Elsevier

Journal of Chromatography A

Volume 1470, 28 October 2016, Pages 76-83
Journal of Chromatography A

Experimental evaluation of chromatographic performance of capillary and microfluidic columns with linear or curved channels

https://doi.org/10.1016/j.chroma.2016.10.004Get rights and content

Highlights

  • We experimentally verified the theoretical relationship between column efficiency N and peak capacity Pc.

  • The relationship was utilized to estimate the column efficiency from gradient analysis.

  • The method was applied to microfluidic (μLC) column efficiency testing.

  • μLC columns with curved channels exhibit a bias from Pc  1 versus N relationship.

Abstract

We prepared 0.3 or 0.15 mm i.d. columns from both fused silica capillaries and planar titanium wafers with machined grooves. Both types of devices were packed with sub-two micron C18 sorbent. Chromatographic efficiency and peak capacity were tested using LC instruments with low extra column dispersion (300 nL2 or 30 nL2 for 0.3 or 0.15 mm i.d. columns, respectively). Micro column testing in gradient mode was less affected by extra column (pre-column) dispersion. To exploit this feature we developed a method for estimation of column efficiency from gradient analysis using the theoretical relationship (Pc  1) = N0.5 × const. The validity of this relationship was experimentally verified using 2.1 mm i.d. and 0.3 mm i.d. columns.

The (Pc  1) versus N relationship was experimentally determined with straight columns, which in turn was employed for the estimation of microfluidic column efficiency. Microfluidic devices with serpentine channels exhibited lower isocratic efficiency than straight capillary columns, but the loss of peak capacity was less significant. The loss chromatographic efficiency due to zone dispersion in serpentine microfluidic channels was more apparent for 0.3 than 0.15 mm i.d. devices. Gradient performance of 0.15 × 100 mm microfluidic columns was comparable to state-of-the-art 2.1 × 100 mm columns packed with the same sorbent.

Introduction

Capillary (cLC) or microfluidic (μLC) liquid chromatography columns are used for high sensitivity liquid chromatography mass spectrometry (LC MS) analyses [1], [2], [3], [4], [5], [6], [7]. Some users prepare capillary columns in-house while others rely on commercially available products. In either case, having methods for the determination of column quality is critical.

Column performance is traditionally expressed as the number of theoretical plates obtained under isocratic elution conditions [8]. However, the experimentally measured column efficiency (N) is adversely affected by the contribution of extra-column (system) dispersion [9], [10], [11], [12], [13], [14]. For separations performed on capillary LC columns of small internal diameter (i.d.), the contribution of the LC system to overall peak dispersion is significant, in many cases precluding a direct measurement of true (intrinsic) column efficiency Nint. Extreme precaution must be taken to reduce the LC system dispersion, including minimization of tubing lengths and i.d., developing zero volume connections, optimization of the injector [15], and detection with near-zero dispersion [15], [16], [17]. “Pinch” injection (where the sample loop is inserted into the flow path for only a brief moment) [15], [18], [19], as well as on-column detection (zones eluting from column are detected directly at outlet frit) have been shown to reduce extra column dispersion [15], [16], [17].

For cLC column testing, the tubing-derived dispersion must be considered carefully. Recent publications by Grinias et al. [20] and Gritti et al. [9] illustrated that within the practical range of cLC flow rates, the connecting tubing dispersion can be predicted from the Aris-Taylor equation or its modified version. Grinias et al. [20] show that the injector and tubing connections are dominant sources of dispersion in cLC. The importance of low-volume connections in maintaining high performance separations on cLC columns was demonstrated by Franklin [21].

Recently, Gritti et al. [9] investigated the contribution of 2.1 mm i.d. column hardware to band dispersion in UHPLC and estimated it to be between 0.1-0.7 μL2 for well-designed frits. While this dispersion level is small, it does reduce the achievable efficiency for analytes with low retention, in particular when using short columns packed with sub two-micron sorbent. To our knowledge, the contribution of column hardware to system dispersion in cLC columns has not been systematically studied yet [22], but it is likely to be more significant than in case of 2.1 or 4.6 mm columns. As a consequence, the experimentally observed efficiency Nobs of cLC column may be lower than the intrinsic column efficiency Nint.

The impact of system dispersion on observed efficiency is illustrated in Table 1 for 100 mm long columns of various internal diameters; the intrinsic efficiency is assumed to be Nint = 25000. As shown in Table 1, the extra column dispersion must be below 493 nL2 for 0.3 mm i.d. columns, and below 31 nL2 for 0.15 mm i.d. columns to experimentally observe an efficiency value that is greater than 95% of Nint. However, reducing cLC system dispersion to such levels is difficult. For example, 1 m of 25 μm i.d. capillary tubing generates a variance of 80 nL2; the same length of 20 μm i.d. capillary gives a variance of 33 nL2 (calculated from Aris-Taylor equation [23] for diffusion coefficient 1 × 10−5 cm2/s, and flow rate 1.5 μL/min).

Gradient elution mode has been also employed to measure a column’s quality [24], [25], [26]. Serving as a surrogate to column efficiency, peak width w or peak capacity Pc becomes the primary indicators of column performance. The advantage of this approach is that well retained analytes are focused on the column inlet at the beginning of analysis, drastically reducing the contribution of sample volume, frits, and pre-column tubing to dispersion [18], [19], [27]. Some band compression also occurs during gradient elution [28], [29]. However, column performance characterization in gradient mode is a less generic approach, since the peak width (peak capacity) also depends on the gradient slope [25], [30].

To this end, Snyder and Dolan [31] proposed a model that can be used to measure the column efficiency in gradient mode (Eq. (1)). When including the gradient compression factor, it is possible to estimate the column efficiency with even better accuracy, albeit with the additional need to determine this factor. Neue et al. evaluated an alternative approach for the estimation of column efficiency for proteins [32]. Both methods are potentially useful to measure the performance of cLC columns in gradient mode.

The goal of this work is to investigate the performance of cLC and μLC columns. To accomplish this objective, we investigated a theoretical relationship between column efficiency and peak capacity, and developed a method for column efficiency estimation using gradient analysis of common small molecule analytes without the need to experimentally determine multiple chromatographic parameters.

Section snippets

Theory

Snyder and Dolan proposed a relationship between peak width w at gradient elution and column efficiency N [31]. Assuming that the injected sample volume is negligible compared to column volume, the temporal peak width w in linear gradient LC can be calculated from Eq. (1), where D is the gradient compression factor [28], [29], [31], [33], t0 is elution time of an unretained peak, N is column plate count (efficiency), and ke is the analyte retention factor at point of elution. This relationship

Materials and reagents

Analytes used in this study, uracil (U), acetophenone (AP), propiophenone (PP), butyrophenone (BP), valerophenone (VP), hexanophenone (HP), reserpine (RES), and bombesin (BOM) were obtained from Sigma (St. Louis, MO, USA). HPLC grade acetonitrile (MeCN) was purchased from Fisher Scientific (Fair Lawn, NJ, USA). A Milli-Q water purification system (Millipore, Bedford, MA, USA) was used for preparation of HPLC mobile phases.

LC instrumentation

Chromatographic experiments were carried using I-Class ACQUITY UPLC

Verification of Pc vs. N relationship

While the definition of peak capacity in gradient elution is generally accepted [25], [30], [31], [35], [36], [37], the validity of Eq. (6b) has not been extensively investigated. In a recent report, we experimentally validated Eq. (6b) using several 2.1 mm i.d. reverse-phase (RP) LC columns of different lengths, operated at various flow rates. During these experiments, the gradient slope was carefully held constant by adjusting the gradient time [9]. Fig. 1 shows the data adopted from reference

Conclusions

In this study we experimentally verified the theoretically predicted relationship between peak capacity Pc and column efficiency N. A column’s Pc  1 value is approximately a function of the square root of N, and it is independent of the column length or experimental flow rate as long as the gradient slope is held constant. This relationship was employed to estimate a capillary column’s efficiency by performing only a gradient test and comparing it to the isocratic and gradient performance of a

Acknowledgements

Authors thank our colleagues at Waters corporation for designing titanium columns and capillary fittings, namely Bob Jencks, David Prentice, Russ Keene and Joseph Michienzi.

References (43)

  • L. Chen et al.

    Fast preparation of a highly efficient organic monolith via photo-initiated thiol-ene click polymerization for capillary liquid chromatography

    J. Chromatogr. A

    (2015)
  • F. Gritti et al.

    Accurate measurement of dispersion data through short and narrow tubes used in very high-pressure liquid chromatography

    J. Chromatogr. A

    (2015)
  • M. Gilar et al.

    Peak capacity in gradient reversed-phase liquid chromatography of biopolymers. Theoretical and practical implications for the separation of oligonucleotides

    J. Chromatogr. A

    (2007)
  • U.D. Neue

    Theory of peak capacity in gradient elution

    J. Chromatogr. A

    (2005)
  • U.D. Neue

    Peak capacity in unidimensional chromatography

    J. Chromatogr. A

    (2008)
  • A.C. Sanchez et al.

    Performance optimizing injection sequence for minimizing injection band broadening contributions in high efficiency liquid chromatographic separations

    J. Chromatogr. A

    (2012)
  • H. Poppe et al.

    Peak width in solvent-programmed chromatography. I. General description of peak broadening in solvent-programmed elution

    J. Chromatogr.

    (1981)
  • U.D. Neue et al.

    Peak compression in reversed-phase gradient elution

    J. Chromatogr. A

    (2006)
  • U.D. Neue et al.

    The determination of the plate count for large molecules under reversed-phase gradient conditions

    J. Chromatogr. A

    (2010)
  • F. Gritti

    General theory of peak compression in liquid chromatography

    J. Chromatogr. A

    (2016)
  • M. Gilar et al.

    Implications of column peak capacity on the separation of complex peptide mixtures in single- and two-dimensional high-performance liquid chromatography

    J. Chromatogr. A

    (2004)
  • Cited by (8)

    • Migration and elution equations in gradient liquid chromatography

      2019, Journal of Chromatography A
      Citation Excerpt :

      the inlet solvent composition (ϕi) is a linear function of time Here Rϕ is the mixing rate [31] (gradient slope [2,32–35], time steepness of the gradient [8], time-based gradient steepness [36]) – a fixed temporal rate of increase of ϕi, and kinit is ki at t = 0. As mentioned in theoretical section, general migration equations, Eqs. (2) and (26.a, 26.b), reproduced in this report from an earlier publication [13] are sufficient for numerical prediction of a solute retention time in any column-based linear chromatography under practically any conditions.

    • A stochastic view on column efficiency

      2018, Journal of Chromatography A
      Citation Excerpt :

      For standard 2 μm VHPLC particles, the results revealed that analytical columns (2.1–4.6 mm i.d.) and capillary columns (< 380 μm i.d.) can both be packed well (h < 2). This result is fully consistent with the past [24] and recent [19,76,77] experimental performances of analytical and capillary columns packed with 2 μm particles. Interestingly, columns of i.d. in between 0.4 and 1.2 mm cannot be packed well (h > 3) with 2 μm particles because the volume fractions of the intermediate and bulk zones in the column are both significant.

    • Chromatographic performance of microfluidic liquid chromatography devices: Experimental evaluation of straight versus serpentine packed channels

      2018, Journal of Chromatography A
      Citation Excerpt :

      In a properly designed experiment with constant generalized gradient slope the relationship between Pc–1 and N can be utilized to predict column efficiency from gradient experiments, or to compare the performance of different μLC columns [17]. In our previous work [17] we experimentally investigated Pc–1 versus N relationship for 0.3 mm i.d. capillary columns; the fitted power function trend was Pc −1 = N0.517 × 0.376. This trend is acceptably close to the expected relationship of Pc−1 versus square root of N.

    View all citing articles on Scopus
    View full text