Patent Application
20070117212
N/A
In the last 50 years, and especially in the last decade, there has been a remarkable trend towards both the automation and the miniaturization of chemical analysis and electromechanical systems. The limits of detection of primary analytical methods have improved by many quantum leaps. Mass spectrometry can detect attomoles of sample using nanospray methods. Nuclear Magnetic Resonance (NMR) can now detect pmoles of analyte using 5 nL microcoil probes (Olson, 1995), a 500-fold improvement over 1980's technology. Capillary electrophoresis with laser-induced fluorescence detection (CE-LIF) can detect zeptomoles of analyte in a volume of picoliters. The tremendous sensitivity of these microscale analytical technologies, however, is useless without the ability to efficiently load and deliver the appropriate microscale samples. For example, capillary electrophoresis analysis generally requires providing several microliters of sample, from which a few nanoliters is drawn. Loading a 1 μL microcoil NMR probe requires filling a 10 μL dead volume. On microfluidic chips, samples typically are introduced to fill entire channels, of which only a small segment may occupy a region of detection or be injected into a separation channel.
An obvious alternative would be to supply small samples and drive them through the conduit either with air or with clean solvent. However, in pressure-driven liquid flow, a sample originating as a short volume segment of a conduit will disperse into a larger volume, with concomitant dilution, proportionally to the volume through which it is moved: the boundary layer at the conduit wall is immobile; however, flow at the center of the conduit is rapid. Although dispersion of small concentrated samples can be significant even within the few-cm distances of a microfluidic chip, the problem is most vividly defined and discussed in the example of flow-NMR, where samples must be transported over distances of several meters.
NMR is a very information-rich spectroscopy, well-established for confirming the structure and purity of newly synthesized compounds or isolated natural products. It has also proven valuable in metabonomics, using pattern recognition software to analyze large numbers of complex spectra. However, the low sensitivity of NMR (1000-fold less than mass spectrometry) is problematic, particularly in LC-NMR where acquisition time is limited and compounds of interest may be a small fraction of the permissible column load. NMR sensitivity has been improved modestly (2-4-fold) using higher field magnets and cryogenically-cooled electronics (cryoprobes). For mass-limited samples, microcoil NMR probes offer up to a 500-fold sensitivity increase (Olson, 1995), however, efficiently loading microcoil probes is a challenge (Kautz, 2001). The detection cell has a volume of 30 nL to 1 μL is recessed 50 cm or more up the narrow bore of the NMR magnet. In contrast to conventional NMR probes, the microcoil probe's axis is oriented transverse to the magnet bore so sample tubes cannot be inserted without removing the probe, and consequently microcoil probes are generally implemented as a flow cell. An additional complication is that any motorized equipment must be located outside the magnet's fringe field, necessitating an additional 1-10 meters of capillary tubing, depending on the magnet's fringe field. The current commercial offering is a compromise with these limitations, using the smallest feasible transfer capillaries to fill a relatively large flowcell. But the challenge of filling a 1 μL observed volume in a 5 μL flow cell through several meters of 50 micron capillary tubing (2 μL/meter) has severely limited microcoil NMR's sensitivity in practice.
The two traditional approaches to flow-NMR (Keifer, 2003a) are direct-injection NMR (Keifer, 2000) and flow-injection analysis-NMR (Keifer, 2003b). The methods differ in how they optimize the necessary steps of clearing, washing and reloading the NMR probe flow cell through the 2-5 meter transfer line while avoiding sample dilution in the dead volumes of the transfer line and NMR probe flow cell. In direct injection NMR, samples are injected into an empty (air-filled) flow cell through a 100 μm i.d. or larger transfer line. Samples can be injected relatively quickly without dilution; however, the percentage of the injected sample that ultimately resides within the NMR coil observed volume during spectral analysis, is low. The need for a wash cycle to reduce sample-to-sample carryover to <1% increases the sample change time. Because it is not feasible to flush 50 micron capillaries longer than 1 meter with air, and larger capillaries have a prohibitively large volume, direct injection methods have only been implemented on microcoil probes manually. Working at the closest approach to the magnet bore, it is possible to fill the flow cell using 8 μL samples.
In flow injection NMR, the flow lines are maintained filled with solvent. Samples are introduced by means of a sample loop valve and delivered to the probe by a liquid chromatographic pump. Because the sample disperses into the carrier solvent during transfer, the final analyte concentration in the NMR coil depends on the sample volume, flow rate, and system dead volume. Sharp gradients of analyte concentration near the NMR coil immediately after injection can cause poor line shape, and an equilibration time of 1-2 min may be required for line shape to sharpen as the analyte diffuses throughout the flow cell. Because the effects of these gradients are more pronounced for dissimilar solvents, the same solvent must be used for both the carrier and sample preparation.
FIA-NMR methods are applicable to microcoil probes, and a high-throughput FIA-NMR method using a commercial microcoil probe with a microfluidic sample loader (Olson, 2004) has been introduced. This method requires 10 μL of sample to deliver at full concentration, or dilutes smaller samples to a dead volume of 10 μL in the course of loading. 50 μL of deuterated solvent was also required per sample to reduce carryover below 1%.
Another approach is segmented flow, in which an immiscible fluid is used to push a small sample as a bolus or “plug” through the fluidic conduit. This approach appears to offer several advantages. Smaller samples could be used, so sample consumption would be lower. Samples would not be diluted, so NMR acquisition time would be faster. Samples could be more accurately positioned in the detection cell, so setup would be straightforward, faster and provide better sensitivity. There would be no “equilibration time” required for lineshape to improve after injection. And for high-throughput operation, a queue of samples could be quickly advanced a short distance, rather than having each new sample delivered the entire conduit distance. While the stability of segmented plugs in the 3-mm vertical flow cells of conventional saddle coil LC/NMR probes is problematic, several preliminary findings with segmented plugs in microcoil NMR probes appear promising. Segmented flow has historically been implemented in clinical analyses and has recently been demonstrated in a microfluidic chip (Ramsey, 2003).
In work on the optimal sizes of microcoil probes, it was shown that samples sandwiched on both sides by the immiscible fluorocarbon fluid FC 43 could be much smaller than samples sandwiched between air bubbles without degradation of the NMR line shape: only twice the coil size instead of 7 times (Behnia, 1998). The utility of this fluorocarbon bracketing was demonstrated in obtaining spectra from the 500-ng eluate of a single solid-phase synthesis bead (Lacey, 2001).
However, substantial challenges remain in putting this approach into practice (Macnaughtan, 2003). Sample plugs are frequently lost or degraded in a variety of ways. Principally, moving sample plugs leave a film of solvent on the wall of the conduit, e.g., capillary, and this film can consume about 2 μL of sample per meter of movement, which is completely prohibitive. All of this lost material can mix with subsequent plugs, resulting in high carryover (Patton et al., 1997). Plugs also have tended either to acquire large discrete breaks in the middle (“fragmentation”) or to form many small breaks (“frothing”) at their ends. Both of these effects increased with increasing capillary size, where the outward pressure of the curved surface of the plug was insufficient to hold the plug against the conduit wall in capillaries over 300 μm diameter. Improvements in these techniques would be greatly appreciated.
The system and method of the invention provide solutions to the problems identified above. This invention is directed to moving small samples through conduits, e.g., the capillary channels or tubing of a microfluidic device, without dilution of the sample or loss of sample to the capillary wall. The transported sample is a small volume of liquid, for example a solution of an analyte for chemical analysis.
In the preferred embodiment of the method of the invention, an aliquot of a liquid that is not miscible with the sample, denominated an “immiscible carrier liquid,” is first introduced into a conduit through a microfluidic device. When an aliquot of the sample is subsequently introduced into the conduit, the sample forms a segment or “plug” in the microfluidic channel or capillary, following the carrier liquid. The carrier liquid is pumped or otherwise caused to flow through the channel, and the sample is carried from one location to another through the microfluidic channels without dilution or dispersion into the immiscible carrier liquid. The interior wall of the conduit (channel) is covalently coated with a suitable coating, so that the carrier liquid wets the conduit wall preferentially to the sample solvent. A film of the carrier liquid will then be retained on the channel wall as the sample plug is moved passed, so that the moving sample plug will not contact the conduit wall. This avoids losses of analyte either by binding of analyte molecules to the conduit wall or by bulk loss of sample as a film on the conduit wall.
Preferably, the microfluidic device is part of a conduit system, with attached tubing to transport samples onto and off of the device. Small liquid samples may thus be transported long distances through microfluidic plumbing and through such microdevices with very low losses, and at relatively high speed. As a specific example, if the immiscible carrier liquid is a fluorocarbon (FC), and the channel surface is fluorine-rich, the carrier liquid will wet the channel wall preferentially to both aqueous and organic (hydrophobic and hydrophilic) solvent samples. The desired effect may be obtained either by making a portion of the system, e.g., the attached tubing, of a fluorine-rich material such as a teflon (PTFE, ETFE, FEP, NGFP, etc.), or a channel wall, e.g., in the microdevice, may be coated with a fluorine-rich layer such as a fluoroalkyl silane coating on glass, silica, or plastic.
Other features and advantages of the invention will be apparent from the following description of the preferred embodiments thereof and from the claims, taken in conjunction with the accompanying drawings, in which:
(B) photograph of a sample plug (dyed DMSO) recovered after automated withdrawal from a vial, delivery through a 2 meter transfer line, and passage through a commercial microcoil NMR probe which included 2 meters of 50 micron fused silica capillary in addition to the NMR flow cell.
The purpose of activities leading to the invention was to develop an automated segmented flow NMR method that could increase throughput, could utilize low-mass samples efficiently (with minimal sample loss) and could accept samples form a 96-well plates. The target application was high-throughput NMR analysis of combinatorial chemistry library. The primary goal of the development activity was to achieve the highest possible throughput that yielded spectra of sufficient quality to be interpretable by automated spectral analysis software. Gains in sensitivity and sample utilization were also desirable to detect, and if possible to identify, contaminants at the 5% level. Sample-to-sample carryover had to be below 1%, the high operating cost of deuterated solvent consumption was to be reduced as much as possible, and the method was to be implementable with commercially available instrumentation. These development activities led to the method of the invention, a zero dispersion method of segmented flow analysis, implemented at the microscale level.
As an example, a segmented flow analysis approach to microcoil NMR analysis (SFA-NMR) is illustrated in
Zero dispersion segmented flow methods (Patton, 1997) have been demonstrated in larger scale systems and are based on the principle that if the carrier fluid has a favorable contact energy with the tubing wall, relative to the sample, a film of carrier is maintained between the wall and the sample as sample plugs are moved through the tubint, or conduit (Patton, 1997; Nord, 1984; Adler, 1973). The combination of a fluorocarbon carrier liquid in Teflon™ tubing was recently demonstrated in continuous flow PCR, a method that is particularly sensitive to carryover (Curcio, 2003). The fluorocarbon FC 43 has been used in building microcoil probes to match the magnetic susceptibility of the copper wire of the coil (Olson, 1995) and has been shown to improve line shape when used to “bracket” small aqueous sample plugs in microcoil NMR (behnia, 1998; Webb, 1996). FC 43 also has a relatively high viscosity (2.8 cs) among fluorocarbons, which favors film formation (Nord, 1984).
However, the performance of segmented flow in microfluidic devices tends to be poor. Many preferred conduit materials, such as glass, fused silica capillary, PEEK tubing or fittings, metal and polypropylene sample tubes or microtiter plates, will retain a film of aqueous or organic solvents and are poorly wetted by fluorocarbon liquids. Treatment of glass with dichlorodimethyl silane, a conventional well-known hydrophobic coating, is able to abrogate this permanent film retention and permit plugs to be moved slowly, but does not realize favorable for zero dispersion.
At the microscale, in conduits below 200 microns, the sample plugs exert an outward force against the conduit walls, and the use of air bubbles to segregate segments cannot be used because bubbles compress to disappearance at the backpressures encountered.
The key to successful practice of the method of the invention for microscale analysis in microfluidic devices, such as the detection cell of an NMR microcoil (or probe) or other microfluidic device such as a microfluidicchip, which are frequently made of glass, fused silica, silicon or other material not easily wettable by fluorocarbon materials, is to ensure the wettability of the conduit wall by applying to the wall a coating that will change its properties. Treatment of glass or silica surfaces with perfluoroalkylsilanes, using covalent bonding methods (Karger, 2002), can transform silica into a favorable material for zero dispersion segmented flow at the microscale level.
For achieving zero dispersion segmented flow in the transport of samples into and out of a microfluidic device, such as a microcoil NMR flow cell, a flexible material that is inherently preferentially wettable by the carrier liquid is connected to the conduit through the device. Teflon™ is a class of exemplary conduit materials to use with a perflourinated liquid carrier fluid, which is immiscible with either aqueous or most organic solvents used for analytical samples. Teflon™, however, has poor mechanical properties for many microfluidic applications: it is difficult to machine or etch channels in, it is resistant to adhesives, castable Teflons are not resistant to fluorocarbon liquids, and Teflon capillary have poor pressure resistance and are difficult to connect to other components. Thus microfluidic devices in practice will include many unfavorable materials for segmented flow.
Thus, in exemplary embodiments of the method of the invention, the immiscible carrier may be a fluorocarbon liquid, which is immiscible with both aqueous and organic solvents (apparently all solvents and analytes other than mixed hydrocarbon-fluorocarbon solvents).
The transfer conduit (or tubing) leading into a microfluidic device for practicing the method of the invention is made of a perfluorinated or highly fluorinated material and the conduit within the device is coated with such a material, so that the conduit inner wall surface is preferentially wetted by the fluorocarbon solvent. In this case, a film of the fluorocarbon carrier liquid is maintained between the conduit wall and the sample as it passes, such that the sample for analysis does not contact the conduit wall. This prevents adsorption of the analyte to the wall directly, or bulk loss of the analyte solution to film formation on the channel wall.
The above features permit efficient transfer of small discrete samples for analysis through conduits leading into, out of and within microfluidic devices and show a marked advantage for successfully processing samples compared to uniformly filling the channels; to injecting plugs of sample in clean sample solvent as a carrier fluid; or to using an immiscible organic solvent as carrier liquid without such coating of the conduit wall as in the method of the invention.
The following examples are presented to illustrate the advantages of the present invention and to assist one of ordinary skill in making and using the same. These examples are not intended in any way otherwise to limit the scope of the disclosure.
Materials and Methods
Materials. Polyimide-clad fused-silica capillaries were obtained from Polymicro Technologies (Phoenix, Ariz.), and the fluoroalkyl silanes were purchased from United Chemical Technologies (Philadelphia, Pa.). Fluorocarbon FC 43 was from 3M Corp (St. Paul, Minn.). Teflon capillary and tubing were obtained from Cole-Parmer (Vernon Hills, Ill.); PEEK capillary, unions, in-line filters, and adapters were from Upchurch (Oak Harbor, Wash.). The 96-well PCR plates were obtained from Nunc (Rochester, N.Y.). The compounds of the test library (uracil, reserpine, erythromycin, chlorpromazine, tolbutamide, indomethacin, haloperidol, 4-acetamidophenol, indapamide, prilocaine, phenylbutazone, and brucine) were from Sigma. Deuterated solvents and the reference standards tetramethylsilane (TMS) and trimethylsilylproprionate (TMSP) were purchased from Cambridge Isotope Labs (Andover, Mass.). All other solvents, buffer salts, and dyes were obtained from Fisher Scientific (Pittsburgh, Pa.) and were used without further purification.
Instrumentation. NMR spectra were acquired on a Varian (Palo Alto, Calif.) Inova spectrometer with an 11.7-T (500 MHz) actively shielded magnet and a flow NMR package consisting of a Gilson (Middleton, Wis.) model 215 sample handler and Varian VAST automation software. The Gilson 215 was fitted with a 100-μL syringe, and supplemental VAST automation programming (Tcl scripts) was written, as described below. A sample loader, model HTSL-1100, from Protasis Corp (Marlborough, Mass.) consisted of a sample loop valve, high-pressure pump, and microprocessor controller. It could either be triggered to deliver a specified volume and rate, or it could be controlled through an RS232 serial connection.
The microcoil NMR probe principally used in these studies was built in-house, as previously described (Kautz, 2001; Olson, 1999). Copper wire (7.5 turns of 50 μm wire) was wrapped on a glass flow cell with 660 μm i.d., 920 μm o.d. (see
Preliminary data were obtained using a commercial microcoil probe, the 1H capLC microflow probe (Olson, 2004) manufactured by Magnetic Resonance Microsensors (MRM, Savoy, Ill.) and distributed by its parent company, Protasis Corp. This probe had a 1.1-μL observed volume (Vobs) in a flow cell volume of ˜3.5 μL, with 50-μm fused-silica inlet and outlet capillaries. When this probe was used, all connections in the sample loop and transfer lines were made using Upchurch PEEK unions.
The sample handler and loader were connected to the NMR probe as shown in
PFOS Silica. Probes were internally coated with tridecafluoro-1,1,2,2-tetrahydrooctyl-1-trichlorosilane (perfluorooctylsilane, PFOS) monolayer. Surfaces were first activated by washing with 1 N NaOH for 1 h, followed by sequential rinses with water, acetone, and chloroform. A fresh 5% (v/v) solution of PFOS in chloroform was flushed slowly through the probe for 1 h, then the probe was rinsed sequentially with chloroform, acetone, and 10% D2O in acetone. Surfaces were exhaustively dried with an acetone rinse followed by air flow overnight, then the probe was stored filled with FC 43.
Fused-silica capillaries were coated with a thicker PFOS gel layer using the trimethoxy form of PFOS according to the manufacturer's protocol (United Chemical Technologies). In brief, fused-silica capillaries were washed with peroxide/sulfuric acid, then activated with 1 N NaOH overnight. A solution was first prepared of 95% methanol and 5% water, with acetic acid added to an apparent pH meter reading of 5. Subsequently, 4% (v/v) tridecafluoro-1,1,2,2-tetrahydrooctyl-1-trimethoxysilane was added with vigorous stirring and allowed to react for ½ h before introduction to the capillaries for 16 h at room temperature. Capillaries were annealed for 2 h at 80° C., flushed with the methanol/water/acetate solution, blown dry with helium, then cured and dried at 110° C. with helium flow. The capillaries were stored filled with fluorocarbon FC 43 until use.
Automation. Automation was controlled using Varian VAST automation programming on the spectrometer host computer (Sparc Ultra 5, Solaris 8, vnmr 6.1C NMR software). NMR acquisition setup macros were written to (1) automatically detect and position an arriving sample and (2) set up a standard spectrum of a sample (16 scans, 1.05-s acquisition time, 16-Kb points). Four sample handler programs (Tcl scripts) were written to (1) form a train of four samples and hold it in the needle line, (2) draw a train from the needle line into the sample loop, (3) change samples by triggering the sample loader to run until stopped by the autodetection macro, and (4) initialize the sample queue by moving a sample train one-half of the distance from the sample loop to the NMR probe. The sample loader was controlled by means of Unix shell scripts, which could be called from either vnmr macros or tcl scripts.
Sample Preparation. The test library of 12 known pharmaceutical entities was prepared as 1-mL aliquots at 30 mM in DMSO-d6 and stored at 4° C. For carryover measurements, standard samples were 4% chloroform-h in DMSO-d6 (standard S1) and 2% acetone-h6, 5% green food color in DMSO-d6 (standard S2). Ninety-six-well plates were typically alternating columns of S1 and S2, with the third and sixth columns replaced by the 12-member model library. For automated NMR runs, all wells contained 3 μL unless otherwise indicated, and plates were covered with adhesive film. Sample plates were placed on 4-mm foam rubber pads on the sample handler racks, carefully positioned under the needle, and taped into position. Two additional vials on the sample handler rack supplied 0.5 mL of FC 43 and 0.1 mL of the wash/autodetection solution, 1% TMSP (60 mM) in DMSO-d6.
NMR Spectroscopy. Sample spectra shown were 16 transients of 16-Kb points, 8000-Hz width, 45° tip angle, auto gain, with no additional relaxation beyond the 1.05-s acquisition time. The spectra were processed by zero-filling to 64-Kb points and Fourier transformed with 1 Hz of exponential line broadening. (Line widths are reported with no line broadening.) Autodetection used single scan spectra with a 60° tip angle, fixed gain, and other parameters as above; the region from −0.5 to +0.5 ppm was monitored for a peak with S/N>10 to detect the TMSP in the wash plugs. The COSY spectrum was a magnitude COSY, 128 increments of 16 transients processed with linear prediction in t1 to 512 points, apodized with sinebell-squared matched to acquisition time in both time domains. Total acquisition time was 10 min.
Preliminary studies were made, observing the movement of DMSO sample plugs using FC 43 as the carrier fluid in capillaries of several different materials. In the Teflon capillary, FC 43 was the continuous phase: DMSO plugs did not contact the capillary wall and could be moved through several meters with no detectable carryover (<0.1%) or losses at all flow rates tested (0-20 μL/min). In plain fused-silica capillary, DMSO was the continuous phase, retention of a DMSO film depleted sample plugs by 2 μL for each meter of movement (in 200-μm capillary). In PFOS-coated silica capillary, neither phase was continuous: a DMSO film was not retained, so sample losses were negligible at modest flow rates (1-10 L/min); however, carryover of minute droplets could occur (10-100 nL/m) if imperfections existed in the coating. It was also found necessary to push the sample train through the NMR probe under positive pressure rather than to pull samples through the system with a syringe or peristaltic pump at the detector outlet, as in a traditional SFA system (Patton, 1997). The flow rate of FC 43 through the microcoil probe with vacuum applied to the outlet capillary was <1 μL/min; changing samples in 30 s would require a flow rate on the order of 10 μL/min, which could be obtained with modest pressures of 150 psi. We therefore pursued a strategy of pulling sample and FC 43 plugs into a sample loop, then pushing the sample train through the transfer line and microcoil NMR probe by positive displacement.
To facilitate adoption of the method, it was implemented by modifying a conventional microVAST installation. The sample handler and loader were connected with the microcoil NMR probe using Teflon capillary, as shown in
The performance of the system was evaluated by loading samples and acquiring spectra from 96-well plates with 3 μL/well of test library compounds (30 mM in DMSO-d6) interspersed with standards for assessing carryover and line shape. Automated analysis completed in 2.5 hr/plate, plus 24 min to initialize the queue in which four trains were drawn and two were injected. This initialization time is reported separately because it applied to the first plate but not to subsequent plates of continuous high-throughput operation. Spectra were output at rates of 1/min along each train of four samples. The sample change and wash was completed in 35 s, NMR acquisition was set to 16 s, and the automation software required 10 s of execution and dead time. Sustainable throughput was 1.5 min/sample due to the time required to draw a new train from the needle line into the sample loop and to advance the queue through the gap between trains (105s).
Spectra of the 12 test library compounds are presented in
Compound identification may at times require advanced 2D spectra, or it may be desirable to analyze samples in protonated solvents requiring solvent suppression. Consequently, care was taken in programming the method so that existing macros for setting up established methods could be inserted, such as for gradient shimming, scout-scan solvent suppression, or 2D spectra. For example, a macro to acquire a magnitude COSY spectrum was added as a single line to one automation queue as described in Kautz, 2001. This flexibility was made a priority in development in order to enable 2D spectra to be acquired in a data-dependent manner, that is, to acquire a COSY or TOCSY spectrum if automated analysis of a 1D spectrum fails to confirm an expected product.
Importantly, the analyte in immiscible sample plugs was not found to disperse with time, so extended stopped flow acquisitions were possible without loss of signal strength due to dilution of the sample in the NMR coil. For example, in a 72-h acquisition of a trace sample (data not shown), the first 8-h block of data acquisition was identical to the last 8-h block. This stability also made it possible to interrupt long high-throughput analyses. One automation run of a 96-well plate was suspended in software, and the microcoil probe was removed from the magnet without disconnecting the transfer line. After using the spectrometer with a different probe for several hours, the microcoil probe was reinstalled, the automation queue was restarted, and analysis of the plate completed without any problems. Sample plugs in Teflon tubing have been stored for over 1 yr refrigerated with out degradation of the plug nor the analyte.
Sample carryover was below 1%, determined by comparing integrals of solvent peaks between alternating samples of 2% acetone and 4% chloroform, with one wash plug between them. A dye test showed that carryover to the wash plugs in the inlet capillary was in the range of 20-30 nL (<1%), suggesting most of the 5% sample-to-wash plug carryover observed at the NMR coil occurred in the residual 75-μm fused-silica segment of the inlet. The line widths obtained when automatically positioning sample plugs, evaluated from single-scan spectra, were between 1.0 and 1.6 Hz, close to the routine line width obtained with this probe (1.2 Hz). Importantly, the cost of deuterated solvent consumption was negligible: 0.4 mL of DMSO-d6/plate at $2/mL, including 100 μL of DMSO-d6 supplied as wash solvent in addition to the 96 3-μL samples.
The efficiency of sample utilization of the SFA-NMR method of this paper was comparable to that attained using capillary isotachophoresis (cITP) (Kautz, 2001; Wolters, 2002b), the most sensitive applied NMR method to date. Moreover, SFA-NMR can relatively quickly load analytes with unknown or zero electrophoretic mobility, which is important for de novo analysis of trace amounts of isolated natural products or drug metabolites. However, most separation and concentration methods cannot be put into practice in situ in the remote confined location of the magnet bore. SFA-NMR permits microseparations and microconcentration to be performed using optimal bench top techniques and instrumentation as long as a 1-μL fraction can be collected for subsequent transfer to the NMR microcoil.
SFA-NMR, as demonstrated above, doubled the throughput, quadrupled the sample efficiency, and reduced deuterated solvent consumption over 20-fold as compared to the commercially supported high-throughput flow NMR methods. Nonetheless, a number of straightforward improvements may still be envisioned. For example, lengthening the transfer line to hold 3 trains without gaps would increase throughput by eliminating the longer sample change time between trains. A larger i.d. sample loop could double the rate of loading the sample loop and could hold more samples in longer trains. Using a 10-port sample loop valve to switch between two loops would eliminate the delay to draw new trains into the sample loop. These and other improvements would increase the throughput of SFA-NMR to over one 96-well plate/h. Using segmented flow to alternately load multiple flow cells (Macnaughtan, 2003; Wolters, 2002c) could additionally increase throughput, reaching essentially continuous NMR data acquisition.
Because sample efficiency is high and no sample is wasted, trace samples may be analyzed. In fact, use of sample may be 100% efficient, as opposed to 10-30% (in commercial Protasis/MRM microinjection) or 0.1% (filling a 200 μm microcoil probe). There is no degradation of sensitivity or resolution if sample plugs as small as 1 μL are picked up by the autosampler and transferred into an NMR probe with a 3.5 μL flowcell/1 μL observe volume.
Rapid sample changes are possible along a queue of sample plugs. Conventionally, a sample loaded in the sample loop must be delivered the entire distance to the NMR coil. With the immiscible plug method according to the invention, small sample plugs may be closely spaced, separated by plugs of the immiscible solvent (segmented flow injection). For example, if 1 μL sample plugs are separated by 0.5 μL of immiscible liquids, samples may be changed by moving the queue only 1.5 μL.
Rapid washing of an NMR detection cell is possible. In conventional DI-NMR and FIA-NMR methods, the flow cell must be flushed with several volumes of clean solvent between samples to reduce sample carryover. With the method of the present invention, one sample plug may be followed closely with one or more small plugs of clean solvent to rinse any traces of sample from surfaces or dead volumes of the plumbing. This “train” of sample and rinse plugs may be less than 2 μL, and subsequent samples may follow immediately, in a flow-through injection scheme.
No relaxation time is required after sample injection before NMR analysis can be carried out. Because the sample plug is of uniform concentration, there are not strong concentration gradients within the sample as in the conventional methods. The linewidth of the NMR spectrum is sharp immediately upon arrival of the sample in the NMR coil.
Because samples are maintained in their original volume of 1-2 μL, sample recovery is greatly facilitated. The photograph shown in
The diffuse leading and trailing edges of sample plugs in the conventional methods make it difficult to determine the optimal positioning of the sample in the NMR cell. However, with the method of the invention (as shown in
Samples may be transferred in larger bore capillary tubing, reducing backpressure and consequent need for specialized pumps and related plumbing equipment. Additionally, samples may be transferred over longer distances without loss. High-end NMR spectrometers have larger magnets with larger fringe fields, which may require the sample handler to be as far away as 10 meters. With the present invention there is no disadvantage in sample efficiency or throughput with longer transfer line lengths. Larger capillaries permit fast transfers even over such distances, where 50 micron capillaries would be prohibitive.
While basic “One-Dimensional” NMR spectra are generally acquired from concentrated samples in a few minutes or less, more information or more dilute samples can require NMR acquisition times of several days or more. Using conventional methods, miscible sample plugs can diffuse out of the detection volume over such long periods of time. However, using the method of the invention, the immiscible sample plugs are stable indefinitely. Plugs stored in Teflon™ tubing have remained intact and undiluted for over a year.
Efficient sample transfer by the immiscible plug method according to the invention depends on the relative contact energies of the sample solvent and immiscible carrier solvent with the channel wall. These contact energies can be modified by chemically modifying the channel wall.
The use of smaller microcoil probes is enabled using the method of the invention. The current (and only) commercial microcoil NMR probe has an NMR coil detection volume (“active volume,” “observe volume,” Vobs) of 1 μl, designed based on the typical size of a capillary LC peak, or of the smallest sample that can be injected using a commercial autosampler, considering the limitations of dilution during transfer. With the present invention, samples of arbitrarily small volume may, be efficiently transferred into smaller microcoil probes. In particular, the most sensitive microcoil NMR probes produced to date are wrapped directly on 200/360 um (i.d/o.d.) capillaries and have an observe volume of 30 nL, but sample transfer into the coils is difficult (Kautz, 2001). Using the method of the invention, samples of approximately 30 nL volumes can be efficiently transferred, making these smaller probes, which are three times more sensitive, feasible for routine samples or high-throughput use.
The above work was performed using fluorocarbon FC-43 as the immiscible carrier liquid. Many other immiscible solvent systems, including other fluorocarbon liquids, are available which may be advantageous for their viscosity, immiscibility with unusual analytes or sample solvents, or to match magnetic susceptibility to a particular sample.
Interfacing to capillary separation or concentration microdevices is easy to carry out using the method and system of the invention. A variety of means have been proposed for microanalysis of trace samples by performing separation and concentration ot sub-microliter volumes. Most of these systems are practical or viable in an openly accessible system on the benchtop, or within a specialized device. Most cannot be adapted to practice in the confined and inaccessible volume of an NMR magnet bore, and so cannot be used in situ for microcoil NMR as we demonstrated for capillary isotachophoresis. Nor has it previously been feasible to transfer the sub-microliter sample fractions produced by these methods into microcoil NMR probes. The immiscible solvent plug injection method of the invention makes these procedures practical.
Traditional and present methods of flow NMR draw samples from microtiter plates such as 96-well plates with 200 μL wells. Arrays of smaller wells such as 384 well plates or 1536-well plates are also in use. A problem in automated sample handling is positioning a needle into the fluid sample volume and withdrawing a small sample completely without drawing any air. With small samples, a significant fraction of small samples must be left in the well, where it is wasted. By the present method, rather than capturing and storing samples in the wells of microtiter plates, samples could be collected at the source of concentration or separation as immiscible plugs in a length of inexpensive teflon tubing filled with the immiscible carrier. The Teflon™ tubing may be easily stored and/or transported to a different laboratory for microcoil NMR analysis or other microfluidic analytical methods.
The present invention also enables a more efficient method of handling small samples in conventional microtiter plates. To draw the entire prepared sample into an autosampler needle without drawing air, an immiscible fluid which is lighter (lower density) than the sample solvent may be added to the sample well together with the prepared sample. This lighter immiscible will float on top of the prepared sample. When the sample is drawn into the needle of the sample handling robot, any excess volume drawn will be the immiscible overlay rather than air, and the sample may be efficiently transferred into the microcoil NMR or other microfluidic device.
While the present invention has been described in conjunction with a preferred embodiment, one of ordinary skill, after reading the foregoing specification, will be able to effect various changes, substitutions of equivalents, and other alterations to the compositions and methods set forth herein. It is therefore intended that the protection granted by Letters Patent hereon be limited only by the definitions contained in the appended claims and equivalents thereof.
This application claims the priority of U.S. Provisional Application No. 60/528,410 filed Dec. 10, 2003 entitled, METHOD FOR EFFICIENT TRANSPORT OF SMALL LIQUID VOLUMES IN MICROFLUIDIC DEVICES, the whole of which is hereby incorporated by reference herein.
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/US04/41923 | 12/10/2004 | WO | 6/9/2006 |
| Number | Date | Country | |
|---|---|---|---|
| 60528410 | Dec 2003 | US |
