Letter
Direct Imaging of 3D Atomic-Scale Dopant-Defect Clustering Processes in Ion-Implanted Silicon S. Koelling,*,†,‡,∥ O. Richard,† H. Bender,† M. Uematsu,§ A. Schulze,†,‡ G. Zschaetzsch,†,‡ M. Gilbert,† and W. Vandervorst†,‡ †
IMEC, Kapeldreef 75, 3001 Leuven, Belgium KU Leuven, IKS, Celestijnenlaan 200D, 3001 Leuven, Belgium § Faculty of Science and Technology, Keio University, 3-14-1 Hiyoshi, 223-8522 Yokohama, Japan ‡
S Supporting Information *
ABSTRACT: The fabrication of nanoscale semiconductor devices for use in future electronics, energy, and health is among others based on the precise placement of dopant atoms into the crystal lattice of semiconductors and their concurrent or subsequent electrical activation. Dopants are built into the lattice by fabrication processes like ion implantation, plasmabased doping, and thermal annealing. Throughout the fabrication processes fundamental phenomena like dopant diffusion, activation, and clustering occur concurrently with damaging and subsequently recovering the crystal lattice. These processes are described by atomic-scale mechanisms of ion−host atom interaction and have an immense impact on the electrical performance of the resulting devices. Insight in their fundamental nature is of utmost importance for optimizing the performance of nanoscale technologies. In this paper, we demonstrate direct three-dimensional imaging of boron clusters and atoms in crystal defects using field ion microscopy. Our approach allows for the first time the complete characterization of the size and crystallographic orientation of boron-decorated crystal defects. This new method opens a path to image a wide variety of dopant-cluster forms and hence to study the formation and dissolution of boron clusters in silicon on the atomic scale. KEYWORDS: Atom probe tomography, field ion microscopy, boron cluster, crystal defects
F
Typical results achievable with these methods are shown in Figures 1 and 2 and Supporting Information Figures S1 and S2. Using FIM we are able to image individual silicon (Figure 1a) and boron atoms (Figure 1c), the crystallographic planes of the silicon lattice as well as the position and orientation of the defect plane (Figure 2a) and the accumulation of boron atoms decorating this defect plane (Figure 2a,b). The presence of boron rich (defect) loops can be confirmed using APT as illustrated in Supporting Information Figure S1. The example shown in Figure 2b shows boron atoms forming a looplike structure containing about 120 boron clusters inside the crystal defect shown in Figure 2a. This concurrent identification and localization of the boron clusters and crystal defects allows for the first time to completely characterize the formation and dissolution of boron clusters in silicon on the atomic scale. We can determine the size and crystallographic orientation of the boron-decorated crystal defect loops that have been previously postulated in a number of publications.9,14 The importance of the ability to directly observe and locate subnanometer size boron clusters within the lattice and its
abricating nanoscale semiconductor devices is a key enabling technology for a variety of industries like the electronics,1 energy,2 and health3 sector. An important step in fabricating these devices is the precise placement of electrically active dopant atoms into the crystal lattice.4−6 As smaller devices are more sensitive to local variations in the dopant distribution resulting from variations in diffusion, clustering, and precipitations, it becomes much more difficult to meet the requirements in terms of device matching, and device variability for the fabrication of nanoscale MOSFETs7 and FinFETs.8 In order to limit these variabilities it becomes imperative to improve the fabrication process through a better understanding of these atomic-scale interactions as they can be identified as one of the sources of variability.5 Despite their crucial role for the electronic properties of nanoscale devices, dopant atoms have not yet been directly imaged in parallel with the crystal structure or within crystal defects. This has limited the modeling of dopant-lattice interactions to effects that could be demonstrated indirectly for example by imaging the strain dopants induce via transmission electron microscopy (TEM)9,10 or investigating the broadening of delta layers via secondary ion mass spectroscopy (SIMS).11 Here we use a novel approach to overcome this limitation based on field ion microscopy (FIM)12 and atom probe tomography (APT).13 © XXXX American Chemical Society
Received: February 3, 2013 Revised: April 17, 2013
A
dx.doi.org/10.1021/nl400447d | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 1. Demonstration of atomic scale imaging using FIM (panels a and c) and APT (panel b) on a silicon tip oriented along a ⟨110⟩ direction: panels a and b show silicon (111) planes as imaged with FIM (a) and APT (b). While both methods are able to reveal the planes, only FIM images show a high regularity of the atomic position within each plane. Panel c shows the impact of a small boron cluster on the FIM image, highlighting the straightforwardness to separate the two species.
crystal.16,17 As the desired electrical impact of the dopants requires their incorporation as a substitutional atom in the crystal, a recovery process like thermal annealing is needed to remove the crystal damage and adequately incorporate the dopants.18 In the majority of cases, this process of damaging and recovering the crystal lattice is imperfect and crystal defects are observed after the recovery process.9,10 During the recovery and any subsequent thermal processes the interactions of dopants with crystal defects result in the diffusion of dopants and formation of clusters.11,19 These processes cause deactivation of dopants in the clusters and a final dopant distribution differing from the original implant profile and can be devastating for the performance of nanoscale devices.18 Hence a fundamental understanding of the cluster and defect formation, dissolution, and their crystallographic nature is of crucial importance for further shrinking semiconductor devices. We analyzed a boron-implanted and thermally annealed silicon sample (methods and Supporting Information Figure S3) using FIM and APT to investigate the three-dimensional distribution of boron atoms in the silicon crystal. In both experiments, a standing voltage of several kilovolts is applied to a cryogenically cooled tip with an apex radius of several 10 nm. This configuration induces an electric field that enables the removal of ionized atoms from the surface of the tip20 and the projection of the ions onto a position sensitive detector.13 By back-projecting the impact positions recorded at the detector on to the tip’s surface, it is possible to reconstruct the threedimensional volume eroded during the analysis on the atomic scale.21 In APT, a laser or voltage pulse superposed to the standing field allows for identifying the ions removed from the tip surface by time-of-flight mass spectrometry and thus permits a three-dimensional mapping of the local composition on a nanometer or even subnanometer scale. This technique has previously been used to image local enrichments of impurity atoms in an iron−aluminum alloy22 or dopant atoms in silicon14,23 upon annealing. However, the spatial resolution of APT is usually limited by local deviations of the tip’s apex radius from the global radius used to reconstruct the data24 and
Figure 2. FIM images of a crystal defect in silicon (a) and the boron clusters decorating the defect (b). Panel a shows the silicon (111) planes cut by a defect plane in the (1̅11) direction marked by the red arrow. The high intensity spots at the right edge of the defect plane are the boron atoms decorating the plane. In Panel b, the entire boron decoration of this defect plane is shown. Separating the boron from the silicon atoms is achieved by filtering based on brightness.
defects cannot be overstated as their study is the key to understanding the atomic scale processes determining the introduction of dopants into nanoscale devices.15 Dopant incorporation is usually based on implantation as this enables the most precise tailoring of the resulting dopant distribution.4 During the implantation process energetic ions enter the crystalline semiconductor, relocate many host atoms, and create defects such as (self-)interstitials and vacancies in the host B
dx.doi.org/10.1021/nl400447d | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 3. Images of a boron-decorated defect loop made with FIM. (a) A part of the loop marking the regions that are exemplary shown in more detail in (b) and (c). In (b) and (c), the depth axis of the reconstruction is stretched by about a factor of 100. The wiggles in (b) and (c) are likely to be caused by the removal single boron atoms and hence allow estimating the number of atoms in each cluster.
the detector. The imaging gas can be seen as probing the local electric field on the surface of the tip. As aforementioned, the differences in the electric field needed to remove silicon or boron from the surface result in the formation of protrusions on the surface,25 as illustrated in Supporting Information Figure S4. While these protrusions limit the achievable resolution in APT,24 they can be used to identify dopant atoms on the surface in FIM. Because of the higher electric field above the protrusion (also shown in Supporting Information Figure S4) more gas atoms are ionized by the boron atoms than by the surrounding silicon atoms and hence boron atoms are significantly brighter in the FIM image as shown in Figures 1b and 2a. We can adjust the pressure of the imaging gas and the evaporation rate of the silicon matrix to obtain imaging conditions in FIM that result in a clear-cut brightness contrast between the boron and silicon atoms as demonstrated in the movie in the Supporting Information. This allows for separating of the boron from the silicon atoms simply by filtering regions of different brightness based on a threshold whereby regions of high brightness are related to boron atoms. This approach is used to highlight the boron-decorated defect loops show in Figures 2b and 3 and Supporting Information Figure S2. We validate our approach by extracting a depth profile of the brightness using the same filter and comparing it to boron depth profiles determined with APT and SIMS. The profiles are in close agreement as shown in Supporting Information Figure S3. This filtering based on brightness makes it possible to separate different atomic species in FIM and hence overcome the lack of elemental identification that usually limits its applicability. The important advantage of FIM over APT is the improved spatial resolution29,30 The advantages of FIM in imaging the positions of single atoms are illustrated in Figure 1 by comparing (111) silicon lattice planes of silicon (spaced 0.31 nm) as imaged with FIM (Figure 1a) and APT (Figure 1b). Both techniques are capable of separately imaging individual lattice-planes. However, the order of the atoms within each plane is strongly disturbed and apparently random in the APT image while the position of each individual atom with respect to its host plane is imaged highly reproducible in FIM. Note that
the thermal motions of the atoms. As dopant clusters may form protrusions on the surface25 and the laser provides thermal energy, the positions of the atoms are reconstructed with an uncertainty that depends on the material system and the temperature of the sample. In particular in APT analysis of semiconductors, where a laser is necessary to allow for a controlled evaporation of the atoms, one does not have sufficient spatial resolution lattice defects based on the displaced position of the lattice atoms. Hence it is not possible to image dopant clusters with concurrent information on their crystallographic position. In FIM, gas atoms are introduced around the tip, which may be ionized by electron transfer to the sample surface induced by the high electric field.12 The ionization rate of the gas is strongly dependent on the local electric field above the sample surface and individual atoms protruding the surface may act as single-atom ion emitters. The resulting variations in ionization of gas atoms allow imaging the position of single surface atoms by projecting the gas ions on an appropriate detector.26 The gas hence acts as an imaging medium and is usually referred to as “imaging gas”. As sample atoms may be evaporated in parallel with the imaging gas, a three-dimensional volume representing the atom distribution within the sample can be constructed by collecting a time sequence (movie; see Supporting Information) of the images from the eroding sample surface.27 Please note that the (field-) evaporated substrate atoms only contribute marginally to the image formation process as tenth or hundredth of gas atoms are generated on the tip for each ionized substrate ion.28 Similar to the data collected in APT these image sequences can be reconstructed into a 3D-volume using the standard APT reconstruction algorithms,21 whereby the time axis of the movie is converted into the depth axis of the volume. Here we demonstrate substantial progress in the imaging of the atomic scale dopant migration and clustering processes in semiconductors using FIM. As shown in Figures 2 and 3 and Supporting Information Figure S2, our approach enables us to directly image and characterize boron atoms, their clustering, and their location relative to crystal defects and crystal habit planes. In FIM, the position of one individual atom is extracted from the impact of hundreds of ions of the imaging gas hitting C
dx.doi.org/10.1021/nl400447d | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
contained between 800 and 1000 boron atoms, which is consistent with the FIM observations. In general the FIM approach is more sensitive than alternative techniques, revealing about twice as many borondecorated loops per volume as TEM and about 10 times as many looplike boron structures as APT. The higher sensitivity of the approach introduced here allows us to work with lower implantation doses in addition to imaging a wider range of defects. The unparalleled sensitivity and spatial resolution of this novel approach in principle enables us to follow the formation of boron clusters within the silicon lattice on the scale of individual lattice as well as dopant atoms. Our results demonstrate that the habit planes of dopant-decorated defects are more diverse as compared to the presently accepted dominance of (111) defects and that the boron population along the edges varies and that individual positions along the edges are not decorated with single boron atoms but with (sub)nanometer sized boron clusters. These detailed atomic scale information on the structure of boron loops are a basis for a refinement of the theories describing the defect-dopant dynamics. In summary, we have shown atomic resolution images of dopant clusters in silicon and presented the first experimental evidence for small boron cluster formation decorating crystal defects in silicon based on an innovative use of FIM. The results obtained indicate formation kinetics favoring a tight bonding of boron atoms linked to a silicon atom from a dislocation loop, which have not been discussed so far in the literature. All the results have been obtained through an innovative use of FIM and are currently not within reach of routine TEM studies due to its limited sensitivity. The availability of this atomic scale metrology provides a way to study the formation dynamics of these clusters and dopant defect interaction in a much more refined way than was feasible hitherto. The resulting insight in the dopant processes is crucial as the electronic properties of nanoscale devices are strongly affected by the distribution of a small number of dopant atoms1,18,30 and introducing dopants in a controlled way has become one of the utmost important technological steps in the future fabrication of nanoscale devices such as FinFETs5 and TFETs.31 The FIM concept introduced here is not restricted to boron in silicon but can also be adopted without changes to analyze other dopant-host cases as long as the impurity has a substantially higher evaporation field than the host atom. Even dopant-host cases with an impurity that has a lower evaporation field than the host atom can in principle be imaged. In this case however, we expect the clusters to appear as dark regions in the volume and anticipate it to be challenging to count the number of atoms contained within each cluster.
the disclike appearance of the atoms in the FIM image reflects differences in the depth and the lateral resolution of both APT and FIM. As the depth resolution is higher than the lateral resolution, each atoms position is constrained significantly more along the depth axis than in the lateral plane. Forming the image of one individual atom by the impact of hundreds of atoms hitting the detector however makes it possible to use the centroid of the impact cloud as a more accurate representation of the atoms position in the lateral plane in FIM than the position of a single impact in APT. Please note that this inherent averaging process is the key contribution to the significantly enhanced spatial resolution of FIM compare to APT, which allows resolving the matrix atoms and therefore the crystalline structure of the material. The high accuracy of imaging single atoms in the plane structure allows us to identify crystal defects by the disturbance they cause in the atomic stacking in the FIM image as shown in Figure 2a. Using the previously discussed brightness filtering, we can image the crystal defect in parallel with the boron clusters decorating the defect as shown in Figure 2b. Hence, we are able to concurrently identify and localize the boron atoms and crystal defects. This enables us to characterize the formation and dissolution of boron clusters in silicon on the atomic scale. Furthermore, we can use the crystallographic information revealed in the FIM images to determine the size and crystallographic orientation of the boron-decorated crystal defect loops as shown in Figure 2b and Supporting Information Figure S2. For the first time, we can identify the habit plane and the direction of the edges of the defect loops based on their orientation relative to imaged crystal planes. Figure 2b indicates the crystal directions and lengths of the edges of the loop residing in a (111)-plane in agreement with the TEM data shown in Supporting Information Figure S5. Supporting Information Figure S2 shows the characterization of a number of other loops residing in (111) and (211)-planes. This information demonstrates further added value of our approach in understanding boron cluster formation, since in TEM we could neither clearly identify defects in the (211)-planes nor the directions of the loop edges or their boron decoration. In addition, the spatial resolution of FIM makes it possible to image the evaporation of individual boron atoms within each cluster. While it is not possible to directly reveal the individual boron atoms in each cluster in the lateral plane, the high-depth resolution allows us to image the movement of the (centroid of) the impact cloud resulting from the evaporation of individual atoms of the cluster. This is depicted in Figure 3b,c, showing the evaporation sequence of individual boron clusters within a defect loop. The evaporation sequence of each cluster reveals steps on the order of the interatomic distances that are likely to be caused by the successive removal of atoms from the cluster during the evaporation of the ions. Using APT we can reveal tubular volumes of high boron concentration in the silicon sample14 as shown in Supporting Information Figure S1. The suggestion is made to compare the boron atoms within these volumes with the boron atoms found in the boron decorated defect loops imaged by FIM. We find that individual boron clusters within the defect loops contain between 5 and 20 atoms. As the loops imaged by FIM (Supporting Information Figure S2) contain 80 to 120 clusters, the total number of boron atoms within the defect loop can be expected to range from about 500 to 2000. After correcting for detection efficiency, the tubular structures identified in APT
■
ASSOCIATED CONTENT
S Supporting Information *
The methods section and author contributions as well as the supplementary Figures S1−S6 and a movie of a FIM measurement as mentioned in the manuscript are included. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. D
dx.doi.org/10.1021/nl400447d | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Present Address
(18) Noda, T.; Ortolland, C.; Vandervorst, W.; Vrancken, C.; Rosseel, E.; Clarysse, T.; Absil, P.; Biesemans, S.; Hoffmann, T. Laser annealed junctions: Pocket profile analysis using an atomistic kinetic monte carlo approach. 2010 Symposium on VLSI Technology (VLSIT) 2010, 73−74. (19) Uematsu, M. Simulation of High-Concentration Boron Diffusion in Silicon during Post-Implantation Annealing. Jpn. J. Appl. Phys. 1999, 38, 3433. (20) Forbes, R. G. Field evaporation theory: a review of basic ideas. Appl. Surf. Sci. 1995, 87−88, 111. (21) Bas, P.; Bostel, A.; Deconihout, B.; Blavette, D. A general protocol for the reconstruction of 3D atom probe data. Appl. Surf. Sci. 1995, 87−88, 298304. (22) Blavette, D.; Cadel, E.; Fraczkiewicz, A.; Menand, A. Threedimensional atomic-scale imaging of impurity segregation to line defects. Science 1999, 286, 2317. (23) Thompson, K.; Flaitz, P. L.; Ronsheim, P.; Larson, D. J.; Kelly, T. F. Imaging of arsenic cottrell atmospheres around silicon defects by Three-Dimensional atom probe tomography. Science 2007, 317, 1370. (24) Koelling, S.; Innocenti, N.; Hellings, G.; Gilbert, M.; Kambham, A. K.; De Meyer, K.; Vandervorst, W. Characteristics of cross-sectional Atom Probe analysis on semiconductor structures. Ultramicroscopy 2011, 111, 540. (25) Blavette, D.; Vurpillot, F.; Pareige, P.; Menand, A. A model accounting for spatial overlaps in 3D atom-probe microscopy. Ultramicroscopy 2001, 89, 145. (26) Homeier, H. H. H.; Kingham, D. R. Effects of local field variations on the contrast of a field-ion microscope. J. Phys. D: Appl. Phys. 1983, 16, L115. (27) Vurpillot, F.; Gilbert, M.; Deconihout, B. Towards the threedimensional field ion microscope. Surf. Interface Anal. 2007, 39, 273. (28) Miller, M.; Cerezo, A.; Hetherington, M.; Smith, G. D. W. Atom Probe Field Ion Microscopy; Clarendon Press: Oxford, 1996. (29) Rendulic, K. D.; Müller, E. W. Twinning of Iridium in a Field Ion Microscope. J. Appl. Phys. 1966, 37, 2593. (30) Ranganathan, S. Field Ion Microscopic Observations of Dislocation Structures at Grain Boundaries. J. Appl. Phys. 1966, 37, 4346. (31) Schulze, A.; Hantschel, T.; Eyben, P.; Verhulst, A. S.; Rooyackers, R.; Vandooren, A.; Mody, J.; Nazir, A.; Leonelli, D.; Vandervorst, W. Observation of diameter dependent carrier distribution in nanowire-based transistors. Nanotechnology 2011, 22, 185701.
∥
(S.K) Fraunhofer-CNT, Kö nigsbrücker Str. 180, 01099 Dresden, Germany.
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS IMEC acknowledges the collaboration with Cameca on the LAWATAP-system. S.K. and A.S. are supported by a Ph.D. grant of the Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-Vlaanderen).
■
REFERENCES
(1) Fuechsle, M.; Miwa, J. A.; Mahapatra, S.; Ryu, H.; Lee, S.; Warschkow, O.; Hollenberg, L. C. L.; Klimeck, G.; Simmons, M. Y. A single-atom transistor. Nat. Nanotechnol. 2012, 7, 242. (2) Oh, J.; Yuan, H. C.; Branz, H. M. An 18.2%-efficient black silicon solar cell achieved through control of carrier recombination in nanostructures. Nat. Nanotechnol. 2012, DOI: 10.1038/ nnano.2012.166. (3) Lee, J. H.; Huh, Y. M.; Jun, Y. W.; Seo, J. W.; Jang, J. T.; Song, H. T.; Kim, S.; Cho, E. J.; Yoon, H. G.; Suh, J. S.; Cheon, J. Artificially engineered magnetic nanoparticles for ultra-sensitive molecular imaging. Nat. Med. 2006, 13, 95. (4) ITRS roadmap 2009: Front end processes (FEP), 2009; url: http://www.itrs.net/Links/2009ITRS/2009Chapters_2009Tables/ 2009_FEP.pdf, accessed Feb. 28, 2011. (5) Vandervorst, W.; Everaert, J. L.; Rosseel, E.; Jurczak, M.; Hoffman, T.; Eyben, P.; Mody, J.; Zschätzsch, G.; Koelling, S.; Gilbert, M.; Poon, T.; del Agua Borniquel, J.; Foad, M.; Duffy, R.; Pawlak, B. J. Conformal Doping of FINFETs: a Fabrication and Metrology Challenge. AIP Conf. Proc. 2008, 1066, 449. (6) Perea, D. E.; Hemesath, E. R.; Schwalbach, E. J.; Lensch-Falk, J. L.; Voorhees, P. W.; Lauhon, L. J. Direct measurement of dopant distribution in an individual vapour-liquid-solid nanowire. Nat. Nanotechnol. 2009, 4, 315. (7) Markov, S.; Binjie, C.; Asenov, A. Statistical Variability in Fully Depleted SOI MOSFETs Due to Random Dopant Fluctuations in the Source and Drain Extensions. IEEE Electron Device Lett. 2012, 33, 315. (8) Leung, G.; Chui, C. Variability Impact of Random Dopant Fluctuation on Nanoscale Junctionless FinFETs. IEEE Electron Device Lett. 2012, 33, 767. (9) Cristiano, F.; Lamrani, Y.; Severac, F.; Gavelle, M.; Boninelli, S.; Cherkashin, N.; Marcelot, O.; Claverie, A.; Lerch, W.; Paul, S.; Cowern, N. Defects evolution and dopant activation anomalies in ion implanted silicon. Nucl. Instrum. Methods Phys. Res., Sect. B 2006, 253, 68. (10) Jones, K. S.; Prussin, S.; Weber, E. R. A systematic analysis of defects in ion-implanted silicon. Appl. Phys. A 1988, 45, 1. (11) Boninelli, S.; Mirabella, S.; Bruno, E.; Priolo, F.; Cristiano, F.; Claverie, A.; De Salvador, D.; Bisognin, G.; Napolitani, E. Evolution of boron-interstitial clusters in crystalline Si studied by transmission electron microscopy. Appl. Phys. Lett. 2007, 91, 031905. (12) Müller, E. W. Das Feldionenmikroskop. Z. Phys. 1951, 131, 136. (13) Blavette, D.; Bostel, A.; Sarrau, J. M.; Deconihout, B.; Menand, A. An atom probe for three-dimensional tomography. Nature 1993, 363, 4325. (14) Duguay, S.; Philippe, T.; Cristiano, F.; Blavette, D. Direct imaging of boron segregation to extended defects in silicon. Appl. Phys. Lett. 2010, 97, 242104. (15) Vandervorst, W.; et al. Semiconductor profiling with sub-nm resolution: Challenges and solutions. Appl. Surf. Sci. 2008, 255, 805. (16) Hobler, G.; Selberherr, S. Two-dimensional modeling of ion implantation induced point defects. IEEE Trans. Comp. Aided Des. 1988, 7, 174. (17) Biersack, J. Basic physical aspects of high energy implantation. Nucl. Instrum. Methods Phys. Res., Sect. B 1988, 35, 205. E
dx.doi.org/10.1021/nl400447d | Nano Lett. XXXX, XXX, XXX−XXX
