- Email: [email protected]
Effective approaches for realizing quantitative analyses and high lateral resolution images on highly insulating samples by Auger electron spectroscopy
Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
Contents lists available at SciVerse ScienceDirect
Journal of Electron Spectroscopy and Related Phenomena journal homepage: www.elsevier.com/locate/elspec
Effective approaches for realizing quantitative analyses and high lateral resolution images on highly insulating samples by Auger electron spectroscopy Jérôme Guillot ∗ , Henri-Noël Migeon Department “Science and Analysis of Materials” (SAM), Centre de Recherche Public – Gabriel Lippmann, 41 rue du Brill, L-4422 Belvaux, Luxembourg
a r t i c l e
i n f o
Article history: Received 26 October 2012 Received in revised form 13 March 2013 Accepted 21 March 2013 Available online xxx Keywords: AES Insulator Ceramic SiAlON SAM Lateral resolution
a b s t r a c t The analysis of insulating materials by Auger electron spectroscopy (AES) remains difficult to achieve because of the accumulation of charges in the sample. Residual charges trapped in the specimen can lead to numerous spectra modifications such as an energy shift of the Auger transitions, enlargement, splitting and deformation of the peaks, intense or lack of secondary and Auger electrons emission. Among the different approaches already developed to circumvent this issue, two of them were considered in this study to carry out AES experiments on a SiAlON insulating ceramic: the charge compensation method and the thin film method. For both of these methods, a systematic approach was used, combining a specific sample preparation with optimized analytical settings to mitigate the charge effects to enable quantitative analysis and high lateral resolution images. The charge compensation method, dedicated to bulk samples, is based on the control of the total secondary electron yield (TSEY) during analysis by optimizing the intensity, the energy and the incident angle of the primary electron beam. Its combination with the metallization of the surface sample and the use of low energy Ar+ ions to compensate the charges allowed the determination of the elemental composition of the three sub-micrometric phases of the ceramic. High lateral resolution (70 nm) Auger maps were also acquired, demonstrating therefore the long-time stability of the surface charge during acquisitions even for small analysis areas. A second method, consisting in thinning the sample down to less than a hundred of nanometres and analyzing it with a high energy electron beam, was implemented too. The results (quantification and imaging) are in good agreement with the analysis of the sample as a bulk. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Due to the rapid progress in downscaling, many materials with a wide range of applications became very important industrial and economic stakes. Thus, recent researches concern the study of material such as ceramics, VLSI devices, polymers, glasses or optical fibres. All these materials belong to the class of highly insulating specimens and, for a long time, electron spectroscopy techniques (i.e. Auger electron spectroscopy (AES), X-ray photoelectron spectroscopy (XPS), reflection electron energy loss spectroscopy (REELS)) were considered to be not applicable to electrically non-conductive sample structures. Indeed, in order to analyze their surface, the specimens are irradiated with X-rays or electrons, resulting in the emission of electrons that are then
∗ Corresponding author. Tel.: +352 470261524. E-mail address: [email protected] (J. Guillot). 0368-2048/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.elspec.2013.03.007
collected and counted. During this process, the previous samples are subjected to charging effects because of their insulating nature. However these analytical techniques are extremely surface sensitive and provide useful elementary, quantitative and chemical information for all elements except hydrogen and helium. Because, in XPS, the sample is irradiated with neutral X-rays, it is often possible, by using a flood electron gun if necessary, to carry out experiments in quite an easier way than for spectroscopies using a primary electron gun which contributes to the charge built up as in AES for instance. That is why XPS is one of the first techniques applied to analyze and to obtain chemical information on insulating samples, this subject being well documented [1–4]. The recent improvement of the size of the X-ray spot can be satisfactory for many problems on flat, homogeneous samples or of laterally uniform composition. However, one of the unique advantages of AES is its capability of providing compositional information with a high spatial resolution. Whereas the scale of the smallest X-ray beam in XPS is several micrometres, the nanometre scale has already been reached in AES.
2
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
With the increasing need to analyze insulating samples at small scales, new methods or approaches allowing AES experiments, need to be developed. Indeed, charging is known to be a major obstacle in Auger for the characterization of insulating materials. The main problems generally encountered are: - a high charge accumulation on the surface leading to the emission of intense secondary electron peaks with the release of excessive negative charges (for surfaces negatively charged) or on the contrary, the lack of electron emission outside the sample (for surfaces positively charged); - the peaks broadening, splitting and energy shifting making more complicated or even impossible the identification of the species and their chemical state; - the sample modification with surface and bulk damaged by the charge accumulation or maybe even the reaction of the surface [5,6]; - a non-homogeneous spatial distribution of the charge accumulation or non-constant in time. To reduce these effects that affect AES imaging and spectroscopy, it is necessary to avoid or at least to limit charge build-up by suitable experimental analysis conditions and by specific sample elaboration and preparation. Reports on the existing methods and their ease of application have already been published and are a good starting point [7,8]. Analyzing insulators is never straightforward, especially with rough surfaces [9] and each sample requires optimized preparation and analytical conditions. The charging effects in Auger result from the low diffusion of charges both on the surface and in the specimen bulk. Samples with a low resistivity behave as resistors and lead to an energy shift of the Auger spectra, on the other hand, highly resistive specimens can be modelled as capacitors. The residual charge on the sample is then expressed as: Qs = Qi − (Qo + Qd )
(1)
where Qs is the sample residual charge, Qi is the incident charge, Qo is the outgoing charge and Qd is the diffusion charge. This charge accumulation leads to an uneven distribution of charges across the surface and to differential charging, which strongly alters the kinetic energy and the direction of the escaping electrons. Although most of the time insulators are negatively charged due to the implantation of electrons from the primary electron beam during the analysis, they can also be positively charged: the key parameter being the total secondary electron yield (TSEY) defined as the ratio of the number of leaving electrons on the number of incident electrons: =
Io =1− Ii
I t
Ii
with It = Ii − Io
(2)
where Ii is the incident electrons current, Io the current of outgoing electrons (i.e. secondary, backscattered and Auger electrons), and It the total current. As described elsewhere [10,11] the TSEY depends mainly on the angle (), the energy (E) of the primary electron beam and the materials. When < 1 the surface is negatively charged whereas it will be positively charged if > 1. Most interesting are the cases when is equal to the unity (for a given a maximum of two energies E1 and E2 (E1 < E2 ) allow to achieve this). The incident electron current is then equal to the outgoing electron one and no charge should build up. This model clearly highlights the main parameters controlling the charge effects; however, it assumes a homogeneous lateral and in-depth distribution of the charges which does not correspond to an insulator submitted to an electron beam during AES analysis. The dependence of the TSEY with and E was recently discussed in details for insulators continuously irradiated by electrons and in terms of time evolution and depth distribution of the
charge [12–14]. The TSEY varies faster with time when the surface is negatively charged. Thus, analytical conditions to obtain a slightly positively charged surface have to be favoured to perform reliable AES characterizations. Furthermore, a low positive charge at the surface of the sample will re-attract the low-energy outgoing secondary electrons that can help for charge stabilization. Moreover, the penetration depth of the primary electrons has to be in the range of the maximum escape depth of the electrons to facilitate the hole – electrons recombination’s and to avoid the creation of a dipole in the sample constituted with a positively charged surface and an excess of negative charges in the bulk [12]. The purpose of this paper is to illustrate two of the common methods used to carry out AES experiments on insulating samples: the charge compensation method applied for samples in the form of bulk and the thin film method. They were applied to characterize a non-conductive SiAlON ceramic specimen. Many works dealing with the characterization of SiAlON with AES can be found in literature [15–21], but rarely with high lateral resolution. The results and the ease of their implementation will be commented. SiAlON are ceramics mainly used as hard cutting tools, in foundries as refractory compounds but they were also recently used for the design of new LED devices. They are commonly obtained by the sintering of Si3 N4 and Al2 O3 powders at high temperature, the addition of a sintering additive such as Yb2 O3 resulting in a denser material. The SiAlON adopt the crystalline structures of Si3 N4 with the substitution of nitrogen atoms by oxygen ones in the anionic lattice whereas silicon atoms are replaced with aluminium ones in the cationic lattice. The resulting material is thus composed of sub-micronic phases: ␣-SiAlON, -SiAlON and a refractory intergranular glass. Controlling the substitution yield between anions and cations in the different phases as well as the proportion of these phases in the ceramics is of prime interest to confer a wide range of variable physical or chemical properties to these materials such as high strength, toughness, high hardness, high temperature corrosion and oxidation resistivity or low thermal expansion. The optimization of the synthesis condition requires the determination of the anions/cations substitution rate and thus the elemental quantification of the sub-micronic insulating phases.
2. Experimental SiAlON samples with identical nominal composition were prepared, as well as reference samples (Si3 N4 bulk, Al2 O3 bulk and Yb2 O3 powders) to accurately determine the relative sensitivity factors of these elements in the specific analytical condition used in this study. These samples were prepared in two different ways, depending on the characterization method used: the charge compensation method and the thin film method. The charge compensation method was applied to bulk SiAlON, Si3 N4 and Al2 O3 specimens, whose surface was first mechanically polished using 3 and 0.25 m diamond paste and sonicated in ethanol. The Yb2 O3 powders were pressed into a pellet in order to have a flat and rather smooth surface, similar to a bulk sample. Afterwards, all the specimens were metallized with a 10 nm gold film in order to increase the diffusion of the charges and improve the homogeneity of their distribution on the surface during the AES analyses. As the escape depth of the studied Auger electrons does not exceed a few nanometres, this coating strongly attenuates the intensity of their signal. The surface was then gently sputtered in situ with an Ar+ bombardment (1 kV, 500 nA, 1 mm × 1 mm) before the AES analysis to open an analytical window by locally removing the gold film. For the thin film method, the preparation was similar to the production of TEM samples. The Yb2 O3 powders were sonicated in ethanol for 15 min and a drop was deposited on a honeycomb
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
3
Fig. 1. Effect of the bombardment of the surface of a silicon wafer with low energy Ar+ ions. The sputtering conditions (E; I) of the beam are specified for each case. The composition of the surface is modified for 50 V and higher energy ion beams. Lower energy Ar+ ions can be used to compensate a negative charge built up without changing the composition of the sample.
copper grid with a carbon film in order to have dispersed and isolated particles present at the surface. With the size of the Yb2 O3 particles ranging from 0.5 to 5 m, the backscattered electron contribution is then more important than in thin films; however this difference was not taken into account for the quantifications. The SiAlON, Si3 N4 and Al2 O3 specimens were cut, polished and then thinned by a mechanical grinding (Dimpel Grinder Gatan) with 3 and 0.25 m diamond paste to reduce their thickness down to around 20 m. A precision ion polishing system with a high milling rate (PIPS, Gatan) was then used (3.5 kV, incident angle ˛ = 8◦ ) until the perforation of the sample was reached and the incident angle was reduced to 2◦ to decrease the thickness of the sample to less than 100 nm. Both sides of the specimens were coated with a 10 nm thick gold layer. To be able to carry out AES spectra, an in situ sputtering with Ar+ ions was performed to open an analytical window as described in the previous method. During the AES analysis, the samples were hung from the edge above vacuum to avoid any contribution of Auger electrons induced by backscattered electrons coming from a sample holder. The mechanical polishing and the argon sputtering of the samples are crucial steps for the conductivity of the specimen and the ability to analyze such insulators. The polishing can result in the entrapment of contaminants or impurities and the formation of nanoscratches or small asperities. They respectively tend to decrease the conductivity [8] and generate a heterogeneous surface charge distribution. Specific care must be taken during the polishing process to avoid or limit these defects. The ion sputtering also introduces defects in the samples such as structural defects, vacancies and implanted ions but favours the diffusion of the charge carriers as it is commonly used with doped semi-conductors [7]. Thus, bombarding bulk insulators, with more energetic Ar+ ions
(3–4 keV), prior to their analysis, generally facilitates the AES spectra acquisitions. The AES characterizations were performed with a Thermo VG Microlab 350 using a field emission gun and a spherical sector analyzer. The energy of the primary beam was set to 5 kV and 2 nA for the analysis of the bulk samples in the charge compensation method, whereas a higher accelerating voltage of 25 kV and a 10 nA primary electron beam, were used to analyze the thin samples and the dispersed powders. The local removal of the gold coating was realized with a VG EX05 argon gun working at 1 kV and 500 nA for a 1 mm2 raster area. The same gun working at low energy (30–100 V and 8.5–74 nA) was used as a flood gun to compensate or decrease the charge accumulation on the insulating bulk specimen in the charge compensation method. 3. Results and discussion 3.1. The charge compensation method for bulk insulators According to Eq. (1) it is theoretically possible to reduce the specimen charging by selecting the appropriate analytical conditions: - reducing the current of the primary electrons will limit the incident charge Qi ; - the optimization of the energy E and the impact angle of the primary beam will affect both the TSEY and the penetration depth of the electrons and will thus modify Qo ; - the diffusion charge Qd can be increased by the metallization of the surface and by using a low energy positive ion beam as a
4
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
flood gun to stabilize and to uniform the charge distribution at the surface. However, the very low escape depth of the Auger electrons and the fast decrease of the Auger signal from the buried layers require the local removal of this metal coating to carry out the spectra. If the sample cannot be metalized and further sputtered because the specimen is inhomogeneous in depth or the surface has to be analyzed, the insulator can also be grounded with a metal plate with an aperture or wrapped in metal foils. The use of a flood gun is not common in AES but it can be achieved with a positive ion beam of low energy in order to compensate the excess of negative charges at the surface and by staying below the static sputtering level. The implantation of the surface with noble gas ions or heavy alkali ones prior to the analysis improves the charge neutralization and decreases the difficulties of the spectra and images acquisition. The Microlab 350 is equipped with an argon ion gun mainly dedicated to surface sputter cleaning and depth profiling. For effective charge neutralization, the ion current density should be as high as possible to compensate or decrease the surface charging but the energy of the ions must be rather low in order to avoid any sputtering of the surface. The maximum achievable ion current was determined in a Faraday cup for energies ranging from 30 V to 100 V. The maximum intensity nosedives with the energy of the ions, the gun being not equipped with a floating column. Below 20 V the current density is too low to obtain efficient charge neutralization. The lack of sputtering was checked for each voltage/current couple by irradiating the surface of a silicon wafer exposed to atmosphere and on their ability to reduce, or not, the contamination and native silicon oxide amount (Fig. 1). The composition in Si, O, C and Ar were carried out up to 3200 s, corresponding to the average acquisition time of narrow AES scans for quantifications. For a 30 V or 40 V ion beam energy, the relative concentrations of each element remain constant in time. Quite a small amount of argon is detected for the bombardment at 40 V and should correspond to adsorbed ions at the surface. At 50 V accelerating voltage, the decrease of the carbon amount highlights the sputtering of the sample and the modification of the surface composition. This carbon fall is not due to its desorption as the electron beam mainly contributes to heat the sample and that this phenomenon is not noticed for the lower ion beam accelerating voltage. For higher beam energy, a faster elimination of the contamination layer occurs, the sputtering of the native silicon oxide layer and the implantation of argon in the sample is also clearly more important. In practice, both 30 and 40 V Ar+ beams were used and can both lead to an effective charge compensation for bulk insulating materials. The size of the rastered area is also adjusted to tune the ionic current density, depending on the sample to be analyzed and the charging effects observed. 3.2. The thin film method The thin film method to analyze insulating materials consists in thinning the specimen and leaving it above vacuum or depositing it on a conducting substrate to allow most of the incident electrons to be transmitted through the sample and not to charge its surface. It is usually achieved with the FIB technique [22], TEM preparation or a microtome cut for organics and biological specimens. The increase of the incident electron beam energy leads to a higher number of electrons passing through the sample, drastically reducing the term Qi in Eq. (1). Moreover, for a given primary current intensity, the lateral resolution is improved by working at high energy with a thin film as the contribution of the backscattered created Auger electrons is insignificant. If applicable, as already discussed in the charge compensation method, the sample metallization (both sides) improves the surface charges mobility and homogeneity.
Fig. 2. Secondary electron image of the SiAlON ceramic. Three phases are identified: ␣-SiAlON (a), -SiAlON (b) and the intergranular glass phase (c).
3.3. Application to SiAlON insulating samples 3.3.1. Charge compensation method The secondary electron image of the surface of the bulk SiAlON reveals the microstructure of the ceramics (Fig. 2). Three phases can be distinguished: the -SiAlON appears as elongated and dark grains, the ␣-SiAlON is soft grey and has a round shape. The round and black grains are cross sections of -SiAlON needles. At last, the intergranular glass phase appears as white and is also the smallest one in size. The AES spectra were first carried out on an area of 400 m × 400 m to spread the electrons on a larger surface and limit the charge effects during the tuning of the acquisition parameters. Tilting the sample around 60◦ , reducing the primary electron energy to 5 kV and limiting its current to 2 nA already allowed the acquisition of an Auger spectrum and the identification of the main elements constituting the surface (Fig. 3). However, a shift of all the Auger peaks of about 110 eV towards the high kinetic energies, compared to the theoretical positions, is visible and highlights the presence of a negative residual charge in the sample. Moreover, a high tail characteristic of the emission of secondary electrons is present at the low kinetic energy. A tilt of the sample to 70◦ completely reverses the surface charge that becomes positive. The spectrum is shifted by −26 eV and an abnormal shape of the spectrum is observed for the low kinetic energies. This phenomenon results from a lack of emission of low energy electrons attracted by the positively charged surface or to a modification of their trajectories after they left the sample. Indeed, a negative or positive electric field elsewhere at the surface of the specimen, resulting from a heterogeneous charge distribution between the area submitted to the impinging electrons, the area previously sputtered with argon to remove the metallic film and the film itself, can deflect the low kinetic energy electrons which are no more detected by the analyzer. With the sample tilted at 60◦ , the addition of low energy Ar+ ions considerably changes the charge at the surface and improves the quality of the AES results. The optimization of the analysis parameters, i.e. the sample tilt angle and the Ar+ ions rastering area, was obtained by minimizing the energy shift and the FWHM of the AES spectra. A shift of the spectra of only 14 eV towards the low kinetic energies is observed and is characteristic of a surface positively charged. An additional low energy electrons flow would certainly improve this charge compensation as suggested elsewhere [23]. All the electrons of more than 30 eV are present in the spectrum and the Si LMM lines can thus be observed. As can be seen in Fig. 4, the narrow scans are neither broadened nor
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
5
Fig. 3. Evolution of the shape of the survey scan with a surface negatively charged (+110 eV shift) (a), and positively charged (−26 eV shift) (b) and (−14 eV shift) (c). The optimization of the analysis conditions and the addition of low energy Ar+ ions to limit the charge enhance the detection of the low energy electrons and allow the Si LMM peaks to be observed.
Fig. 4. Narrow scans of the N KLL, O KLL and Si KLL Auger transitions acquired with a high energy resolution (E/E = 0.1) and using the ion gun for charge compensation. The peaks are neither broadened nor split, highlighting the effectiveness and long-time stability of the charge compensation process.
split, showing that the charge state is stable for a longer acquisition time. Punctual AES analysis were performed using a focused electron beam (spot size of 30 nm measured on gold islands deposited on a conductive carbon substrate) to determine the elemental composition of the three sub-micronic phases. From the survey spectra (Fig. 5) one can already notice that the ␣-SiAlON phase contains more aluminium and ytterbium than the -SiAlON one and that the intergranular glass grains are the richest in oxygen and ytterbium. More accurate quantifications were obtained from the narrow scans. They are summarized in Table 1 and are the average of ten AES points acquired for each phase. The SiAlON phase is isomorphe to Si3 N4 with the substitution of Si atoms (resp. N atoms) by Al atoms (resp. O atoms). The relative concentration of nitrogen is equal to the oxygen one (around 5 at.%). With the same substitution rate, both in the anionic and cationic lattices, the electro neutrality is fulfilled and no additional ion is inserted in this phase. The anions/cations ratio is only 0.7 instead of 0.75 in Si3 N4 but this discrepancy is in the range of AES
quantifications. On the contrary, the aluminium amount in the ␣SiAlON is slightly higher than the oxygen one. The substitution rate in the cationic lattice (Si and Al) is more important than in the anionic one (N and O) thus Yb3+ ions (around 2 at.%) are occupying
Table 1 Elemental composition of the ceramic determined by AES and by using the charge compensation method (bulk SiAlON) and the thin film one (thinned SiAlON). The uncertainty arises from the standard deviation of the measurements. Relative composition (at.%) Si Bulk SiAlON ␣-SiAlON -SiAlON Glass phase Thinned SiAlON ␣-SiAlON -SiAlON Glass phase
Al
O
N
Yb
34.7 ± 1.0 36.3 ± 0.9 26.4 ± 0.8
7.5 ± 0.4 4.8 ± 0.3 9.5 ± 1.0
5.9 ± 0.4 5.3 ± 1.2 21.2 ± 1.6
49.9 ± 1.6 53.6 ± 1.4 33.1 ± 2.8
2 ± 0.4 0 9.8 ± 0.7
33.8 ± 0.9 36.2 ± 0.7 24.4 ± 0.8
8 ± 0.7 5.2 ± 0.3 10.1 ± 0.4
7.6 ± 0.8 6.1 ± 0.6 23.9 ± 1.0
48.9 ± 0.9 52.5 ± 1.2 31.3 ± 1.4
1.7 ± 0.8 0 10.3 ± 0.6
6
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
Fig. 5. AES survey spectra acquired on the different submicronic phases of the bulk SiAlON ceramic by optimizing the TSEY and using a low energy Ar+ beam to compensate the negative charge build-up.
Fig. 6. Scanning Auger images of Si, O, N and Yb from the bulk SiAlON sample with the charge compensation method. The lateral resolution is 70 nm. The displayed images are the normalized intensity ratio (peak − background)/background to limit the intensity variations due to the topography and to the backscattering contributions.
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
7
Fig. 7. Scanning Auger images of Si, O, N and Yb of the thinned SiAlON sample. The displayed images are the normalized intensity ratio (peak − background)/background to limit the intensity variations due to the topography and to the backscattering contributions.
interstitial holes in the latter one to satisfy valency requirements. Scanning Auger Microscopy (SAM) images of the distribution of Si, N, O and Yb were also acquired (Fig. 6). The three submicronic phases are well defined and rather easy to identify, proving that the residual charge is homogeneously distributed in the analysis area and also that it is stable for long time periods, the acquisition of the four images requiring around eight hours. The lateral resolution, determined as the width at 16–84% of the intensity amplitude of a linescan across a sharp interface between two phases, is 70 nm. These images demonstrate the feasibility of analyzing, by AES and with a high lateral resolution, bulk insulating samples by combining low energy positive ions with optimized analytical settings and surface preparation. Such good results were obtained because two specific criteria are fulfilled in this ceramic: first, the sample having been mirror polished there is no more irregularities at the surface. They usually lead to various volume excitation and surface angles, so to various TSEY at different points of the surface, resulting in a heterogeneous charge distribution. With a smooth surface, the diffusion of the charge and the efficiency of the ion flood gun are improved too. Secondly, the ceramic is composed of three different phases but their elemental composition are really similar, especially the ␣-SiAlON and -SiAlON. One could expect to have a comparable insulating behaviour for these phases. 3.3.2. Thin film method A thin insulator irradiated with electrons behaves as a conductor and AES analysis can thus be performed with a higher energetic beam. Auger images were carried out on the thinned SiAlON sample with an electron beam of 25 kV which allows the achieving of
a lateral resolution of 40 nm (Fig. 7). The punctual AES quantifications, calculated using the relative sensitivity factors corresponding to these analytical conditions, are in good agreement with the ones determined on the bulk SiAlON (Table 1). The main difference is a slightly higher silicon and nitrogen concentration observed in the glass phase of the bulk insulator. This overestimation should come from the diameter of the electron beam which is in the range of the size of the glass phase. Auger electrons coming from the neighbouring ␣-SiAlON and -SiAlON richer in Si and N are then also taken into account in the quantification. 4. Conclusions This study demonstrates the applicability of methods preventing charging in AES in realizing quantitative analyses and elemental images on highly insulating specimen such as ceramics, at high lateral resolution. It thus offers the opportunity to further investigate by AES a wide class of engineering materials and even characterize submicronic phases in insulators. As an illustration, the characterization of a SiAlON insulating ceramic was successfully performed with AES by using two different approaches: the charge compensation method for bulk insulators and the thin film method. The first one is mainly dependent on the analysis conditions and on the optimization of the intensity, energy and incident angle of the primary electron beam. It is sometimes enough to modify the TSEY of the sample and to decrease the surface charge sufficiently to obtain AES spectra. The addition of a low energy positive ion beam, even with a non-floating column gun, makes easier the charge compensation, stabilizes the
8
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
charge over a long period and improves the results. The quantification off sub-micronic phases and the acquired SAM images demonstrate that AES analyses can be performed on bulk insulators with a high lateral resolution (70 nm) with this approach. The thin film method is based on the sample preparation performed with mechanical polishing and ion sputtering as for TEM preparation, with FIB devices or ultra-microtome cutting tools for soft materials. The sample behaves as a conductor since the majority of the primary electrons are transmitted through it. Therefore, there is no more restriction to perform electron based spectroscopy. Whereas the charge compensation will be implemented for as-received bulk materials and raw surfaces that cannot be cut or prepared, the thinning of the sample makes its characterization easier but is to be considered only for specimens whose every point in the bulk is characteristic of the whole sample. Acknowledgments The authors gratefully acknowledge Mrs. A. El Moul for the preparation of the samples. This study was partially funded by the Fonds National de la Recherche, Luxembourg (FNR). References [1] A. Cros, J. Electron Spectrosc. Relat. Phenom. 59 (1992) 1–14.
[2] Y. Uwamino, T. Ishizuka, H. Yamatera, J. Electron Spectrosc. Relat. Phenom. 23 (1981) 55–62. [3] S. Kohiki, T. Ohmura, K. Kusao, J. Electron Spectrosc. Relat. Phenom. 31 (1983) 85–90. [4] J. Cazaux, J. Electron Spectrosc. Relat. Phenom. 113 (2000) 15–33. [5] C.G. Pantano, T.E. Madey, Appl. Surf. Sci. 7 (1981) 115–141. [6] J. Cazaux, Microsc. Microanal. Microstruct. 6 (1995) 345–362. [7] D.R. Baer, A.S. Lea, J.D. Geller, J.S. Hammond, L. Kover, C.J. Powell, M.P. Seah, M. Suzuki, J.F. Watts, J. Wolstenholme, J. Electron Spectrosc. Relat. Phenom. 176 (2010) 80–94, and references therein. [8] S. Hofmann, J. Electron Spectrosc. Relat. Phenom. 59 (1992) 15–32. [9] J.W. Park, J. Vac. Sci. Technol. A 15 (1997) 292–293. [10] M.P. Seah, in: John Wiley and Sons Ltd (Ed.), Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, 2nd ed., Wiley, Chichester, 1990, pp. 550–552. [11] M.P. Seah, S.J. Spencer, J. Electron Spectrosc. Relat. Phenom. 109 (2000) 291–308. [12] J. Cazaux, Nucl. Instrum. Meth. Phys. Res. B 244 (2006) 307–322. [13] J. Cazaux, Ultramicroscopy 108 (2008) 1645–1652. [14] J. Cazaux, J. Electron Spectrosc. Relat. Phenom. 176 (2010) 58–79. [15] J.J. Dubray, C.G. Pantano, M. Meloncelli, E. Bertran, J. Vac. Sci. Technol. A 9 (1991) 3012–3018. [16] B.S.B. Karunaratne, J. Mater. Synth. Process. 6 (1998) 433–439. [17] L.D. Laude, C. Ogeret, A. Jadin, K. Kolev, Appl. Surf. Sci. 127–129 (1998) 848–851. [18] R.C. McCune, R.E. Chase, E.L. Cartwright, Surf. Coat. Technol. 53 (1992) 189–197. [19] E. Paparazzo, Surf. Interface Anal. 31 (2001) 1110–1111. [20] L. Yu, D. Jin, Surf. Interface Anal. 31 (2001) 338–342. [21] L. Yu, D. Jin, Surf. Interface Anal. 31 (2001) 1112–1113. [22] S. Wannaparhun, S. Seal, K. Scammon, V. Desai, Z. Rahman, J. Phys. D: Appl. Phys. 34 (2001) 3319–3326. [23] H. Iwai, H. Namba, T. Morohashi, R.E. Negri, A. Ogata, T. Hoshi, R. Oiwa, J. Surf. Anal. 5 (1999) 161–164.
Contents lists available at SciVerse ScienceDirect
Journal of Electron Spectroscopy and Related Phenomena journal homepage: www.elsevier.com/locate/elspec
Effective approaches for realizing quantitative analyses and high lateral resolution images on highly insulating samples by Auger electron spectroscopy Jérôme Guillot ∗ , Henri-Noël Migeon Department “Science and Analysis of Materials” (SAM), Centre de Recherche Public – Gabriel Lippmann, 41 rue du Brill, L-4422 Belvaux, Luxembourg
a r t i c l e
i n f o
Article history: Received 26 October 2012 Received in revised form 13 March 2013 Accepted 21 March 2013 Available online xxx Keywords: AES Insulator Ceramic SiAlON SAM Lateral resolution
a b s t r a c t The analysis of insulating materials by Auger electron spectroscopy (AES) remains difficult to achieve because of the accumulation of charges in the sample. Residual charges trapped in the specimen can lead to numerous spectra modifications such as an energy shift of the Auger transitions, enlargement, splitting and deformation of the peaks, intense or lack of secondary and Auger electrons emission. Among the different approaches already developed to circumvent this issue, two of them were considered in this study to carry out AES experiments on a SiAlON insulating ceramic: the charge compensation method and the thin film method. For both of these methods, a systematic approach was used, combining a specific sample preparation with optimized analytical settings to mitigate the charge effects to enable quantitative analysis and high lateral resolution images. The charge compensation method, dedicated to bulk samples, is based on the control of the total secondary electron yield (TSEY) during analysis by optimizing the intensity, the energy and the incident angle of the primary electron beam. Its combination with the metallization of the surface sample and the use of low energy Ar+ ions to compensate the charges allowed the determination of the elemental composition of the three sub-micrometric phases of the ceramic. High lateral resolution (70 nm) Auger maps were also acquired, demonstrating therefore the long-time stability of the surface charge during acquisitions even for small analysis areas. A second method, consisting in thinning the sample down to less than a hundred of nanometres and analyzing it with a high energy electron beam, was implemented too. The results (quantification and imaging) are in good agreement with the analysis of the sample as a bulk. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Due to the rapid progress in downscaling, many materials with a wide range of applications became very important industrial and economic stakes. Thus, recent researches concern the study of material such as ceramics, VLSI devices, polymers, glasses or optical fibres. All these materials belong to the class of highly insulating specimens and, for a long time, electron spectroscopy techniques (i.e. Auger electron spectroscopy (AES), X-ray photoelectron spectroscopy (XPS), reflection electron energy loss spectroscopy (REELS)) were considered to be not applicable to electrically non-conductive sample structures. Indeed, in order to analyze their surface, the specimens are irradiated with X-rays or electrons, resulting in the emission of electrons that are then
∗ Corresponding author. Tel.: +352 470261524. E-mail address: [email protected] (J. Guillot). 0368-2048/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.elspec.2013.03.007
collected and counted. During this process, the previous samples are subjected to charging effects because of their insulating nature. However these analytical techniques are extremely surface sensitive and provide useful elementary, quantitative and chemical information for all elements except hydrogen and helium. Because, in XPS, the sample is irradiated with neutral X-rays, it is often possible, by using a flood electron gun if necessary, to carry out experiments in quite an easier way than for spectroscopies using a primary electron gun which contributes to the charge built up as in AES for instance. That is why XPS is one of the first techniques applied to analyze and to obtain chemical information on insulating samples, this subject being well documented [1–4]. The recent improvement of the size of the X-ray spot can be satisfactory for many problems on flat, homogeneous samples or of laterally uniform composition. However, one of the unique advantages of AES is its capability of providing compositional information with a high spatial resolution. Whereas the scale of the smallest X-ray beam in XPS is several micrometres, the nanometre scale has already been reached in AES.
2
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
With the increasing need to analyze insulating samples at small scales, new methods or approaches allowing AES experiments, need to be developed. Indeed, charging is known to be a major obstacle in Auger for the characterization of insulating materials. The main problems generally encountered are: - a high charge accumulation on the surface leading to the emission of intense secondary electron peaks with the release of excessive negative charges (for surfaces negatively charged) or on the contrary, the lack of electron emission outside the sample (for surfaces positively charged); - the peaks broadening, splitting and energy shifting making more complicated or even impossible the identification of the species and their chemical state; - the sample modification with surface and bulk damaged by the charge accumulation or maybe even the reaction of the surface [5,6]; - a non-homogeneous spatial distribution of the charge accumulation or non-constant in time. To reduce these effects that affect AES imaging and spectroscopy, it is necessary to avoid or at least to limit charge build-up by suitable experimental analysis conditions and by specific sample elaboration and preparation. Reports on the existing methods and their ease of application have already been published and are a good starting point [7,8]. Analyzing insulators is never straightforward, especially with rough surfaces [9] and each sample requires optimized preparation and analytical conditions. The charging effects in Auger result from the low diffusion of charges both on the surface and in the specimen bulk. Samples with a low resistivity behave as resistors and lead to an energy shift of the Auger spectra, on the other hand, highly resistive specimens can be modelled as capacitors. The residual charge on the sample is then expressed as: Qs = Qi − (Qo + Qd )
(1)
where Qs is the sample residual charge, Qi is the incident charge, Qo is the outgoing charge and Qd is the diffusion charge. This charge accumulation leads to an uneven distribution of charges across the surface and to differential charging, which strongly alters the kinetic energy and the direction of the escaping electrons. Although most of the time insulators are negatively charged due to the implantation of electrons from the primary electron beam during the analysis, they can also be positively charged: the key parameter being the total secondary electron yield (TSEY) defined as the ratio of the number of leaving electrons on the number of incident electrons: =
Io =1− Ii
I t
Ii
with It = Ii − Io
(2)
where Ii is the incident electrons current, Io the current of outgoing electrons (i.e. secondary, backscattered and Auger electrons), and It the total current. As described elsewhere [10,11] the TSEY depends mainly on the angle (), the energy (E) of the primary electron beam and the materials. When < 1 the surface is negatively charged whereas it will be positively charged if > 1. Most interesting are the cases when is equal to the unity (for a given a maximum of two energies E1 and E2 (E1 < E2 ) allow to achieve this). The incident electron current is then equal to the outgoing electron one and no charge should build up. This model clearly highlights the main parameters controlling the charge effects; however, it assumes a homogeneous lateral and in-depth distribution of the charges which does not correspond to an insulator submitted to an electron beam during AES analysis. The dependence of the TSEY with and E was recently discussed in details for insulators continuously irradiated by electrons and in terms of time evolution and depth distribution of the
charge [12–14]. The TSEY varies faster with time when the surface is negatively charged. Thus, analytical conditions to obtain a slightly positively charged surface have to be favoured to perform reliable AES characterizations. Furthermore, a low positive charge at the surface of the sample will re-attract the low-energy outgoing secondary electrons that can help for charge stabilization. Moreover, the penetration depth of the primary electrons has to be in the range of the maximum escape depth of the electrons to facilitate the hole – electrons recombination’s and to avoid the creation of a dipole in the sample constituted with a positively charged surface and an excess of negative charges in the bulk [12]. The purpose of this paper is to illustrate two of the common methods used to carry out AES experiments on insulating samples: the charge compensation method applied for samples in the form of bulk and the thin film method. They were applied to characterize a non-conductive SiAlON ceramic specimen. Many works dealing with the characterization of SiAlON with AES can be found in literature [15–21], but rarely with high lateral resolution. The results and the ease of their implementation will be commented. SiAlON are ceramics mainly used as hard cutting tools, in foundries as refractory compounds but they were also recently used for the design of new LED devices. They are commonly obtained by the sintering of Si3 N4 and Al2 O3 powders at high temperature, the addition of a sintering additive such as Yb2 O3 resulting in a denser material. The SiAlON adopt the crystalline structures of Si3 N4 with the substitution of nitrogen atoms by oxygen ones in the anionic lattice whereas silicon atoms are replaced with aluminium ones in the cationic lattice. The resulting material is thus composed of sub-micronic phases: ␣-SiAlON, -SiAlON and a refractory intergranular glass. Controlling the substitution yield between anions and cations in the different phases as well as the proportion of these phases in the ceramics is of prime interest to confer a wide range of variable physical or chemical properties to these materials such as high strength, toughness, high hardness, high temperature corrosion and oxidation resistivity or low thermal expansion. The optimization of the synthesis condition requires the determination of the anions/cations substitution rate and thus the elemental quantification of the sub-micronic insulating phases.
2. Experimental SiAlON samples with identical nominal composition were prepared, as well as reference samples (Si3 N4 bulk, Al2 O3 bulk and Yb2 O3 powders) to accurately determine the relative sensitivity factors of these elements in the specific analytical condition used in this study. These samples were prepared in two different ways, depending on the characterization method used: the charge compensation method and the thin film method. The charge compensation method was applied to bulk SiAlON, Si3 N4 and Al2 O3 specimens, whose surface was first mechanically polished using 3 and 0.25 m diamond paste and sonicated in ethanol. The Yb2 O3 powders were pressed into a pellet in order to have a flat and rather smooth surface, similar to a bulk sample. Afterwards, all the specimens were metallized with a 10 nm gold film in order to increase the diffusion of the charges and improve the homogeneity of their distribution on the surface during the AES analyses. As the escape depth of the studied Auger electrons does not exceed a few nanometres, this coating strongly attenuates the intensity of their signal. The surface was then gently sputtered in situ with an Ar+ bombardment (1 kV, 500 nA, 1 mm × 1 mm) before the AES analysis to open an analytical window by locally removing the gold film. For the thin film method, the preparation was similar to the production of TEM samples. The Yb2 O3 powders were sonicated in ethanol for 15 min and a drop was deposited on a honeycomb
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
3
Fig. 1. Effect of the bombardment of the surface of a silicon wafer with low energy Ar+ ions. The sputtering conditions (E; I) of the beam are specified for each case. The composition of the surface is modified for 50 V and higher energy ion beams. Lower energy Ar+ ions can be used to compensate a negative charge built up without changing the composition of the sample.
copper grid with a carbon film in order to have dispersed and isolated particles present at the surface. With the size of the Yb2 O3 particles ranging from 0.5 to 5 m, the backscattered electron contribution is then more important than in thin films; however this difference was not taken into account for the quantifications. The SiAlON, Si3 N4 and Al2 O3 specimens were cut, polished and then thinned by a mechanical grinding (Dimpel Grinder Gatan) with 3 and 0.25 m diamond paste to reduce their thickness down to around 20 m. A precision ion polishing system with a high milling rate (PIPS, Gatan) was then used (3.5 kV, incident angle ˛ = 8◦ ) until the perforation of the sample was reached and the incident angle was reduced to 2◦ to decrease the thickness of the sample to less than 100 nm. Both sides of the specimens were coated with a 10 nm thick gold layer. To be able to carry out AES spectra, an in situ sputtering with Ar+ ions was performed to open an analytical window as described in the previous method. During the AES analysis, the samples were hung from the edge above vacuum to avoid any contribution of Auger electrons induced by backscattered electrons coming from a sample holder. The mechanical polishing and the argon sputtering of the samples are crucial steps for the conductivity of the specimen and the ability to analyze such insulators. The polishing can result in the entrapment of contaminants or impurities and the formation of nanoscratches or small asperities. They respectively tend to decrease the conductivity [8] and generate a heterogeneous surface charge distribution. Specific care must be taken during the polishing process to avoid or limit these defects. The ion sputtering also introduces defects in the samples such as structural defects, vacancies and implanted ions but favours the diffusion of the charge carriers as it is commonly used with doped semi-conductors [7]. Thus, bombarding bulk insulators, with more energetic Ar+ ions
(3–4 keV), prior to their analysis, generally facilitates the AES spectra acquisitions. The AES characterizations were performed with a Thermo VG Microlab 350 using a field emission gun and a spherical sector analyzer. The energy of the primary beam was set to 5 kV and 2 nA for the analysis of the bulk samples in the charge compensation method, whereas a higher accelerating voltage of 25 kV and a 10 nA primary electron beam, were used to analyze the thin samples and the dispersed powders. The local removal of the gold coating was realized with a VG EX05 argon gun working at 1 kV and 500 nA for a 1 mm2 raster area. The same gun working at low energy (30–100 V and 8.5–74 nA) was used as a flood gun to compensate or decrease the charge accumulation on the insulating bulk specimen in the charge compensation method. 3. Results and discussion 3.1. The charge compensation method for bulk insulators According to Eq. (1) it is theoretically possible to reduce the specimen charging by selecting the appropriate analytical conditions: - reducing the current of the primary electrons will limit the incident charge Qi ; - the optimization of the energy E and the impact angle of the primary beam will affect both the TSEY and the penetration depth of the electrons and will thus modify Qo ; - the diffusion charge Qd can be increased by the metallization of the surface and by using a low energy positive ion beam as a
4
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
flood gun to stabilize and to uniform the charge distribution at the surface. However, the very low escape depth of the Auger electrons and the fast decrease of the Auger signal from the buried layers require the local removal of this metal coating to carry out the spectra. If the sample cannot be metalized and further sputtered because the specimen is inhomogeneous in depth or the surface has to be analyzed, the insulator can also be grounded with a metal plate with an aperture or wrapped in metal foils. The use of a flood gun is not common in AES but it can be achieved with a positive ion beam of low energy in order to compensate the excess of negative charges at the surface and by staying below the static sputtering level. The implantation of the surface with noble gas ions or heavy alkali ones prior to the analysis improves the charge neutralization and decreases the difficulties of the spectra and images acquisition. The Microlab 350 is equipped with an argon ion gun mainly dedicated to surface sputter cleaning and depth profiling. For effective charge neutralization, the ion current density should be as high as possible to compensate or decrease the surface charging but the energy of the ions must be rather low in order to avoid any sputtering of the surface. The maximum achievable ion current was determined in a Faraday cup for energies ranging from 30 V to 100 V. The maximum intensity nosedives with the energy of the ions, the gun being not equipped with a floating column. Below 20 V the current density is too low to obtain efficient charge neutralization. The lack of sputtering was checked for each voltage/current couple by irradiating the surface of a silicon wafer exposed to atmosphere and on their ability to reduce, or not, the contamination and native silicon oxide amount (Fig. 1). The composition in Si, O, C and Ar were carried out up to 3200 s, corresponding to the average acquisition time of narrow AES scans for quantifications. For a 30 V or 40 V ion beam energy, the relative concentrations of each element remain constant in time. Quite a small amount of argon is detected for the bombardment at 40 V and should correspond to adsorbed ions at the surface. At 50 V accelerating voltage, the decrease of the carbon amount highlights the sputtering of the sample and the modification of the surface composition. This carbon fall is not due to its desorption as the electron beam mainly contributes to heat the sample and that this phenomenon is not noticed for the lower ion beam accelerating voltage. For higher beam energy, a faster elimination of the contamination layer occurs, the sputtering of the native silicon oxide layer and the implantation of argon in the sample is also clearly more important. In practice, both 30 and 40 V Ar+ beams were used and can both lead to an effective charge compensation for bulk insulating materials. The size of the rastered area is also adjusted to tune the ionic current density, depending on the sample to be analyzed and the charging effects observed. 3.2. The thin film method The thin film method to analyze insulating materials consists in thinning the specimen and leaving it above vacuum or depositing it on a conducting substrate to allow most of the incident electrons to be transmitted through the sample and not to charge its surface. It is usually achieved with the FIB technique [22], TEM preparation or a microtome cut for organics and biological specimens. The increase of the incident electron beam energy leads to a higher number of electrons passing through the sample, drastically reducing the term Qi in Eq. (1). Moreover, for a given primary current intensity, the lateral resolution is improved by working at high energy with a thin film as the contribution of the backscattered created Auger electrons is insignificant. If applicable, as already discussed in the charge compensation method, the sample metallization (both sides) improves the surface charges mobility and homogeneity.
Fig. 2. Secondary electron image of the SiAlON ceramic. Three phases are identified: ␣-SiAlON (a), -SiAlON (b) and the intergranular glass phase (c).
3.3. Application to SiAlON insulating samples 3.3.1. Charge compensation method The secondary electron image of the surface of the bulk SiAlON reveals the microstructure of the ceramics (Fig. 2). Three phases can be distinguished: the -SiAlON appears as elongated and dark grains, the ␣-SiAlON is soft grey and has a round shape. The round and black grains are cross sections of -SiAlON needles. At last, the intergranular glass phase appears as white and is also the smallest one in size. The AES spectra were first carried out on an area of 400 m × 400 m to spread the electrons on a larger surface and limit the charge effects during the tuning of the acquisition parameters. Tilting the sample around 60◦ , reducing the primary electron energy to 5 kV and limiting its current to 2 nA already allowed the acquisition of an Auger spectrum and the identification of the main elements constituting the surface (Fig. 3). However, a shift of all the Auger peaks of about 110 eV towards the high kinetic energies, compared to the theoretical positions, is visible and highlights the presence of a negative residual charge in the sample. Moreover, a high tail characteristic of the emission of secondary electrons is present at the low kinetic energy. A tilt of the sample to 70◦ completely reverses the surface charge that becomes positive. The spectrum is shifted by −26 eV and an abnormal shape of the spectrum is observed for the low kinetic energies. This phenomenon results from a lack of emission of low energy electrons attracted by the positively charged surface or to a modification of their trajectories after they left the sample. Indeed, a negative or positive electric field elsewhere at the surface of the specimen, resulting from a heterogeneous charge distribution between the area submitted to the impinging electrons, the area previously sputtered with argon to remove the metallic film and the film itself, can deflect the low kinetic energy electrons which are no more detected by the analyzer. With the sample tilted at 60◦ , the addition of low energy Ar+ ions considerably changes the charge at the surface and improves the quality of the AES results. The optimization of the analysis parameters, i.e. the sample tilt angle and the Ar+ ions rastering area, was obtained by minimizing the energy shift and the FWHM of the AES spectra. A shift of the spectra of only 14 eV towards the low kinetic energies is observed and is characteristic of a surface positively charged. An additional low energy electrons flow would certainly improve this charge compensation as suggested elsewhere [23]. All the electrons of more than 30 eV are present in the spectrum and the Si LMM lines can thus be observed. As can be seen in Fig. 4, the narrow scans are neither broadened nor
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
5
Fig. 3. Evolution of the shape of the survey scan with a surface negatively charged (+110 eV shift) (a), and positively charged (−26 eV shift) (b) and (−14 eV shift) (c). The optimization of the analysis conditions and the addition of low energy Ar+ ions to limit the charge enhance the detection of the low energy electrons and allow the Si LMM peaks to be observed.
Fig. 4. Narrow scans of the N KLL, O KLL and Si KLL Auger transitions acquired with a high energy resolution (E/E = 0.1) and using the ion gun for charge compensation. The peaks are neither broadened nor split, highlighting the effectiveness and long-time stability of the charge compensation process.
split, showing that the charge state is stable for a longer acquisition time. Punctual AES analysis were performed using a focused electron beam (spot size of 30 nm measured on gold islands deposited on a conductive carbon substrate) to determine the elemental composition of the three sub-micronic phases. From the survey spectra (Fig. 5) one can already notice that the ␣-SiAlON phase contains more aluminium and ytterbium than the -SiAlON one and that the intergranular glass grains are the richest in oxygen and ytterbium. More accurate quantifications were obtained from the narrow scans. They are summarized in Table 1 and are the average of ten AES points acquired for each phase. The SiAlON phase is isomorphe to Si3 N4 with the substitution of Si atoms (resp. N atoms) by Al atoms (resp. O atoms). The relative concentration of nitrogen is equal to the oxygen one (around 5 at.%). With the same substitution rate, both in the anionic and cationic lattices, the electro neutrality is fulfilled and no additional ion is inserted in this phase. The anions/cations ratio is only 0.7 instead of 0.75 in Si3 N4 but this discrepancy is in the range of AES
quantifications. On the contrary, the aluminium amount in the ␣SiAlON is slightly higher than the oxygen one. The substitution rate in the cationic lattice (Si and Al) is more important than in the anionic one (N and O) thus Yb3+ ions (around 2 at.%) are occupying
Table 1 Elemental composition of the ceramic determined by AES and by using the charge compensation method (bulk SiAlON) and the thin film one (thinned SiAlON). The uncertainty arises from the standard deviation of the measurements. Relative composition (at.%) Si Bulk SiAlON ␣-SiAlON -SiAlON Glass phase Thinned SiAlON ␣-SiAlON -SiAlON Glass phase
Al
O
N
Yb
34.7 ± 1.0 36.3 ± 0.9 26.4 ± 0.8
7.5 ± 0.4 4.8 ± 0.3 9.5 ± 1.0
5.9 ± 0.4 5.3 ± 1.2 21.2 ± 1.6
49.9 ± 1.6 53.6 ± 1.4 33.1 ± 2.8
2 ± 0.4 0 9.8 ± 0.7
33.8 ± 0.9 36.2 ± 0.7 24.4 ± 0.8
8 ± 0.7 5.2 ± 0.3 10.1 ± 0.4
7.6 ± 0.8 6.1 ± 0.6 23.9 ± 1.0
48.9 ± 0.9 52.5 ± 1.2 31.3 ± 1.4
1.7 ± 0.8 0 10.3 ± 0.6
6
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
Fig. 5. AES survey spectra acquired on the different submicronic phases of the bulk SiAlON ceramic by optimizing the TSEY and using a low energy Ar+ beam to compensate the negative charge build-up.
Fig. 6. Scanning Auger images of Si, O, N and Yb from the bulk SiAlON sample with the charge compensation method. The lateral resolution is 70 nm. The displayed images are the normalized intensity ratio (peak − background)/background to limit the intensity variations due to the topography and to the backscattering contributions.
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
7
Fig. 7. Scanning Auger images of Si, O, N and Yb of the thinned SiAlON sample. The displayed images are the normalized intensity ratio (peak − background)/background to limit the intensity variations due to the topography and to the backscattering contributions.
interstitial holes in the latter one to satisfy valency requirements. Scanning Auger Microscopy (SAM) images of the distribution of Si, N, O and Yb were also acquired (Fig. 6). The three submicronic phases are well defined and rather easy to identify, proving that the residual charge is homogeneously distributed in the analysis area and also that it is stable for long time periods, the acquisition of the four images requiring around eight hours. The lateral resolution, determined as the width at 16–84% of the intensity amplitude of a linescan across a sharp interface between two phases, is 70 nm. These images demonstrate the feasibility of analyzing, by AES and with a high lateral resolution, bulk insulating samples by combining low energy positive ions with optimized analytical settings and surface preparation. Such good results were obtained because two specific criteria are fulfilled in this ceramic: first, the sample having been mirror polished there is no more irregularities at the surface. They usually lead to various volume excitation and surface angles, so to various TSEY at different points of the surface, resulting in a heterogeneous charge distribution. With a smooth surface, the diffusion of the charge and the efficiency of the ion flood gun are improved too. Secondly, the ceramic is composed of three different phases but their elemental composition are really similar, especially the ␣-SiAlON and -SiAlON. One could expect to have a comparable insulating behaviour for these phases. 3.3.2. Thin film method A thin insulator irradiated with electrons behaves as a conductor and AES analysis can thus be performed with a higher energetic beam. Auger images were carried out on the thinned SiAlON sample with an electron beam of 25 kV which allows the achieving of
a lateral resolution of 40 nm (Fig. 7). The punctual AES quantifications, calculated using the relative sensitivity factors corresponding to these analytical conditions, are in good agreement with the ones determined on the bulk SiAlON (Table 1). The main difference is a slightly higher silicon and nitrogen concentration observed in the glass phase of the bulk insulator. This overestimation should come from the diameter of the electron beam which is in the range of the size of the glass phase. Auger electrons coming from the neighbouring ␣-SiAlON and -SiAlON richer in Si and N are then also taken into account in the quantification. 4. Conclusions This study demonstrates the applicability of methods preventing charging in AES in realizing quantitative analyses and elemental images on highly insulating specimen such as ceramics, at high lateral resolution. It thus offers the opportunity to further investigate by AES a wide class of engineering materials and even characterize submicronic phases in insulators. As an illustration, the characterization of a SiAlON insulating ceramic was successfully performed with AES by using two different approaches: the charge compensation method for bulk insulators and the thin film method. The first one is mainly dependent on the analysis conditions and on the optimization of the intensity, energy and incident angle of the primary electron beam. It is sometimes enough to modify the TSEY of the sample and to decrease the surface charge sufficiently to obtain AES spectra. The addition of a low energy positive ion beam, even with a non-floating column gun, makes easier the charge compensation, stabilizes the
8
J. Guillot, H.-N. Migeon / Journal of Electron Spectroscopy and Related Phenomena 187 (2013) 1–8
charge over a long period and improves the results. The quantification off sub-micronic phases and the acquired SAM images demonstrate that AES analyses can be performed on bulk insulators with a high lateral resolution (70 nm) with this approach. The thin film method is based on the sample preparation performed with mechanical polishing and ion sputtering as for TEM preparation, with FIB devices or ultra-microtome cutting tools for soft materials. The sample behaves as a conductor since the majority of the primary electrons are transmitted through it. Therefore, there is no more restriction to perform electron based spectroscopy. Whereas the charge compensation will be implemented for as-received bulk materials and raw surfaces that cannot be cut or prepared, the thinning of the sample makes its characterization easier but is to be considered only for specimens whose every point in the bulk is characteristic of the whole sample. Acknowledgments The authors gratefully acknowledge Mrs. A. El Moul for the preparation of the samples. This study was partially funded by the Fonds National de la Recherche, Luxembourg (FNR). References [1] A. Cros, J. Electron Spectrosc. Relat. Phenom. 59 (1992) 1–14.
[2] Y. Uwamino, T. Ishizuka, H. Yamatera, J. Electron Spectrosc. Relat. Phenom. 23 (1981) 55–62. [3] S. Kohiki, T. Ohmura, K. Kusao, J. Electron Spectrosc. Relat. Phenom. 31 (1983) 85–90. [4] J. Cazaux, J. Electron Spectrosc. Relat. Phenom. 113 (2000) 15–33. [5] C.G. Pantano, T.E. Madey, Appl. Surf. Sci. 7 (1981) 115–141. [6] J. Cazaux, Microsc. Microanal. Microstruct. 6 (1995) 345–362. [7] D.R. Baer, A.S. Lea, J.D. Geller, J.S. Hammond, L. Kover, C.J. Powell, M.P. Seah, M. Suzuki, J.F. Watts, J. Wolstenholme, J. Electron Spectrosc. Relat. Phenom. 176 (2010) 80–94, and references therein. [8] S. Hofmann, J. Electron Spectrosc. Relat. Phenom. 59 (1992) 15–32. [9] J.W. Park, J. Vac. Sci. Technol. A 15 (1997) 292–293. [10] M.P. Seah, in: John Wiley and Sons Ltd (Ed.), Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, 2nd ed., Wiley, Chichester, 1990, pp. 550–552. [11] M.P. Seah, S.J. Spencer, J. Electron Spectrosc. Relat. Phenom. 109 (2000) 291–308. [12] J. Cazaux, Nucl. Instrum. Meth. Phys. Res. B 244 (2006) 307–322. [13] J. Cazaux, Ultramicroscopy 108 (2008) 1645–1652. [14] J. Cazaux, J. Electron Spectrosc. Relat. Phenom. 176 (2010) 58–79. [15] J.J. Dubray, C.G. Pantano, M. Meloncelli, E. Bertran, J. Vac. Sci. Technol. A 9 (1991) 3012–3018. [16] B.S.B. Karunaratne, J. Mater. Synth. Process. 6 (1998) 433–439. [17] L.D. Laude, C. Ogeret, A. Jadin, K. Kolev, Appl. Surf. Sci. 127–129 (1998) 848–851. [18] R.C. McCune, R.E. Chase, E.L. Cartwright, Surf. Coat. Technol. 53 (1992) 189–197. [19] E. Paparazzo, Surf. Interface Anal. 31 (2001) 1110–1111. [20] L. Yu, D. Jin, Surf. Interface Anal. 31 (2001) 338–342. [21] L. Yu, D. Jin, Surf. Interface Anal. 31 (2001) 1112–1113. [22] S. Wannaparhun, S. Seal, K. Scammon, V. Desai, Z. Rahman, J. Phys. D: Appl. Phys. 34 (2001) 3319–3326. [23] H. Iwai, H. Namba, T. Morohashi, R.E. Negri, A. Ogata, T. Hoshi, R. Oiwa, J. Surf. Anal. 5 (1999) 161–164.