Anal. Chem. 2010, 82, 4194–4200
Analysis of Boronic Acids by Nano Liquid Chromatography-Direct Electron Ionization Mass Spectrometry Cornelia Flender,*,† Peter Leonhard,† Christian Wolf,† Matthias Fritzsche,† and Michael Karas‡ Merck KGaA, Frankfurter Strasse 250, 64293 Darmstadt, Germany, and Institute of Pharmaceutical Chemistry, University of Frankfurt, Max-von-Laue Strasse 9, 60438 Frankfurt, Germany A new method, based on a direct-electron ionization (EI) interface, is presented for the analysis of compounds insufficiently amenable to usual MS methods. The instrumentation is composed of a nano liquid chromatograph (LC) and a mass spectrometer (MS) directly coupled by a transfer capillary. The eluent is directly introduced into the heated electron impact ion source of the MS. Significant advantages are the generation of reproducible spectra and the ability to ionize highly polar compounds. Boronic acids are used as coupling reagents to produce drugs, agrochemicals, or herbicides. The purity of educts is of high importance because impurities in the educt are directly associated with impurities in the product. Because of their high polarity and tendency to form boroxines, boronic acids require derivatization for GC analysis. The presented nano-LC-EI/MS method is easily applicable for a broad range of boronic acids. The method shows good detection limits for boronic acids up to 200 pg, is perfectly linear, and shows a very high robustness and reproducibility. A mixture of compounds could easily be separated on a monolithic RP18e column. The method represents a new, simple, robust, and reproducible approach for the detection of polar analytes. It is a good candidate to become a standard method for industrial applications. There is a long-standing desire for combining the advantages of liquid chromatography (LC) and electron ionization mass spectrometry (EI-MS). LC is a common technique, established for a broad range of applications. EI-MS is almost exclusively used as a gas chromatography (GC) coupled detector. EI comprises many advantages compared to other ionization techniques, e.g., its low technical complexity, the generation of reproducible spectra independent of the environment and, therefore, the ability to buildup mass spectral databases. There have been many efforts to efficiently couple a liquid chromatograph to a mass spectrometer in order to obtain high quality mass spectra. In the early 70’s, Tal’roze et al. pioneered in * To whom correspondence should be addressed. E-mail: Cornelia.Flender@ merck.de. Fax: +49-6151-726286. † Merck KGaA. ‡ University of Frankfurt.
4194
Analytical Chemistry, Vol. 82, No. 10, May 15, 2010
coupling a liquid sample injector directly to an MS.1 Their work triggered a variety of LC/MS interface developments.2-5 The first commercially available interface was developed by McFadden and Schwartz.6 A rotating moving belt carries the eluent through several vacuum lock chambers into the ion source of the mass spectrometer. The system had some fundamental disadvantages. It suffered from a lack of reproducibility and sensitivity and was mechanically complex. On the basis of the work of Willoughby and Browner,7 the particle beam interface8 was developed. The effluent is nebulized in a desolvation chamber. After vaporization of the solvent, the mixture of solvent and particles is drawn through several pumped chambers to finally reach the mass spectrometer. With a mechanically simple setup, classical EI spectra could be obtained. Reproducibility issues that existed in earlier developments could be overcome,9 but sensitivity issues still remained.10 Another approach has been presented by Amirav and Granot.11 The eluent is vaporized inside a channel supersonic nozzle and then supercooled in a supersonic expansion. Thus, further dissociation of the vaporized compounds is avoided. The socalled “cold-EI” spectra obtained by this method are slightly different from standard EI spectra. They show an enhanced molecular ion and less fragments in the lower molecular mass regions.12 In the past few years, columns and capillaries with an internal diameter of 50 µm and less have become available. This offered the possibility to work with flow rates as low as 50 nL/min and to establish a direct coupling between LC and EI-MS. Cappiello et (1) Tal’roze, V. L.; Skurat, V. E.; Gorodetskii, I. G.; Zolotoi, N. B. Russ. J. Phys. Chem. 1972, 46, 456–458. (2) Scott, R. P. W.; Scott, C. G.; Munroe, M.; Hess, J. J. Chromatogr. 1974, 99, 395–405. (3) Arpino, P. J.; Baldwin, M. A.; McLafferty, F. W. Biomed. Mass Spectrom. 1974, 1, 80–82. (4) Lovins, R. E.; Ellis, S. R.; Tolbert, G. D.; McKinney, C. R. Anal. Chem. 1973, 45, 1553–1556. (5) Niessen, W. M. A. Chromatographia 1986, 21, 277–287. (6) McFadden, W. H.; Schwartz, H. L.; Evans, S. J. Chromatogr. 1976, 122, 389–396. (7) Willoughby, R. C.; Browner, R. F. Anal. Chem. 1984, 56, 2626–2631. (8) Creaser, C. S.; Stygall, J. W. Analyst 1993, 118, 1467–1418. (9) Bonfanti, L.; Careri, M.; Mangia, A.; Manini, P.; Maspero, M. J. Chromatogr., A 1996, 728, 359–369. (10) Magi, E.; Ianni, C. Anal. Chim. Acta 1998, 359, 237–244. (11) Amirav, A.; Granot, O. J. Am. Soc. Mass Spectrom. 2000, 11, 587–591. (12) Granot, O.; Amirav, A. Int. J. Mass Spectrom. 2005, 244, 15–28. 10.1021/ac1004585 2010 American Chemical Society Published on Web 04/22/2010
Figure 1. Direct-EI ion source (by courtesy of Achille Cappiello, University of Urbino).
al. developed a direct liquid introduction system13,14 based on 100-400 nL/min flow rates, which allows one to transfer the whole LC eluent into the ion source of the mass spectrometer. Figure 1 shows the instrumental setup of the interface. A transfer capillary with a diameter of 10-25 µm, on the front end coupled to the LC column, is directly attached to the mass spectrometer. The end of the capillary extends 2 mm ±10% into the ion source. This has been experimentally determined as the optimum position for ionization.15 As the eluent proceeds into the heated ion source, solvents and solutes evaporate, are being ionized in the gas phase, and proceed to the mass analyzer. This system has been evaluated and applied to a variety of substances15-20 and showed good results in terms of limit of detection (LOD), linearity, and reproducibility. A major advantage of the presented LC-MS interface has been investigated by Cappiello et al. Since the ionization takes place in the gas phase, experiments are not subject to matrix effects,21 like commonly seen in electrospray ionization (ESI). Matrix effects are caused by coeluting substances or matrix interferences, e.g., in environmental or human plasma samples, and cause signal suppression or enhancement of the target substance peak. Various factors like mobile phase composition, background noise, or flow instabilities can also influence the ionization of the analyte.22 If substances are ionized in the gaseous phase under vacuum conditions, like (13) Cappiello, A.; Famiglini, G.; Mangani, F.; Palma, P. J. Am. Soc. Mass Spectrom. 2002, 13, 265–273. (14) Cappiello, A.; Famiglini, G.; Palma, P. Anal. Chem. 2003, 75, 496–503. (15) Cappiello, A.; Famiglini, G.; Palma, P.; Mangani, F. Anal. Chem. 2002, 74, 3547–3554. (16) Cappiello, A.; Famiglini, G.; Mangani, F.; Palma, P.; Siviero, A. Anal. Chim. Acta 2003, 493, 125–136. (17) Famiglini, G.; Palma, P.; Siviero, A.; Rezai, M. A.; Cappiello, A. Anal. Chem. 2005, 77, 7654–7661. (18) Cappiello, A.; Famiglini, G.; Palma, P.; Pierini, E.; Trufelli, H.; Maggi, C.; Manfra, L.; Mannozzi, M. Chemosphere 2007, 69, 554–560. (19) Cappiello, A.; Famiglini, G.; Pierini, E.; Palma, P.; Trufelli, H. Anal. Chem. 2007, 79, 5364–5372. (20) Famiglini, G.; Palma, P.; Pierini, E.; Trufelli, H.; Cappiello, A. Anal. Chem. 2008, 80, 3445–3449. (21) Cappiello, A.; Famiglini, G.; Palma, P.; Pierini, E.; Termopoli, V.; Trufelli, H. Anal. Chem. 2008, 80, 9343–9348. (22) Cappiello, A. Advances in LC-MS Instrumentation, 1 ed.; Elsevier: Amsterdam, 2007.
in EI, the probability of molecule-molecule or molecule-ion reactions is close to zero, which means that usually no signal interferences occur.23 This method is fundamentally sound, nevertheless the analysis of boronic acids implies specific challenges. Boronic acids are an important educt for a variety of reactions in industrial production processes, among them the Suzuki reaction. This reaction describes a cross coupling between organoboron compounds and halides, catalyzed by a palladium complex.24 It is one of the most famous reactions in organic synthesis, used for synthesis of drugs, agrochemicals, polymers, and herbicides. The Suzuki coupling has many advantages compared to other techniques. It operates under mild conditions, even at room temperature. The educts for Suzuki reaction are stable in water and air and are nontoxic. The method implies a high chemoselectivity, other functional groups remain largely unaffected.25 In the Suzuki reaction, the purity of products becomes more and more important and so do the analytical methods to identify the impurities that are being produced as byproducts during synthesis steps or that are even present in educts. There is a wide range of methods available. Every method has its advantages and disadvantages. The methods most commonly used for impurity profiling of boronic acids suffer from the ability to efficiently vaporize the components, low ionization efficiency, or production of qualifiable results. GC, as a well established method in industry, shows limit of detections (LODs) in the low femtogram regions for many volatile substances. However, polar substances or substances with high molecular weight cause problems in the GC injector and on the GC column. Boron acid detection by GC requires derivatization, e.g., with pinacole,26-28 because aromatic boron acids especially tend to form boroxines in the GC injector. Derivatization is an additional time-consuming preparation step. Derivatization reagents alter the structure of the educt and can cause unwanted side reactions.29 Furthermore, derivatization complicates quantification because it is difficult to determine which percentage of substance has been converted. LC/MS offers different atmospheric pressure ionization (API) methods, which are commonly accepted for the analysis of mostly high molecular weight compounds. However, some small molecular weight compounds are difficult to ionize. Using these methods, the structural information provided is often restricted to the molecular ion and/or adducts and is not sufficient to clearly identify compounds.30,31 The objective of the presented research is to introduce a new method for the analysis of small polar compounds, (23) Speranza, M. Int. J. Mass Spectrom. Ion Processes 1992, 118/119, 395– 447. (24) Miyaura, N.; Yamada, K.; Suzuki, A. Tetrahedron Lett. 1979, 20, 3437– 3440. (25) Suzuki, A. J. Organomet. Chem. 1999, 576, 147–168. (26) Longstaff, C.; Rose, M. E. Org. Mass Spectrom. 1982, 17, 508–518. (27) Singhawangcha, S.; Chen Hu, L.-E.; Poole, C. F.; Zlatkis, A. J. High Resolution Chromatogr. Chromatogr. Commun. 1978, 1, 304–306. (28) Rose, M. E.; Longstaff, C.; Dean, P. D. J. Chromatogr. 1982, 249, 174– 179. (29) Birkemeyer, C.; Kolasa, A.; Kopka, J. J. Chromatogr., A 2003, 993, 89– 102. (30) Williams, C. M.; Stein, B. K.; Brenton, A. G.; Mosely, J. A.; Hurst, G.; Lubben, A. T.; Bristow, A. W. T. Analysis of Boronic Acids Without Chemical Derivatisation, 18th IMSC conference, Bremen, Germany, September 2nd, 2009. (31) Llewellyn, G.; Stein, B. K. Boronic Acid Analysis by Mass SpectrometrysI: Cis-diol Derivatisation for EI and CI Analysis: Application Note No 6, EPSRC National Mass Spectrometry Centre, Swansea, 2008.
Analytical Chemistry, Vol. 82, No. 10, May 15, 2010
4195
Table 1. SIM Ions for LOD Determination compound
molecular mass
SIM ions
4-propyl-phenylboronic acid 4-ethyl-phenylboronic acid 2-thienylboronic acid phenetylboronic acid 2-(2′-methoxybenzyloxy)-phenylboronic acid cis-propenylboronic acid methylboronic acid
164 150 128 150 258 86 60
164, 135, 117 150, 135, 117, 106, 91 128, 110, 84 91 121, 91 86, 68 60
Table 2. Concentration of Mixture Compounds compound
concentration, µg/mL
dilution 1, µg/mL
dilution 2, µg/mL
dilution 3, µg/mL
cis-propenylboronic acid 2-thienylboronic acid 4-propyl-phenylboronic acid 1-bromo-4-propyl-benzene 1-bromo-4-ethylbenzene 1,4-dibromo-benzene 1-allyl-4-bromo-benzene 1-(4-bromo-phenyl)-propan-1-one
2100 2040 550 270 270 330 330 280
1400 1360 367 180 180 220 220 187
700 680 184 90 90 110 110 94
350 340 92 45 45 55 55 47
using the example of boronic acids as an educt in the Suzuki reaction. The principal idea of this method is to improve the efficiency of industrial analysis processes relative to existing methods. EXPERIMENTAL SECTION Material. Acetonitrile (LC grade), toluole (Suprasolv), and ethane-1,2-diol (reagent grade) were purchased at Merck (Darmstadt, Germany). Water was purified by a Milli-Q water purification system from Millipore (Billerica, USA). Methylboronic acid (97%), phenetylboronic acid (97%), 2-(2′-methoxybenzyloxy)phenylboronic acid (97%), cis-propenylboronic acid (97%), and 2-thienylboronic acid (98%) were purchased at Merck. 4-Propyl-phenylboronic acid (97%), 1-bromo-4-propyl-benzene (99%), 1-bromo-4-ethylbenzene (97%), 1,4-dibromo-benzene (98%), 1-allyl-4-bromo-benzene, and 1-(4-bromo-phenyl)-propan-1-one were purchased at SigmaAldrich (Munich, Germany). Nano-LC. The experiments were carried out using an Agilent Nano-HPLC consisting of two nano pumps, a microwell plate sampler, and degasser of the 1100 and 1200 series. In order to be able to inject small sample sizes, the well plate sampler was connected to an external 6-port-valve from Rheodyne (Oak Harbor, USA). An external injection loop providing a capacity of 113 nL was attached. Chromatographic separations were carried out using a monolithic RP18e column research sample, 50 µm ID × 600 mm, from Merck. The loading pump was run at 1 µL/min to carry the sample from the autosampler into the loop of the external valve. The analytical pump was run at 200 nL/min, to take over the sample flow from the injection valve to the analytical column to finally reach the detector. The flush time was set at 1.5 min to transfer the sample from the autosampler to the injection loop. Direct-EI. The external valve was connected to an Agilent MSD 5975B by a 15 µm ID transfer capillary. The end of the transfer capillary was directly introduced into the ion source (see Figure 1), with no need of further interface adjustment. Inside the heated ion source, the eluent vaporizes explosively and is being ionized by an electron beam. A flow rate of 200 nL/min leads to a vacuum pressure of approximately 1 × 10-5 Torr inside 4196
Analytical Chemistry, Vol. 82, No. 10, May 15, 2010
a b c d e f g h
the vacuum chamber. The MSD was equipped with a high temperature EI ion source. It was kept at 350 °C during analysis. The quadrupole temperature was kept at 150 °C. The ionization mode was EI+ with an electron energy of 70 eV. In scan mode masses from 59 to 500 amu were acquired with a threshold of 100 counts and a scan rate of 1.69 Hz. In SIM mode the dwell time was set as 100 ms.32 GC/MS. Comparative measurements were carried out on a Waters GCT Premier. The separation was performed on a Varian VF-5 ms column (30 m × 0.25 mm ID DF ) 0.25 µm). GC parameters included a temperature program from 100 °C (1 min) to 320 °C (33 min) with a heating rate of 8 °C/min. Helium was used as a carrier gas. The column flow was 1 mL/min with an injector temperature of 280 °C. The ionization mode was EI+ with an electron energy of 70 eV. The acquisition ranged from 25 to 800 amu. The scan rate was 3 Hz. The ion source was heated to 250 °C, and the transfer line was heated to 280 °C. Method Validation. Important validation parameters like limit of detection (LOD), limit of quantification (LOQ), linearity, and precision were examined. Stock solutions of 1 mg/mL of each compound listed in Table 1 were prepared in 80% acetonitrile and 20% water. The stock solution was dissolved to obtain sample concentrations of 500, 100, 80, 50, 30, 10, 5, and 1 µg/mL. LODs were determined empirically in Scan and SIM mode under authentic chromatographic conditions with a monolithic RP18e column and an eluent composition of 80% acetonitrile and 20% water. Seven injections per concentration of 10 nL each were performed, which matches a switching of valve position of 0.05 min for each injection. LODs were calculated on the basis of the regression equation and the minimal detectable amount of sample with a signal-to-noise ratio (S/N) of 3. LOQs were calculated as the minimal detectable amount with a S/N of 10. LODs and LOQs were determined in Scan and in SIM mode. Table 1 lists the characteristic ions of each substance, which have been used for SIM detection. Linearity was calculated on the basis of peak areas. Precision was determined as RSD of peak areas. (32) Eveleigh, L. J.; Ducauze, C. J. J. Chromatogr., A 1997, 765, 241–245.
Table 3. Results of Method Validation compound 4-propyl-phenylboronic acid 4-ethyl-phenylboronic acid phenetylboronic acid 2-(2′-methoxybenzyloxy)phenylboronic acid cis-propenylboronic acid 2-thienylboronic acid methylboronic acid
LOD (pg) scan/SIM
LOQ (pg) scan/SIM
linear regression equation
R2
232/2.5 387/2.5 386/3.6 337/3.3
772/8.3 1289/8.4 1285/12.1 1124/10.9
y ) 10698.0x - 167579 y ) 7797.0x - 133169 y ) 7428.4x - 108795 y ) 8393.9x - 156893
0.9998 0.9999 0.9998 0.9986
636/5.0 586/10.6 4275/726
2120/16.5 1893/35.2 14249/2422
y ) 1271.2x - 26940 y ) 4019.7x - 100102 y ) 50.5x - 3449.6
0.9970 0.9998 1
Chromatographic Separation. A chromatographic separation was carried out to show that the method is suitable for separation and detection of possible educts and byproducts of the Suzuki reaction. A mixture of the compounds listed in Table 2 was prepared. The alphabetical numbering for each compound corresponds to Figure 3. The separation was carried out under isocratic conditions with 65% acetonitrile and 35% water; 10 nL of the mixture was injected. Three replicates were examined on different days within 1 week
to ensure the method has a good day-to-day reproducability. The experiment was repeated (a) with a monolithic RP18e column of the same diameter and length but from a different batch and (b) at three different dilution steps of 2/3, 1/3, and 1/6 of the initial concentration to show the robustness and linearity of the application. On the basis of the above parameters, an impurity profile of a process sample was recorded both with direct-EI and GC/MS and they were compared. The sample was taken from a production process, containing 4-propyl-phenylboronic acid and unknown
Figure 2. EI-mass spectrum of 4-propylphenylboronic acid.
Figure 3. TIC of boron and bromine compounds separated on monolithic column, alphabetical order corresponds to Table 2. Analytical Chemistry, Vol. 82, No. 10, May 15, 2010
4197
RESULTS AND DISCUSSION
Table 4. Regression Coefficients of Mixture Compounds compound
regression coefficient
cis-propenylboronic acid 2-thienylboronic acid 4-propyl-phenylboronic acid 1-(4-bromo-phenyl)-propan-1-one 1,4-dibromo-benzene 1-allyl-4-bromo-benzene 1-bromo-4-ethylbenzene 1-bromo-4-propylbenzene
0.9978 0.9909 0.9954 0.9895 0.9932 0.9979 0.9945 0.9998
a b c h f g e d
amounts of impurities of unknown structure. Since impurities are usually expected in a range of 0.1-4%, a solution containing 20 mg/mL in 80% acetonitrile and 20% water was prepared for the LC/direct-EI run. Ten nanoliters of the solution was injected by the external injector. For the GC/MS run, a sample of 7 mg/mL was prepared in toluene. Ethane-1,2-diol was added as a derivatization reagent. The mixture was incubated at 80 °C for 30 min. One microliter was injected with a split ratio of 1:25. The absolute injected amounts were approximately the same for both methods.
The direct-EI interface has been developed for the analysis of small molecules in a wide range of polarities, including many that are not detectable by conventional LC-MS methods. Boronic acids are too polar for being amenable for direct GC/MS detection without derivatization. Common LC/MS techniques require specific properties for ionization like protonable atoms within the organic molecule. APCI (atmospheric pressure chemical ionization), ESI, and other “soft-ionization” techniques primarily produce molecule ions and adducts. EI spectra are far more informative than spectra obtained with other ionization techniques, and moreover, EI spectra are reproducible and independent of method parameters and matrix influences. The applicability of the method for analysis of boronic acids was tested determining LOD, LOQ, linearity, and precision of several boronic acid samples of different nature. LOD and LOQ were determined on the basis of a S/N ratio of 3 and 10, respectively. For most boronic acids, satisfactory LODs were achieved (compare Table 3). The results imply that direct-EI is the method most suitable for detection of phenylboronic acids. Methylboronic acid is the smallest molecule with the lowest
Figure 4. Impurity profile of 4-propyl-phenylboronic acid recorded with direct-EI, top picture; TIC, pictures 1a-5a; LC/MS spectrum of each peak. 4198
Analytical Chemistry, Vol. 82, No. 10, May 15, 2010
Figure 5. TIC of 4-propyl-phenylboronic acid and impurities recorded with GC/MS, pictures 1b-7b; GC/MS spectrum of each peak. Analytical Chemistry, Vol. 82, No. 10, May 15, 2010
4199
molecular weight and shows the worst LOD and LOQ in both SIM and scan modes. The ability to detect molecules with directEI depends on three major parameters. On one hand, there is the need for efficient desolvatization and vaporization, which depends on the polarity of molecules and their affinity to the solvent molecules. On the other hand, there is the ionization efficiency (electron ionization cross-section) which depends on the size of the molecule, the binding energy inside the molecule, the kinetic energy, and the differential dipole oscillator strength for the subshell.33 As a third point, the molecular weight is playing an important role. If the molecular weight of the substance is too low and in a similar area as the molecular weight of the solvents, it is possible that large parts of the fragment ion masses are part of the solvent cutoff area. Methylboronic acid with its low molecular mass and its apparently low electron ionization cross section is, therefore, not suitable for detection with direct-EI. Linearity was determined starting at the calculated LOD in Scan mode for each substance and up to 10 ng. The areas were evaluated and show values above R2 ) 0.997. The precision was calculated as RSD of seven replicates per substance and concentration. The value never exceeds 12%. The mass spectra obtained show a good spectral quality, as shown in Figure 2. Due to the fact that all solvent mass peaks appear in the lower mass region, no masses below 59 amu were scanned. Formation of boroxines was not observed in any of the measurements. Preliminary tests that have been carried out lead to the assumption that formation of boroxines depends on the amount of water in the mobile phase. With an amount of at least 10% water, boroxine formation is inhibited. Some possible educts and byproducts of the Suzuki reaction are listed in Table 2. The compounds were separated on a monolithic RP18e microcolumn. Figure 3 shows the obtained chromatogram. Three replicates have been carried out on different days within 1 week. The results show relative standard deviations of 2-6% for the peak areas (EIC for peaks a-c, TICs for peak d-h) of each substance, which is a good day-to-day reproducibility. These results have been verified with another column of the same kind and length but from a different batch. The relative difference in peak area comparing both columns never exceeds 7%. This comparison illustrates that the established method offers a high robustness and reproducibility. The dilutions of the sample mix showed a linear correlation for all investigated substances. Table 4 shows the linear regression coefficients, calculated on the basis of peak areas (again EIC for peaks a-c, TIC for peaks d-h) of four dilutions (see Table 2). A satisfactory linearity was obtained for all investigated components. The developed method was applied to a process sample which was taken from an industrial production process and compared to a GC/MS method. Figures 4 and 5 show the chromatograms of the sample runs with direct-EI and GC/MS, respectively. With (33) Kim, Y. K.; Rudd, M. E. Phys. Rev. A: At., Mol., Opt. Phys. 1994, 50, 3953– 3967.
4200
Analytical Chemistry, Vol. 82, No. 10, May 15, 2010
both methods, impurities could be detected and clearly identified. The content of the impurities show 0.2-2.5% of the total amount. Figure 4 shows the primary component 4-propyl-phenylboronic acid (peak 1a) and four impurities. Peak 2a shows the deboronated primary component with a molecular mass of 120 and the characteristic loss of an ethyl group which leads to a fragment mass of 91. Peak 3a represents butylated hydroxytoluene, which is often used as an antioxidant in production processes or as additive in plastics. Peak 4a is produced as a byproduct during synthesis. An also expected byproduct of the Grignard-based boronic acid synthesis is Peak 5a with two strong losses of ethyl. In GC/MS, partially similar impurities could be detected (see Figure 5). Substance 1b and 7b represent the same substances as peak 2a and 5a, respectively. Peak 3b is the glycol derivatized target compound. Peak 5b is an often seen and, therefore, reasonable impurity. In difference to the LC-EI experiment, in the GC/MS acquisition, a diboronic acid could be detected (peak 6b). Since the underlying separation mechanisms of both methods are based on different principles, the same substances do not have the same retention times or order. Compounds that cannot be detected with one of the methods but are detected with the other method are probably subject to signal overlapping by the main component. Since the main component is highly concentrated, it is possible that low concentration impurities with similar retention times do not show up in the chromatogram. CONCLUSION The method of direct coupling between LC and MS has proven to be efficient and useful for the detection and impurity profiling of boronic acids. It was proven that a wide range of boronic acids with different structures can be analyzed with a good sensitivity. Monolithic columns are perfectly suitable for nano-LC separations. They show reproducible, batch-to-batch consistent results and a high selectivity for the target compounds. The separation of a process sample demonstrated that LC-EI/MS is competitive to GC/MS. The obtained spectra are highly distinctive and easily interpretable due to reproducible fragmentation patterns. High reproducibility, precision, and linearity and the simple configuration of the instrument allows the method to be employed as a standard procedure in industrial manufacturing processes. ACKNOWLEDGMENT The authors thank Gerard Rozing and Uwe Effelsberg from Agilent Technologies for the instrumentation and many helpful discussions. A special thank you to Oliver Blessmann for providing GC/MS data. We thank Stephan Altmaier for column research samples and Mario Anton and Bernhard Meyer for supporting the project. Received for review February 19, 2010. Accepted April 12, 2010. AC1004585
