazine, and 1 mg L−1 of diuron. These were loaded using the syringe pump, equating to 3, 5, 10, 15, and 25 min load times at a flow rate of 0.5 µL min−1. As before, this step was acquired by the MS to evaluate the retention of the compounds during loading. As an example, Fig. 6 displays a chromatogram obtained for a 2.5 µL load volume (5 min extraction time). The calibration curve for loading volume can be found at Electronic Supplementary Material Fig. S4. Results showed a quadratic fit (r2 from 0.990 to 0.999) for the three compounds tested, as a slightly increased response was observed at higher concentrations. Retention time %RSDs for ofloxacin, atrazine, and diuron were 2.41%, 2.74%, and 5.51%, respectively.

Fig. 6

Modified PLOT-MLC as combined concentrator, separator column and ESI emitter: overlapped extract ion chromatograms (EICs) for ofloxacin (1), atrazine (2), and diuron (3) obtained using setup C (Fig. 1). Vertical short dashed lines show frame time for 5 min extraction/loading time (2.5 µL injection volume). Horizontal long dashed line shows step-gradient applied indicating % ACN through PLOT-MLC on the right axis. Workflow and experimental conditions are specified at “Experimental” section. MS instrumental parameters are shown in Electronic Supplementary Material Table S1

A direct comparison between efficiencies obtained using setup B and C was not possible, as the latter involved large sample volume loading and on-capillary focusing. However, taking the chromatogram in Fig. 6 as that obtained by injection of a large sample volume and an application of on-capillary focusing, and eluted with an isocratic elution, peak efficiencies of N = 2846, N = 2000, and N = 1600 for ofloxacin, atrazine, and diuron, were achieved. That represents a ninefold improvement for ofloxacin and 2–1.5-fold improvement for atrazine and diuron, with respect to the efficiencies obtained employing setup B, albeit under different conditions. We consider the main reason for the improvement, achieved despite the large volume sample loading, as coming from the elimination of tubing and unions, plus the use of MLC as emitter, altogether significantly reducing extra-column broadening. It is worth noting that the better efficiencies obtained using set up C corresponded to an injection of 0.375 ng of ofloxacin and atrazine and 2.5 ng of diuron, 75 and 125 times higher amounts for ofloxacin/atrazine and diuron, respectively, than amounts injected using setup B. Peak width variations across the loading volumes investigated, measured as full width at half maximum (FWHM), showed variations of less than 0.1%. This implied that there was sufficient column capacity to focus the compounds of interest on the start of the column prior to their subsequent elution in isocratic mode. Here, it is important to note that, although the obtained efficiencies were lower than those possible with single open tubular sub-10 µm capillary columns, these are the best reported in the literature using MLC to date (see Electronic Supplementary Material Fig. S5).

Low mass loadability of single-channel PLOT LC columns is a considerable limitation, so, to evaluate this regarding the PLOT-MLC, a standard solution of 10 mg L−1 of atrazine prepared in 100% water was continuously loaded on the PLOT-MLC at 0.5 µL min−1 to determine a breakthrough volume (monitoring m/z 216.10, corresponding to atrazine positive ion). Repeat experiments gave an average breakthrough point at ~ 46 min, indicating a mass loadability of 233.75 ± 5.30 ng for atrazine. Figure S6 in Electronic Supplementary Material shows the extracted ion trace for atrazine during this loading study. Here, it is not possible to compare these data with the same single channel PLOT column as reported by Hara et al. [1] as information on mass loadability was not specifically reported. Nevertheless, based on the theoretical work carried out by Tock et al. [39] and the experimental data provided by Hara et al. [1], mass loadability for a 60-cm-long column, 5.7 µm i.d. with a silica layer of 306-nm thickness, can be estimated as < 1 ng. The much greater surface area of the C18-silica layer formed on 65-cm-long MLC, which is ~ 92 times higher than that for the same layer within a single 5.7 µm i.d. channel, clearly provides a substantial increase in mass loadability, allowing the use of relatively large injection volumes without significantly comprising chromatographic performance.


A homogenous open tubular porous C18-silica layer has been synthesised upon the internal walls of a multi-lumen capillary (PLOT-MLC). The modified PLOT-MLC demonstrated reversed-phase selectivity and was first evaluated for its use as off-line in-capillary micro extractor, and later as a capillary separator column. Most importantly, for the first time, the modified PLOT- MLC has been successfully and reproducibly demonstrated as a combined concentrator, separator column, and ESI emitter, allowing the direct electrospray of the eluted compounds following an extraction/concentration step, into the MS without the need for connecting capillary. Results obtained demonstrated that a stable electrospray was generated, based upon the use a CE–MS sprayer. It was confirmed that the multi-channel structure significantly increases mass loadability compared to single-channel capillaries, facilitating the injection of large sample volumes, of the order of microliters, although the i.d. of each channel was just 4.2 µm. In addition, the multi-channel structure allowed the capillary to be applied at higher flow rates (greatly reduced backpressure than sub-10-µm channel capillaries of similar lengths), which simplifies system requirements and increases detection options. Specifically, mass spectrometry detection can be coupled directly, reducing the importance of high separation efficiencies for many applications. To-date, no other system has reported extraction, separation, and direct ESI emission within a MLC capillary format. Future work will be focused on improving separation efficiencies and exploring practical applications.



The authors wish to acknowledge the Australian Research Council for funding (Grant IC140100022). We would also like to acknowledge the help of Christopher Broinowski, Dr. Sandrin T. Feig, and Dr. Karsten Goemann of the Central Science Laboratory, University of Tasmania, and Petr Smejkal from ACROSS, for the provision of technical support.

Compliance with Ethical Standards

Conflict of interest

The authors declare no conflict of interest.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.

Supplementary material

10337_2018_3629_MOESM1_ESM.docx (864 kb)
Supplementary material 1 (DOCX 864 KB)


  1. 1.
    Hara T, Futagami S, Eeltink S, De Malsche W, Baron GV, Desmet G (2016) Very high efficiency porous silica layer open-tubular capillary columns produced via in-column sol–gel processing. Anal Chem 88(20):10158–10166CrossRefGoogle Scholar
  2. 2.
    Causon TJ, Shellie RA, Hilder EF, Desmet G, Eeltink S (2011) Kinetic optimisation of open-tubular liquid-chromatography capillaries coated with thick porous layers for increased loadability. J Chromatogr A 1218(46):8388–8393. CrossRefGoogle Scholar
  3. 3.
    Jorgenson JW, Guthrie EJ (1983) Liquid chromatography in open-tubular columns. Theory of column optimization with limited pressure and analysis time, and fabrication of chemically bonded reversed-phase columns on etched borosilicate glass capillaries. J Chromatogr A 255(C):335–348. CrossRefGoogle Scholar
  4. 4.
    Kazarian AA, Sanz Rodriguez E, Deverell JA, McCord J, Muddiman DC, Paull B (2016) Wall modified photonic crystal fibre capillaries as porous layer open tubular columns for in-capillary micro-extraction and capillary chromatography. Anal Chim Acta 905:1–7. CrossRefGoogle Scholar
  5. 5.
    Pesek JJ, Matyska MT (1997) Column technology in capillary electrophoresis and capillary electrochromatography. Electrophoresis 18(12–13):2228–2238. CrossRefGoogle Scholar
  6. 6.
    Pesek JJ, Matyska MT (2000) Open tubular capillary electrokinetic chromatography in etched fused-silica tubes. J Chromatogr A 887(1–2):31–41. CrossRefGoogle Scholar
  7. 7.
    Collins DA, Nesterenko EP, Brabazon D, Paull B (2013) In-process phase growth measurement technique in the fabrication of monolithic porous layer open tubular (monoPLOT) columns using capacitively coupled contactless conductivity. Analyst 138(9):2540–2545. CrossRefGoogle Scholar
  8. 8.
    Peng L, Zhu M, Zhang L, Liu H, Zhang W (2016) Preparation and evaluation of 3 m open tubular capillary columns with a zwitterionic polymeric porous layer for liquid chromatography. J Sep Sci 39(19):3736–3744. CrossRefGoogle Scholar
  9. 9.
    Wang H, Yao Y, Li Y, Ma S, Peng X, Ou J, Ye M (2017) Preparation of open tubular capillary columns by in situ ring-opening polymerization and their applications in cLC–MS/MS analysis of tryptic digest. Anal Chim Acta 979:58–65. CrossRefGoogle Scholar
  10. 10.
    Wang D, Hincapie M, Rejtar T, Karger BL (2011) Ultrasensitive characterization of site-specific glycosylation of affinity-purified haptoglobin from lung cancer patient plasma using 10 µm i.d. porous layer open tubular liquid chromatography-linear ion trap collision-induced dissociation/electron transfer dissociation mass spectrometry. Anal Chem 83(6):2029–2037. CrossRefGoogle Scholar
  11. 11.
    Rogeberg M, Wilson SR, Greibrokk T, Lundanes E (2010) Separation of intact proteins on porous layer open tubular (PLOT) columns. J Chromatogr A 1217(17):2782–2786CrossRefGoogle Scholar
  12. 12.
    Rogeberg M, Vehus T, Grutle L, Greibrokk T, Wilson SR, Lundanes E (2013) Separation optimization of long porous-layer open-tubular columns for nano-LC–MS of limited proteomic samples. J Sep Sci 36(17):2838–2847CrossRefGoogle Scholar
  13. 13.
    Thakur D, Rejtar T, Wang D, Bones J, Cha S, Clodfelder-Miller B, Richardson E, Binns S, Dahiya S, Sgroi D, Karger BL (2011) Microproteomic analysis of 10,000 laser captured microdissected breast tumor cells using short-range sodium dodecyl sulfate-polyacrylamide gel electrophoresis and porous layer open tubular liquid chromatography tandem mass spectrometry. J Chromatogr A 1218(45):8168–8174. CrossRefGoogle Scholar
  14. 14.
    Tock PPH, Stegeman G, Peerboom R, Poppe H, Kraak JC, Unger KK (1987) The application of porous silica layers in open tubular columns for liquid chromatography. Chromatographia 24(1):617–624CrossRefGoogle Scholar
  15. 15.
    Guo Y, Colón LA (1995) A Stationary phase for open tubular liquid chromatography and electrochromatography using sol–gel technology. Anal Chem 67(15):2511–2516. CrossRefGoogle Scholar
  16. 16.
    Guo Y, Colón LA (1995) Modification of the inner capillary surface by the sol–gel method: application to open tubular electrochromatography. J Microcolumn Sep 7(5):485–491. CrossRefGoogle Scholar
  17. 17.
    Guo Y, Colón LA (1996) Open tubular liquid chromatography using a sol–gel derived stationary phase. Chromatographia 43(9–10):477–483CrossRefGoogle Scholar
  18. 18.
    Detobel F, Eghbali H, De Bruyne S, Terryn H, Gardeniers H, Desmet G (2009) Effect of the presence of an ordered micro-pillar array on the formation of silica monoliths. J Chromatogr A 1216(44):7360–7367. CrossRefGoogle Scholar
  19. 19.
    Detobel F, De Bruyne S, Vangelooven J, De Malsche W, Aerts T, Terryn H, Gardeniers H, Eeltink S, Desmet G (2010) Fabrication and chromatographic performance of porous-shell pillar-array columns. Anal Chem 82(17):7208–7217. CrossRefGoogle Scholar
  20. 20.
    Forster S, Kolmar H, Altmaier S (2013) Preparation and kinetic performance assessment of thick film 10–20 µm open tubular silica capillaries in normal phase high pressure liquid chromatography. J Chromatogr A 1315:127–134. CrossRefGoogle Scholar
  21. 21.
    Hara T, Mascotto S, Weidmann C, Smarsly BM (2011) The effect of hydrothermal treatment on column performance for monolithic silica capillary columns. J Chromatogr A 1218(23):3624–3635. CrossRefGoogle Scholar
  22. 22.
    Hara T, Desmet G, Baron GV, Minakuchi H, Eeltink S (2016) Effect of polyethylene glycol on pore structure and separation efficiency of silica-based monolithic capillary columns. J Chromatogr A 1442:42–52. CrossRefGoogle Scholar
  23. 23.
    Hara T, Kobayashi H, Ikegami T, Nakanishi K, Tanaka N (2006) Performance of monolithic silica capillary columns with increased phase ratios and small-sized domains. Anal Chem 78(22):7632–7642. CrossRefGoogle Scholar
  24. 24.
    Vehus T, Roberg-Larsen H, Waaler J, Aslaksen S, Krauss S, Wilson SR, Lundanes E (2016) Versatile, sensitive liquid chromatography mass spectrometry-Implementation of 10 µm OT columns suitable for small molecules, peptides and proteins. Sci Rep. Google Scholar
  25. 25.
    Xu L, Xu B, Zhao ZY, Yang HP, Tang C, Dong LY, Liu K, Fu L, Wang XH (2017) Preparation and characterization of micro-cell membrane chromatographic column with N-hydroxysuccinimide group-modified silica-based porous layer open tubular capillary. J Chromatogr A 1516:125–130. CrossRefGoogle Scholar
  26. 26.
    Knob R, Kulsing C, Boysen RI, Macka M, Hearn MTW (2015) Surface-area expansion with monolithic open tubular columns. TrAC Trends Anal Chem 67:16–25. CrossRefGoogle Scholar
  27. 27.
    Rogers B, Gibson GTT, Oleschuk RD (2011) Bundled capillary electrophoresis using microstructured fibres. Electrophoresis 32(2):223–229. CrossRefGoogle Scholar
  28. 28.
    Sun Y, Nguyen NT, Kwok YC (2009) Enhanced electrophoretic DNA separation in photonic crystal fiber. Anal Bioanal Chem 394(6):1707–1710. CrossRefGoogle Scholar
  29. 29.
    Tůma P, Opekar F, Samcová E, Štulík K (2013) The use of a multichannel capillary for electrophoretic separations of mixtures of clinically important substances with contactless conductivity and UV photometric detection. Electrophoresis 34(14):2058–2064. CrossRefGoogle Scholar
  30. 30.
    Mugo SM, Huybregts L, Mazurok J (2014) A porous layer open tubular monolith on microstructured optical fibre for microextraction and online GC–MS applications. Anal Methods 6(5):1291–1295. CrossRefGoogle Scholar
  31. 31.
    Daley AB, Wright RD, Oleschuk RD (2011) Parallel, fluorous open-tubular chromatography using microstructured fibers. Anal Chim Acta 690(2):253–262. CrossRefGoogle Scholar
  32. 32.
    Brandtzaeg OK, Røen BT, Enger S, Lundanes E, Wilson SR (2017) Multichannel open tubular enzyme reactor online coupled with mass spectrometry for detecting ricin. Anal Chem 89(17):8667–8673. CrossRefGoogle Scholar
  33. 33.
    da Silva MR, Brandtzaeg OK, Vehus T, Lanças FM, Wilson SR, Lundanes E (2017) An automated and self-cleaning nano liquid chromatography mass spectrometry platform featuring an open tubular multi-hole crystal fiber solid phase extraction column and an open tubular separation column. J Chromatogr A. Google Scholar
  34. 34.
    Meyer RF, Champlin PB, Hartwick RA (1983) Theory of multicapillary columns for HPLC. J Chromatogr Sci 21(10):433–438. CrossRefGoogle Scholar
  35. 35.
    Belov YP, Ulyanova MM, Sidelnikov VN (2005) Multicapillary columns for chromatography. Am Lab 37(6):42–46Google Scholar
  36. 36.
    Fu Y, Gibson GTT, McGregor C, Oleschuk RD (2015) Fabrication of a polymer nozzle array in a microstructured fibre as a nanoelectrospray emitter for mass spectrometry. Can J Chem 93(4):477–484. CrossRefGoogle Scholar
  37. 37.
    Su S, Gibson GTT, Mugo SM, Marecak DM, Oleschuk RD (2009) Microstructured photonic fibers as multichannel electrospray emitters. Anal Chem 81(17):7281–7287. CrossRefGoogle Scholar
  38. 38.
    Fu Y, Gibson GTT, Proulx A, Croteau A, Schneider BB, Covey TR, Oleschuk RD (2015) Polymer micronozzle array for multiple electrosprays produced by templated synthesis and etching of microstructured fibers. Anal Chem 87(1):747–753. CrossRefGoogle Scholar
  39. 39.
    Tock PPH, Duijsters PPE, Kraak JC, Poppe H (1990) Theoretical optimization of open-tubular columns for liquid chromatography with respect to mass loadability. J Chromatogr A 506(C):185–200. CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Australian Centre for Research on Separation Science (ACROSS), and ARC Training Centre for Portable Analytical Separation Technologies (ASTech), School of Natural SciencesUniversity of TasmaniaHobartAustralia

Personalised recommendations