J. Phys. Chem. B 2010, 114, 769–774
769
Sample Cooling or Rotation Improves C60 Organic Depth Profiles of Multilayered Reference Samples: Results from a VAMAS Interlaboratory Study P. Sjo¨vall,† D. Rading,‡ S. Ray,§ L. Yang,§ and A. G. Shard*,§ Chemistry and Materials Technology, SP Technical Research Institute of Sweden, PO Box 857, SE-50115 Borås, Sweden, ION-TOF GmbH, Heisenbergstrasse 15, 48149 Mu¨nster, Germany, and National Physical Laboratory, Hampton Road, Teddington, Middlesex TW11 0LW, U.K. ReceiVed: October 5, 2009; ReVised Manuscript ReceiVed: NoVember 25, 2009
We demonstrate two methods to improve the quality of organic depth profiling by C60 sputtering using multilayered reference samples as part of a VAMAS (Versailles project on Advanced Materials and Standards) interlaboratory study. Sample cooling was shown previously to be useful in extending the useful depth over which organic materials can be profiled. We reinforce these findings and demonstrate that cooling results in a lower initial sputtering yield to approximately -40 °C, but the improvement in useful profiling depth continues as the sample is cooled further, even though there is no further reduction in the initial sputtering yield. We report, for the first time, the use of sample rotation in organic depth profiling and demonstrate that the initial sputtering yield at room temperature is maintained throughout the depth of the samples used in this study. Useful profiling depth and good depth resolution are both associated with a constant sputtering yield. The fact that rotation results in the maintenance of depth resolution underlines the fact that depth resolution is often limited by the development of ion-beam-induced topography. Constant sputtering yield results in a constant secondary-ion yield, after transient processes have occurred, and this allows simple quantification methods to be applied to organic depth profiling data. Introduction The sputter depth profiling of organic materials using cluster ion beams is a technique that has developed rapidly over the past few years.1–6 The potential of this technique has been explored in a wide number of phenomenological investigations that have shown promise for the three-dimensional reconstruction7–9 of the distribution of organic species in polymer films,10–13 water ice,14–16 and biological samples.17,18 These investigations have also demonstrated that some organic materials are not usefully profiled using normal experimental conditions and identified potential strategies to overcome these difficulties. The most promising approaches include using alternative cluster ion sources, such as massive argon clusters19 and, for some materials, low-energy cesium and oxygen ions;20 sputtering at low incidence angles;21 and cooling the sample with liquid nitrogen.3 Another procedure that has not yet been adequately explored for organic depth profiling is sample rotation. This approach is commonly used in the sputter depth profiling of inorganic systems using monatomic ions, following the work of Zalar.22 Sample rotation reduces the development of roughness during sputtering, and because roughening has been identified as one of the limiting factors in organic depth profiling, it is highly appropriate to use sample rotation to improve organic depth profiling. One of the difficulties in comparing the variety of approaches reported in the literature is the range of different materials used in those investigations. Recently, this issue was addressed by the production of multilayered reference samples23 by NPL * Corresponding author. Tel.: +44 (0) 20 8943 6193. Fax: +44 (0) 20 8943 6453. E-mail:
[email protected]. † SP Technical Research Institute of Sweden. ‡ ION-TOF GmbH. § National Physical Laboratory.
(National Physical Laboratory, Middlesex, U.K.) as part of an interlaboratory study conducted under the auspices of VAMAS (Versailles Project on Advanced Materials and Standards) TWA2, surface chemical analysis. The study demonstrated that excellent repeatability (better than 5% variation in depth resolution, sputtering yields, and relative secondary-ion intensities) could be obtained by many participating institutions.24 Good comparability in depth resolution and sputtering yields between institutions that had well-defined procedures and similar instruments was also found. In this article, we describe results from the reference materials taken on similar instruments using two of the methods described above, namely, cooling the sample and rotating the sample during sputtering. Both methods deliver significant improvements to the quality of the depth profile. Experimental Section The construction of the samples is described in detail elsewhere.23 In brief, silicon wafers with the (100) orientation were cleaved to ∼10 mm × 10 mm squares. Particulates on the surface were removed gently using lint-free tissue, and the wafers were soaked overnight in isopropanol prior to drying and deposition of the Irganox layers. Wafers were placed face down in a holder positioned above two crucibles, one containing Irganox 1010 and the other Irganox 3114 (CIBA, Macclesfield, U.K.), in an Edwards AUTO306 vacuum coater. A quartz crystal microbalance (QCM) was positioned to the side of the holder and used to monitor the thickness of each Irganox layer. The thicknesses of various films deposited from each crucible were established using spectroscopic ellipsometry (M2000, Woollam, NE) in calibration runs, and the ellipsometric thickness used to obtain calibration factors for the QCM. Layered samples were constructed by heating each crucible alternately; the film thicknesses of the deposited layers for samples used in this investigation are listed in Table 1. The samples were large
10.1021/jp9095216 2010 American Chemical Society Published on Web 12/18/2009
770
J. Phys. Chem. B, Vol. 114, No. 2, 2010
Sjo¨vall et al.
TABLE 1: Layer Thicknessesa (nm) for the Samples Used in This Studyb layer layer layer layer layer layer layer layer layer
1: Irganox 1010 1M: Irganox 3114 2: Irganox 1010 2M: Irganox 3114 3: Irganox 1010 3M: Irganox 3114 4: Irganox 1010 4M: Irganox 3114 5: Irganox 1010
sample A
sample B
sample C
49.3 2.8 48.3 3.5 99.1 2.8 99.7 2.8 98.9
48.5 2.8 48.2 2.7 95.9 2.7 96.5 2.8 96.1
49.7 3.1 50.3 4.2 99.7 3.1 99.7 3.1 99.7
a QCM, calibrated to ellipsometric thickness. Within one sample the precision is (0.3 nm, between samples the uncertainty in ellipsometric thickness is 2%. b Layer 1 is at the film surface, and layer 5 is adjacent to the silicon substrate.
enough that multiple analyses could be carried out on separate areas of a single sample. These samples were very smooth and of homogeneous thickness, and therefore, they were ideal for the investigation and optimization of parameters that can affect depth profiling. Three different secondary-ion mass spectrometers were employed in this work. All were manufactured by ION-TOF GmbH (Mu¨nster, Germany) and equipped with C60 sources mounted at 45° with respect to the sample surface and Bi ion sources mounted at 45° with respect to the sample surface and opposite the C60 source. Data were obtained using cycles of C60n+ sputtering in direct current followed by analysis with a pulsed 25 keV Bi3+ beam within the center of the sputtered area. Care was taken to ensure that the analysis area avoided the edges of the sputtered area where the C60n+ dose was not constant. The NPL sputtered sample A in a TOFSIMS IV instrument using 20 keV C602+ ions at room temperature and without rotation. The sputtered area was 400 µm × 400 µm, and the analysis area was 100 µm × 100 µm. SP Technical Research Institute of Sweden sputtered sample B in a TOFSIMS IV instrument using 10 keV C60+ and 20 keV C602+ ions without rotation and at room temperature. Analyses at 10 keV were carried out at a variety of different temperatures (from 20 to -80 °C). Samples were cooled using a coldfinger connected to a liquid-nitrogen reservoir. A resistive heater was used in a controlled manner, using the instrument software, to obtain the desired temperature. The sample temperature was monitored using a thermocouple in contact with the sample. The sputtered area was 500 µm × 500 µm, and the analysis area was 200 µm × 200 µm. ION-TOF GmbH sputtered sample C in a TOFSIMS 5 instrument using 10 keV C60+ and 20 keV C602+ ions with the sample mounted on a specially adapted holder. The sample was rotated at 14 Hz during the analysis. The sputtered area was 700 µm × 700 µm at 20 keV and 750 µm × 750 µm at 10 keV, and the analysis area was 200 µm × 200 µm in both cases. An analysis using 20 keV C602+ ions without rotation was also performed. The C60n+ ion beam current was between 0.2 and 1 nA, and the pulsed Bi3+ beam current was more than 3 orders of magnitude lower than the C60n+ ion beam current. Care was taken to measure the C60n+ ion beam current before and after each experimental session. Where drifts in the beam current were found, a linear interpolation between the initial and final currents was used to estimate the beam current during an individual depth profiling experiment. All instruments obtained closely similar results using 20 keV C602+ ions in terms of sputtering yields and depth resolution; the variation was within
Figure 1. Organic depth profiles through the multilayer samples used in this work showing the normalized intensity of the 564 u secondary ion as a function of C60n+ dose. Data points are shown as crosses; fits are shown as solid lines. (a) Profiles taken using 20 keV C602+ energy from the reference sample and the rotation sample at room temperature. (b) Profiles taken from the cooling sample at different temperatures without rotation.
the limits set by the samples themselves. Results reported herein for this energy without rotation are from sample A. Results A comparison between example profiles taken under different conditions is shown in Figure 1. In these graphs, we plot the intensity of the characteristic Irganox 3114 fragment at 564 u (C33H46N3O5-) as a function of C60n+ ion dose. Also shown are fits using Dowsett et al.’s response function25 as described previously.2 These plots graphically demonstrate the improvement in depth profiles that can be made using sample rotation or cooling to -80 °C. The four Irganox 3114 marker layers are clearly distinguished and have comparable widths and shapes, implying a minimal degradation in depth resolution during the sputtering process. In both cases, a small peak in signal is observed at the organic/silicon interface (more pronounced for the low-temperature sample) that we attribute to an ambient
Improved Organic Depth Profiling by C60 Sputtering
J. Phys. Chem. B, Vol. 114, No. 2, 2010 771
Figure 3. Depth resolution of organic depth profiling, expressed as fwhm of intensities from Irganox 3114 secondary ions, as a function of depth. Markers indicate the width of the 564 u signal, and the range of widths from other characteristic Irganox 3114 secondary ions are indicated by bars. (a) 20 keV C602+ ion sputtering, (0) with and (]) without rotation. (b) 10 keV C60+ ion sputtering as a function of temperature (0, 0 °C showing typical range with bars; ], -20 °C; 4, -40 °C; [, -60 °C; 9, -80 °C). At the two highest temperatures, the fwhm of the fourth layer cannot be reliably measured; the lines included in the graph are merely indicative of the degradation in depth resolution.
Figure 2. (a) Depths of marker layers as a function of the dose required to reach them. Data are shown for 10 keV C60+ (], standard conditions; 0, with rotation; and 4, with cooling) and for 20 keV C602+ ([, standard conditions; and 9, with rotation. Lines represent fits to the data, as described in the text. (b) Initial sputtering yield, S0 (]), and cross section for sputtering decay, σDS (0), at 10 keV C60+ energy as a function of temperature.
background deposition of Irganox 3114 on the samples in the evaporator prior to the commencement of sample production. From the negligible drift in QCM readings during this period, the thickness of this inadvertent Irganox 3114 layer is less than 0.4 nm. Sputtering Yields. Close inspection of Figure 1 reveals some interesting features. The dose required to reach the first Irganox 3114 layer is approximately identical at 20 keV C602+ regardless of whether rotation is used. With rotation, however, the deeper layers are revealed with a lower sputter dose, implying that the initial sputtering yield is maintained. Upon cooling of a sample, the dose required to reach the first layer increases, implying a lower initial sputtering yield. In Figure 2a, we plot the depth of each layer against the dose required to obtain a maximum in the 564 u secondary-ion intensity for that layer: At 10 keV under normal conditions, the dose required to reach the final layer is very uncertain, as indicated by the error bars. We use a modified form of a previous equation for sputtering yield variation2,23 to fit these limited data sets with two parameters that describe the initial sputtering yield and the dose-dependent decay in sputtering yield
D)
S0 [1 - exp(-σDSF)] σDS
(1)
where D is the eroded depth as a function of C60n+ ion dose, F. S0 is the initial sputtering yield volume, and σDS is the cross section for sputtering decay; with the equation in this form, we
assume that the final sputtering yield is zero. We excluded highly uncertain points from the fit. The fits to the data clearly describe the majority of the important behavior, and it is clear that the initial sputtering yield at both 10 and 20 keV is unaffected by the use of rotation. The decline in sputtering yield is dramatically affected by the use of rotation; σDS changing from 0.26 nm2 per ion without rotation at 20 keV to
