RESUMEN

Applying the recently introduced patchwork model for porous media, we present a new step forward in the modelling of eddy dispersion in chromatographic columns. The logarithmic law describing the velocity dependency emerging from this patchwork model is supplemented with a retention factor dependency via first principles modelling of the variations in flow resistance and retention capacity caused by the packing disorder. Furthermore, it is shown the derived expression is also able to fit the eddy dispersion originating from the wall effect on the packing. When applied to literature data of eddy dispersion, the newly introduced logarithmic law has a goodness of fit that is at least equal to that of Knox' empirical power law (R2>0.98). The main difference is that, whereas Knox' power law requires a separate fit for each component due to the retention factor dependency, the present model simultaneously fits all plate height curves measured on one chromatographic column, using only two parameters with a clear physical meaning.

Asunto(s)

RESUMEN

We report on a steady-state based, and hence highly accurate numerical modelling study of the effect of the top and bottom wall in the current generation of micro-pillar array columns. These have a mesoporous retention layer that not only covers the pillar walls but also the bottom wall. Our results show that the performance of these columns can in general not be improved by also covering the top wall with the same layer, despite the increased column symmetry this approach would offer. The reason for this is that the local species retardation caused by a retentive layer is much stronger than the pure flow arresting effect of an uncovered wall. At least, this has a crucial impact in high aspect-ratio systems such as micro-pillar array columns because these require a small inter-pillar distance to promote mass transfer together with a large channel depth to enable a sufficiently high flow rate. On the other hand, a notable improvement could be made if micro-pillar array would be produced without having a retentive layer at the bottom. At Péclet number Pe = 50 and aspect ratio AR = 5 for flow-channels, this gain amounts up to about 4.5 h-units at a zone retention factor k'' = 2 and 1.75 h-units at k'' = 16 (gain scales almost linearly with Pe). To verify these results, we also considered another high aspect-ratio system with a simplified geometry: the open-tubular channel with a flat-rectangular cross-section. This led to very similar observations, thus confirming the findings for the micro-pillar array. The results produced in the present study also allow us to conclude that the classic modelling paradigm adopted in chromatography, which is based on the independency and hence additivity of the hCm- and hCs-contributions, can lead to large modelling errors in chromatographic systems with a high aspect-ratio, even when their geometry is so simple as that of a straight open-tubular channel with constant cross-section. Indeed, when both zones are treated independently, the analysis misses how the vertical diffusion through the retentive layer helps suppressing the vertical gradients in the mobile zone. The diffusion through this layer occurs in a ratio of k''Ds/Dm (Dm being the diffusion coefficient in mobile phase zone and Ds being the diffusion coefficient in stationary phase zone), such that at high retention factors this diffusion contribution even becomes the dominant one.

Asunto(s)

RESUMEN

We have investigated the possibility to establish a theoretical plate height expression for the band broadening in the most widely used micro-pillar array column format, i.e., a cylindrical pillar array wherein the pillar walls and the channel bottom are coated with a thin layer of meso­porous material. Assuming isotropic diffusion in the shell-layer, it was found that the vertical diffusive transport along the porous shell-layer covering the pillar walls significantly suppresses the band broadening originating from the vertical migration velocity gradients. As the vertical transport in the shell-layer increases linearly with the retention equilibrium constant K, this leads to an anomalous dependency on the retention factor. Indeed, instead of increasing with k'' and following the classic (1+ak''+bk''2)/(1 + k'')2-dependency governing a classic Taylor-Aris system, the variation of the mobile zone mass transfer resistance term hCm in a 3D pillar array with bottom-wall retention goes through a maximum (resp. factor 1.5 (k''=4) and 2 (k''=16) difference between observed and classic Taylor-Aris behaviour). This effect increases with increasing pillar heights and increasing reduced velocities. Because of this complex k''-dependency, it proves very cumbersome to establish a general plate height equation covering all conditions. Instead, a plate height expression was established that is limited up to k''=4, but remains accurate for higher k''-values for cases where the ratio of pillar height over inter-pillar distance remains below 5. It can however be anticipated the proposed analytical model is only valid in a rather limited range around the presently considered external porosity of ε=0.5.

Asunto(s)

RESUMEN

Using a two-zone moment analysis (TZMA) method based on Brenner's generalized dispersion theory for two-dimensional (2D) and three-dimensional (3D) periodic media, we investigated the mechanisms for dispersion in particulate media for liquid chromatography. This was done using a set of plate height data covering an unprecedented wide range of retention factors, diffusion coefficients, and velocities, all computed with unequaled accuracy. Applying Giddings' additivity test, based on alternatingly making the diffusion coefficient in the mobile and stationary zones infinitely large, the dispersion data clearly indicate a lack of additivity. Although this lack could be directly understood by identifying the existence of multiple parallel mass transfer paths, the additivity assumption interestingly overestimates the true C term band broadening (typically by more than 10%, depending on conditions and dimensionality of the system). However, Giddings originally asserted the occurrence of parallel paths would always lead to an underestimation of the dispersion. The origin of the lack of additivity is analyzed in detail and qualitatively explained. Finally, we also established a generic framework for the modeling of the effect of the reduced velocity and the retention coefficient on the C term in ordered chromatographic media. This led to the introduction of a new expression for the mobile zone mass transfer term, which, unlike the currently used literature expression, contains the complete k″ dependency.

RESUMEN

We investigated the possibility of reducing the effect of precolumn band broadening (PreCBB) by sandwiching the sample between two small plugs of an immiscible liquid. It has been found that in cases of severe PreCBB, improvements in peak efficiency can amount up to 20 times for the early-eluting compounds. For smaller degrees of PreCBB, the gain on the efficiency of early-eluting compounds is smaller (order of 50%), yet it is still significant. It has been verified that the presence of the immiscible fluid sandwich does not affect the repeatability of the analysis nor the linearity of the calibration curves used for analyte quantitation. It is also shown that the main effect of the two sandwich plugs is the minimization of the dispersion in the precolumn transfer tubing itself, which makes the method fundamentally different from pure on-column focusing methods such as the performance optimizing injection sequence (POISe) method. It is further demonstrated that both halves of the sandwich are needed, since the beneficial effect is clearly much smaller when only one plug is present. A drawback of the method is that some of the late-eluting peaks are slightly adversely affected by the presence of the sandwich liquid in the case where 127 µm i.d. tubing was used. The mechanism for this peak deterioration effect is at present still unclear but only occurs under gradient conditions and is clearly linked to the size of the sandwich plugs (the smaller the plugs, the smaller the adverse effect) and the internal diameter of the tubing used between the injection valve and the column.

RESUMEN

We report on a new homogenization approach to solve, with drastically improved speed and accuracy, the general advection-diffusion equation in hierarchical porous media with localized diffusion and adsorption/desorption processes, thus opening the way to a much deeper understanding of the band broadening process in chromatographic systems. The proposed robust and efficient moment-based approach allows us to compute the exact local and integral concentration moments and hence provides exact solutions for the effective velocity and dispersion coefficients of migrating solute particles. Innovative to the proposed method is also that it not only produces the exact effective transport parameters of the long-time asymptotic solution, but also their entire transient. The analysis of the transient behaviour can be used, for example, to properly identify the time and length scales needed to achieve the macro-transport conditions. If the hierarchical porous media can be represented as the periodic repetition of a unit lattice cell, the method only requires the solution of the time-dependent advection-diffusion equations for the zeroth order and first-order exact local moments, exclusively on the unit cell. This implies an enormous reduction of the computational efforts and a significant improvement of the accuracy of the results when compared to the direct numerical simulation (DNS) approaches which require flow domains that are long enough to achieve steady-state conditions, and hence often cover tens to hundreds of unit cells. The reliability of the proposed method is verified by comparing its predictions with DNS results, in one, two and three dimensions, in both transient and asymptotic conditions. The influence of top and bottom no-slip walls on the separation performance of chromatographic columns with micromachined porous and nonporous pillars is discussed in detail.

Asunto(s)

RESUMEN

We report on a generic mathematical study of the transient plate height regime in bundles of polydisperse capillaries with diffusional bridging. I.e., we have investigated how the plate height in such systems can be expected to vary with the residence time, or equivalently, with the column length, before reaching a long-time limit wherein the plate height becomes constant. This is important because, in case of systems with a large number of capillaries (N), the transient regime solution is practically much more relevant than the long-time limit solution. The problem has been investigated using a simplified, yet representative geometry that is amenable to a semi-analytical solution, only requiring a numerical integration of the velocity field across a line (2D geometries) or a square or rectangle (3D geometries). The availability of this semi-analytical approach allows to consider the required large number of different random capillary diameter drawings (order of 10,000) needed to establish a sufficiently averaged result (deviation from true mean on the order of 1%). The general solution could be simplified into a set of two-parameter expressions describing either the long-time limit (τ→∞) as well as that of an infinite number of capillaries (N→∞). The combinations of N and τ where these expressions are valid to within an accuracy of 5% could be delimited with simple analytical expressions as well. It has also been observed that the long-time limit solution for the additional plate height (Δhτ→∞) remains dependant on the number of capillaries N, even when N tends to infinity. It is found that Δhτ→∞ can be expected to increase in proportion with N-1 (2D-case) or with ln(N-1) (3D-case). This dependency of Δhτ→∞ on N is a result that is not obtained when modelling the capillary bundle using a representative binary capillary system approach as has been done up-till-now in literature.

Asunto(s)

RESUMEN

Using a recent generalization of the Aris dispersion method, we have derived exact analytical expressions for the long-time limit dispersion in 2-D multi-capillary packings with diffusional bridging (binary channel system). Both the plug flow and the parabolic flow case are considered. The expressions are mathematically exact over the entire range of possible values of the degree of polydispersity (σ), the retention equilibrium constant K and the diffusion coefficients in the mobile and stationary zone. They are validated by comparing them to the dispersion data obtained by numerically solving the partial differential equation describing the general advection-diffusion mass balance. A correlation coefficient of R2 > 0.99995 is obtained. The form of the obtained analytical expression shows the dispersion arising from the polydispersity effect in multi-capillary systems can be split in two contributions, one related to the actual velocity difference between the adjacent channels and one related to the fact that the average of the intra-capillary C-term band broadening is larger in any σ≠0-case than in the σ=0-case. When leaving the small σ-assumption, a new factor ((1- σ2)/(1+3σ2)2) emerges, common to both the solution for the plug flow and the parabolic case. The analysis shows the parabolic flow not only leads to a larger intra-capillary dispersion (σ=0) than the plug flow case as known from literature but is also more sensitive to the polydispersity effect (up to some 10 to 20%).

Asunto(s)

RESUMEN

The origin of the peak skewness that can be observed when applying the deconvolution method to isolate the diffusion process from the flow processes for peak parking experiments conducted under conditions of slow radial equilibration and strong trans-column velocity gradients was investigated. Numerical simulations were carried out for a variety of trans-column velocity profiles and a broad range of experimental conditions and system parameters were investigated. Results show that, under the aforementioned conditions, the traditionally employed variance subtraction method displays a consistent error which follows the dynamics of the diffusive relaxation during both the peak parking and the flow steps. It is also found that, under the same conditions, the peak deconvolution method is bound to produce deconvoluted "parking-only" peaks that are strongly asymmetric, despite the perfectly symmetric nature of the pure diffusion process marking this parking step. It is shown that this asymmetry is acquired during the flow step following the parking stop. During this step, parked and non-parked peaks are deformed in different ways, despite being subjected to the same trans-column velocity profile. This different deformation cannot be filtered away with the deconvolution or the variance subtraction method, hence introducing an error. Solutions to alleviate the peak skewness and the variance error consist of parking the peak close to the inlet or the outlet or exiting the parked peak through the column inlet (flow reversal method). Under the considered conditions, these approaches could reduce the error on the measured effective diffusion coefficient up to 87%. Carrying out the variance subtraction or the deconvolution process with a peak that has also been parked for a substantially long parking time instead of using a "no-parking" peak as is customary done, is another option to counter the effect.

Asunto(s)

RESUMEN

We report on the performance of three classes of evolutionary algorithms (genetic algorithms (GA), evolution strategies (ES) and covariance matrix adaptation evolution strategy (CMA-ES)) as a means to enhance searches in the method development spaces of 1D- and 2D-chromatography. After optimisation of the design parameters of the different algorithms, they were benchmarked against the performance of a plain grid search. It was found that all three classes significantly outperform the plain grid search, especially in terms of the number of search runs needed to achieve a given separation quality. As soon as more than 100 search runs are needed, the ES algorithm clearly outperforms the GA and CMA-ES algorithms, with the latter performing very well for short searches (<50 search runs) but being susceptible to convergence to local optima for longer searches. It was also found that the performance of the ES and GA algorithms, as well as the grid search, follow a hyperbolic law in the large search run number limit, such that the convergence rate parameter of this hyperbolic function can be used to quantify the difference in required number of search runs for these algorithms. In agreement with one's physical expectations, it was also found that the general advantage of the GA and ES algorithms over the grid search, as well as their mutual performance differences, grow with increasing difficulty of the separation problem.

Asunto(s)

RESUMEN

The additivity assumption underlying Giddings' coupling model for the eddy-dispersion in laminar flows through heterogeneous media is critically analyzed and a potential solution for its non-additivity in the high velocity limit is presented. Whereas the unit cell in Giddings' model only consists of a single velocity bias step, the unit dispersion cell of the newly proposed model comprises two consecutive velocity bias steps. Consequently, the unit cell of this new model allows to account for the occurrence of an internal velocity bias rectification at high reduced velocities and is therefore additive in both the low and high velocity limit. First, a mathematical expression for the velocity- and diffusion-dependency of the model's dispersion characteristics has been established. Subsequently, the physical behavior of the model is discussed. It is shown the relation between the eddy-dispersion plate height h and the reduced velocity ν can be expected to display a local maximum in systems where the transversal dispersion purely occurs by molecular diffusion, as is the case in perfectly ordered flow-through media. In disordered media, where the transversal dispersion also contains a significant advective component, the model predicts a velocity-dependency that is qualitatively similar to that described by Giddings' coupling model but, all other conditions being equal, converges to a significantly smaller horizontal asymptote at high reduced velocity. The latter might shed new light on earlier eddy-dispersion studies pursuing a quantitative agreement between experimental data and the Giddings model.

Asunto(s)