X-ray Photoelectron Spectroscopy Analysis of Gold Surfaces after Removal of Thiolated DNA Oligomers by Ultraviolet/Ozone Treatment

Abstract

Well ordered films of molecular DNA can be formed by the attachment of thiolated DNA oligonucleotides to a supporting gold substrate. The gold substrate represents a significant fraction of the total cost of preparing such films and it is thus important to determine whether such substrates can be reused. Here we investigate with X-ray Photoelectron Spectroscopy the suitability of UV/ozonolysis previously employed to remove alkanethiols from gold, for removing 40-mer, single and double stranded synthetic DNA. We find that while UV/O₃ can indeed remove thiolated DNA from gold slides, the treatment times required permit the implantation of additional organic contaminants.

Introduction

Self-assembled monolayers (SAMs) are nanoscale structures often formed by the adsorption of organic molecules from solution onto a solid surface. The molecules attach to the surface via a chemically specific “headgroup” and steric-type interactions between neighboring bound molecules lead to the formation of dense, organized structures, one molecule thick [1]. It is generally recognized that SAMs represent an important ‘enabling technology’ with potential applications in areas such as micro-contact printing, organic electronics and highly specific bio-sensors [1]. The first SAM systems studied were organic disulphides adsorbed onto a gold surface [2]. Today gold remains a common substrate and the binding molecular headgroup is usually a thiol. Indeed, the most extensively studied type of SAM is that formed by the adsorption of alkanethiols onto a gold surface. The latter being typically prepared by the evaporative desposition of Au onto a thin chromium layer, which itself is bound to a silicon, glass or mica substrate.

Perhaps unsurprisingly, the pre-conditioning (or cleanliness) of the gold substrate prior to its treatment has been found, at least for alkanethiol derived monolayers, to play an important role in determining the properties of the final SAM [3–6]. Various cleaning methods have been studied and/or employed to remove impurities from gold surfaces prior to their use in SAM preparation. These have included standard cleaning agents such as piranha solutions [7], a novel mixture of NH₄OH and H₂O₂ [8], thermal desorption [9], plasma treatment [10], and most commonly, UV/ozonolysis (combined exposure to UV and ozone) [11–15]. Frequently, the effectiveness of such techniques has been demonstrated by removing an entire alkanethiol SAM from the gold, though this has been observed to decrease as the length of the adsorbed alkanthiol increases [16] (i.e., more time is required to remove greater mass). Nevertheless, since high quality gold slides are comparatively difficult and/or expensive to manufacture or otherwise acquire, techniques, such UV-ozonolysis, that are routinely employed to prepare lightly contaminated gold surfaces for SAM formation, have also been used to recover gold slides for reuse.

Studies on Au-SAM cleaning procedures have generally been concerned with the removal of alkanethiols or other smaller contaminant molecules [9–11, 13–15], but little is known about the efficacy of these methods for other types of contaminations. It is therefore of interest to find out if these present techniques can clean the gold substrates contaminated with more complex molecules. Here we have chosen to investigate the suitability of the ozone/UV method for the removal of thiolated DNA of intermediate length (here a 40 base pair long oligonucleotide), from a gold surface.

The adsorption via a terminal thiol onto a gold surface, has often been used to form well-characterised films of DNA [17–19], principally with the aim of developing bio-sensor arrays [20], but also for fundamental studies, including investigations of radiation damage and electron transport [21, 22]. Radiation damage is thought to proceed via the creation by primary ionizing radiation of numerous reactive secondary species, of which secondary electrons represent the most numerous species [23, 24]. These secondary electrons have very low energy. Their energy distribution peaks around 9–10 eV and they have a high cross section for interaction with matter [25]. Hence their effective ranges is short; so that their interactions with bio-molecules can be studied using electron impact techniques on thin sample films, and for such work, self-assembled monolayers (SAMs) offer certain advantages, in particular their (relatively) easy preparation, uniformity, and high density and supposed high order. Previous researchers from our laboratory have adsorbed thiolated DNA onto gold to determine electron attenuation lengths and effective damage cross sections in DNA SAMs [26] and to measure the electron stimulated desorption (ESD) of neutral [27, 28] and of anionic [22] species from DNA. In this latter work, single- and double-stranded (ss and ds, respectively) DNA oligonucleotides of 40 base pairs, were attached to gold by either a single terminal thiol (1S) or by five thiols (5S) positioned with a spacing of ~8 base pairs along the sugar-phosphate backbone. These DNA samples were chosen so as to form films in which molecules either lay in the plane of the gold surface (5S), or were more highly ordered and with molecules oriented away from the surface, at some small angle to the normal (1S). A comparison of the ESD yield of OH⁻ from these different films led to the conclusion that the anion derived from dissociation events occurring at the phosphate groups within the DNA backbone [22]. In all these types of experiments, there is an obvious interest to know if gold substrates can be cleaned and be reused.

Method

The DNA SAM samples were prepared on gold substrates supplied by Arrandee, Werther, Germany. Prior to use, these substrates were cleaned by (1) rinsing in ethanol and pure (i.e., doubly distilled and deionized) water (resistivity of 18 MΩ, pH=5.5 at room temperature), (2) exposition to ozone/UV radiation for 30 min (see below for details), (3) rinsing with pure water to remove the impurities due to reactions during step (3), (4) exposition to ozone/UV for another 30 min, and finally (5) rinsing. SAMS were prepared using phosphothiolated DNA obtained from substitution by sulfur of the oxygen doubly bonded to phosphorous. The following three different samples were prepared with the 40-mers oligonucleotides 5′-GGT ACC AGG CCT ACT ACG ATT TAC GAG TAT AGC GAG CTC G-3′ (G indicates the base guanine, C the cytosine, A the adenine, and T the thymine) with and without their complementary strands. These samples were purchased from University Core DNA services at the University of Calgary. A sulfur (1S) was substituted at one end of one backbone in the single (ss) stranded configuration (1S-ssDNA) and 5 sulfur atoms (5S) were substituted in the backbone in the ss and ds configurations (5S-ssDNA and 5S-dsDNA).

SAM samples were prepared by immersing the cleaned gold substrates in a 0.4 molar phosphate buffer solution (PBS) containing DNA, for at least 12 h. The DNA concentrations were 2.7, 12.7, and 3.5 M for 5S-ssDNA, 1S-ssDNA, and 5S-dsDNA, respectively. The samples were rinsed thoroughly with pure water and dried under a stream of N2 gas. These three DNA forms and their proposed binding to a Au substrate are illustrated schematically in Figure 1.

Figure 1.

Schematic drawing of DNA SAM films used in this study.

Once the films were formed, the UV/ozone cleaning technique was applied to remove the DNA SAMs from the gold substrates. Samples were exposed to ozone generated by a UV light in a stainless steel box while oxygen was continuously flowing into the box. The exposure time was varied from 15 to 90 minutes. The procedure was carried out in a laboratory hood at room temperature. The distance between the samples and the light was kept to be about 1 cm. After ozone exposure, the plates were rinsed carefully with pure water.

The samples were then transferred via a load-lock into ultra-high vacuum (UHV) for analysis by X-ray photoelectron spectroscopy (XPS). The XPS spectrometer (Perkin Elmer ESCA 5.2) employed is part of a multifunctional UHV apparatus that has been described elsewhere [29]. For these experiments, the X-ray source (15 kV, 350 W) was composed of an Al anode and Al window. The X-ray beam was incident at 72° relative to the sample surface and the energy of ejected electrons was measured with a hemispherical electrostatic analyser positioned normal to the samples surface. The pass energy of the electron energy analyser, which was set to 11.75 eV for all measurements, delivers XPS spectral resolutions of ΔE ~ 2 eV as estimated from the full width at half maximum (FWHM) of the S 2p and structures within the N 1s photoelectron lines.

Results and Discussion

Figure 2 demonstrates how the N 1s, S 1s, O 1s, Au 4p^3/2 and C 1s photoelectron peaks obtained from 5S-ds DNA samples vary with the period of exposure to the UV/O₃ cleaning agent. Simple linear backgrounds have been subtracted from each data set, which have been vertically offset for clarity. The cleaned samples were exposed for the indicated periods and “washed” (rinsed with water) during treatment, at the 15, 30, 60 and 90 minute exposure points.

Figure 2.

The effect of increasing exposure to the UV/O₃ on 5S-ds DNA films. Medium resolution X-ray photoelectron spectra show changes occurring in vicinity of the a) N1s, b) S2p, c) Au 4p_3/2 and O 1s and d) C 1s photoelectron lines. The cleaned samples were exposed for the indicated periods and “washed” (rinsed with water) after treatment at the 15, 30, 60 and 90 minute exposure points. Spectra were recorded for periods of 25 minutes (N1s and S2p), 12.5 mins (Au 4p_3/2 and O 1s) and 6 minutes (C1s).

Nitrogen atoms are found within all the DNA bases and since the contamination of gold surface with nitrogen containing molecules is rare, the N 1s photoelectron line has along with the weaker P 1s line, been used in earlier studies as a key indicator of the successful deposition of DNA onto a gold [17, 19] Indeed these two photoelectron lines has been used to a accurately measure DNA coverage [18, 30] and so determine the kinetics of the DNA self assembly. For the freshly deposited 5S-ds DNA film studied in Fig. 2, the maximum photoelectron yield of the N 1s manifold is observed at a binding energy (BE) of 399 eV, which, considering the possible multiple sites for nitrogen atoms within long DNA samples, is consistent with earlier XPS measurements of DNA SAM films (e.g., [17,18,21,30]). It is apparent from Fig. 2a) that increasing exposure of 5S-ds DNA films to the UV/O₃ cleaning treatment results in the progressive removal of N containing molecular species from the sample.

Previously, the observation of S 2p photoelectron line has been used to identify the successful attachment and/or removal of alkanthiols [9, 10] and DNA [31] onto gold substrates. Such studies report binding energies of ~ 162 eV for S 2p photo-electrons ejected from Au-bound S atoms (i.e., the -S-Au moiety), ~ 164 eV for unbound S and ~ 168 eV for sulphates and other S-O bond containing species. It is clear in Fig. 2b) that when an appreciable S 2p signal is observed in the present experiments, it can be linked to the binding of a S atom to the gold surface.

XPS measurements of the Au 4p_3/2, O 1s and C 1s on panels c) and d) provide complementary information on the cleanliness of the gold surface, and the degree of oxidative and carboniferous contamination. Data for the Au 4f _7/2,5/2 photoelectron lines, the strongest such lines for gold, were also recorded but are not shown, since the dependence of their intensities on exposure to the UV/O₃ cleaning procedure was essentially that of the 4p_3/2 line. It is evident from the lowest curves in panels c) and d) that even the fresh, unused and UV/O₃ cleaned Au slides show appreciable contamination by C and O containing impurities. However comparison with “fresh” gold slides (i.e., unused Au slides that were not exposed to the UV/O₃ treatment), indicates that this contamination can not be attributed to the cleaning and the formation oxides and/or the injection carboniferous species, both of which have been reported [12], since it is largely present prior to UV/O₃ exposure.

While the spectral resolution used in the present experiments does not permit a thorough investigation of changes to C and O surface chemistry when the DNA SAMS are cleaned, certain points can be drawn from the data of Fig. 2c) and d). In particular for the C 1s feature, the removal of DNA via UV/O₃ treatment, results in an apparent shift of the maximum of the line to lower BE, from ~ 285.5 eV to 284.5 eV. This change can be interpreted as the result of reduction in the proportion of C atoms bound to N (and also O), as typically found in DNA and associated with C 1S binding energies > 285 eV [21], and an increase in the proportion of C-C, C=C and C-H bound carbon atoms, that are associated with a C 1s of slightly lower BE. A similar, but less pronounced shift to lower BE is also seen for the O 1s, where careful inspection reveals a relative loss of signals at BE >532 eV, that might be associated with the PO₄⁻ in the backbone and even tight-bound water molecules [21]. The O 1s BE from ‘clean’ films is consistent with photo-emission from C-O bound oxygen atoms and is typical of oxygen containing, organic, Au-surface contaminants [32].

Original X-ray photoelectron spectra of the type presented in Fig. 2, were processed using the “XPSPEAK” data analysis freeware [33]. Processing included the subtraction of Shirley-type backgrounds from each spectrum and the fitting of XPS peaks with (where appropriate multiple) Guassian-Lorentzian functions. The total integrated areas for each peak, so obtained from cleaned 5S-ds and 5S-ss and 1S-ss films are plotted in Figures 3, 4 and 5, respectively. The data presented for the 5S-ds film (Fig. 3) additionally includes the result of exposing the SAM to UV/O3 cleaning for 90 min, but without regular washing.

Figure 3.

Signal intensities for the elements N, S, Au, O and C obtained by integration of the XPS peaks for N 1s, S 2p, Au 4p_3/2, O 1s and C1s, contained in Fig. 2 for 5S-ds DNA films following cleaning with UV/O₃ and regular rinses, for the indicated periods. For data marked ’90 min, uw’ no rinsing was performed. In this graph “clean” means unused (new), UV/O₃ cleaned gold slides.

Figure 4.

Signal intensities for the elements N, S, Au, O and C obtained by integration of the XPS peaks for N 1s, S 2p, Au 4p^3/2, O 1s and C1s, for 5S-ss DNA films following cleaning with UV/O₃ for the indicated periods. In this graph “clean” means unused (new), UV/O₃ cleaned gold slides.

Figure 5.

Signal intensities for the elements N, S, Au, O and C obtained by integration of the XPS peaks for N 1s, S 2p, Au 4p^3/2, O 1s and C1s, for 1S-ss DNA films following cleaning with UV/O₃ for the indicated periods. In this graph “clean” means unused (new), UV/O₃ cleaned gold slides.

The peak-integrated signal of the N 1s and S 2p lines are plotted in Fig. 6 as a function of duration of exposure to the UV/O₃ cleaning agent (with regular washing). These signals are compared to the ‘baseline’ values obtained from unused, UV/O3 cleaned gold slides (horizontal lines in the figure)

Figure 6.

Variation with UV/O₃ treatment time of the total integrated N 1s and S 2p photoelectron signals. Dashed horizontal lines indicate the values of these signals observed on unused, UV/O₃ cleaned gold slides.

While the data presented in Figures 3–6 are subject to normal, random experimental errors, several regularities can be discerned.

1. The intensity of the N 1S signal from DNA SAMS, which is related to DNA coverage [17, 19] varies in the order, 1S-ss > 5S-ds ≫ 5S-ss. This is broadly consistent with the view that the number density of DNA molecules should be largest for singly-tethered 1S-ss molecules. While a similar density of 5S-ds and 5S-ss might be expected on the Au surface, the double stranded molecules should contain approximately twice as many N atoms, hence generating a larger XPS signal. We attribute the surprising near-equality of the 1S-ss and 5S-ds N 1s intensities, (Fig. 6) to the orientation of the latter film being near parallel to the surface (see Fig. 1). This orientation ensures that N-atom containing components lie close to the film-vacuum interface, which assuming similar electron attenuation length between samples, increases the probability that N 1s photo-electrons will be detected. This effect is apparently sufficient to offset the larger quantity of N-atoms in the thicker 1S-ss samples, since in this case, many of these will be situated deep within the layer. We note in support of this picture that the Au 4p_3/2 photo-electron intensities follow the reverse sequence, viz., 5S-ss > 5S-ds ≫ 1S-ss, i.e., a larger proportion of Au 4p_3/2 photoelectrons are scattered (and are thus not detected) when traversing the thick 1S-ss layer than is the case of samples oriented parallel to the surface.

2. In all three cases during cleaning, the strongest N 1s signal is observed from intact DNA SAM films. Increased exposure to the UV/O₃ cleaning agent rapidly reduces the N 1s intensity, but even cleaning periods as long as 90 minutes are insufficient to reduce the intensity to the levels observed from previously unused gold slide (Fig. 6). It is somewhat surprising however that the UV/cleaning process appears to be roughly equally effective in each of the cases studied here. Previous work with alkanthiols [16] has indicated that larger alkanthiols are less easily photooxidised than are smaller molecules. While the DNA molecules in all the studies samples are all 40-mers, we had anticipated that there could be significant differences in cleaning rates due to differing film thicknesses and density. This appears not to be the case.

3. For both the 5S-ds and 5S-ss DNA films, the intensity of the S 2p photoelectron line is relatively small, but increases with initial cleaning until it falls to the same levels seen for UV/O₃ cleaned, new gold slides (Fig. 6). Such a behavior can be understood as resulting from the comparatively low number density of S atoms (5 atoms per DNA molecule), which for intact DNA films will be at the bottom of the DNA layer, attached to the gold. Subsequent cleaning results in the removal of the DNA layer (as demonstrated by the lost of N 1s signal), permitting easier detection of the S 2p photoelectrons. Sulfur atoms make up an even smaller fraction of the 1S-ss film, which as argued above, is expected to be both thicker and denser than the 5S-containing films. It is surprising then to see here that the S 2p signal from the intact film is very intense. We attribute this unexpected result to surface contamination. Previous work has shown that the purity of synthetic DNA oligonucleotide samples can vary with manufacturer and sample-lot and that sulfur-containing dithiothreitol, a reductant used to cleave disulfide precursors is a common impurity [31]. Indeed the XPS spectrum of the 1S-ss film (not shown) displays a much broader S 2p manifold, showing a structure centered at a BE of 164 eV, which consistent with the existing literature [9, 10, 31], we attribute to the presence of unbound S atoms. Interestingly, initial exposure of the 1S-ss film to UV/O₃ cleaning results in a greatly diminished S 2p signal, the intensity of which is much closer to the expected proportion of S atoms in the film, relative to those in the 5S-SAMS.

4. The importance of regular washing during UV/O₃ cleaning that has been identified in earlier studies [14,15]. Data presented in Fig. 3 for a 5S-ds sample that has been cleaned for 90 mins without washing demonstrates a similar effect. The sample exhibits a much higher N 1s signal than a washed sample, indicating a higher coverage of DNA or DNA degradation products than would otherwise be expected. Similarly the S 2p photoelectron is unusually intense. Indeed, the XPS spectrum of this sample show considerable signal at a BE of 168 eV, consistent with the presence of (water soluble) sulphates within the film [9, 10, 31].

5. The variation with duration of cleaning, of the intensities of the Au 4p_3/2, O 1s and C 1s photoelectron lines is comparatively complicated. Nevertheless, close inspection of Figures 3–5 indicates that at short times, UV/O₃ cleaning invariably results in the removal of C- and O-containing-species from the film, and a reduction in effective film thickness, as evidenced by increases in the Au 4p_3/2 signal. These observations are consistent with the removal of DNA from the film and the observations described above for the S 2p and N 1s signals. Unfortunately continued exposure to the cleaning agent results in the incorporation of new C- and O-containing-species. Similar effects have been reported elsewhere [12]. Significantly perhaps, there is a surprisingly good correlation between the quantity of O containing species on a sample and the intensity photoelectrons coming from the gold substrate, as illustrated in Fig. 7 which plots the Au 4p_3/2 intensity against O 1s intensity for 1S-ss films during cleaning.

Figure 7.

Correlation between the intensities of Au4p3/2 and O 1s photoelectron lines measured from neat and cleaned 1S-ss DNA films.

Thus we see from Figures 3–6, that the prolonged exposure of the DNA SAM to the UV/O₃ results in new sample contamination and that this is observed at times shorter than that necessary for the complete removal of N-containing species from the gold slide. We must thus conclude that the UV/O3 cleaning procedure is not satisfactory for the reconditioning of gold surfaces from slides of DNA SAM. As noted elsewhere, the improper preparation of gold substrates has measurable effects on the uniformity of alkanethiol SAMS [3–6]. It is highly likely that the same is true for DNA SAMS, so that use of recycled, UV/O₃ cleaned gold should be avoided in experiments with SAM films of DNA.

Thermal desorption for removal of thiol molecules has been shown to be a successful method [9]. This technique was briefly examined in our laboratory. The film of 5S-ssDNA was heated red hot under vacuum conditions for 5 minutes. As shown in Fig. 8 heating DNA films to elevated temperatures seems to produce superior results, however some defects in the gold plates were observed. In order to optimize the effectiveness of this method conditions such as temperature and duration of heating should be very well controlled. Such investigations are currently ongoing in our laboratory.

Figure 8.

Signal intensities for the element N1s obtained by XPS for DNA SAM, heated DNA SAM film to red hot for 5 mins and heated unused (new) gold slide.

In summary, we advise that the results of any presumptive cleaning procedure be rigorously tested before slides are reused. The present sequence of experiments was initiated to address our questions regarding sample uniformity in electron stimulated desorption measurements. During the course of the present work we have observed that UV/O₃ cleaning rates alter appreciably through very simple experimental changes, for example by moving samples from 3 cm, to 1cm away from the UV source. If gold-slide cleaning is to be used, its effectiveness must be quantified (with techniques such as XPS) under exactly the same operating conditions as will be routinely employed.