Chromatographic study of the structural properties of mesoporous silica layers deposited on radially elongated pillars

Journal of Chromatography A, 1595 (2019) 58–65

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Chromatographic study of the structural properties of mesoporous silica layers deposited on radially elongated pillars Shunta Futagami a,b , Takeshi Hara a,c , Heidi Ottevaere b , Herman Terryn d , Gino V. Baron a , Gert Desmet a , Wim De Malsche a,∗ a

Department of Chemical Engineering, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium Department of Applied Physics and Photonics, Brussels Photonics (B-PHOT), Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium c Division of Metabolomics, Medical Institute of Bioregulation, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan d Department of Materials and Chemistry, Vrije Universiteit Brussel, Pleinlaan2, 1050 Brussels, Belgium b

a r t i c l e

i n f o

Article history: Received 29 December 2018 Received in revised form 12 February 2019 Accepted 14 February 2019 Available online 15 February 2019 Keywords: Radially elongated pillar array column Sol-gel processing Mesoporous silica layer Retention Nano-LC Peak capacity

a b s t r a c t We report on the ability to change the layer properties of porous layered radially elongated pillar (PLREP) array columns and its relevance to the separation efficiency. The adjustment of the preparation condition resulted in the formation of a 1.2-fold thicker layer than the layer produced in the preceding study. The mesoporosity of the layer was controlled by changing the hydrothermal treatment temperature from 105 ◦ C to 80 ◦ C. Chromatographic characterization was performed on a commercial nano-LC system using the octadecylsilylated PLREP columns having the aforementioned characteristics, i.e. (1) different layer thickness (df = 180–220 nm) and (2) different mesoporosity (dp = 7.6–11.2 nm, Pore volume (Vp ) = 0.733–0.838 cc/g and Surface area (SA) = 364–611 m2 /g). For isocratic separations of an alkylphenone mixture, the change in both the layer thickness and the mesoporosity caused no significant difference in the column efficiency, while the thicker layer and the reduction of mesopore size resulted in a 1.3-fold increase and a 1.4-fold increase in the retention capacity, respectively. Based on the result of the examination using scanning electron microscopy and argon physisorption technique, the formar enhancement was in agreement with the increase in the layer thickness, and the latter one was attributed to the larger surface area. When applying a column with 16.5 cm long to gradient separations, the combination of the thicker layer and the smaller mesopores provided the peak capacity of 365 for the alkylphenone mixture at a 180 min gradient, while the combination of the thinner layer and the larger mesopores provided the peak capacity of 315. For peptide separations, it appeared that the thicker layer was still favorable, however, the lager mesopores were more advantageous for MWs of larger than 1000, providing a conditional peak capacity of 245 for a commercially available peptide mixture because of less content of small pores which hinder the diffusion of large molecules in pores in the layer. © 2019 Elsevier B.V. All rights reserved.

1. Introduction Silica materials have been widely used as separation support for HPLC columns because of their high mechanical strength, shape stability in a change of mobile phase composition, and their versatility due to the large number of available functional groups [1]. Traditionally, silica supports were provided only in a spherical shape, however, since Minakuchi and co-workers synthesized and applied co-continuous silica skeleton structure called monolith in 1990s [2–4], the column format has gained in popularity owing to its high permeability and high column efficiency [5–8].

∗ Corresponding author at: Pleinlaan 2, B-1050, Brussels, Belgium. E-mail address: [email protected] (W. De Malsche). https://doi.org/10.1016/j.chroma.2019.02.035 0021-9673/© 2019 Elsevier B.V. All rights reserved.

The macropore structure of monolithic silica is formed using the phase separation based on the spinodal decomposition of waterrich phase and silica-rich phase, and the unique structure can remain via sol-gel transition of the silica phase. In addition, it is possible to tune the mesoporosity by applying Ostwald ripening [9–11], independently of the control of the macropore structure. In order to achieve a high column performance, the preparation methods to control the bimodal structure (to particularly reduce the heterogeneity of macropore structure) have been intensively studied so far [3,8,12–15]. Besides the fabrication of the monolithic silica structure, the aforementioned sol-gel processing can also be applied to the preparation of a silica layer on the wall of open tubular (OT) capillaries via a so-called “wetting transition” [16,17]. This phenomenon takes place when the preparation conditions are selected such that the domain size (= through-pore size + skeleton size) of bulk monolithic

S. Futagami, T. Hara, H. Ottevaere, et al. / J. Chromatogr. A 1595 (2019) 58–65

silica (e.g. prepared in a plastic tube) is larger than the size of a confined space (e.g. capillary inner diameter (i.d.)), and accordingly it leaves a layered structure on a wall that can interact with and anchor to the resulting silica gel. The first silica-layered OT capillary based on this technology was reported by Kanamori and co-workers [18]. They demonstrated the formation of a silica layer in a 10 ␮m i.d. capillary while studying the effect of the dimensions of the confined space on the resultant gel morphology, and therefore, initially their work was not extended to the preparation of silica-layered OT columns for HPLC. The first attempt to use this technique for the preparation of silicalayered OT columns was made by Forster and co-workers [19–21]. They prepared a uniform silica layer in 10–20 ␮m i.d. capillaries, and even managed to control the layer thickness by adjusting the feed composition. Recently we developed preparation methods which enables uniform layer formation in as small as 5 ␮m i.d. capillaries [22,23]. With our methods, long silica-layered OT columns up to 2.5 m can be prepared without losing the uniformity, and correspondingly results in high performance as expected from Golay-Aris theory [24]. We also recently showed the adjustability of the mesopore size as well as the layer thickness [25]. This sol-gel-based silica layer preparation technique has been studied not only for OT capillaries, but also for pillar array columns. Pillar array columns are a column format where microfabricated pillars on a substrate are used as the support structure of the stationary phase. Although the precisely controlled pillar shape, order, and interpillar distance enable high column efficiency, the nonporous nature of the pillars provides limited retention and sample loading capacities [26–28]. In order to overcome this issue, an interesting and feasible approach appears to be the use of sol-gel processing to deposit a silica layer on the pillars [17,29–31]. In these studies, an increase in surface area compared to that of nonporous pillars was reported. However, these studies did not concern a variation of the properties of the layer, i.e., the layer thickness and mesopore size, while a number of studies focused on similar structural properties of monolithic silica and silica-layered OT capillaries as mentioned earlier. In this contribution, we address controlling those characteristics (layer thickness and mesopore size) to further enhance the separation efficiency of silica-layered pillar array columns. The silica layer was grown on radially elongated pillars (REPs), a pillar shape that provides a superior column efficiency compared to classical cylindrical pillars [27,32,33]. The layer thickness was increased by adjusting the amount of the silica precursor and polyethylene glycol (PEG) in the feed solution and lowering the gelation temperature, which determines at which stage of the phase separation the structure is frozen. The mesopore size was controlled by changing the temperature of the hydrothermal treatment (HT) to form mesopores. After the C18 - and endcapping silylation steps, the prepared REP columns were tested with an alkylphenone mixture and a peptide mixture to examine the chromatographic performance. Because of the limited amount of silica in the columns [8,10], bulksilica rods were also prepared under the same conditions as for the sol-gel REP columns. Argon physisorption measurements of these corresponding rods were carried out to characterize the mesoporosity of the layers. 2. Experimental

59

none and octanophenone were purchased from Sigma-Aldrich (Overijse, Belgium). Trifluoroacetic acid (TFA, LC–MS Ultra grade) was purchased from Fluka (Buchs, Switzerland). Octadecyldimethyl-N,N-dimethylaminosilane (ODS-DMA) was purchased from ChemPur Feinchemikalien und Forschungsbedarf GmbH (Karlsruhe, Germany). Acetonitrile (ACN, HPLC supragradient grade) was purchased from Biosolve B.V. (Valkenswaard, the Netherlands). N-(Trimethylsilyl)-imidazole (TMSI) was obtained from Merck Schuchardt OHG (Hohenbrunn, Germany). Deionized water was produced in-house with a Milli-Q water purification system of Merck Millipore (Billerica, MA, USA). PTFE filters (0.20 ␮m × 25 mm) were purchased from Macherey-Nagel (Düren, Germany). MassPREP Peptide Mixture was purchased from Waters (Milford, MA, USA). Fused-silica capillaries of 20 ␮m i.d. (90 ␮m outer diameter (o.d.)) and 40 ␮m i.d. (105 ␮m o.d.) were purchased from Polymicro Technologies (Phoenix, AZ, USA). 2.2. Sample preparation Thiourea, acetophenone, propiophenone, butyrophenone, valerophenone, hexanophenone, heptanophenone and octanophenone were dissolved in ACN to prepare stock solutions of 2000 ppm. The stock solutions were mixed and diluted to obtain a mixture containing the 8 components (100 ppm each) in water/ACN = 50/50 (v/v). MassPREP Peptide Mixture was dissolved in 1000 ␮L of 0.1% aqueous TFA (v/v) to prepare a stock solution containing 1˜ pmol/␮L of each compound in the mixture (MassPREP Peptide Mixture contains 1˜ nmol of each compound). The stock solution was diluted twofold for the measurements. 2.3. Microfabrication A pillar array of 16.5 cm long and 1 mm wide (three lanes of 5.5 cm long channels connected by two turns) containing REPs with aspect ratio 20 (100 ␮m in the lateral direction, 5 ␮m in the axial direction and 2.5 ␮m interpillar distance) was patterned on a 475 ␮m thick Boron doped (resistivity: 0.001-0.025) p-type (100) silicon wafer (Okmetic, Finland) with mid-UV photolithography (photoresist, Olin 907-12), followed by a dry etching step (Adixen AMS100DE, Alcatel Vacuum Technology, Culemborg, The Netherlands) to etch the 300 nm thick SiO2 hard mask underneath. Next, 115 ␮m deep capillary insertion channels were defined by a second lithography step and subsequent Bosch-type etching step (Adixen AMS100SE). The photoresist was removed by using oxygen plasma and nitric acid, and the pillars defined in the SiO2 mask were subsequently Bosch etched to reach a depth of 18 ␮m (the total depth of the capillary insertion channels became about 130 ␮m). A flow distributor was placed at the capillary-pillar bed interface to ensure a good flow distribution over the entire width of the column. The Si substrate was subsequently anodically bonded to a 500 ␮m thick Mempax glass wafer (Schott, Germany) with an EVG EV-501 wafer bonder (EV Group Inc., Schaerding, Austria). Next, the chip was diced 100 ␮m deep from both sides of the wafer and subsequently cleaved, which exposed the channels to enable insertion of the interfacing capillaries. Finally, 15 cm long fused-silica capillaries (90 ␮m o.d. and 20 ␮m i.d.) were inserted and glued for the following mesoporous silica layer preparation. 2.4. Preparation of the mesoporous silica layer

2.1. Chemicals and materials Toluene (HPLC grade, ≥99.8%), urea, tetramethoxysilane (TMOS), metyltrimethoxysilane (MTMS), 1 M aqueous acetic acid solution, polyethylene glycol (PEG) of molecular weight (MW) = 10,000 g/mol, thiourea, acetophenone, propiophenone, butyrophenone, valerophenone, hexanophenone, heptanophe-

The mesoporous silica layer was prepared via the on-chip sol-gel processing procedure as described in [31]. In order to increase the layer thickness, the amount of TMOS/MTMS (VT /VM = 75/25) was increased by 30%, compared to the previously fabricated columns (cf. ref [31].). On the other hand, the amount of PEG in the feed solution was decreased by 20% and a 5 ◦ C lower temperature was

60

S. Futagami, T. Hara, H. Ottevaere, et al. / J. Chromatogr. A 1595 (2019) 58–65

Fig. 1. Scanning electron micrographs of PLREPs with a thicker layer. Magnifications: (a, d) 4000-fold and (b, c, e, f) 30000-fold. Figures (b) and (c) are zoom-in images of (a). Figures (e) and (f) are zoom-in image of (d). The thickness of the porous silica layer on the pillars, on the glass lid and on the silicon bottom face is 220 nm, 200 nm and 210 nm, respectively. The top side of (a), (b), (d) and (e) is the glass lid covering the silicon pillars.

applied to the polycondensation process in a water bath to compensate for the increase in the amount of the silica precursor. In the following HT process, the columns were treated at either 80 ◦ C or 105 ◦ C for 15 h (the temperature was raised from 40 ◦ C at a constant rate of 5.5 ◦ C/h). The columns were then flushed with methanol for 48 h. Subsequently, the columns were heated from 40 ◦ C to 380 ◦ C in 10 h and kept at 380 ◦ C for 24 h for calcination and burning off the glue. After removal of the capillaries, silica debris remaining in the capillary insertion channels was flushed away by methanol. Next, the columns were dried and 30 cm long fusedsilica capillaries (105 ␮m o.d. and 40 ␮m i.d.) were inserted and glued for C18 -modification with 10% ODS-DMA/toluene (v/v) and endcapping with 10% TMSI/toluene (v/v). After the surface functionalization, chromatographic measurements were conducted for the fabricated columns with a commercially available nano-LC system (see Section 2.5). 2.5. Measurements LC measurements were performed on a nano-LC system as described in our preceding study [34]. The nano-LC system was composed of a membrane degasser, LPG-3600 pump, WPS-3000 autosampler, FLM-3200 column oven, and VWD-3400 RS UV detector (3 nL flow cell) from Thermo Fisher Scientific (Germering, Germany). A 4 nL internal sample injector (C4N-4344) from Valco (Schenkon, Switzerland) was employed for injections of the mixture containing thiourea and 7 alkylphenones. Injections of MassPREP Peptide Mixture were carried out with a 1 ␮L loop and a built-in 10-port valve of FLM-3200 column oven. The detection wavelength was set to 254 nm (for the mixture containing thiourea and 7 alkylphenones) and 220 nm (for MassPREP Peptide Mixture). The column temperature was set to 30 ◦ C. The mobile phases were prepared with water and ACN (for the LC measurement with mixture containing thiourea and 7 alkylphenones), and with 0.1% aqueous TFA (v/v) and 0.1% TFA in ACN (v/v, for the LC measurement with MassPREP Peptide Mixture). Scanning electron microscopy was carried out with a field-emission scanning electron microscope (JSM-7100 F) from JEOL Ltd. (Tokyo, Japan) to visualize the silica

layer. The layer thickness on the pillars, on the glass lid and on the silicon bottom face was measured using Image J software. Argon physisorption measurements of the corresponding bulk-silica rods were conducted to characterize the mesoporosity of the layers by applying the non-local density functional theory (NLDFT) [35–37], using an Autosorb-1-MP instrument from Quantachrome corporation (Boynton Beach, FL, USA). 3. Results and discussion 3.1. SEM observation Fig. 1 shows representative SEM images of the mesoporous silica layer prepared by the newly developed thicker layer condition. The layer thickness on the pillars, on the glass lid and on the silicon bottom face was measured at three different locations in a column (measured at 10 points each for the layer at one location). The average value of the layer thickness on the pillars is 220 nm (relative standard deviation (RSD) = 15%), which is a 1.2-fold increase from the value obtained in the preceding study (layer thickness: 180 nm) [31]. The thicknesses on the glass lid and on the silicon bottom face are 200 nm (RSD = 19%) and 210 nm (RSD = 16%), respectively. 3.2. Argon physisorption measurements of bulk-silica rods As was the case in our preceding work [22,25,31], the effects of changing the layer thickness condition and the HT temperature on the mesoporosity of the layer were studied by argon physisorption measurements for corresponding silica rods. In earlier work, a good agreement in pore size distribution was obtained between the results from inverse size exclusion chromatography with a silica material prepared in a confined space via sol-gel processing and the results from physisorption of a bulk-silica rod [14]. Fig. 2 presents the results of the argon physisorption measurements for three different bulk-silica rods, Thin-105, Thick-105 and Thick-80, where “Thin” or “Thick” denotes the difference in the layer thickness in the corresponding PLREP columns. The following number

S. Futagami, T. Hara, H. Ottevaere, et al. / J. Chromatogr. A 1595 (2019) 58–65

dV(r) (cc/Å/g)

0.05

61

0.8

0.04 0.6

dp = 10.0 nm Vp = 0.733 cc/g SA = 364 m2/g s = 307 m2/cc

0.03 0.02 0.01

0.4 0.2

0.00

0.0 0

50

100

150

200

0.06

1.0

(b)

dV(r) (cc/Å/g)

0.05

0.8

0.04

dp = 11.2 nm Vp = 0.837 cc/g SA = 428 m2/g s = 332 m2/cc

0.03 0.02 0.01

0.6 0.4 0.2

0.00

0.0 0

50

100

150

200

Cumulave pore volume (cc/g)

Radius of pore (Å)

Fig. 3. Isocratic separation of the mixture containing thiourea and 7 alkylphenones (C6 H5 CO(CH2 )n –1 H, n = 2–8). Peaks: (1) thiourea, (2) acetophenone, (3) propiophenone, (4) butyrophenone, (5) valerophenone, (6) hexanophenone, (7) heptanophenone, and (8) octanophenone. Mobile phase: 60% ACN/water (v/v). Flow rate: 0.20 ␮L/min. Injection volume: 4 nL. Temperature: 30 ◦ C, Detection: 254 nm.

0.06

1.0

(c)

dV(r) (cc/Å/g)

0.05

0.8

0.04 0.6

dp = 7.6 nm V p = 0.838 cc/g SA = 611 m2/g s = 473 m2/cc

0.03 0.02 0.01

0.4 0.2

0.00

0.0 0

50

100

150

200

Cumulave pore volume (cc/g)

Radius of pore (Å)

Fig. 4. Relationship between the natural logarithms of the retention factors for octanophenone and mobile phase compositions obtained with ODS-modified nonporous REP and PLREP columns. Symbols: nonporous REP (䊏), PLREP-Thin-105 (blue ), PLREP-Thick-105-#1 (red 䊉), PLREP-Thick-105-#2 (), PLREP-Thick-80-#1 (green ) and PLREP-Thick-80-#2 (). The symbol sizes correspond to the min-max error bars (n = 3). Temperature: 30 ◦ C. Detection: 254 nm.

Fig. 2. Cumulative pore volume curves and mesopore size distributions obtained by argon physisorption measurements of the bulk-silica rods. Samples: (a) silica rod Thin-105, (b) silica rod Thick-105 and (c) silica rod Thick-80. Cumulative pore volume curves and mesopore size distributions were given according to NLDFT (cylindrical pores, NLDFT equilibrium model).

(80 or 105) represents the HT temperature in degrees Celsius. The specific surface area per unit layer volume (s) was estimated via εs =

Vpore Vpore + 1/silica

s = SA ×

Vpore εs

(1)

(2)

where εs is the intra-layer porosity, Vpore is the specific pore volume (cc/g), silica is the silica skeleton density (g/cc) and SA is the specific surface area (m2 /g). The exact value of silica of the TMOS/MTMS mixture employed in this study is unknown. For relative comparison of the preparation conditions, silica is taken as 2.2 g/cc, which is the silica of pure amorphous silica [38]. Between silica rods Thin105 and Thick-105 there are only minor differences in the mesopore size (dp , nm) and the s-value, while between silica rods Thick-105 and Thick-80 there are considerable differences in the dp - and svalues. This comparison demonstrates that the amount of silica precursor in the feed solution has a limited influence on the meso-

porosity, while the HT temperature is a predominant variable to control the mesoporosity. 3.3. Chromatographic performance in isocratic mode Fig. 3 shows chromatograms for a mixture containing thiourea and 7 alkylphenones obtained with PLREP-Thin-105, PLREP-Thick105-#1 and PLREP-Thick-80-#1 (PL stands for “porous layered”) at a flow rate of 0.20 ␮L/min. As can be clearly seen in Fig. 3, high resolution separation for the alkylphenone mixture was achieved with the three PLREP columns regardless of the layer thickness and the HT temperature. The retention behavior was compared for the columns by measuring the retention factors for octanophenone (the most retained compound in the mixture) with varying the mobile phase composition, and the natural logarithms of the retention factors are plotted (see Fig. 4). In order to calculate the retention factors in Fig. 4, the migration times in the connecting capillaries (nonporous) were subtracted from the elution times because the ˜ connecting capillaries had 50% of the column (chip) volume. When comparing PLREP-Thin-105 and PLREP-Thick-105-#1 which were produced with the same HT condition, PLREP-Thick-105-#1 provided a 35% larger retention capacity with the 20% thicker layer (shown in Fig. 1) than PLREP-Thin-105. When comparing PLREPThick-105-#1 and PLREP-Thick-80-#1, produced with different HT temperatures, PLREP-Thick-80-#1 provided a 40% larger retention

62

S. Futagami, T. Hara, H. Ottevaere, et al. / J. Chromatogr. A 1595 (2019) 58–65

Fig. 5. Comparison of the plate heights for octanophenone obtained with ODS-modified PLREP columns. Symbols: PLREP-Thin-105 (blue , k = 3.50), PLREP-Thick-105-#1 (red 䊉, k = 4.76), PLREP-Thick-105-#2 (, k = 4.90), PLREPThick-80-#1 (green , k = 6.81) and PLREP-Thick-80-#2 (, k = 6.85). The symbol sizes correspond to the min–max error bars (n = 3). The blue, red, and green lines are the fitted van Deemter curves of PLREP-Thin-105, PLREP-Thick-105-#1 and PLREP-Thick-80-#1, respectively. Mobile phase: 60% acetonitrile/water (v/v). Injection volume: 4 nL. Temperature: 30 ◦ C. Detection: 254 nm.

Table 1 van Deemter coefficients for octanophenone with the mobile phase of 60% ACN/water (v/v). column

koctanophenone

A (m)

B (m2 /s)

C (s)

Thin-105 Thick-105-#1 Thick-105-#2 Thick-80-#1 Thick-80-#2

3.50 4.76 4.90 6.85 6.81

2.32 × 10–7 3.62 × 10–7 2.35 × 10–7 3.01 × 10–7 2.77 × 10–7

4.39 × 10–11 6.64 × 10–11 7.34 × 10–11 5.38 × 10–11 4.07 × 10–11

2.52 × 10–3 2.01 × 10–3 2.11 × 10–3 2.08 × 10–3 2.57 × 10–3

capacity than PLREP-Thick-105-#1 because of the higher surface area (see Fig. 2). These results are in agreement with the anticipation that increasing layer thickness and the larger s-value can lead to the enhancement of retention capacity of PLREP columns, as recently demonstrated with porous layered open-tubular (PLOT) columns [25]. As shown in our preceding study of integrating a pillar array column into commercial nano-LC systems [34,39,40], extra-column dispersion at the coupling points has a large influence on the total dispersion observed at the detector, due to the relatively low dispersion occurring in pillar array columns. With the currently employed column length, capillary insertion and gluing method, chip-to-chip variability of the quality of the chip-to-capillary interfaces is unavoidable. Therefore, all the values related to dispersion (i.e. plate heights and peak capacities) were calculated with the peak widths at half height to compare the values with as little extracolumn dispersion at the coupling points as possible. Plate height (H) was calculated by H=

L t

5.54 ( w R )

2

(3)

50%

where L is the column length (L = 16.5 cm in this study), and plotted against the linear velocity (u) in Fig. 5. All the curves in Fig. 5 have similar profile and Hmin -values range from 0.9 ␮m to 1.1 ␮m. Considering the fact that the aforementioned extra-column dispersion at the coupling points is not completely negligible even at half height, the observed variability of 0.2 ␮m around Hmin is acceptable as a chip-to-chip deviation. The experimentally obtained values were then fitted to the van Deemter equation. H = A+

B + Cu u

(4)

Fig. 6. Gradient separations of the mixture containing thiourea and 7 alkylphenones. (a) Representative chromatograms at tG = 180 min. (b) Comparison of conditional peak capacity (nc ) values obtained at different gradient times with a fixed flow rate of 0.25 ␮L/min. Symbols: PLREP-Thin-105 (blue ), PLREP-Thick105-#1 (red 䊉) and PLREP-Thick-80-#1 (green ). Mobile phase: (A) water, (B) ACN. Gradient: 15–80% B. Temperature: 30 ◦ C. Detection: 254 nm. Injection volume: 4 nL.

The calculated coefficients, A, B and C, are listed in Table 1 and the van Deemter curves of PLREP-Thin-105 (blue), PLREP-Thick105-#1 (red) and PLREP-Thick-80-#1 (green) are given in Fig. 5. The coefficients listed in Table 1 are in good agreement with the coefficients for coumarin 480 (k = 1.97) in 60% methanol obtained by on-chip detection reported in our preceding study [31] (the exponents of the C-term values listed in Table 1 of [31] were misprinted as –8 and should be –3). 3.4. Chromatographic performance in gradient mode PLREP-Thin-105, PLREP-Thick-105-#1 and PLREP-Thick-80-#1 were then compared in gradient mode, which is more often employed than isocratic mode in practice. The column performance in gradient mode was evaluated by conditional peak capacity, nc calculated by nc =

tZ – tA +1 w50%¯

(5)

where tA is the elution time of the first peak, tZ is the elution time of the last peak. The first comparison in gradient mode was performed with the alkylphenone mixture. Representative chromatograms are presented in Fig. 6(a), and nc -values obtained with a fixed flow rate of 0.25 ␮L/min are plotted against the gradient times in Fig. 6(b). As can be expected from Figs. 4 and 5 (i.e. similar H-u curves for the three columns and the retention capacity order of PLREP-Thick-80#1 > PLREP-Thick-105-#1 > PLREP-Thin-105), PLREP-Thick-80-#1

S. Futagami, T. Hara, H. Ottevaere, et al. / J. Chromatogr. A 1595 (2019) 58–65

63

Fig. 8. Comparison of the peak widths of the peptides at half height. Peptide numbers: (1) RASG-1 (MW = 1000), (2) angiotensin fragment 1–7 (MW = 898), (3) bradykinin (MW = 1060), (4) angiotensin II (MW = 1046), (5) angiotensin I (MW = 1296), (6) renin substrate (MW = 1758), (7) enolase T35 (MW = 1872), (8) enolase T37 (MW = 2827), (9) melittin (MW = 2846). Because of the low intensity, enolase T35 is not included in the comparison. Symbols: PLREP-Thin-105 (blue ), PLREP-Thick-105-#1 (red 䊉) and PLREP-Thick-80-#1 (green ).

Fig. 7. Gradient separations of the peptide mixture. (a) Representative chromatograms at tG = 180 min. (b) Comparison of conditional peak capacity (nc ) values obtained at different gradient times with a fixed flow rate of 0.25 ␮L/min. Symbols: PLREP-Thin-105 (blue ), PLREP-Thick-105-#1 (red 䊉) and PLREPThick-80-#1 (green ). Mobile phase: A: water/formic acid = 100/0.05 (v/v), B: acetonitrile/water/formic acid = 80/20/0.05. Gradient: 0–50% B. Temperature: 30 ◦ C. Detection: 220 nm. Injection volume: 1 ␮L.

provided the highest peak capacity (nc = 365) of the three columns followed by PLREP-Thick-105-#1 (nc = 340) and PLREP-Thin-105 (nc = 315) at tG = 180 min. These results imply that the mesopores of all three conditions, studied in detail with the corresponding bulk-silica rods shown in Fig. 2, are large enough for separation of small molecules without loss of column efficiency, and thus simply a thicker layer and larger s-value provide better peak capacity. Similar comparison of the three columns was performed with a peptide mixture to examine the difference of the performance for large molecules. Fig. 7(a) and (b) show representative chromatograms and nc -values. The depressions of the baseline at the beginning of the chromatograms in Fig. 7(a) are due to the small difference between the injected volume (1 ␮L) and the column vol˜ ␮L). In this study, the small peak after the passage of the umes (1.4 depressions was used as the tA in Eq. (5). The mixture contained 9 peptides, however, only the 7th peptide showed much lower intensity than the other components (barely seen in Fig. 7(a) between 120 min and 130 min). Therefore, the 7th peptide was not taken into account for the calculations of nc -values. Fig. 7(b) shows a different trend from that in Fig. 6(b). PLREP-Thick-80-#1, which provided the highest peak capacity for the alkylphenone mixture, provided a lower peak capacity (nc = 215) than the other two columns for the peptide mixture at tG = 180 min. Instead, the other thick layer column, PLREP-Thick-105-#1 (nc = 245) provided the highest peak capacity. The higher peak capacity of PLREP-Thick-105-#1 than that of PLREP-Thin-105 (nc = 225) indicates the relevance of the surface

area to the peak capacity for peptides. However, in contrast to the alkylphenone separations, the mesopore size also plays a crucial role in peptide separations, as suggested by the lower peak capacity of PLREP-Thick-80-#1 than that of the other two columns. In order to examine the detailed relationship between the molecular weights (MWs) of the peptides, the layer properties and the performance, the peak widths at half height obtained at tG = 180 min are plotted in Fig. 8. The peak widths of the 1 st (RASG1, MW = 1000) and 2nd (angiotensin fragment 1–7, MW = 898) peptides are similar among all the three columns, and even PLREP-Thick-80-#1 provided narrower widths than the other two columns. The difference in the widths becomes more significant from the 3rd peptide (bradykinin, MW = 1060). PLREP-Thick-105#1 provided narrowest peak widths of the three columns for molecules having a larger MW than 1000, while PLREP-Thick-80-#1 provided the widest. These wider peak widths observed for PLREPThick-80-#1 are seemingly due to a higher content of small pores in its layer than in the layers of PLREP-Thin-105 and PLREP-Thick-105#1, causing considerable obstruction of molecular diffusion in the layer. From the results of the argon physisorption measurements shown in Fig. 2, silica rod Thick-80 had a more than 2.2-times as large cumulative pore volume (0.29 cc/g) as silica rods Thin-105 (0.11 cc/g) and Thick-105 (0.13 cc/g) in the range of pores smaller than 6 nm. In the case of monolithic silica capillary columns, it was ˜ g/cc for pores smaller reported that a cumulative pore volume of 0.3 than 6 nm considerably lowered performance for peptides, while a ˜ cc/g did not cause an observcumulative pore volume of up to 0.2 able difference in chromatographic performance [14]. This suggests that it is mandatory to apply the higher HT (105 ◦ C) to adequately reduce the volume of small pores with <6 nm in a PLREP column for separations of molecules larger than MW of 1000. According to our results, we anticipate that the preparation of PLREP columns with a larger mesopore size would be meaningful for biochemical research such as shotgun proteomics with an LC/MS system, which generally includes a number of peptides with larger MW than 1000 as target compounds. Following the preceding studies on Ostwald ripening of silica materials, it was demonstrated that extension of the HT time or increase in the HT temperature can lead to further mesopore enlargement [9,14].We are currently exploring a preparation condition to form mesopores >15 nm in the silica layer, which is commonly used for larger molecule separations with particulate columns. A theoretical study on kinetic performance of PLOT columns reported by Causon and co-workers showed that for 1–6 ␮m i.d.

64

S. Futagami, T. Hara, H. Ottevaere, et al. / J. Chromatogr. A 1595 (2019) 58–65

PLOT columns, a good compromise between kinetic performance and retention capacity is found in the range of 0.5 ≤ ı ≤ 1.1, where ı is the ratio of the layer thickness to the flow-through pore diameter [24]. When the native (non-layered) interpillar distance of 2.5 ␮m is simply applied to the calculation of this ı-value, the thicker layer case of this study corresponds to a ı-value of 0.11, which is as large as the maximum values reported so far for PLOT columns [22,23]. The ı-value of 0.5 (the minimum of the aforementioned ı-range) is supposed to be the layer thickness of 630 nm in the current REP column design. Based on this comparison and the fact that a REP column can be regarded as a folded OT column [27,33], a further increase in the layer thickness is presumed to enhance the performance of PLREP columns. 4. Conclusions Modification of the layer thickness and the mesoporosity of the deposited silica layer on REPs was demonstrated, and concomitant effects on chromatographic performance were studied. A 1.2-fold increase in the layer thickness compared to that described in our earlier work was achieved by adjusting the feed composition and lowering the gelation temperature. The mesoporosity of three different silica layers were examined by argon physisorption measurements of corresponding bulk-silica rods. The result revealed that there is little influence of the layer thickness on the mesoporosity, while the HT temperature predominantly determines the mesoporosity and HT at 80 ◦ C provided smaller mesopores and larger surface area than at 105 ◦ C. PLREP columns prepared by three different conditions were first examined in isocratic mode with an alkylphenone mixture. In terms of column efficiency, similar H-u curves having Hmin of around 1.0 ␮m were obtained with all the tested columns. In contrast, the retention capacity was increased 1.3-fold by the thicker layer compared to the thinner layer, and 1.4-fold by the HT at 80 ◦ C compared to the HT at 105 ◦ C. Since the column efficiency (H-value) was similar among the columns, the advantage in the retention capacity was directly reflected in higher peak capacities for gradient separations of the alkylphenone mixture, and the maximum peak capacity of 365 was obtained with the thicker layer treated at 80 ◦ C. The peak capacity for larger molecules was subsequently evaluated by gradient separations of a peptide mixture. The thicker layer provided better performance also for the peptide mixture. However, the HT at 105 ◦ C was preferred because of the larger mesopores unlike the case of the alkylphenone mixture, and resulted in the maximum peak capacity of 245. These results show that the thicker layer presented in this study is always favorable, but also that better separation efficiency can be obtained by changing the HT temperature to adjust the mesoporosity for the size of targeted molecules. The detailed investigation of the peak widths in the gradient separations of the peptide mixture suggests that the HT at 80 ◦ C yields a better separation efficiency for MWs of up to 1000, while the HT at 105 ◦ C is required for MWs of larger than 1000 to have a less quantity of small pores which is associated with the diffusion hindrance in pores in the layer. Conflict of interest W.D.M. and G.D. are co-founders (and have shares) of the spinoff company PharmaFluidics that has a commercial activity around pressure-driven pillar array column-based chromatography. Acknowledgements W.D.M. greatly acknowledges a research grant (679033EVODIS ERC-2015-STG) from the European Research Council (ERC). The

authors thank PharmaFluidics for technical support, fruitful discussions and providing nonporous REP columns for sol-gel processing. W.D.M., H.O. and S.F. appreciate the financial support of the Vrije Universiteit Brussel. References [1] L.R. Snyder, J.J. Kirkland, J.W. Dolan, Introduction to Modern Liquid Chromatography, 3rd ed., John Wiley&Sons, Inc., New York, 2010, http://dx. doi.org/10.1007/s13361-010-0021-8. [2] H. Minakuchi, K. Nakanishi, N. Soga, N. Ishizuka, N. Tanaka, Octadecylsilylated porous silica rods as separation media for reversed-phase liquid chromatography, Anal. Chem. 68 (1996) 3498–3501, http://dx.doi.org/10. 1021/ac960281m. [3] H. Minakuchi, K. Nakanishi, N. Soga, N. Ishizuka, N. Tanaka, Effect of skeleton size on the performance of octadecylsilylated continuous porous silica columns in reversed-phase liquid chromatography, J. Chromatogr. A. 762 (1997) 135–146, http://dx.doi.org/10.1016/S0021-9673(96)00944-2. [4] H. Minakuchi, K. Nakanishi, N. Soga, N. Ishizuka, N. Tanaka, Effect of domain size on the performance of octadecylsilylated continuous porous silica columns in reversed-phase liquid chromatography, J. Chromatogr. A. 797 (1998) 121–131, http://dx.doi.org/10.1016/S0021-9673(97)00947-3. [5] K. Cabrera, Applications of silica-based monolithic HPLC columns, J. Sep. Sci. 27 (2004) 843–852, http://dx.doi.org/10.1002/jssc.200401827. [6] Q. Luo, J.S. Page, K. Tang, R.D. Smith, MicroSPE-nanoLC-ESI-MS/MS using 10-␮m-i.d. Silica-based monolithic columns for proteomics, Anal. Chem. 79 (2007) 540–545, http://dx.doi.org/10.1021/ac061603h. [7] K. Horie, Y. Sato, T. Kimura, T. Nakamura, Y. Ishihama, Y. Oda, T. Ikegami, N. Tanaka, Estimation and optimization of the peak capacity of one-dimensional gradient high performance liquid chromatography using a long monolithic silica capillary column, J. Chromatogr. A. 1228 (2012) 283–291, http://dx.doi. org/10.1016/j.chroma.2011.12.088. [8] T. Hara, G. Desmet, G.V. Baron, H. Minakuchi, S. Eeltink, Effect of polyethylene glycol on pore structure and separation efficiency of silica-based monolithic capillary columns, J. Chromatogr. A. 1442 (2016) 42–52, http://dx.doi.org/10. 1016/j.chroma.2016.03.009. [9] K. Nakanishi, Pore structure control of silica gels based on phase separation, J. Porous Mat. 4 (1997) 67–112, http://dx.doi.org/10.1023/A:1009627216939. [10] T. Hara, Study on Preparation and Characterization of Monolithic Silica Capillary Columns for High Separation Efficiency in High Performance Liquid Chromatography, Justus-Liebig-Universität Gießen, 2012. [11] A. Galarneau, J. Iapichella, D. Brunel, F. Fajula, Z. Bayram-Hahn, K. Unger, G. Puy, C. Demesmay, J.L. Rocca, Spherical ordered mesoporous silicas and silica monoliths as stationary phases for liquid chromatography, J. Sep. Sci. 29 (2006) 844–855, http://dx.doi.org/10.1002/jssc.200500511. [12] M. Motokawa, H. Kobayashi, N. Ishizuka, H. Minakuchi, K. Nakanishi, H. Jinnai, K. Hosoya, T. Ikegami, N. Tanaka, Monolithic silica columns with various skeleton sizes and through-pore sizes for capillary liquid chromatography, J. Chromatogr. A. 961 (2002) 53–63, http://dx.doi.org/10.1016/S00219673(02)00133-4. [13] T. Hara, H. Kobayashi, T. Ikegami, K. Nakanishi, N. Tanaka, Performance of monolithic silica capillary columns with increased phase ratios and small-sized domains, Anal. Chem. 78 (2006) 7632–7642, http://dx.doi.org/10. 1021/ac060770e. [14] T. Hara, S. Mascotto, C. Weidmann, B.M. Smarsly, The effect of hydrothermal treatment on column performance for monolithic silica capillary columns, J. Chromatogr. A. 1218 (2011) 3624–3635, http://dx.doi.org/10.1016/j.chroma. 2011.04.008. [15] K. Cabrera, A New generation of silica-based monoliths HPLC columns with improved performance, JC GC North. Am. 30 (2012) 30–35. [16] H. Tanaka, Interplay between wetting and phase separation in binary fluid mixtures: roles of hydrodynamics, J. Phys. Condens. Matter. 13 (2001) 4637–4674, http://dx.doi.org/10.1088/0953-8984/13/21/303. [17] F. Detobel, H. Eghbali, S. De Bruyne, H. Terryn, H. Gardeniers, G. Desmet, Effect of the presence of an ordered micro-pillar array on the formation of silica monoliths, J. Chromatogr. A. 1216 (2009) 7360–7367, http://dx.doi.org/10. 1016/j.chroma.2009.03.010. [18] K. Kanamori, K. Nakanishi, K. Hirao, H. Jinnai, Three-dimensional observation of phase-separated siloxane sol-gel structures in confined spaces using laser scanning confocal microscopy (LSCM), Colloids Surf. A: Physicochem. Eng. Aspects 241 (2004) 215–224, http://dx.doi.org/10.1016/j.colsurfa.2004.04. 022. [19] S. Forster, H. Kolmar, S. Altmaier, Synthesis and characterization of new generation open tubular silica capillaries for liquid chromatography, J. Chromatogr. A. 1265 (2012) 88–94, http://dx.doi.org/10.1016/j.chroma.2012. 09.054. [20] S. Forster, H. Kolmar, S. Altmaier, Performance evaluation of thick film open tubular silica capillary by reversed phase liquid chromatography, J. Chromatogr. A. 1283 (2013) 110–115, http://dx.doi.org/10.1016/j.chroma. 2013.01.107. [21] S. Forster, H. Kolmar, S. Altmaier, Preparation and kinetic performance assessment of thick film 10-20 ␮m open tubular silica capillaries in normal phase high pressure liquid chromatography, J. Chromatogr. A. 1315 (2013) 127–134, http://dx.doi.org/10.1016/j.chroma.2013.09.059.

S. Futagami, T. Hara, H. Ottevaere, et al. / J. Chromatogr. A 1595 (2019) 58–65 [22] T. Hara, S. Futagami, S. Eeltink, W. De Malsche, G.V. Baron, G. Desmet, Very high efficiency porous silica layer open-tubular capillary columns produced via in-column sol–gel processing, Anal. Chem. 88 (2016) 10158–10166, http://dx.doi.org/10.1021/acs.analchem.6b02713. [23] T. Hara, Y. Izumi, M. Nakao, K. Hata, G.V. Baron, T. Bamba, G. Desmet, Silica-based hybrid porous layers to enhance the retention and efficiency of open tubular capillary columns with a 5 ␮m inner diameter, J. Chromatogr. A. 1580 (2018) 63–71, http://dx.doi.org/10.1016/j.chroma.2018.10.023. [24] T.J. Causon, R.A. Shellie, E.F. Hilder, G. Desmet, S. Eeltink, Kinetic optimisation of open-tubular liquid-chromatography capillaries coated with thick porous layers for increased loadability, J. Chromatogr. A. 1218 (2011) 8388–8393, http://dx.doi.org/10.1016/j.chroma.2011.09.047. [25] T. Hara, S. Futagami, W. De Malsche, G.V. Baron, G. Desmet, Exploring the effect of mesopore size reduction on the column performance of silica-based open tubular capillary columns, J. Chromatogr. A. 1552 (2018) 87–91, http:// dx.doi.org/10.1016/j.chroma.2018.03.050. [26] W. De Malsche, J. Op De Beeck, S. De Bruyne, H. Gardeniers, G. Desmet, Realization of 1 × 106 theoretical plates in liquid chromatography using very long pillar array columns, Anal. Chem. 84 (2012) 1214–1219, http://dx.doi. org/10.1021/ac203048n. [27] M. Callewaert, G. Desmet, H. Ottevaere, W. De Malsche, Detailed kinetic performance analysis of micromachined radially elongated pillar array columns for liquid chromatography, J. Chromatogr. A. 1433 (2016) 75–84, http://dx.doi.org/10.1016/j.chroma.2015.12.086. [28] M. Isokawa, K. Takatsuki, Y. Song, K. Shih, K. Nakanishi, Z. Xie, D.H. Yoon, T. Sekiguchi, T. Funatsu, S. Shoji, M. Tsunoda, Liquid chromatography chip with low-dispersion and low-pressure-drop turn structure utilizing a distribution-controlled pillar array, Anal. Chem. 88 (2016) 6485–6491, http:// dx.doi.org/10.1021/acs.analchem.6b01201. [29] F. Detobel, S. De Bruyne, J. Vangelooven, W. De Malsche, T. Aerts, H. Terryn, H. Gardeniers, S. Eeltink, G. Desmet, Fabrication and chromatographic performance of porous-shell pillar-array columns, Anal. Chem. 82 (2010) 7208–7217, http://dx.doi.org/10.1021/ac100971a. [30] W. De Malsche, S. De Bruyne, J. Op De Beeck, S. Eeltink, F. Detobel, H. Gardeniers, G. Desmet, Separations using a porous-shell pillar array column on a capillary LC instrument, J. Sep. Sci. 35 (2012) 2010–2017, http://dx.doi. org/10.1002/jssc.201200279.

65

[31] S. Futagami, T. Hara, H. Ottevaere, G.V. Baron, G. Desmet, W. De Malsche, Preparation and evaluation of mesoporous silica layers on radially elongated pillars, J. Chromatogr. A. 1523 (2017) 234–241, http://dx.doi.org/10.1016/j. chroma.2017.06.062. [32] J. Op De Beeck, M. Callewaert, H. Ottevaere, H. Gardeniers, G. Desmet, W. De Malsche, On the advantages of radially elongated structures in microchip-based liquid chromatography, Anal. Chem. 85 (2013) 5207–5212, http://dx.doi.org/10.1021/ac400576s. [33] G. Desmet, M. Callewaert, H. Ottevaere, W. De Malsche, Merging open-tubular and packed bed liquid chromatography, Anal. Chem. 87 (2015) 7382–7388, http://dx.doi.org/10.1021/acs.analchem.5b01579. [34] S. Futagami, T. Hara, H. Ottevaere, H. Terryn, G.V. Baron, G. Desmet, W. De Malsche, Study of peak capacities generated by a porous layered radially elongated pillar array column coupled to a nano-LC system, Analyst. (2019), http://dx.doi.org/10.1039/c8an01937a. [35] M. Thommes, R. Skudas, K.K. Unger, D. Lubda, Textural characterization of native and n-alky-bonded silica monoliths by mercury intrusion/extrusion, inverse size exclusion chromatography and nitrogen adsorption, J. Chromatogr. A. 1191 (2008) 57–66, http://dx.doi.org/10.1016/j.chroma.2008. 03.077. [36] M. Thommes, Physical adsorption characterization of nanoporous materials, Chemie-Ingenieur-Technik. 82 (2010) 1059–1073, http://dx.doi.org/10.1002/ cite.201000064. [37] M. Thommes, K.A. Cychosz, Physical adsorption characterization of nanoporous materials: progress and challenges, Adsorption. 20 (2014) 233–250, http://dx.doi.org/10.1007/s10450-014-9606-z. [38] U.D. Neue, HPLC columns: theory, technology, and practice, Instrum. Sci. Technol. 26 (1998) 439–440, http://dx.doi.org/10.1080/10739149808001913. [39] W. De Malsche, S. De Bruyne, J. Op De Beek, P. Sandra, H. Gardeniers, G. Desmet, F. Lynen, Capillary liquid chromatography separations using non-porous pillar array columns, J. Chromatogr. A. 1230 (2012) 41–47, http:// dx.doi.org/10.1016/j.chroma.2012.01.060. [40] M. Callewaert, K. Maeno, S. Sukas, H. Thienpont, H. Ottevaere, H. Gardeniers, W. De Malsche, Integration of uniform porous shell layers in very long pillar array columns using electrochemical anodization for liquid chromatography, Analyst. 139 (2014) 618–625, http://dx.doi.org/10.1039/c3an02023a.