LETTER pubs.acs.org/ac
Surface Mass Spectrometry of Two Component DrugPolymer Systems: Novel Chromatographic Separation Method Using Gentle-Secondary Ion Mass Spectrometry (G-SIMS) Ryosuke Ogaki,†,§ Ian S. Gilmore,*,†,‡ Morgan R. Alexander,† Felicia M. Green,‡ Martyn C. Davies,† and Joanna L. S. Lee‡ †
Laboratory of Biophysics and Surface Analysis, School of Pharmacy, University of Nottingham, Nottingham, United Kingdom, NG7 2RD ‡ National Physical Laboratory, Teddington, Middlesex, United Kingdom, TW11 0LW
bS Supporting Information ABSTRACT: In recent years, there has been an increase in the use of time-of-flight secondary ion mass spectrometry (TOF-SIMS) for characterizing material surfaces. A great advantage of SIMS is that the analysis is direct and has excellent spatial resolution approaching a few hundred nanometers. However, the lack of the usual separation methods in mass spectrometry such as chromatography or ion mobility combined with the complexity of the heavily fragmented ions in the spectra means that the interpretation of multicomponent spectra in SIMS is very challenging indeed. The requirements for highdefinition imaging, with say 256 256 pixels, in around 10 min analysis time places significant constraints on the instrument design so that separation using methods such as ion mobility with flight times of milliseconds are incompatible. Clearly, traditional liquid and gas chromatographies are not at all possible. Previously, we developed a method known as Gentle-SIMS (G-SIMS) that simplifies SIMS spectra so that the dominant ions are simply related to the structure of the substances analyzed. The method uses a measurement of the fragmentation behavior under two different primary ion source conditions and a control parameter known as the g-index. Here, we show that this method may be used “chromatographically” to separate the mass spectra of a drug molecule from the matrix polymer. The method may be used in real-time and is directly compatible with the majority of TOF-SIMS instruments. The applicability to other imaging mass spectrometeries is discussed.
O
ver the last several decades, various materials including biodegradable polymers have become established in a range of medical applications, particularly in controlled release drug delivery systems.13 The rapid growth in the field of pharmaceutical biotechnology has led to an increase in demand for the development of more sophisticated delivery systems in order to accommodate and safely deliver various therapeutic agents in a controlled manner to the target site in vivo.4 A greater understanding of the physical and chemical surface properties of drug delivery systems aids in the design and development of such products with the desired therapeutic properties, for example, surface properties of therapeutic nanoparticles for cancer treatment.5 Time of flight secondary ion mass spectrometry (TOF-SIMS) is particularly suitable for the characterization of drug delivery systems due to its molecular specificity, good detection limits, surface sensitivity, and lateral resolution.6 In the last couple of decades, the technique has been employed extensively to characterize a variety of drug delivery systems such as multilayer polymer/drug beads,7 mannitol microspheres loaded with chitosan/tripolyphosphate nanoparticles,8 and drug eluting stent coating containing various loadings of sirolimus drug in poly (lactic-co-glycolic acid) (PLGA) matrix.9 r 2011 American Chemical Society
In SIMS, primary ions with typically tens of keV impact the surface depositing the energy over a few tens of nanometers depth causing an energetic collisional process, the details of which depend on the ion energy, number of atoms in the ion, and their atomic number. Secondary ions are liberated from the surface which, for an organic substance, are predominantly heavily fragmented or structurally rearranged. A small number are more directly related to the structure of the substance. Consequently, the mass spectra are hard to interpret. In the early development of SIMS, for organic materials, libraries of spectra were developed for fingerprint matching.10 These have now grown to cover approximately a thousand substances. A recent analysis of the PubChem substance database shows that there are over 70 million substances in the mass range relevant to SIMS.11 Consequently, experimental libraries will always be useful but contain a very limited subset of materials. A further complication has been the almost complete change from atomic to cluster primary ions in studies of complex molecules. The Received: February 9, 2011 Accepted: April 15, 2011 Published: April 15, 2011 3627
dx.doi.org/10.1021/ac200347a | Anal. Chem. 2011, 83, 3627–3631
Analytical Chemistry nonlinear enhancement of molecular fragments12 helps sensitivity and specificity, but the spectra are not closely comparable to the older library data,10 largely acquired with atomic ions. These requirements led to our development of Gentle-SIMS (G-SIMS),13 which is a library independent method that simplifies the SIMS spectrum so that the most structurally significant ions are clearly prominent and the degraded and rearranged fragments ions are suppressed. Details on G-SIMS and relevant references are given elsewhere.13,14 Very briefly, consider a single primary ion impact at a surface of molecules. Energy is deposited at the surface with a region of high energy density at the point of impact and a lower energy density further away. Secondary ions ejected from the high energy density region will be more degraded than those from the periphery of the zone of emission, where the energy density is lower. For a given energetic condition of the primary ion beam (particle atomic number and energy), there will be a characteristic population of intact and fragmented secondary ion fragments. At lower energetic conditions, the population of intact fragments is expected to increase and vice versa. This behavior was found experimentally, and a description was developed13 in terms of a population of fragments described by a partition function with a characteristic surface temperature, Tp. The G-SIMS theory13 shows that, for two spectra with peak intensities S1 and S2 and mass Mx acquired with different ion beam conditions and consequently different surface temperatures, T1 and T2 (where T1 < T2), it is possible to extrapolate to an equivalent spectrum at a much lower temperature. The G-SIMS spectrum, Gx, is simply computed as g S1x Gx ¼ Mx S1x ð1Þ S2x where g is an extrapolation index, known as the g-index. Typically, this has a value of 13 for atomic primary ions, but we have shown recently15 that the g-index may be less than 5 when cluster primary ions for spectrum S1 were used. The mass term, Mx, simply scales up the intensity at higher masses where the secondary ion yields in the SIMS process are weaker. G-SIMS has been successfully applied to a wide variety of pure substances including polymers13,16,17 and complex molecules.18 As discussed previously, the lack of any separation method in SIMS means that the mass spectra for multicomponent spectra are further complicated. Since one may expect that different substances will have rather different fragmentation behavior under ion beam irradiation, it could, in principle, be possible to use the g-index parameter, as a separation parameter. In effect, this would be a chromatogram with separation based on fragmentation energy. To evaluate this concept, we use two binary component drugpolymer model systems each consisting of an anesthetic drug, either codeine or bupivacaine (as a hydrochloride compound), which are dispersed in poly (lactic acid) (PLA). The drugs are chosen because of their commercial relevance, and they are known to undergo different fragmentation processes in electron ionization mass spectrometry.19 Poly (lactic acid) (PLA) is used for the matrix in which the drug is dispersed since it is a widely used biodegradable polymer, which is already well characterized by TOF-SIMS and G-SIMS.16 To check the sensitivity of the method, each system was analyzed at three drugpolymer concentrations of 1:10, 1:20, and 1:40 by weight. Details of the sample preparation method and characterization are provided in the Supporting Information. The surfaces of the
LETTER
spun cast blends were analyzed by X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM). Quantification of the amount of the three elements detected at the surface by XPS was achieved using the C1s, O1s, and N 1s core levels. This indicated that the amount of the drug at the surface is between 33 and 75% of the bulk composition (Supplementary Table 1, Supporting Information). This suggests a systematic desegregation of both codeine and bupivacaine away from the surface to provide a polymerdrug distribution vertically through the sample. Changes in these profiles can have a significant influence on the drug release profile. The slightly elevated surface carbon content and reduced oxygen level is consistent with adventitious hydrocarbon contamination. AFM shows that, as far as SIMS is concerned, the samples are relatively smooth and have no significant lateral structure to suggest phase separation (Figure S1, Supporting Information). Some pits were observed in the films, ascribed to voids formed during drying of the blends. The chloroform used as a solvent has a high volatility and may contribute toward the nonuniformity of the surfaces. SIMS spectra were acquired using an ION-TOF IV instrument (ION-TOF GmbH, Germany) with a dual source column equipped with Csþ and Arþ. This ensures that the ion beams are coincident at the sample surface and analyze the same area. This is, of course, important for samples with any heterogeneity. For both primary ion species, the ion beam was digitally rastered with a 128 by 128 data point array over the same area of 250 μm by 250 μm using a beam current of less than 1 pA with an energy of 11 keV. The order of analysis was Csþ (lower fragmentation condition) followed by Arþ (higher fragmentation condition) with a combined ion dose kept well-below 1 1016 ions m2 to avoid ion induced damage.20 An important experimental requirement for G-SIMS, and indeed any relative quantification, is good control of the instrument and high repeatability of the relative intensities. SIMS spectra pairs were acquired from four different central areas separated by 1 mm. A repeatability of the relative intensities for flat reference materials of typically 5% was achieved. The mass spectra were recorded with high mass resolution typically above 7000 at m/z 29 (SiHþ). The mass scale was calibrated using the procedure in ref 21 giving a typical error of 20 ppm, which is good for SIMS. High mass resolution and good mass scale calibration accuracy are important in assigning the correct species for each mass peak. An electron gun was used for charge compensation, and the electron dose was kept below 3 1018 electrons/m2 to avoid electron induced damage.22 The intensities S1x and S2x are formed from summing the Poisson corrected intensities from the Cs and Ar spectra, respectively, for each mass peak Mx. These peaks and associated integration widths were identified automatically from the Csþ spectra using a threshold intensity of 200 counts (peak area) or more for m/z 0 to 150 and 100 counts or more for m/z 150 to 350. All the identified chemical formula for secondary ions, discussed later, are within 20 ppm of the measured mass. In Figure 1, we show the positive ion SIMS spectra of the two drugpolymer blends using Cs primary ions, that is the S1x intensities, since these are less fragmented than the spectra using Ar primary ions and have higher intensities. The peaks in the spectra have been colored red and blue, and we shall return to this later. For now, there is no direct way, a priori, to distinguish which peaks are from the drug molecule and which are from the polymer. One would, laboriously, need to compare each peak with library spectra with some hope, of course, that the pure 3628
dx.doi.org/10.1021/ac200347a |Anal. Chem. 2011, 83, 3627–3631
Analytical Chemistry
LETTER
Figure 1. Positive ion SIMS spectra of (a) codeine in PLA (1:10) and (b) bupivacaine in PLA (1:10). The peaks are colored red and blue for a gmax value less than and greater than the separation parameter, gsep, discussed in the text. The values of gsep are 6 and 4 for codeine in PLA and bupivacaine in PLA, respectively.
Figure 2. Chromatographic separation of SIMS mass spectra of a two component drug polymer system at 1:10 concentration using G-SIMS with a g separation parameter varying from 1 to 40 for (a) codeine in PLA and (b) bupivacaine in PLA. The pixel width is set to 1 mass unit, unless there is more than one peak at nominal mass. In which case, a thinner pixel width is selected so that all ions are displayed. Horizontal lines are shown at g = 6 and g = 4 for codeine in PLA and bupivacaine in PLA, respectively. These are the gsep separation values used to separate drug and polymer peaks in Figure 1.
substance spectrum is contained in the library. Alternatively, if the substances are known, reference spectra of pure substances could be acquired, or the usual procedures for interpreting mass spectra may also be used by studying mass losses between peaks. Hopefully, this would lead to some indication of the relationships between secondary ions and may allow classification of drug and polymer peaks. In the case when the substances are unknowns, then unfortunately, a mass accuracy of 20 ppm is insufficient for the direct identification of the chemical composition of each secondary ion.11,23 Earlier, we described the g-index which controls the amount of fragmentation in the G-SIMS spectrum. At g = 1, the G-SIMS spectrum is simply the S2 spectrum (with a simple linear mass scaling) using Ar primary ions, and at g = 0, it is the less degraded S1 spectrum using Cs primary ions. At higher g values, the G-SIMS spectrum consists of a higher proportion of fragments that only require a little energy to be generated, such as molecular ions, and less of the ions that require more energy such as polymers which require at least two bonds to be broken to eject the fragment. At g = 40, there are no structurally degraded or rearranged ions. In Figure 2, for each drugpolymer system, we plot an intensity map using a thermal color scale of the normalized G-SIMS intensities (to the maximum intensity) at each integer
value of g from 1 (high fragmentation) to 40 (very low fragmentation) for each mass, Mx. The intensity map, with the ordinate inverted, is similar to a chromatogram used in GC/MS where the separation parameter (ordinate scale) is retention time or ion mobility mass spectroscopy where the separation parameter is the ion mobility. We call this the g-ogram, and the separation parameter is the g-index. The g-ogram provides a convenient way to visualize a lot of information. For each mass peak, Mx, there are vertical streaks of intensity. Those that are bright to begin with and fade away as g increases (down the image) are from processes that involve more energy and fragmentation. Conversely, those peaks that are dark and become bright at higher g values are from processes requiring less energy and fragmentation. Visually, we see from both the g-ograms that there are two classes of mass peak that may be separated by a horizontal separation line at a value of g called gsep. We find to separate codeine in PLA and bupivacaine in PLA, that gsep = 6 and gsep = 4, respectively. For each peak, Mx, there is a g value where its intensity is a maximum, known as gmax. It is trivial to identify which peaks have gmax < gsep and which peaks have gmax g gsep, and we color the peaks in Figure 1 red and blue, respectively. As can be seen from Figure 2, the precise value of gsep is not too critical and values (1 from those selected would make little difference to the separation shown in Figure 1. 3629
dx.doi.org/10.1021/ac200347a |Anal. Chem. 2011, 83, 3627–3631
Analytical Chemistry Table 1. List of Main Characteristic Secondary Ions Observed from the SSIMS Positive Ion Spectra of Bupivacaine and Codeinea
a
The spectra can be found in the Supporting Information.
Previously,16 in a detailed study of homopolyesters, we identified all the key ions in the PLA mass spectrum (see Table 1 of ref 16). These correspond directly to the red peaks in Figure 1a,b and are consistent between the two figures as expected. Similarly, the peaks colored blue directly relate to the SIMS mass spectrum of pure drug shown in the Supporting Information. The characteristic drug peaks are listed in Table 1. Even though the mass peaks from the drug molecule and polymer overlap across the mass scale and in some cases have similar intensity and similar mass (near neighbors), the separation has worked excellently; considering there is no a priori information. Using this method, Figure 1 was labeled with the identified peaks. The PLA monomer unit is denoted by the letter M. Interestingly, the relative intensities of the codeine related ions are significantly different to those of the pure substance shown in Figure S2, Supporting Information. In the polymer matrix, the [Co H]þ ion is strongest and the [Co þ OH]þ ion is almost a factor of 10 weaker. In contrast, the [Co þ OH]þ ion is strongest in the spectrum of pure substance. This difference is most likely to result from the well know matrix effect resulting from
LETTER
competitive ionization processes. For bupivacaine, which appears to have approximately 10 times the ion yield, this matrix effect is not observed. We find that the g-ogram separation method gives consistent results for each of the three drugpolymer concentrations of 10:1, 20:1, and 40:1. We also note that interpretation using the g-ogram method is less affected by matrix effects than comparison with library spectra of pure substances. In conclusion, we show a promising new method to separate the mass spectra of a drugpolymer binary mixture into the separate substances. The method uses the g-index derived from the G-SIMS experiment as the separation parameter which relates to the amount of fragmentation or energy involved in the emission process. The data is visualized in a chromatogram called the g-ogram with the gindex and the mass, Mx. We show this is very effective for both codeine in PLA and bupivacaine in PLA at drugpolymer concentrations of 10:1, 20:1, and 40:1 by weight. Of course, two drugs in PLA would not be individually separated. This method may also be useful for separating surface contamination peaks from substrate peaks which is important in many industrial analyses. It is also likely to be generally applicable to separating molecules, such as additives, and polymers. The sensitivity of this is being tested. An attractive feature of this approach for analysts is that it can be done in real time. Recently, the G-tip24 was innovated which allows G-SIMS spectra to be acquired from a single ion gun. Either bismuth or manganese primary ions are electrostatically selected from a binary component source. In principle, this allows the G-SIMS spectrum to be acquired in real-time without compromising high spatial resolution specification. Therefore, the g-ogram may also be displayed in real-time as the acquisition proceeds. The gsep value may then be adjusted separating the mass spectrum; this could be of significant advantage to analysts. The real-time G-SIMS spectrum and g-ogram are from the entire analysis area. The counting statistics are too low to generate G-SIMS spectra and g-ograms individually for each pixel. However, as is usual in data acquisition, selected ion images may be displayed in real-time. More generally, this approach could work with other imaging mass spectrometries such as MALDI where the laser frequency or intensity may be adjusted or plasma methods such as plasma assisted desorption ionization25 (PADI) where it has been shown that the plasma power affects fragmentation26 and possibly also for low temperature plasma (LTP).27 This interesting possibility is being investigated.
’ ASSOCIATED CONTENT
bS
Supporting Information. Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Present Addresses §
Interdisciplinary Nanoscience Center (iNANO), Aarhus University, Aarhus, Denmark, 8000.
’ ACKNOWLEDGMENT R.O. gratefully acknowledges BBSRC for funding during his PhD and subsequently the Danish Agency for Science, 3630
dx.doi.org/10.1021/ac200347a |Anal. Chem. 2011, 83, 3627–3631
Analytical Chemistry
LETTER
Technology and Innovation. This work also forms part of the Chemical and Biological Metrology Programme supported by the National Measurement System (NMS) of the UK Department of Business, Innovation and Skills (BIS). The authors are grateful to Dr. Martin Seah for stimulating discussion and for suggestion of the name “g-ogram”.
’ REFERENCES (1) Lu, Y.; Chen, S. C. Adv. Drug Delivery Rev. 2004, 56, 1621–1633. (2) Langer, R. Science 1990, 249, 1527–1533. (3) Allen, T. M.; Cullis, P. R. Science 2004, 303, 1818–1822. (4) Keith, C. T.; Borisy, A. A.; Stockwell, B. R. Nat. Rev. Drug Discovery 2005, 4, 71–78. (5) Davis, M. E.; Chen, Z.; Shin, D. M. Nat. Rev. Drug Discovery 2008, 7, 771–782. (6) Vickerman, J. C.; Briggs, D. ToF-SIMS: surface analysis by mass spectrometry; IM: Chichester, 2001. (7) Belu, A. M.; Davies, M. C.; Newton, J. M.; Patel, N. Anal. Chem. 2000, 72, 5625–5638. (8) Grenha, A.; Seijo, B.; Serra, C.; Remunan-Lopez, C. Biomacromolecules 2007, 8, 2072–2079. (9) Mahoney, C. M.; Fahey, A. J.; Belu, A. M. Anal. Chem. 2008, 80, 624–632. (10) Vickerman, J. C.; Briggs, D.; Henderson, A. The Static SIMS Library, version 2; SurfaceSpectra, Manchester, UK, 2003. (11) Green, F. M.; Gilmore, I. S.; Seah, M. P. Anal. Chem. 2011, 83, 3239–3243. (12) Seah, M. P. Surf. Interface Anal. 2007, 39, 890–897. (13) Gilmore, I. S.; Seah, M. P. Appl. Surf. Sci. 2000, 161, 465–480. (14) Gilmore, I. S.; Seah, M. P. Appl. Surf. Sci. 2004, 231, 224–229. (15) Seah, M. P.; Green, F. M.; Gilmore, I. S. J. Phys. Chem. C 2010, 114, 5351–5359. (16) Ogaki, R.; Green, F. M.; Li, S.; Vert, M.; Alexander, M. R.; Gilmore, I. S.; Davies, M. C. Surf. Interface Anal. 2008, 40, 1202–1208. (17) Straif, C. J.; Hutter, H. Anal. Bioanal. Chem. 2009, 393, 1889–1898. (18) Gilmore, I. S.; Seah, M. P. Appl. Surf. Sci. 2003, 203, 551–555. (19) Watson, D. G. Pharmaceutical analysis: a textbook for pharmacy students and pharmaceutical chemists; Churchill Livingstone: Edinburgh, 1999. (20) Gilmore, I. S.; Seah, M. P. Surf. Interface Anal. 1996, 24, 746–762. (21) Green, F. M.; Gilmore, I. S.; Seah, M. P. J. Am. Soc. Mass Spectrom. 2006, 17, 514–523. (22) Gilmore, I. S.; Seah, M. P. Appl. Surf. Sci. 2002, 187, 89–100. (23) Kind, T.; Fiehn, O. BMC Bioinf. 2006, 7. (24) Green, F. M.; Kollmer, F.; Niehuis, E.; Gilmore, I. S.; Seah, M. P. Rapid Commun. Mass Spectrom. 2008, 22, 2602–2608. (25) Ratcliffe, L. V.; Rutten, F. J. M.; Barrett, D. A.; Whitmore, T.; Seymour, D.; Greenwood, C.; Aranda-Gonzalvo, Y.; Robinson, S.; McCoustra, M. Anal. Chem. 2007, 79, 6094–6101. (26) Salter, T. L.; Green, F. M.; Faruqui, N.; Gilmore, I. S. Private communication, 2011. (27) Harper, J. D.; Charipar, N. A.; Mulligan, C. C.; Zhang, X. R.; Cooks, R. G.; Ouyang, Z. Anal. Chem. 2008, 80, 9097–9104.
3631
dx.doi.org/10.1021/ac200347a |Anal. Chem. 2011, 83, 3627–3631
