1.1
2D-HPLC - Method Development for Successful Separations

Dwight R. Stoll Ph.D.

Gustavus Adolphus College, Department of Chemistry, 800 West College Avenue, St. Peter, MN 56082, USA

1.1.1 Motivations for Two-Dimensional Separation

Historically, much of the research devoted to multidimensional separations and their application to real analytical problems has been focused on dealing with complex samples. These have traditionally been described as containing hundreds or thousands of compounds and are often derived from natural sources such as plant extracts or body fluids (e.g. blood or urine). Increasingly, however, we observe that multidimensional separation can be exquisitely effective for dealing with samples containing analytes that are difficult to separate but are not complex by the traditional definition. Since this distinction can have a big impact on how one approaches method development, we start here by explicitly differentiating the two cases.

1.1.1.1 Difficult-to-Separate Samples

The difficulty associated with a separating a particular sample may originate from its sheer complexity (i.e. thousands of compounds). In this case relying on chromatographic separation alone will not be enough to fully separate the mixture, and some other source of selectivity will be needed (e.g. sample preparation, and/or selective detection such as mass spectrometry). However, it is now common to encounter samples that contain only a few compounds but are difficult to separate simply due to the high degree of similarity of the compounds in the mixture. For example, a mixture may only contain six compounds, but if two of those six compounds are enantiomers (1a and 1b), then fully separating the mixture using a single column may be difficult even if the separation of compounds 2-5 from 1a/1b is straightforward. Such situations are encountered more frequently now compared to the past, in part due to the development of small-molecule drugs with multiple chiral centers [1], and the increasing recognition of the importance of both the D- and L- enantiomers of amino acids ([2], see Chapter 1.7), for example.

1.1.1.2 Complex Samples

As stated above, traditionally complex samples have been thought of as containing hundreds or thousands of different compounds. These samples often come from nature, but not always. For example, surfactants and polymers produced by chemical synthesis can result in highly heterogeneous mixtures of thousands of different compounds. Historically, the analysis of such samples by multidimensional chromatography has been mainly focused on so-called comprehensive methods of separation that yield a kind of global profile or "fingerprint" of the contents of the sample. However, in cases where only one or a few particular molecules in the sample are of importance to the analysis, simpler multidimensional separation methods such as heartcutting can be adequate, and even preferred.

1.1.1.3 Separation Goals

As is often discussed in the multidimensional separation literature, and below, the process of developing a multidimensional separation method is one full of compromises. For example, conditions that favor shorter analysis times do not lead to the best detection sensitivity, and vice versa. Therefore, it is important for the analyst to identify - at the very beginning of method development - what are the characteristics or performance metrics for the method that are most important to him/her. For example, if achieving baseline resolution of six critical pairs of analytes is critically important for the method to be successfully applied, then method development decisions should support this objective, even if it comes at the cost of increased analysis time, and/or lower detection sensitivity.

1.1.2 Choosing a Two-Dimensional Separation Mode

All two-dimensional separations can be executed either "offline" or "online." In the offline mode, one or more fractions of 1D effluent are collected in some kind of storage device such as a set of vials or a wellplate. These fractions are then injected at some later time (minutes to years) into another LC system (i.e. the same LC system running different conditions from the 1D separation, or a different LC system altogether), either with or without intermediate processing of these fractions. For example, in proteomics applications of 2D-LC, it is common to desalt the fractions, or dry them down by evaporation to remove organic solvent, before analysis by the 2D separation [3]. In the online mode, fractions collected from the 1D column are either processed immediately by direct injection into the 2D column, or stored for a short time (seconds to hours) in some kind of device (typically capillary loops or sorbent-based traps) that is internal to the instrument. An example of an instrument configuration commonly used for this purpose is shown in Figure 1.1.1. In this case, the interface valve situated between the 1D and 2D columns has two positions. Switching between them changes the roles of loops 1 and 2 between collecting 1D effluent and introducing the fraction of the 1D effluent into the 2D flow stream, effectively injecting that material into the 2D column.

Figure 1.1.1 Illustration of an instrument configuration typically used for 2D-LC.

Source: Dr. Gabriel Leme.

As commercially available equipment for 2D-LC separation has become more sophisticated and reliable, the trend in the industry has been to move away from offline separations because of challenges associated with implementation of offline separations for large numbers of samples, and with degradation and contamination of 1D effluent fractions when they are handled external to the instrument [4]. Given this trend, I have chosen to focus entirely on online 2D-LC for the rest of this chapter. Readers interested in learning more about offline 2D-LC are referred to review articles dedicated to this topic [5, 6].

1.1.2.1 Analytical Goals Dictate Choice of Mode

Starting in the late 1970s, different groups began developing the modes of 2D-LC separation we have come to know as "heartcutting" and "comprehensive" [4, 7]. In the most recent decade, two additional modes have been developed, which are now known as "multiple heartcutting" and "selective comprehensive" 2D separations. Each of these four modes will be discussed in some detail in Section 1.1.2.2. At this point, though, I want to emphasize that choosing which separation mode you will use should be driven by the overall goals of the analysis. For example, if you have a complex sample and you want to learn as much as you can about that sample (i.e. identify hundreds of compounds), then the comprehensive mode of 2D separation will almost always be the best choice. However, if you are only interested in a few target compounds in the sample - even if the sample matrix is highly complex - then a more targeted mode of 2D separation such as heartcutting or multiple heartcutting will likely be the best approach. In practice, time spent on each 2D separation is one of the most precious resources of the 2D-LC instrument, and allocating effort to 2D separations that are not necessary to achieve the overall analytical goals of the analysis is costly (in terms of both time and supplies), wasteful, and adds unnecessary complexity to the method.

1.1.2.2 Survey of Four 2D Separation Modes

The vast majority of 2D-LC applications being developed today fit into one of the four modes of 2D separation illustrated in Figure 1.1.2. In the single-heartcut mode (A; LC-LC), a single fraction of 1D effluent containing analytes of interest is captured at the outlet of the 1D column and transferred to the 2D column where this submixture of the original sample will be further separated if the separation mechanisms employed in the first and second dimensions are complementary. Perhaps the biggest advantage of the LC-LC mode is that the time that can be dedicated to separation of the 1D effluent fraction in the second dimension is not strictly limited. This provides tremendous flexibility in terms of choosing parameters for the 2D separation, including flow rate, column dimensions, and injection volume. The biggest disadvantage of LC-LC, however, is that the scope of the analysis is limited. We are restricted to the analysis of compounds that can be captured in a single fraction of 1D effluent. Nevertheless, the LC-LC approach has been used to great effect in application areas ranging from identification of small-molecule pharmaceutical impurities [8] to the detection of drug metabolites in plasma [9].

Figure 1.1.2 Illustration of four different modes of 2D-LC separation.

The extreme opposite of LC-LC in terms of analytical scope is the comprehensive mode of 2D separation (D; LC × LC). As the illustration shows, in this case, fractions of 1D effluent are collected and transferred - one at a time, in a regular, serial fashion - to the 2D separation. Typically, this results in a long string of many (tens to hundreds) 2D chromatograms collected in a single detector datafile. This long data...