Anal. Chem. 2001, 73, 1292-1295
Desorption-Ionization Mass Spectrometry Using Deposited Nanostructured Silicon Films Joseph D. Cuiffi*
Nanofabrication Facility, The Pennsylvania State University, University Park, Pennsylvania 16802 Daniel J. Hayes, Stephen J. Fonash, Kwanza N. Brown, and Arthur D. Jones†
Mass Spectrometry Center, The Pennsylvania State University, Whitmore Laboratory, University Park, Pennsylvania 16802.
We present a method for desorption ionization on silicon based on novel column/void-network-deposited silicon thin films. A number of different peptides and proteins in the e6000 Daltons range are analyzed by time-of-flight mass spectrometry in this demonstration of our approach. A variety of sample preparation conditions, including the use of chemical additives, surface treatments, and sample purification are used to show the potential of mass analysis using deposited column/void-network silicon films for high throughput proteomic screening. Obtaining mass spectra for synthetic and biological samples using matrix-assisted laser desorption ionization (MALDI) mass spectrometry offers soft ionization capabilities that preserve molecular mass information over a broad molecular mass range. These features have made MALDI a popular technique since its inception.1 However, for analysis of low-mass analytes (m/z < 500), irreproducible and heterogeneous cocrystallization, suppression of ionization by electrolytes and other additives, and interference from matrix ions have limited the utility of MALDI in automated high-throughput combinatorial and chip-array analyses. Active efforts to improve this process have led to alternatives such as successful desorption ionization from particles suspended in a liquid such as glycerol.2-6 More recently, desorption ionization was achieved without a matrix by using electrochemically etched conventional porous silicon.7 This matrix-free approach, which has drawn considerable attention, utilizes a silicon support material that is etched from bulk silicon wafers. The etched porous silicon suffers, however, from spacial nonuniformity, degradation due to trapped etchants, and manufacturability issues. Surface morphology is recognized as an important determinant of MALDI ionization yields.8-10 Therefore, substrates that would avoid these issues * Fax: (814) 865-7173. E-mail:
[email protected]. † Mass Spectrometry Center. (1) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299. (2) Dale, M. J. K.; Knochenmuss, R.; Zenobi, R. Anal. Chem. 1996, 68, 3321. (3) Han, M. S. J. J. Am. Soc. Mass Spectrom. 2000, 11, 644. (4) Schu ¨ renburg, M.; Dreisewerd, K.; Hillenkamp, F. Anal. Chem. 1999, 71, 221. (5) Sunner, J.; Dratz, E.; Yu-Chie, C. Anal. Chem. 1995, 67, 4335. (6) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T. Rapid Commun. Mass Spectrom. 1988, 2, 151. (7) Wei, J.; Buriak, J. M.; Siuzdak, G. Nature 1999, 399, 243. (8) Cohen, S. L.; Chait, B. T. Anal. Chem. 1996, 68, 31.
1292 Analytical Chemistry, Vol. 73, No. 6, March 15, 2001
and present a more controlled and homogeneous morphology without a matrix promise to give reproducible ion yields, mass accuracy, high resolution, and if manufacturable, a significant impact on efforts such as proteomic screening. This report describes a matrix-free method based on a deposited silicon thin film. This film is a unique columnar/voidnetwork silicon film material that is prepared by plasma-enhanced chemical vapor deposition (PECVD).11 The film, as deposited, requires no etching and can be fabricated on a variety of substrates including glass and plastics. The morphology of the film is best described as vertical rodlike columns on the order of 10 nm in diameter clustered into larger columnar features on the order of 100 nm that are contained within a continuous void space. Detailed specifics of columnar/void-deposited silicon material morphology and processing are presented in Kalkan et al.11 Our deposited film approach offers advantages over electrochemically etched silicon in cost, contamination control, uniformity, and control of porosity. In this study, we demonstrate the use of this novel material for the detection of analytes in the range of 0-6000 Daltons. EXPERIMENTAL SECTION Substrate Preparation. Our columnar/void-network silicon films are deposited by plasma enhanced chemical vapor deposition (PECVD) using a Plasma Therm, electron cyclotron resonance (ECR) high-density plasma source. This technique produces a nano-structured columnar/void silicon film at low substrate temperatures (100 °C). For our study, poly(ethylene terephthalate)(PET) and glass(Corning 1737) substrates were coated with between 500 and 10 000 Å of this deposited columnar/voidnetwork silicon film. Depositions were controlled to give a range of porosities (void densities) with and without a silicon nitride base layer. Postdeposition surface modifications to the films were explored, including growth of a thin silicon dioxide layer, silanization with 3-aminopropyltriethoxysilane (from Sigma), and lightmediated surface functionalizations with 1-hexyne, 5-hexyn-1-ol, 1-decyne, and 9-decen-1-ol (all g97% pure, from Lancaster). All mass spectra presented in this report were obtained using (9) Sadeghi, M.; Vertes, A. Appl. Surf. Sci. 1998, 129, 226. (10) Westman, A.; Huthfehre, T.; Demirev, P.; Sundqvist, B. U. R. J. Mass Spectrom. 1995, 30, 206. (11) Kalkan, A. K.; Bae, S.; Li, H.; Hayes, D.; Fonash, S. J. Appl. Phys. 2000, 88, 555. 10.1021/ac001081k CCC: $20.00
© 2001 American Chemical Society Published on Web 02/10/2001
Figure 1. Mass spectrum of a 1-µL drop of a 1 µM (1 pmol) despro3,[ala2,6]-bradykinin (m/z 920) solution on a columnar/void-network silicon thin film substrate.
columnar/void-deposited silicon prepared at 8m Torr process pressure.11 Other substrates, such as gold, glass, thermal silicon oxide, and silicon wafers were also used for comparison. Sample Preparation. The proteins, peptides and ammonium citrate used in this study were obtained from Sigma, and the trypsin was frozen, sequencing grade, from Promega. Organic solvents such as methanol, acetonitrile, and DMSO were added in some cases for signal comparison (all HPLC grade, from EM Science). In the case of ubiquitin, test solutions for the mass desorption study were made using HPLC-purified compounds and deionized water or ammonium bicarbonate (0.1 M, pH 7.8) for the tryptic digest reactions of ubiquitin. The matrix-free preparation of the sample on the surface was done by allowing a 0.5-1 µL drop of the sample to air-dry on the surface. On-column desalting was performed on Zip Tip (Millipore Co.) C18 resin pipet tips. Sample Analysis. All samples were analyzed using a Perseptive Biosystems (Framingham, MA) Voyager-DE STR mass spectrometer using 337-nm light from a nitrogen laser. Substrates were attached to the face of the conventional MALDI target using double-sided tape. Analyses were performed in linear mode with instrument parameters identical to normal MALDI operation, except that no low-mass cutoff was employed. Spectra given in this report are shown with an average of g50 laser shots. Significant differences between theoretical and reported mass-tocharge ratios are a result of a 0.7-mm thickness of the substrateattached to the MALDI sample plate and can be corrected through proper calibration. RESULTS AND DISCUSSION Our results not only show the usefulness of our deposited thinfilm nano-structured silicon for laser desorption but also give insight into some of the mechanisms governing desorption and ionization on these films. In Figures1-4, we first demonstrate the performance of our films as a laser desorption substrate. Figure 1 shows the detection of des-pro3,[ala2,6]-bradykinin at m/z 920 with surrounding peaks (at m/z 943 and 959), which corresponds to sodium and potassium attachment. This plot demonstrates the utility of our deposited film material with the absence of noise below m/z 500 and clear detection of the peptide. Figures 2 and 3 give spectra of a mixture of peptides on our film that was coated on glass (Figure 2) and plastic (Figure 3). Figure 4 provides the
Figure 2. Mass spectrum of a mixture of peptides, all in the picomole range, including des-arg1-bradykinin (m/z 905), angiotensin I (m/z 1297), and neurotensin (m/z 1673) on a glass substrate coated with columnar/void-network silicon thin film.
Figure 3. Mass spectrum of a mixture of peptides, all in the picomole range, including des-arg1-bradykinin (m/z 905), (m/z 1184), angiotensin I (m/z 1297), glu1-fibrinopeptide B (m/z 1571), and neurotensin (m/z 1673) on a plastic substrate coated with columnar/ void-network silicon thin film.
Figure 4. Mass spectrum obtained from a 1-µL drop of a 5 µM (5 pmol) solution of thyrocalcitonin (m/z 3605), insulin (m/z 5735), and other smaller peptides.
spectra of proteins with masses ranging from 3000 to 6000 Daltons, in which insulin (m/z 5735) was detected. Although the lower mass range of the spectrum in Figure 4 is not given for simplicity, it is important to note that the larger molecules were still able to be detected when competing for energy and charge in the presence of smaller peptides, such as bradykinin and angiotensin, which were present in the mixture. At the time of this study, the limit of detection for des-pro,3[ala2,6]-bradykinin we had attained was 50 fmol. Analytical Chemistry, Vol. 73, No. 6, March 15, 2001
1293
Figure 5. Plot showing the minimum laser power/pulse that is necessary to obtain a mass signal for des-pro3,[ala2,6]-bradykinin versus columnar/void-network silicon thin film reflectivity.
Although the mechanisms for ionization are quite complex, energy transfer from the incident laser to the sample molecules is a very important process in our technique, as well as in MALDI. No observable peptide signal was obtained if our sample molecules were placed on a metal surface, where the laser light reflects efficiently, or on a glass surface, where the laser light is poorly absorbed. The molecules did, however, desorb and ionize on a UV-absorbing silicon wafer with or without a silicon dioxide surface layer. The results indicate that the critical element of this process is the coupling of the laser light into the substrate. The UV reflectance of our deposited nano-structured column/voidnetwork silicon films can be tailored by adjusting the deposition parameters.11 With our high density plasma deposition process, highly repeatable normal-incidence UV reflectance between 10 and 50% at 337 nm can be obtained by varying the process parameters, which varies the porosity of the film. For comparison, the UV reflectance of a silicon wafer at 337 nm with or without a silicon dioxide coating ranges between 40 and 70%. Figure 5 shows the relationship between the minimum laser power necessary to detect bradykinin and the 337-nm reflectivity of our deposited silicon films. The low reflectance possible with our films (seen in Figure 5) is advantageous, because lower laser power reduces kinetic energy transfer and may improve high-resolution spectra. Because of the efficient UV absorption of silicon, a depth of about 50 nm into a crystal silicon wafer is necessary for complete UV laser absorption. The void space in our deposited silicon films alters the absorption characteristics, as compared to a dense crystal-silicon wafer. However, we found that after approximately 100 nm of our deposited columnar/void film, we no longer saw a signal dependence on film thickness. These results indicate that very thin films and, consequently, short deposition times are adequate for useful films, supporting the manufacturability of these films for commercial use. To observe the effects of laser power with our deposited Si films, we characterized the laser desorption mass spectrometry detection of bradykinin using various laser powers. Figure 6 shows the trends in bradykinin counts and signal-to-noise ratios versus laser power. At low laser energies, the bradykinin signal increases with increasing power. However, as the laser power continues to increase, the molecules begin to break down during desorption or extraction, which lowers the counts and the signal-to-noise ratio. Figure 1 gives the spectrum that corresponds to the highest signalto-noise ratio point on this plot. The data in Figure 6 demonstrates 1294 Analytical Chemistry, Vol. 73, No. 6, March 15, 2001
Figure 6. Detection characteristics of des-pro3,[ala2,6]-bradykinin versus laser pulse power. The circle (O) and the diamond (]) correspond to relative des-pro3,[ala2,6]-bradykinin counts and RMS signal-to-noise ratios, respectively.
Figure 7. Mass spectrum obtained from a column desalted tryptic digest of ubiquitin (1 mM predigest) with the addition of ammonium citrate (250 µM ammonium citrate, 1:1 mixture).
that a laser power range suitable for sample detection in this case is between 6 and 10 mJ/pulse, with a spot size of ∼200 mm. Solution tryptic digests of ubiquitin were conducted to determine the suitability of these columnar/void-network films for peptide mass mapping. As we show in Figure 7, by eliminating matrix compound contamination, low-mass data can be collected for small peptides and molecules using our mass analysis with silicon films (MASiF) approach. The addition of ammonium citrate to post-digestion reaction mixtures was used here and dramatically improved the ability to detect peptide fragments. Figure 7 demonstrates 10 peaks corresponding to predicted ubiquitin tryptic digest fragments or common products of incomplete digestion. Light and chemically mediated molecular attachment to the surface was performed on the material used for Figure 1 to also study the effects of mediation and of hydrophobicity and hydrophilicity of the column/void-network films on signal acquisition. To achieve the various modified surfaces, 6- and 10-carbon-chain molecules were attached with both methyl- and alcohol-terminated ends. It was observed that all such modified films required an increased laser power for detection but displayed no change in the reflectivity. This indicates that the extra molecular layer between the analyte and nano-structured surface reduces the energy transfer efficiency, but still enables desorption and ionization to occur. Other differences in the signals obtained using hydrophobic and hydrophilic surfaces are difficult to generalize and are probably related to the analyte distribution, crystallization
rate, and footprint that result when a drop dries on the surface. However, clear mass spectra can be obtained from surfacefunctionalized samples, which may lead to novel surface selectivity detection devices. To date, similar conclusions have been drawn from our studies of adding organic solvents to our molecular samples before their application to the deposited columnar/void-network silicon film surface. Primarily, our samples were prepared with HPLC-purified molecules dissolved in deionized water; however, we also examined the effects of various additional solvents. In general, it was found that these additional solvents affected the way the drop dried and, therefore, modified the state of the analyte on the surface. We found this to be a very important, but not yet fully understood, phenomenon. A specific example of these effects can be found in samples that contained >25% acetonitrile. Drops from such a sample have low surface tension and quickly spread over a large surface area, reducing signal intensity. Some solvents such as DMSO have extremely low vapor pressures and will not dry within a reasonable amount of time at room-temperature conditions. Purity of the organic solvents also must be considered because of the sensitivity of this technique to low-mass contaminants that can be included in the signal. Our deposited column/void-network films are extremely effective in absorbing and fixing atomic and molecular species, including atmospheric species. We have observed that over a period of days after deposition, low-mass noise begins to appear along with the desired spectrum obtained using our films.
However, such contamination can be avoided with careful storage and handling techniques. We always immediately implement these procedures to avoid inadvertent absorption when our films are taken from the vacuum deposition chamber. Defining the surface state of our film in a vacuum and not in an etching bath yields controllability that is very important for clinical and drug development applications in which contamination issues can be most critical. CONCLUSIONS A nonmatrix laser desorption ionization technique has been developed based on deposited nanostructured thin film silicon. The vacuum preparation of these deposited films avoids the contamination issues that are involved in etching and exposure to wet solutions and electrodes. These films, which can be deposited on plastics or glass, can be sealed after fabrication and only exposed at the time of sample preparation. We demonstrate that these nanostructured deposited films enable molecular detection in the range of 0-6000 Daltons with little or no lowmass noise. We establish that the efficient optical coupling of the laser energy into the substrate provided by these unique films is very important in the desorption and ionization process.
Received for review September 12, 2000. Accepted November 28, 2000. AC001081K
Analytical Chemistry, Vol. 73, No. 6, March 15, 2001
1295
