Inter-Diffusion of Iridium, Platinum, Palladium and Rhodium with Germanium Improved Materials for the Next Generation of Electronic Devices

https://doi.org/10.1595/205651318X696639 Johnson Matthey Technol. Rev., 2018, 62, (2), 211–230

www.technology.matthey.com

Inter-Diffusion of Iridium, Platinum, Palladium and Rhodium with Germanium Improved materials for the next generation of electronic devices

Adrian Habanyama* platinum, palladium and rhodium. Our approach Department of Physics, Copperbelt University, was essentially twofold. Firstly, conventional thin PO Box 21692, Jambo Drive, Riverside, Kitwe film couples were used to study the sequence 10101, Zambia; Department of Physics, of phase formation in the germanide systems. University of Cape Town, Rondebosch 7700, Conventional thin film couples were also used to South Africa identify and monitor the dominant diffusing species during the formation of some of the germanides *Email: [email protected] as these can influence the thermal stability of a device. Secondly, we observed and analysed Craig M. Comrie several aspects of the lateral diffusion reactions in iThemba LABS, National Research Foundation, these four systems, including activation energies PO Box 722, Somerset West 7129 South and diffusion mechanisms. Lateral diffusion Africa; Department of Physics, University of couples were prepared by the deposition of thick Cape Town, Rondebosch 7700, South Africa rectangular islands of one material on to thin films of another material. Rutherford backscattering spectrometry (RBS) and microprobe-Rutherford The down-scaling of nanoelectronic devices to backscattering spectrometry (µRBS) were ever smaller dimensions and greater performance used to analyse several aspects of the thin film has pushed silicon-based devices to their physical and lateral diffusion interactions respectively. limits. Much effort is currently being invested in X-ray diffraction (XRD) and scanning electron research to examine the feasibility of replacing microscopy (SEM) were also employed. Si by a higher mobility semiconductor, such as germanium, in niche high-performance metal 1. Introduction oxide semiconductor (MOS) devices. Before Ge can be adopted in industry, a suitable contact Ge has several attractive properties such as high material for the active areas of a transistor must mobility of charge carriers and very low carrier be identified. It is proposed that platinum group freeze-out temperatures (1–4). There is currently metal (pgm) germanides be used for this purpose, much research on high mobility semiconductors, in a similar manner as metal silicides are used such as Ge, with the view of using them to replace in Si technology. Implementation of Ge-based Si in niche high-performance MOS devices (5–7). technology requires a thorough understanding of Before Ge can be adopted by industry a suitable the solid-state interactions in metal/Ge systems contact material to the active areas (source, drain in order to foresee and avoid problems that may and gate) of a transistor must be identified. It is be encountered during integration. We present a proposed that pgm germanides be used for this systematic study of the solid-state interactions in purpose, similar to the manner in which metal germanide systems of four of the pgms: iridium, silicides are used in present Si technology (6, 8).

Our work investigates the solid-state interactions (23) reported a scanning tunnelling microscopy in germanide systems of four of the pgms: Ir, Pt, (STM) and spectroscopy study of the formation of Pd and Rh, in thin film and lateral diffusion couples. Pt/germanide phases on Ge (111). This study gave In the design of transistors the contact material a demonstration of the structural dependence of should be stable over a wide temperature range. electronic properties in the Pt-Ge system. Schottky Conventional thin film couples are well suited for barrier diodes have been used in many applications investigating the phase formation sequence and such as gates for metal semiconductor field-effect temperature stability of the phases of a system. We transistors (MESFET), solar cells and detectors have also used thin film couples to identify some (24–27). A reduction of the PtGe/Ge electron of the dominant diffusing species during the phase Schottky barrier height by rapid thermal diffusion formation. For device integrity it is important to of phosphorus was reported by Henkel et al. (28). identify the dominant diffusing species during the The results showed that rapid thermal diffusion formation of the respective germanides as this can from a solid diffusion doping source was effective in influence their thermal stability. reducing Schottky barrier heights of Pt germanide The samples used for studying lateral diffusion Schottky barrier diodes on Ge. Chawanda et al. reactions were composed of a thick island of one (29) investigated the change in the current-voltage material on top of a much thinner film of another (I–V) electrical properties of Pt Schottky contacts material. Upon annealing the island material on Ge (100) at different annealing temperatures. would react with the underlying film through Their results showed that the as-deposited barrier vertical diffusion, going through a sequence of heights had values that were near the band gap phases until the most island-material rich phase of Ge for Pt/Ge (100) Schottky diodes resulting in is formed. Since no further vertical reaction with good Schottky source/drain contact materials in the underlying film is possible, the most island- p-channel Ge-MESFETs for the hole injection from material rich phase may then grow laterally until source into inverted p-channel (30). Chawanda it attains a critical width, after which other phases et al. (31) also studied the electrical properties appear and grow simultaneously. This is a case of of Pd Schottky contacts with Ge (100). I–V and multiple phase formation as would be found in bulk capacitance-voltage (C–V) measurements were diffusion couples. Lateral diffusion couples thus performed under various annealing conditions. provide the transition between thin film and bulk Only one Pd germanide phase, PdGe, was formed. behaviour. A hole trap at 0.33 eV above the valence band was Since the island material is abundant for diffusion observed after annealing at 300°C. In another in the lateral diffusion couples, phase formation and study Chawanda et al. (32) investigated the change reaction kinetics can be studied to a greater extent in the electrical properties of Ir Schottky barrier than in thin film planar structures. Lateral diffusion diodes on Ge (100). Electrical characterisation of structures can be used to simulate bulk diffusion these contacts using I–V and C–V measurements couples because phase formation could extend to was performed under various annealing conditions. lengths of around 100 µm (9). In kinetic studies Thermal stability of the Ir/Ge (100) sample was of thin film planar structures the diffusion lengths observed up to an annealing temperature of are typically less than 0.5 µm. One can therefore 500°C. The results also showed that the onset study the transition from thin film to bulk diffusion temperature for agglomeration (binding of primary couple behaviour. The study of lateral diffusion particles leading to phase formation) in 20 nm Ir/ couples is particularly well suited for dealing with Ge (100) samples occurs between 600–700°C. the challenges of achieving the required lateral Gaudet et al. (7) carried out a systematic study abruptness of semiconductor junctions. Excessive of thermally induced reactions of 20 transition diffusion of the substrate element, in this case metals with Ge substrates. They monitored Ge, during germanide formation could result in metal/Ge reactions in situ during ramp anneals overgrowth and bridging in devices (10). at 3°C s–1 using time-resolved XRD and diffusion Various early techniques were developed to light scattering. They also carried out resistance study lateral diffusion couples (11–20). In later measurements. Their results showed that the pgms studies, µRBS was used (9, 21, 22). The major Pd and Pt were among the six most promising advantage of this technique is its ability to give candidates for microelectronic applications, the depth information. other candidates being nickel, cobalt, copper and Some previous work has been carried out in the iron. Ni is the most used metal for reducing contact research field of pgm/Ge junctions. Saedi et al. resistance. An example of previous research in the

212 © 2018 Johnson Matthey https://doi.org/10.1595/205651318X696639 Johnson Matthey Technol. Rev., 2018, 62, (2) area of Ni/Ge junctions is the work reported by plane in monitoring the direction of flow of atoms. A Peng et al. (33) on the I–V characteristics of Ni/ thin layer of Ti acted as an inert marker interposed Ge (100) Schottky diodes and the Ni germanide between coupling layers of metal and Ge; the Ge induced strain after subjecting the Schottky layer being at the surface of the sample. Upon contacts to rapid thermal annealing in the annealing of this structure both Ge and metal atoms temperature range of 300–600°C. Their results could have diffused across the marker at different showed that the orthorhombic structure of NiGe rates causing it to shift in position towards the induces epitaxial tensile strain on Ge substrates dominant diffusing species. The marker Ti signal due to the difference in lattice constants. They also was monitored by RBS for different annealing suggested that the increase in barrier height with times. The dominant diffusing species (DDS) during increasing annealing temperature may have been phase formation was determined by observing the due to the conduction band edge shift by the strain relative shift of the marker. after the germanidation process. The lateral diffusion couples were also prepared Hallstedt et al. (34) studied the phase by electron beam evaporation at a base pressure in transformation and sheet resistance of Ni on the low 10–5 Pa range. A thin film of one material was single crystalline SiGe(C) layers after annealing deposited first. An ordinary Si wafer with an array treatments at 360–900°C. The role of strain of 390 × 780 mm2 rectangular windows (referred relaxation or compensation in the reaction of Ni to as a Si mask), made by photolithographic on Si1–x–yGexCy layers due to Ge or carbon out- techniques and selective etching, was then brought diffusion to the underlying layer during the phase into contact with the substrate without breaking transformation was investigated. Formation of vacuum. Island material was deposited through crystalline Ni(SiGe) was complete at 400–450°C the mask to form structures consisting of metal but the thermal stability decreased rapidly with islands on Ge films and Ge islands on metal films. increased Ge amount due to agglomeration. This Figure 1 shows a schematic illustration of the thermal behaviour was shifted to higher annealing sample preparation setup. The source metal or Ge temperatures when C was incorporated. Ni(SiGeC) sample pieces were each placed in one of the three layers formed at 500–550°C after which there crucibles on a water-cooled copper hearth. The was Ge segregation to the underlying layer and positions of the crucibles could be adjusted from the C accumulation at the interface. Thanailakis outside of the electron-beam evaporation chamber et al. (35) established a relationship between as- so that each material to be evaporated could be deposited Pd/Ge (111) and Ni/Ge (111) Schottky placed in the path of the electron beam in turn. Each barrier height values, the metal work functions and crucible was shielded from the adjacent one by a the density of surface states of the Ge substrate. 2 cm high partition to prevent cross-contamination during deposition. Above the crucibles was a shutter 2. Experimental that could be opened and closed from outside the chamber using a bar magnet. The sample changer, A study of the phase formation sequence, using which accommodated a maximum of three sample conventional thin film couples, was carried out prior holders, hovered above the shutter. The sample to the lateral diffusion study. This study was carried changer could be rotated using an external handle out using RBS and XRD for phase identification. in such a way as to place one sample holder at a Thermally oxidised single crystal Si wafers with a time in the line of sight of the target vapour. In (100) crystal orientation were used as substrates this way, it was possible to prepare up to three in all studies. In the studies of the Ir-Ge system, a sets of different samples (each on one sample thin layer of titanium (2 nm) had to be deposited holder) in one experimental run. Between the onto the SiO2 prior to the deposition of the coupling shutter and the sample changer hung the Si mask materials; this Ti layer reacted with SiO2 forming a holder which had a provision where a Si wafer ‘glue’ without which the structure could not adhere. with several rectangular 390 × 780 mm2 openings Electron beam vacuum deposition of coupling could be placed. The mask holder could be swung layers of metal and Ge was carried out at pressures from side to side without breaking the vacuum. in the low 10–5 Pa range. Likewise, its height could be adjusted externally. In a further preliminary investigation a marker The main features of the bottom compartment of technique was used to monitor atomic mobility the chamber were a system of vacuum pumps and during phase formation. The term marker refers to a cryopanel. A baffle valve is closed between the a material deposited in the sample as a reference upper and lower compartments to ensure that the

Mask holder rotation handle

Sample holder rotation handle

Sample holder Sample changer Thickness monitor Samples Shutter Position of mask

Aperture

Crucibles Leak valve

Cryopanel Turbo pump

Ti-sublimation VacIon pump pump

Fig. 1. Schematic illustration of the high-vacuum system used for electron-beam evaporation of thin films and lateral diffusion couples. The upper section contains the sample holder, thickness monitor and electron gun. A baffle valve is closed between the upper and lower sections to ensure that the lower section of the system is maintained under vacuum when the system is not in use or during sample changing

lower section of the system is maintained under were recorded along with position information. vacuum when the system is not in use or during Typically, 128 × 128 spectra were generated in sample changing. each run and once it had been established that After removing the samples with islands from no variation in composition was observed parallel the evaporation chamber they were cleaved into to the interface, spectra were summed along this twelve identical samples, each with two or three direction. This reduced the analysis to a one- islands. Furnace-annealing was used to activate dimensional traverse of 128 spectra perpendicular solid-state interaction after which the samples to the interface, thereby optimally exploiting beam were analysed. The samples were examined using position while improving on statistics. The RBS SEM to distinguish the various reaction regions and data was analysed using RUMP (36) simulations. measure their diffusion lengths. Representative samples were selected for further analysis by 3. Results mRBS. The distribution of elements as a function of lateral position was obtained using mRBS. This 3.1 Iridium-Germanium System technique also provided information regarding 3.1.1 Thin Film Couples the elemental distribution as a function of depth and the thickness of the films. A 2 MeV a-beam Ir-Ge conventional thin film couple samples for focused down to a pixel size of about 1 × 1 µm was the study of the phase formation sequence had scanned across a well-defined area of the samples. the structure: SiO2/Ge (550 nm)/Ir (90 nm). The This area, typically of 400 × 400 mm2, was chosen samples were annealed in vacuum for various to include all reaction regions observed in the SEM periods of time at temperatures ranging from 350°C micrographs. Sample orientation was adjusted in to 800°C. The results showed the appearance such a way that the interfaces of the regions of of the compounds IrGe and Ir4Ge5 at annealing interest lay horizontally in line with the original temperatures of around 350°C. Observation of the island edge so that the microprobe beam scanned separate formation of either compound as a first parallel to the original island interface. RBS spectra phase could not be achieved. An XRD spectrum

214 © 2018 Johnson Matthey https://doi.org/10.1595/205651318X696639 Johnson Matthey Technol. Rev., 2018, 62, (2) obtained following annealing at 400°C for 80 minutes is shown in Figure 2. Energy, MeV 1.4 1.6 1.8 250 The figure shows the presence of IrGe and Ir4Ge5 together with unreacted Ir indicating co-existence.

After these two phases, Ir3Ge7 formed; from our 200 Ti data it was not possible to tell whether all the Ir –2 –2 –2 2 –2 was consumed before Ir3Ge7 appeared. The phase 5

150 cm Ir SiO IrGe is the most Ge rich in the Ir-Ge system, it was Ge

4 4 Si <> IrGe Ge 180 cm the final phase observed and only formed above 150 cm Ir cm 35 275 800°C. There was no evidence of the presence of 100 Ti

Ir3Ge4 during the whole reaction. Normalised yield A 1.2 nm layer of Ti acted as an inert marker 50 interposed between coupling layers of Ir and Ge to monitor atomic mobility during phase formation. 0 This structure was annealed at 400°C for different 300 350 400 450 500 times. Figure 3 is a diagram showing the results Channel obtained from a RUMP simulation (36) on an RBS Fig. 3. A diagram showing the results obtained spectrum of a sample annealed at 400°C for 20 from a RUMP simulation on an RBS spectrum minutes. The results show the Ti marker as lying of a sample annealed at 400°C for 20 minutes. between the phase Ir4Ge5 and the unreacted Ge at The inset shows the result of the simulation with 15 –2 the surface. thickness in units of 10 cm Figure 4 shows the Ti ‘glue’ and Ti marker signals before and after annealing. The Ti marker signal shifted by 0.02 MeV to higher energies, i.e. towards the surface. From these results a conclusion can be Energy, MeV drawn if certain assumptions are made. Firstly, if 1.25 1.30 1.35 1.40 1.45 1.50 1.5 IrGe was the first phase to form, then Ge would be As deposited the sole moving species for both IrGe and Ir4Ge5 400ºC, 20 min formation. This is so because had Ir been a moving A species during either IrGe or Ir4Ge5 formation, it 1.0 B would have been observed to diffuse across the marker, which was not the case. Secondly, if Ir4Ge5 ‘Glue’ Ti were the first phase to form, then all we canbe sure of is that Ge was the sole moving species 0.5 Marker Ti Normalised yield

300

Si 0 250 Ge 300 320 340 360 380 Ir Channel IrGe 200 Ir4Ge5 Fig. 4. Ti ‘glue’ and Ti marker signals before and after annealing. The marker signal shifted by 0.02 150 MeV to higher energies (A to B), while the Ti ‘glue’ signal did not shift 100 Yield, arbitrary units Yield, arbitrary

50 during Ir4Ge5 formation. During IrGe formation 0 there would be no discernible indicator as to which 0 20 40 60 80 species was moving, the marker being at the 2q, º interface between Ir4Ge5 and Ge. The only firm Fig. 2. X-ray diffraction spectrum of a sample of conclusion we can draw from the marker results is composition SiO2/Ge (550 nm)/Ir (90 nm) after that Ge was the sole diffusing species during Ir4Ge5 annealing at 400°C for 80 minutes formation.

3.1.2 Lateral Diffusion Couples Energy, MeV 1.2 1.4 1.6 1.8 2.0 The lateral diffusion couple results with Ir islands on Ge films showed little germanide growth Ir upon annealing with a reaction region that was 2.0 Region A (Ge island) too narrow to properly resolve and monitor. The Region B Ir4Ge5 reverse configuration with Ge islands on Ir films 1.5 Region C Region D showed substantial lateral diffusion. Figure 5 Unreacted Ir film Ir3Ge7 shows an SEM micrograph of part of a lateral 1.0 IrGe4 diffusion couple, with a Ge island (250 nm) on an Ge

Normalised yield Ir Ir film (25 nm), which had been annealed at 800°C for 30 minutes. 0.5 Four different reaction regions were observed and are labelled A to D in Figure 5. The original edge 0 of the island is clearly visible as the right edge of 300 350 400 450 500 the white vertical strip in region C. This sample Channel then underwent mRBS in order to determine the Fig. 6. Superposition of selected RBS spectra composition and thickness of the various regions. from each of the reaction regions and from the An area to be scanned was chosen to include all unreacted Ir film. Ir peak heights of the various regions observed. phases and surface positions of Ge and Ir are A representative RBS spectrum picked from each indicated of the four reaction regions labelled A to D is shown in Figure 6. A spectrum picked from the unreacted

8 Original interface Ir film is also included in Figure 6. Peak heights of the spectra taken from regions D, C and B show the phases Ir

Ge , Ir Ge and IrGe respectively. x 4 5 3 7 4 6 The germanides in these three regions are seen Ge + at the surface position. On the other hand, the Ir IrGe4 IrGe4 peak from region A lies below the surface position, 4 indicating that there was no Ir at the surface. Region A therefore consisted of a layer of unreacted Ir3Ge7 Ge on top of the IrGe4 phase. Composition, IrGe 2 Decomposition Ir Ge The RBS data from the scanned area were analysed 4 5 to get stoichiometric information from integrated counts of the Ir and Ge peaks. Figure 7 shows 0 0 100 200 300 400 Lateral position, mm

Fig. 7. Stoichiometric information of a Ge island (250 nm) on an Ir film (25 nm) annealed at 800°C for 30 minutes, derived from integrated counts of the Ir and Ge peaks as a function of position

A B C D the results as a function of lateral position. The

figure shows a region of the phase Ir3Ge7 inside the

original island interface. The width of this Ir3Ge7 region was observed to increase with annealing time. This suggests a gradual decomposition of the

phase IrGe4 into Ir3Ge7 by the reaction 3IrGe4 → 30 mm Ir3Ge7 + 5Ge. As in the thin film work, the major difference Fig. 5. SEM micrograph of part of a Ge island (250 between the lateral diffusion couples prepared by nm) on an Ir film (25 nm) annealed at 800°C annealing at temperatures below 800°C and those for 30 minutes showing all the different reaction at and above 800°C was the absence of the IrGe regions observed 4 phase below 800°C; hence no region of IrGe4

Si 3.2 Platinum-Germanium System 250 Ge Pt3Ge2 3.2.1 Thin Film Couples 200

Pt-Ge conventional thin film couple samples for 150 the study of the phase formation sequence had the structure: SiO2/Ge (500 nm)/Pt (120 nm) 100 15 –2 or equivalently, SiO2/Ge (2270 × 10 cm )/ Yield, arbitrary units Yield, arbitrary 15 –2 Pt (780 × 10 cm ). These were annealed in 50 vacuum for various periods of time at temperatures ranging from 150°C to 300°C. The results showed 0 0 20 40 60 80 the appearance of Pt Ge as the first phase formed 2 2q, º at an annealing temperature of around 190°C. Figure 8 shows an RBS spectrum obtained Fig. 9. X-ray diffraction spectrum of a sample of following annealing at 190°C for 2 hours, together composition SiO2/Ge (500 nm)/Pt (120 nm) after annealing at 220°C for 80 minutes. The non- with its RUMP simulation. The inset in Figure 8 congruent phase Pt3Ge2 is observed to form second shows that at this stage of the reaction the sample consisted of Si covered with SiO2 as the substrate, followed by Ge, then Pt2Ge and some unreacted our data it was not possible to tell whether all the

Pt at the surface. The results shown in Figure 8 Pt was consumed before Pt3Ge2 appeared. The were consistent with results obtained from the XRD next detected phase was PtGe at 250°C. The last analysis performed on this same sample. phase observed was PtGe2 which formed from the

Pt3Ge2 was observed as the second phase to interaction of PtGe with unreacted Ge. form at 220°C. Figure 9 shows an XRD spectrum A 1.2 nm layer of Ti acted as an inert marker of a sample annealed at 220°C for 80 minutes interposed between coupling layers of Pt (300 × 1015 –2 15 –2 indicating the presence of Pt3Ge2. The result was cm ) and Ge (430 × 10 cm ) to monitor atomic consistent with the result of a RUMP simulation on mobility during phase formation. Figure 10 shows RBS data obtained from the same sample. From

Energy, MeV Energy, MeV 1.0 1.2 1.4 1.6 1.8 1.2 1.4 1.6 1.8 300 250 Ti –2

–2 –2 –2

250 2 –2 –2 –2 2 –2 200 cm Pt Ge SiO 2 cm Ge Ge Ge SiO Si <> cm 70 Ge 2 Pt Pt 110 cm 2 Si <> 255 200 cm 55 Pt PtGe 235 cm 2175 cm 2175 600 Pt 275 cm 150 150 100 100 Normalised yield Normalised yield 50 50

0 0 250 300 350 400 450 500 200 250 300 350 400 450 500 Channel Channel Fig. 8. RBS spectrum of a sample of composition Fig. 10. RBS spectrum and RUMP simulation of 15 –2 15 –2 a marker structure of composition SiO / Pt (300 SiO2/Ge (2270 × 10 cm )/Pt (780 × 10 cm ) 2 15 –2 15 –2 after annealing at 190°C for 2 hours, together with × 10 cm )/Ti (1.2 nm)/Ge (430 × 10 cm ) its RUMP simulation. The inset shows the result of after annealing at 250°C for 20 minutes. The inset the simulation with thickness in units of 1015 cm–2. shows the result of the RUMP simulation with thickness in units of 1015 cm–2 The phase Pt2Ge is observed to form first

217 © 2018 Johnson Matthey https://doi.org/10.1595/205651318X696639 Johnson Matthey Technol. Rev., 2018, 62, (2) the results obtained from a RUMP simulation on an It is therefore unlikely for the first process to RBS spectrum of a sample annealed at 250°C for have taken place above the marker. The second 20 minutes. The inset shows the Ti marker as lying reaction is the one most likely to have taken place between two layers of Pt2Ge. We observe some above the marker. The PtGe was therefore formed unreacted Ge at the surface and some unreacted from Pt2Ge and Ge above the marker where there Pt below the marker. The results show a significant was no discernible indicator as to which species 15 –2 amount (235 × 10 cm ) of PtGe above the was moving during its growth. The phase Pt3Ge2 marker, between Pt2Ge and unreacted Ge. There is was not observed to form in the presence of the no PtGe below the marker. Ti marker; PtGe was formed after Pt2Ge skipping

Figure 11 shows the Ti marker signals before Pt3Ge2, which was observed in the absence of the and after annealing at 250°C for 20 minutes. The Ti marker during the study of the phase formation

Ti marker signal shifted from A to B by 0.025 MeV sequence using the sample structure: SiO2/Ge to lower energies, i.e. towards the substrate. It is (500 nm)/Pt (120 nm). clear that both Pt and Ge atoms migrated across the marker to interact, forming Pt2Ge on either 3.2.2 Lateral Diffusion Couples side of the marker. The amount of Pt that crossed the marker and was observed above it, in units of The lateral diffusion couple results for Pt islands 1015 cm–2, is shown in Equation (i): on Ge films showed little lateral diffusion upon annealing with a reaction region that was too # Pt = (2/3) × (55) (from Pt2Ge) + (1/2) × narrow to resolve and monitor properly. The (235) (from PtGe) = 154 (i) reverse configuration with Ge islands on Pt films The amount of Ge that crossed the marker and showed substantial lateral germanide growth. was observed below it is shown in Equation (ii): Figure 12 shows an SEM micrograph of part of a lateral diffusion couple of a Ge island (145 nm) on # Ge = (1/3) × (110) (from Pt2Ge) = 37 (ii) a Pt film (35 nm) annealed at 450°C for 24 hours. The atomic diffusion ratio of Pt to Ge is therefore The figure shows a part of the island that is close about 4 to 1. to one of its corners. Five different regions, labelled There are two possible mechanisms by which the A to E, are observed where A starts in the middle PtGe above the marker could have been formed. of the island and E is the outermost region. The

These are: Pt2Ge → PtGe + Pt and Pt2Ge + Ge → original edge of the island is clearly visible between 2PtGe. No PtGe was seen below the marker; the regions C and D. first of these two processes did not take place in The sample in Figure 12 underwent mRBS analysis that region as it would have left some PtGe there. to obtain stoichiometric information by RUMP simulation of the Pt and Ge peaks. The scanned

Energy, MeV 1.25 1.30 1.35 1.40 1.45 1.50 1.5

B A 1.0

0.5 C E Normalised yield A B D

30 mm

0 300 320 340 360 380 Channel Fig. 12. SEM micrograph of a Ge island (145 nm) Fig. 11. Upon annealing at 250°C for 20 minutes, on a Pt film (35 nm) annealed at 450°C for 24 the marker Ti signal shifts by about 0.025 MeV hours, showing representative regions in a part of from channel 342 to channel 335 (A to B) the island which is close to one of the corners

218 © 2018 Johnson Matthey https://doi.org/10.1595/205651318X696639 Johnson Matthey Technol. Rev., 2018, 62, (2) area was chosen to include all regions observed in the SEM micrograph. Figure 13 shows the results Energy, MeV 1.0 1.2 1.4 1.6 1.8 as a function of lateral position. 2.0 The regions B, C, D and E are found to consist of Pt PtGe , Pt Ge , PtGe and unreacted Pt respectively. 2 2 3 Region A Region A appeared to be composed of a mixture of 1.5 Region B PtGe and unreacted Ge. To show this more clearly, 2 Region C PtGe a representative spectrum picked from each of the Region D Pt2Ge3 five regions labelled A to E is shown in Figure 14. 1.0 Region E PtGe2 Downward pointing arrows are used to indicate the surface positions of Ge and Pt. The spectrum from Normalised yield Shoulder region E shows a peak of unreacted Pt and no Ge. 0.5 Ge Pt Peak heights of the spectra taken from regions D,

C and B show the phases PtGe, Pt2Ge3 and PtGe2 respectively. The Ge in these three regions is seen 0 at the surface position. It is seen from the solid 250 300 350 400 450 Channel line in the figure that the region A consisted of Fig. 14. Superposition of selected RBS spectra from unreacted Ge and the phase PtGe . The Pt peak 2 each of the five regions. Pt peak heights of the of the solid line lies at the surface position. This various phases and the surface positions of Ge and shows that there was some Pt at the surface. The Pt are indicated ‘shoulder’ marked in the figure however indicates that there was less Pt at the surface than deeper down. Region A therefore consisted of PtGe2 at the signals. This is by virtue of their having atomic bottom while at the top there was unreacted Ge numbers that lie relatively close to each other in the together with the phase PtGe2. periodic table. At the same time, samples needed to comprise of elemental layers thick enough to 3.3 Palladium-Germanium System induce an appreciable X-ray yield. The structure used was SiO2/Ge (500 nm)/Pd (70 nm). 3.3.1 Thin Film Couples In this system, reaction was induced at relatively low temperatures. Figures 15 and 16 show RBS Of major concern while determining the best spectral data for samples annealed at temperatures thickness of our Pd-Ge sample structure was the of 100°C and 150°C for 2 hours and 80 minutes likelihood of overlap between Pd and Ge RBS respectively. Data from XRD analysis were also obtained, these are displayed alongside the corresponding RBS data. Layer thicknesses obtained 3.0

Original interface by RUMP simulations (solid lines) are shown. Rather Ge + straightforward behaviour is observed in this 2.5 PtGe2 system with the two congruent phases Pd2Ge and PtGe x 2 PdGe being the only ones observed. Pd Ge was the 2.0 2 first to form at around 100°C. Pt2Ge3 1.5

PtGe 3.3.2 Lateral Diffusion Couples 0.1 Samples for lateral diffusion study were prepared Composition, PtGe by deposition of thick Ge islands on thin Pd films. 0.5 This configuration was chosen on the basis of the Pt results observed in the Ir-Ge and Pt-Ge systems. 0 0 50 100 150 200 250 Several lateral diffusion samples were annealed Lateral position, mm at various temperatures for different lengths of time. The Pd-Ge system exhibited relatively low Fig. 13. Stoichiometric information of a Ge island temperature reaction. It was therefore necessary on a Pt film annealed at 450°C for 24 hours derived from RUMP simulation of the Pt and Ge to carry out the investigation for this system at peaks, as a function of lateral position much lower temperatures than those used for the other systems.

(a) (b) Energy, MeV 300 0.8 1.0 1.2 1.4 1.6 Si 250 Ge 100 Pd Pd2Ge Pd 200 80 240 cm–2

150 60 Pd2Ge 220 cm–2 100 40 Ge –2 Normalised yield Yield, arbitrary units Yield, arbitrary 2035 cm 50 20 SiO2 Si <> 0 0 20 40 60 80 0 2q, º 150 200 250 300 350 400 450 Channel

Fig. 15. X-ray diffraction and corresponding RBS spectrum, of a sample of composition SiO2/Ge (500 nm)/ Pd (70 nm) after annealing at 100°C for 2 hours. The inset shows the result of the RUMP simulation with 15 –2 thickness in units of 10 cm . The phase Pd2Ge is observed to form first

(a) (b) Energy, MeV 0.8 1.0 1.2 1.4 1.6 300

Si 100 150ºC, 80 min 250 Ge Simulation of Pd-Ge/Pd-Ge/Ge/Si-O/Si Pd2Ge PdGe 80 Pd2Ge 200 360 cm–2 60 150 PdGe 300 cm–2 100 40 Ge

Normalised yield –2 Yield, arbitrary units Yield, arbitrary 1860 cm

50 20 SiO2 Si <> 0 0 20 40 60 80 0 2q, º 150 200 250 300 350 400 450 Channel

Fig. 16. X-ray diffraction and corresponding RBS spectrum of the sample of composition SiO2/Ge (500 nm)/ Pd (70 nm) after annealing at 150°C for 80 minutes. The inset shows the result of the RUMP simulation 15 –2 with thickness in units of 10 cm . The phase PdGe is observed to form after Pd2Ge

Figure 17 shows an SEM micrograph of one The spectrum from region C shows a peak of representative sample with a 100 nm thick Ge unreacted Pd and no Ge. The peak height of the island on a 20 nm thick Pd film, annealed at 325°C spectrum taken from the region B shows the phase for 2 hours, showing three distinct regions labelled Pd2Ge. This phase is seen at the surface position. A to C. Areas which were chosen to include all From the solid line in the figure, it can be seen that the reaction regions observed were scanned on region A consisted of unreacted Ge and PdGe. The the nuclear microprobe for analysis by mRBS. A Pd peak of the solid line lies at the surface position, spectrum picked from each of the three regions showing that there was some Pd at the surface. of the Pd-Ge lateral diffusion sample is shown in The ‘shoulder’ marked in the figure indicates that Figure 18. there was less Pd at the surface than deeper

Energy, MeV 0.8 1.0 1.2 1.4 1.6 0.8 C Pd

Pd2Ge B 0.6 Region A (Ge island) Region B A Region C (Pd film) 0.4 PdGe Ge

Normalised yield 0.2 Shoulder Pd 30 mm

0 200 250 300 350 400 Fig. 17. SEM micrograph of a Ge island (100 nm) Channel on a Pd film (20 nm) annealed at 325°C for 2 hours showing the different reaction regions Fig. 18. Superposition of selected RBS spectra from each of the regions of the Pd-Ge lateral diffusion sample. Pd peak heights of the various phases and down. Region A therefore consisted of PdGe at the surface positions of Ge and Pd are indicated bottom while at the top there was unreacted Ge intermingled with PdGe. reported on in this study; therefore great care was 3.4 Rhodium-Germanium System taken to abate this. The sample structure used in this study was SiO2/Ge (500 nm)/Rh (60 nm). 3.4.1 Thin Film Couples Figures 19, 20 and 21 show XRD results alongside RBS spectra with RUMP simulations for

Only four equilibrium phases exist in the Rh-Ge samples with the structure SiO2/Ge (500 nm)/Rh system: Rh2Ge, Rh5Ge3, RhGe and Rh17Ge22. The (60 nm) annealed between 320°C and 400°C. Our chance of getting excessive overlap of RBS peaks RBS data strongly suggest the formation of the was greater in this system than in any other yet non-congruent phase Rh2Ge as the first phase but

(a) (b) Energy, MeV 300 0.8 1.0 1.2 1.4 1.6 Si 250 Ge 100 320ºC, 2 h Simulation of Rh-Ge/Rh-Ge/Ge/Si-O/Si RhGe

200 80 Rh2Ge 265 cm–2 150 60 RhGe 345 cm–2 100 40

Normalised yield Ge Yield, arbitrary units Yield, arbitrary –2 50 1890 cm 20 SiO2 0 Si <> 0 20 40 60 80 0 2q, º 150 200 250 300 350 400 450 Channel

Fig. 19. X-ray diffraction and corresponding RBS spectrum of a sample of composition SiO2/Ge (500 nm)/ Rh (60 nm) after annealing at 320°C for 2 hours. The inset shows the result of the RUMP simulation with thickness in units of 1015 cm–2

(a) (b) Energy, MeV 300 0.8 1.0 1.2 1.4 1.6 Si 250 Ge 100 330ºC, 2 h Simulation of Rh-Ge/Ge/Si-O/Si RhGe 200 80 –2

150 RhGe

60 740 cm

–2 100 40 Ge Normalised yield Yield, arbitrary units Yield, arbitrary 50 cm 1800 20 SiO2 Si <> 0 0 20 40 60 80 0 2q, º 150 200 250 300 350 400 450 Channel

Fig. 20. X-ray diffraction and corresponding RBS spectra of a sample of composition SiO2/Ge (500 nm)/Rh (60 nm) after annealing 330°C for 2 hours. The inset shows the result of the RUMP simulation with thickness 15 –2 in units of 10 cm . From RBS data the non-congruent phase Rh2Ge is observed to convert to RhGe, Rh2Ge X-ray peaks could not be observed

(a) (b) Energy, MeV 300 0.8 1.0 1.2 1.4 1.6 Si 250 Ge 100 400ºC, 20 min Simulation of Rh-Ge/Ge/Si-O/Si Rh17Ge22 22 200 80 –2 Ge 17 810 cm 150 60 Rh

–2

100 40 Ge Normalised yield Yield, arbitrary units Yield, arbitrary

1720 cm 1720 50 20 SiO2 Si <> 0 0 20 40 60 80 0 2q, º 150 200 250 300 350 400 450 Channel

Fig. 21. X-ray diffraction and corresponding RBS spectra of a sample of composition, SiO2/Ge (500 nm)/Rh (60 nm) after annealing at 400°C for 20 minutes. The inset shows the result of the RUMP simulation with 15 –2 thickness in units of 10 cm . The phases Rh17Ge22 is observed

there is no firm evidence of this from the X-ray 3.4.2 Lateral Diffusion Couples data. RBS data showed that the Rh2Ge started to form around 280°C and later gave way to RhGe. Samples for lateral diffusion study were prepared

Figure 21 shows that Rh17Ge22 was formed after by deposition of thick Ge islands on thin Rh films. the RhGe phase. Only RhGe and Rh17Ge22 peaks This configuration, as for the Pd-Ge lateral diffusion were observed in the X-ray results, probably samples, was chosen on the basis of the results because the Rh2Ge layer was too thin to induce a observed in the Ir-Ge and Pt-Ge systems. Several good X-ray yield. samples were annealed at various temperatures

222 © 2018 Johnson Matthey https://doi.org/10.1595/205651318X696639 Johnson Matthey Technol. Rev., 2018, 62, (2) while time was monitored in the usual way. two regions are seen at the surface position. It can Shown in Figure 22 is an SEM micrograph of one be seen from the solid line in the figure that the representative sample, with a 100 nm Ge island region A consisted of unreacted Ge and the phase on a 20 nm Rh film, annealed at 600°C for 15 Rh17Ge22. The Rh peak of the solid line lies below minutes. The figure shows four reaction regions the surface position. This shows that there was no labelled A to D. Areas which were chosen to include Rh at the surface. Region A consisted of unreacted all the reaction regions observed were scanned on Ge overlaying the phase Rh17Ge22. For very long the nuclear microprobe for analysis by mRBS. annealing times at relatively high temperatures RBS spectra picked from each of the four regions, (around 600°C and above) the phase RhGe was A, B, C and D are shown in Figure 23. The observed to slowly stretch across the original spectrum from region D shows a peak of unreacted interface into the island region. Rh and no Ge. Peak heights of the spectra taken from regions C and B show the phases RhGe and 4. Discussion Rh17Ge22 respectively. The germanides in these In this work we used conventional thin film as well as lateral diffusion couples to study the germanide systems of four pgms: Ir, Pt, Pd and Rh.

D 4.1 Iridium-Germanium System C 4.1.1 Thin Film Couples

B Using conventional thin film couples, IrGe and

Ir4Ge5 where observed to be the first phases A to form in the Ir-Ge system and co-existed at annealing temperatures of around 350°C. Ir3Ge7 formed after these two phases while the IrGe 100 mm 4 phase only appeared above 800°C. By interposing a thin layer of Ti (1.2 nm) to act Fig. 22. SEM micrograph of a Ge island (100 nm) as an inert marker between coupling layers in the on an Rh film (20 nm) annealed at 600°C for 15 Ir-Ge system, the direction of atomic mobility was minutes successfully monitored during the initial stages of the reaction. In the marker samples, IrGe and

Ir4Ge5 were again found to coexist from the first Energy, MeV stages of reaction. The movement of the marker 1.2 1.3 1.4 1.5 1.6 1.7 1.8 0.4 indicated that Ge was the sole moving species Rh during Ir4Ge5 formation. It was not certain whether Region A (Ge island) RhGe Ge was also the sole moving species during IrGe 0.3 Region B formation. Region C Rh17Ge22 Region D (Rh film) 0.2 4.1.2 Lateral Diffusion Couples Ge Rh The phases observed to form in lateral diffusion couples of the Ir-Ge system were the same as those Normalised yield 0.1 observed in the thin film study on this system with

the exception of IrGe, i.e. IrGe4, Ir3Ge7 and Ir4Ge5. 0 The phase Ir3Ge7 was seen to stretch across the 300 350 400 450 original island interface at all temperatures. As in

Channel the results of the thin film couples, the phase IrGe4 Fig. 23. Superposition of selected RBS spectra was only observed to nucleate at temperatures from each of the four regions of the Rh-Ge lateral above 800°C. diffusion sample. Rh peak heights of the various The graph in Figure 24 is a plot of the growth in phases and surface positions of Ge and Rh are width with annealing time for a sample annealed indicated at 800°C. In this figure the growth in width of

40 –27.0

Ge + Ir4Ge5 Ir IrGe4 Ir Ge –27.5 IrGe4 3 7 30 –28.0

Ir4Ge5 X b b –28.5 Ea = 1.6 ± 0.1 eV 20 lnK Xb + Xg –29.0

Anneal time, min 10 –29.5

–30.0 0 10.5 11.0 11.5 12.0 12.5 –100 –50 0 50 100 150 200 1/k T eV–1 Reaction length, mm

Fig. 24. Plot of growth width with time of anneal Fig. 25. Arrhenius plot, lnKb versus 1/kbT, showing temperature dependence of Ge diffusion rate for the phases Ir3Ge7 and Ir4Ge5, for a sample with Ge island (250 nm) on an Ir film (25 nm) annealed through Ir4Ge5, yielding an average activation energy of 1.6 ± 0.1 eV at 800°C. Xb refers to the growth width of the Ir3Ge7 region while Xg refers to that for Ir4Ge5

This was because the mechanisms at play during the Ir3Ge7 phase region is labelled as Xb while Ir3Ge7 growth were not the same at 800°C as at the width of the Ir4Ge5 region is labelled as Xg. It temperatures below. At 800°C the Ir3Ge7 observed should be pointed out that the origin in Figure 24 inside the island region was from the decomposition does not correspond to the original island interface of IrGe4 while that outside was formed by the but refers to the interface between the IrGe4 and interaction between Ir4Ge5 and the outward diffusing

Ir3Ge7 regions. The growth curve for the region Ge atoms. Below 800°C the Ir3Ge7 outside was also labelled as Ir3Ge7 was a result of Ir3Ge7 grown formed by the interaction of Ge and Ir4Ge5 but that from Ir4Ge5 and that from the decomposition of inside was observed as a result of exposure as Ge

IrGe4. The growth characteristics observed are was consumed from the source region. On the other parabolic with time. hand, Ir4Ge5 was formed by the interaction of Ge

The growth of the phases Ir3Ge7 and Ir4Ge5 were with the unreacted Ir film, both below and above monitored at temperatures of 700°C, 750°C and 800°C and thus at all three temperatures. 800°C. Different annealing times were chosen to obtain a reasonable range of growth widths at 4.2 Platinum-Germanium System each of the three temperatures. The squares of the growth widths were plotted against the times of 4.2.1 Thin Film Couples annealing at each temperature and the diffusional growth constants, Kb, were obtained from the In our work on the Pt-Ge system, Pt2Ge was the slopes. The logarithms of the diffusional growth first phase formed. The second phase observed constants were then plotted against the reciprocals was Pt3Ge2. The next phase detected was PtGe of the product, kbT, of the Boltzmann constant and the last was PtGe2. The non-congruent phase and the absolute temperatures. The resulting Pt2Ge3 was skipped between the last two phases.

Arrhenius plot for the phase Ir4Ge5 is shown in The marker technique was also applied to the Pt-Ge

Figure 25. The average activation energy, Ea, was system. Pt was found to be the dominant diffusing obtained from the slope of the straight line fit of species during Pt2Ge formation. Some Ge diffusion the Arrhenius plot. The value determined for the was also observed to take place. The atomic diffusion diffusion process in the Ir4Ge5 phase was Ea = 1.6 ratio of Pt to Ge was measured as being about 4 to 1. ± 0.1 eV. Unfortunately the results from the samples 4.2.2 Lateral Diffusion Couples annealed at the temperatures 700°C, 750°C and 800°C could not be used to obtain a value Three phases were observed to form in the lateral of the activation energy for the Ir3Ge7 phase. diffusion couples of the Pt-Ge system: PtGe2 and

Pt2Ge3 inside the original island region and PtGe the phases PtGe2, Pt2Ge3 and PtGe are shown in outside. The Pt2Ge and Pt3Ge2 phases which were Figure 27. observed in the thin film study of the system The activation energies, Ea, were obtained from were absent in the lateral diffusion couples. The the slopes of the straight line fits. The average graph in Figure 26 is a plot of the various growth activation energies determined from the lateral widths against the time of annealing for a sample growth rates of Pt2Ge3 and PtGe were 0.9 ± 0.1 annealed at 550°C. The origin in this figure does eV and 1.3 ± 0.4 eV, respectively. The activation not correspond to the original island interface but energy corresponding to the apparent lateral refers to the interface between the regions labelled growth rate of the PtGe2 region was 1.5 ± 0.2 eV. as A and B in Figure 12. The growth widths of the phase regions, PtGe2, Pt2Ge3 and PtGe are labelled 4.3 Palladium-Germanium System as Xa, Xb and Xg respectively. It must be pointed out that whereas the lateral 4.3.1 Thin Film Couples growth of the Pt2Ge3 and PtGe in the regions labelled as C and D in Figure 12 respectively are The only phases observed to form in the thin film due to reaction mechanisms, the growth of region study of the Pd-Ge system were the two congruent

B is due to exposure of PtGe2 by the consumption of phases PdGe and Pd2Ge. overlaying Ge. The growth characteristics observed for all regions are parabolic with time. 4.3.2 Lateral Diffusion Couples The lateral widths, Xa, Xb and Xg, were monitored at the temperatures 450°C, 500°C and 550°C. The two phases, PdGe and Pd2Ge, which were Different annealing times were chosen to obtain a observed to form in the thin film study of the Pd- reasonable range of growth widths at each of these Ge system were also the only two observed in the three temperatures. The squares of the growth lateral diffusion couples. The growth region outside widths were plotted against the times of annealing the original island interface, labelled as region at each temperature and the diffusional growth B in Figure 17, consisted of the phase Pd2Ge. constants, Kβ, were obtained from the slopes. The The original island region, labelled as region A in logarithms of the diffusional growth constants were Figure 17, consisted of PdGe at the bottom while then plotted against the reciprocals of the product, at the top there was unreacted Ge intermingled kbT, of the Boltzmann constant and the absolute with PdGe. There was no region of completely temperatures. The resulting Arrhenius plots for exposed PdGe without intermingled unreacted Ge, it was therefore not possible to obtain data for the growth or exposure rates of the phase PdGe. 6

Ge + 5 PtGe Pt PtGe2 2 Pt2Ge3 PtGe –27 4

–28 X 3 a PtGe X +X a b Ea =1.3±0.4eV –29 Xa +X b +X Anneal time, h g 2 b lnK –30 1 PtGe2 Pt2Ge3 Ea =1.5±0.2eV Ea =0.9±0.1eV 0 –31 –50 0 50 100 150 200 250 300 Reaction length, mm –32 13 14 15 16 17 Fig. 26. Plot of annealing time against reaction –1 1/kb T eV widths for the phases PtGe2, Pt2Ge3 and PtGe in a sample with Ge (145 nm) on Pt (35 nm) at a constant annealing temperature of 550°C. The Fig. 27. Arrhenius plot, lnKb versus 1/kbT, showing temperature dependence of the Ge diffusion rate growth widths of the phase regions, PtGe2, Pt2Ge3 through Pt2Ge3, PtGe and PtGe2 and PtGe are labelled as Xa, Xb and Xg respectively

The growth of the Pd2Ge region was monitored at determined from the plot in Figure 29 was the temperatures 275°C, 300°C and 325°C. The Ea = 1.0 ± 0.1 eV. periods of annealing were chosen so as to obtain a reasonable range of growth widths at each of 4.4 Rhodium-Germanium System the three temperatures; results are presented in Figure 28. Parabolic growth characteristics were 4.4.1 Thin Film Couples observed. Arrhenius plots were obtained from the data presented in Figure 28, in the same way as Our RBS data strongly suggested the formation of explained for the Ir-Ge and Pt-Ge lateral diffusion Rh2Ge as the first phase in the Rh-Ge system but couples. Figure 29 is an Arrhenius plot showing there was no firm evidence of this from the X-ray the temperature dependence of the Ge diffusion data because this phase was too thin to give a good rate through Pd2Ge. The average activation energy X-ray yield. This was a consequence of the fact that the samples had to be kept thin enough to avoid

excessive RBS peak overlap. The Rh2Ge phase 20 appeared to give way to RhGe while Rh17Ge22 was Ge + Pd formed as the last phase. PdGe Pd2Ge

15 4.4.2 Lateral Diffusion Couples

275ºC The phase Rh17Ge22 was observed inside the original island region in lateral diffusion couples of 10 the Rh-Ge system while RhGe grew outside in most 300ºC cases. Under certain conditions, the latter phase Anneal time, h was observed to slowly stretch across the original 5 325ºC interface into the island region. This suggested a

slow decomposition of Rh17Ge22 into RhGe. There were four distinct regions observed in the lateral 0 –10 0 10 20 30 40 50 diffusion couples of the Rh-Ge system; these Reaction length, mm are represented schematically in Figure 30. The relative position of the original island interface Fig. 28. A plot of reaction length against the time and different phases are shown. Knowledge of the of annealing for the phase Pd Ge in Ge (100 nm) 2 position at which the original interface lay was vital on Pd (20 nm) at temperatures 275°C, 300°C and 325°C in the analysis of the reactions taking place between different phase regions. This was particularly so in this system with wide reaction regions and a slight shift of the reaction interface between the –29.0 Rh17Ge22 and RhGe regions with respect to the original interface for different times of annealing. –29.5 Our RBS data from the thin film study of this system strongly suggested the formation of the –30.0 non-congruent phase Rh Ge but this phase was Pd2Ge 2 Ea =1.1±0.1eV not observed in the lateral diffusion couples. The b –30.5 growths of Rh17Ge22 and RhGe in the lateral diffusion lnK

–31.0

Original interface –31.5

Ge + –32.0 Rh17Ge22 Rh Rh17Ge22 RhGe 19.0 19.5 20.0 20.5 21.0 21.5 –1 1/kb T eV

Fig. 29. Arrhenius plot, lnKb versus 1/kbT, showing temperature dependence of Ge diffusion rate Fig. 30. Diagram showing different phase regions observed in lateral diffusion couples of the Rh-Ge through Pd2Ge, yielding an average activation energy of 1.0 ± 0.1 eV system

226 © 2018 Johnson Matthey https://doi.org/10.1595/205651318X696639 Johnson Matthey Technol. Rev., 2018, 62, (2) couples were monitored at the temperatures 450°C, 5. Summary and Conclusion 500°C and 600°C. The temperature range and annealing times were chosen in such a way that The results of our thin film study are summarised in the decomposition of Rh17Ge22 into RhGe was not Table I. Temperatures at which the first reactions significant. Results for carefully chosen annealing were observed to begin are indicated. times at 500°C are presented in Figure 31. The The lateral diffusion samples used in this study growth characteristics were observed to be parabolic. were prepared with the configuration of Ge islands Arrhenius plots obtained from the data shown in evaporated onto metal films. In all systems, the Figure 31 are presented in Figure 32, showing germanide phases were seen to spread out from the temperature dependence of the Ge diffusion the source region in decreasing order of Ge rate through RhGe and Rh17Ge22. An average content. The growth characteristics observed in all activation energy of 1.2 ± 0.3 eV was obtained for phases of the four systems studied were parabolic both Rh17Ge22 and RhGe. with time. This is indicative of a diffusion controlled

80 –25 Ge + Rh Rh17Ge22 Rh17Ge22 RhGe –26 60 RhGe

–27 Ea =1.2±0.3eV b Rh Ge 40 17 22 lnK –28 Ea =1.2±0.3eV

Anneal time, min 20 –29

–30 13 14 15 16 17 0 –1 –50 0 50 100 150 200 1/kb T eV Reaction length, mm Fig. 32. Arrhenius plots, lnKb versus 1/kbT, showing Fig. 31. Plots of annealing time against reaction temperature dependence of Ge diffusion rate length for the phases Rh Ge and RhGe in Ge through Rh17Ge22 and RhGe, yielding an average 17 22 (100 nm) on Rh (20 nm) at 500°C activation energy of 1.2 ± 0.3 eV for both phases

Table I Summary of the Phase Formation Sequence Results for the Four Systems Studied, with the Temperatures at which the First Reactions were Observed to Start

Temperature at which System Phase formation sequence first reaction begins

1st IrGe and Ir4Ge5 (co-existing) Ir-Ge 2nd Ir3Ge7 350°C 3rd IrGe4

1st Pt2Ge 2nd Pt Ge Pt-Ge 3 2 190°C 3rd PtGe 4th PtGe2 1st Pd Ge Pd-Ge 2 100°C 2nd PdGe

1st Rh2Ge Rh-Ge 2nd RhGe 280°C 3rd Rh17Ge22

227 © 2018 Johnson Matthey https://doi.org/10.1595/205651318X696639 Johnson Matthey Technol. Rev., 2018, 62, (2) process which, as modelled by Kidson (37), results insight and to make comprehensive suggestions on in parabolic growth even in multiphase systems. the ideally suited choice of pgm/Ge combinations Table II gives a summary of the various phases for particular applications in semiconductor observed in the lateral diffusion couples of each technology applications such as gates for metal system and the corresponding activation energies semiconductor field-effect transistors, solar cells obtained. and detectors.

The magnitudes of the activation energies, Ea, calculated for all phases, as shown in Table II, Acknowledgements suggest that the lateral diffusion reactions in all four systems were not driven by surface diffusion The authors wish to thank the University of but rather by diffusion through the interior of the Cape Town, the South African National Research lateral diffusion couples; typical values for surface Foundation and the University Science, Humanities diffusion being around 0.6 eV (38). and Engineering Partnerships in Africa (USHEPIA) In the current design and processing of transistors for financial assistance. They also wish to thank the the contact material should exhibit low sheet and Materials Research Group at the iThemba Labs at contact resistances, form at a low temperature Faure, South Africa, for the use of their facilities, and be stable over a wide temperature range. A Miranda Waldron in the Electron Microscope unit at previous systematic study of the thermally induced the University of Cape Town, Ms Terry Davies of the reaction of a large number of transition metals with X-ray unit in the Geological Science Department, Ge substrates revealed that NiGe and PdGe are the University of Cape Town. most promising candidates when taking the above requirements into account (7, 8). From Table I we References see that Pd germanides form at lower temperatures than any of the other pgm germanides. Pt 1. C. O. Chui, H. Kim, D. Chi, B. B. Triplett, P. C. germanides were also found to be promising McIntyre and K. C. Saraswat, ‘A Sub-400/spl deg/C candidates for microelectronic applications (7). Germanium MOSFET Technology with High-/spl There has been very little work carried out on Kappa/Dielectric and Metal Gate’, International the electrical properties of Ir/Ge junctions (32). Electron Devices Meeting, San Francisco, USA, 8th–11th December, 2002, Institute of Electrical Our survey of previous work in this research field and Electronics Engineers Inc, Piscataway, USA, showed no evidence of any systematic study of the pp. 437–440 electrical properties of Rh/Ge junctions. Our work 2. A. Ritenour, S. Yu, M. L. Lee, N. Lu, W. Bai, A. Pitera, has looked at several aspects of the interfacial E. A. Fitzgerald, D. L. Kwong and D. A. Antoniadis, phase growth and inter-diffusion kinetics at ‘Epitaxial Strained Germanium p-MOSFETs pgm/Ge junctions. However, more work needs to with HfO/sub 2/Gate Dielectric and TaN Gate be carried out regarding the electrical properties of Electrode’, International Electron Devices Meeting, these junctions. Our current results could then be Washington, DC, USA, 8th–10th December, 2003, used in conjunction with the results based on the Institute of Electrical and Electronics Engineers study of electrical properties in order to draw more Inc, Piscataway, USA, pp. 18.2.1–18.2.4

Table II Summary of the Germanide Phases Seen to Spread Out from the Source Region in Their Decreasing Order of Germanium Content During Lateral Diffusion, with Corresponding Activation Energies Obtained

System Phases observed during lateral diffusion and activation energies obtained

IrGe Ir Ge Ir Ge Ir-Ge 4 3 7 4 5 Ea = 1.6 ± 0.1 eV

PtGe2 Pt Ge PtGe Pt-Ge 2 3 Ea = 1.5 ± 0.2 eV Ea = 0.9 ± 0.1 eV Ea = 1.3 ± 0.4 eV PdGe Pd Ge Pd-Ge 2 Ea = 1.0 ± 0.1 eV Rh Ge RhGe Rh-Ge 17 22 Ea = 1.2 ± 0.3 eV Ea = 1.2 ± 0.3 eV

3. H. Shang, H. Okorn-Schmidt, K. K. Chan, M. Copel, 18. J. C. Liu, J. W. Mayer and J. C. Barbour, J. Appl. J. A. Ott, P. M. Kozlowski, S. E. Steen, S. A. Cordes, Phys., 1988, 64, (2), 651 H.-S. P. Wong, E. C. Jones and W. E. Haensch, ‘High 19. J. C. Liu, J. W. Mayer and J. C. Barbour, J. Appl. Mobility p-Channel Germanium MOSFETs with a Phys., 1988, 64, (2), 656 Thin Ge Oxynitride Gate Dielectric’, International 20. J. C. Liu and J. W. Mayer, J. Mater. Res., 1990, 5, Electron Devices Meeting, San Francisco, USA, (2), 334 8th–11th December, 2002, Institute of Electrical and Electronics Engineers Inc, Piscataway, USA, 21. P. J. Ding, R. Talevi, W. A. Lanford, S. Hymes and pp. 441–444 S. P. Murarka, Nucl. Instr. Meth. Phys. Res. Sect. B: Beam Int. Mater. Atoms, 1994, (1–4), 85, 167 4. C. O. Chui, S. Ramanathan, B. B. Triplett, P. C. 22. H. S. de Waal, “The Effect of Diffusion Barriers, McIntyre and K. C. Saraswat, IEEE Electron Dev. Stress and Lateral Diffusion on Thin-Film Phase Lett., 2002, 23, (8), 473 Formation”, PhD thesis, University of Stellenbosch, 5. C. Claeys and E. Simoen, “Germanium-Based South Africa, 1999 Technologies: From Materials to Devices”, Elsevier 23. A. Saedi, B. Poelsema and H. J. W. Zandvliet, Surf. BV, Oxford, UK, 2007, 449 pp Sci., 2011, 605, (5–6), 507 6. D. P. Brunco, B. De Jaeger, G. Eneman, J. Mitard, 24. E. Hökelek and G. Y. Robinson, Solid State G. Hellings, A. Satta, V. Terzieva, L. Souriau, F. E. Electron., 1981, 24, (2), 99 Leys, G. Pourtois, M. Houssa, G. Winderickx, E. Vrancken, S. Sioncke, K. Opsomer, G. Nicholas, 25. E. H. Rhoderick and R. H. Williams, “Metal- M. Caymax, A. Stesmans, J. Van Steenbergen, Semiconductor Contacts”, 2nd Edn., Monographs P. W. Mertens, M. Meuris and M. M. Heyns, J. in Electrical and Electronic Engineering, Vol. 19, Electrochem. Soc., 2008, 155, (7), H552 Clarendon Press, Oxford, UK, 1988, 252 pp 26. G. A. Baraff and M. Schlüter, Phys. Rev. B, 1986, 7. S. Gaudet, C. Detavernier, A. J. Kellock, P. 33, (10–15), 7346 Desjardins and C. Lavoie, J. Vac. Sci. Technol. A, 2006, 24, (3), 474 27. S. Asubay, Ö. Güllü and A. Türüt, Appl. Surf. Sci., 2008, 254, (11), 3558 8. J. A. Kittl, K. Opsomer, C. Torregiani, C. Demeurisse, S. Mertens, D. P. Brunco, M. J. H. Van 28. C. Henkel, S. Abermann, O. Bethge, G. Pozzovivo, Dal and A. Lauwers, Mater. Sci. Eng.: B, 2008, S. Puchner, H. Hutter and E. Bertagnolli, J. 154–155, 144 Electrochem. Soc., 2010, 157, (8), H815 29. A. Chawanda, C. Nyamhere, F. D. Auret, W. 9. R. S. Nemutudi, C. M. Comrie and C. L. Churms, Mtangi, M. Diale and J. M. Nel, J. Alloys Compd., Thin Solid Films, 2000, 358, (1–2), 270 2010, 492, (1–2), 649 10. S.-L. Zhang and M. Östling, Crit. Rev. Solid State 30. H. B. Yao, C. C. Tan, S. L. Liew, C. T. Chua, C. K. Mater. Sci., 2003, 28, (1), 1 Chua, R. Li, R. T. P. Lee, S. J. Lee and D. Z. Chi, 11. L. R. Zheng, L. S. Hung, J. W. Mayer, G. Majni and ‘Material and Electrical Characterization of Ni- and G. Ottaviani, Appl. Phys. Lett., 1982, 41, (7), 646 Pt-Germanides for p-channel Germanium Schottky 12. L. R. Zheng, L. S. Hung and J. W. Mayer, J. Vac. Source/Drain Transistors’, Sixth International Sci. Technol. A, 1983, 1, (2), 758 Workshop on Junction Technology, Shanghai, China, 15th–16th May, 2006, Institute of Electrical 13. L. R. Zheng, L. S. Hung and J. W. Mayer, Thin Solid and Electronics Engineers Inc, Piscataway, USA, Films, 1983, 104, (1–2), 207 pp. 164–169 14. S. H. Chen, L. R. Zheng, J. C. Barbour, E. C. Zingu, 31. A. Chawanda, C. Nyamhere, F. D. Auret, W. Mtangi, L. S. Hung, C. B. Carter and J. W. Mayer, Mater. T. T. Hlatshwayo, M. Diale and J. M. Nel, Physica B, Lett., 1984, 2, (6), 469 2009, 404, (22), 4482 15. B. Blanpain, J. W. Mayer, J. C. Liu and K. N. Tu, J. 32. A. Chawanda, S. M. M. Coelho, F. D. Auret, W. Appl. Phys., 1990, 68, (7), 3259 Mtangi, C. Nyamhere, J. M. Nel and M. Diale, J. 16. B. Blanpain, J. W. Mayer, J. C. Liu and K. N. Tu, Alloys Compd., 2012, 513, 44 Phys. Rev. Lett., 1990, 64, (22–28), 2671 33. C.-Y. Peng, Y.-H. Yang, C.-M. Lin, Y.-J. Yang, C. 17. B. Blanpain, ‘Lateral Diffusion Couples and F. Huang and C. W. Liu, ‘Process Strain Induced Their Contribution to Understanding Thin Film by Nickel Germanide on (100) Ge Substrate’, Reactions’, in “Crucial Issues in Semiconductor 9th International Conference on Solid-State and Materials and Processing Technologies”, eds. Integrated-Circuit Technology, Beijing, China, S. Coffa, F. Priolo, E. Rimini and J. M. Poate, 20th–23rd October, 2008, Institute of Electrical Springer Science+Business Media, Dordrecht, The and Electronics Engineers Inc, Piscataway, USA, Netherlands, 1992, pp 421–425 pp. 681–683

34. J. Hållstedt, M. Blomqvist, P. O. Å. Persson, L. 36. L. R. Doolittle, Nucl. Instr. Meth. Phys. Res. Sect. Hultman and H. H. Radamson, J. Appl. Phys., B: Beam Int. Mater. Atoms, 1986, 15, (1–6), 227 2004, 95, (5), 2397 37. G. V. Kidson, J. Nucl. Mater., 1961, 3, (1), 21 35. A. Thanailakis and D. C. Northrop, Solid State 38. L. Eckertová, “Physics of Thin Films”, 2nd Edn., Electron., 1973, 16, (12), 1383 Plenum Press, New York, USA, 1986

The Authors

Adrian Habanyama is a Senior Professor Craig Comrie is an Lecturer at the Copperbelt Emeritus Associate Professor University, Zambia. His research at the University of Cape Town, areas are solid-state physics South Africa. His research areas and nanotechnology. are solid-state physics and nanotechnology.