Ion beam sputtering of germanium Energy and angular distributionof sputtered and scattered particlesRene Feder, Carsten Bundesmann, Horst Neumann, Bernd RauschenbachLeibniz-Institut fr Oberchenmodizierung e.V., Permoserstrae 15, D-04318 Leipzig, Germanyarti cle i nfoArticle history:Received 4 November 2013Received in revised form 5 May 2014Accepted 6 May 2014Available online 14 June 2014Keywords:Ion beam sputteringESMSAngular distributionEnergetic distributionabstractThe energy and angular distributions of scattered and sputtered particles produced by ion beam sputter-ing of a Ge target under variation of geometrical (incidence angle of primary ions and emission angle ofsecondary particles) and ion parameters (ion species (Ar, Xe) and energy (0.51.5 keV) are presented.Several sets of Ge thin lms are deposited and their thickness is measured by prolometry to deter-mine the angular particle ux distribution of the sputtered particles. The particle ux distributions areof cosine-like shape and tilted in forward direction and the tilt of the maximum position increases withdecreasing energy of the primary ions and increasing incidence angle.The energy distributions of the sputtered and the scattered ions are measured with an energy-selectivemass spectrometer. The average energy of the sputtered ions increases with increasing incidence angle ofthe primary ions and with increasing emission angle, but is nearly unaffected by the species of the pri-mary ions and their energy. The energy distribution of the scattered Ar ions reveals high energetic max-ima that originate in direct scattering between Ar/Ge and Ar/Ar and which shift with increasing emissionangle to higher energies. For Xe ion bombardment, there are only maxima for Xe/Xe scattering observed.All experimental data are compared with Monte Carlo simulations done with the well-known TRIM.SPcode. 2014 Elsevier B.V. All rights reserved.1. IntroductionThe ion beam sputter deposition (IBD) technique is a PVD tech-niquefortheproductionofhighqualitythinlmswithtailoredproperties. InIBD, theenergyandmassoftheprimaryions, themass of the target atoms and the process geometry lead to differ-ent angular and energy distributions of the sputtered and scatteredparticlesandthereforetodifferentthinlmproperties[13]. Asystematic analysis of the properties of these lm forming particlesis necessary for further process adaption.The present report focuses on the energy and angular distribu-tions of the sputtered and scattered particles for ion beam sputter-ing of a Ge target. The ux distributions of sputtered Ge particles,the energy distributions of sputtered Ge ions and the energy distri-butions of scattered primary ions are measured under variation ofthe process geometry (incidence and emission angle), the primaryionenergy (0.51.5 keV)and theion species(Ar, Xe). Thesedataarecomparedwithsimulationresults, basedontheMonteCarlocode TRIM.SP [4].Gewaschosenastargetmaterial becauseitisamonatomicsemiconductorandit isknownthat semiconductorsturntoanamorphous state under ion bombardment, while a metal target likeAgstayspolycrystalline[5]. Therefore, differentresultstothosefound for Ag [6] can be expected. Additionally, Ge does not havetherestrictionsknownforSi regardingthemeasurementswiththeenergy-selectivemass spectrometer (ESMS) [7]. For Si, theESMS is not able to differentiate between the mass of Si and themass of N2-molecules from the residual gas, what leads to an over-lay of the energy distributions. Besides, Ge is important for infraredoptics and micro electronics.Untilnow, therearenostudiesontheenergydistributionofsputteredandscatteredparticles fromGe for lowenergy ionbombardment and only a fewother semiconductors where stud-ied. Pellet et al. [8,9] studied the angular resolved energy spectraof particles sputtered from a Si target, but only under a xed ionincidenceangle. Other studies, likeGoehlichet al. [10,11] onlyfocus on metal targets. There also exist theoretical studies predict-ingtheanisotropicenergydistributionofsputteredparticlesforoblique incidence [12].TheangulardistributionsofGeparticlessputteredfromaGetarget have been reported by Andersen et al. [13] and Chini et al.http://dx.doi.org/10.1016/j.nimb.2014.05.0090168-583X/ 2014 Elsevier B.V. All rights reserved.Corresponding author. Tel.: +49 (0)341 235 4021; fax: +49 (0)341 235 2313.E-mail address: rene.feder@iom-leipzig.de (R. Feder).Nuclear Instruments and Methods in Physics Research B 334 (2014) 8895ContentslistsavailableatScienceDirectNuclear Instruments and Methods in Physics Research Bj our nal homepage: www. el sevi er . com/ l ocat e/ ni mb[14]. Both studies show over-cosine angular distributions, even forprimary ion energies down to 0.6 keV. The absolute sputter yieldsof Gefor different energies andprimaryions werestudiedbyRosenberg et al. [15] and Laegreid et al. [16].Recently, we have reported results for sputtering of Ag targetswithArandXeionsundervariationof ionincidenceangleandprimary ion energy carried out with the same setup [6,7]. The cor-relationof theprimaryandsecondaryprocess parameters andelectrical andoptical propertiesof theAglmshavealsobeenshown [3].For the energy distribution of sputtered atoms from a collisioncascadetheThompsonrelationship[17]canbeused, predictingthe energy of a sputtered particle to be proportional to E/(E + U)3,where U is the surface binding energy of the target atoms.The collision cascade theory also predicts an isotropic distribu-tion of recoil atoms in the target if it is bombarded by ions at nor-mal incidence. Inthiscaseacosine-typeangulardistributionofsputtered particles is predicted [18]. Additionally, simulationsandexperimental resultsindicateanenergydependenceof theangular distribution [1921], because for low primary ion energiesthecollisioncascadeisnotcompletelydeveloped. Consequently,the angular distribution of recoil atoms is not isotropic, resultinginachangedangulardistribution(heart-, under-orover-cosinetypes) [5,13].2. Experimental conditions and simulationFig. 1shows a schematic sketchof the vacuumdepositionchamber. The set up provides the possibility tovary the primaryion incidence angle (a) and the polar emission angle (b). Therefore,the target and the ion source are mounted on rotary tables with anidentical rotation axis. Additionally, a sample holder can bemounted in the chamber for thin lm deposition. The ion sourceis an in house development RF type broad beam ion source [22].Amoredetailedviewontheexperimental setupisgivenelse-where [6,7].Ar ions and Xe ions with energies between 0.5 keV and 1.5 keVare used to sputter the Ge target for different incidence angles (0,30 and 60). For the determination of the particle ux, polar emis-sion angles between 40 and 90 in steps of 10 are investigated.Due to the dimensions of the ion source and the ESMS, the emis-sion angle for the measurements of the energy distribution is lim-ited to 60, 30 or 0 for primary ion incidence angles of 0, 30 or60 respectively.For thedeterminationof theparticleuxdistributions, thesputtered Ge is collected on Si substrates, like described for Ag inpreviouswork [6,7]. Thestickingcoefcient oftheGeisabout1[23]. ThethinGelmsareall amorphousandthethicknessisbetween 10 nm and 100 nm. Prolometry is used to determine lmthicknesses by measuring the step height between the lm and thesubstrate. Forstepgeneration, apartofthesubstrateiscoveredduringthedepositionprocess. Theaverageparticleuxcanbecalculated using the lm thickness and the sputter time. The massdensity needed for this calculation is measured using RBS (Ruther-ford Backscattering Spectrometry).An energy selective mass spectrometer (ESMS) is used to mea-sure the energy distributions of sputtered and scattered ions [6,7].The ESMS operates in a mass range from 1 amu up to 512 amu andanenergyrangeupto500 eVwitharesolutionof 1 amuand0.5 eV, respectively. ForareasonableinterpretationoftheESMSsignal, the transmission probability of the ESMS and the ionizationprobability of the sputtered particles must be taken into account.The transmission probability was simulated by Zeuner et al. [24]foranotherESMSthatusesquitethesimilarionoptics. ThereisunfortunatelynoworkreportingontheionisationprobabilityofsputteredGe, but there are generalizedconsiderations like in[25]. Taking both into account, a relative error of about 10% is esti-mated. Due to this small deviation, the ESMS signal is taken as theenergy distribution and the relative error is taken into account inthe calculation of the average energies of the sputtered ions.All experimental data are compared with simulations which aredoneusingtheMonteCarlocodeTRIM.SP(versiontrvmc95)[4].TheinputparametersaretakenfromEckstein[26]. Thenumberof simulated primary ions is 108for each simulation.3. Results and discussion3.1. Angular distributions of sputtered Ge particlesIn Fig. 2, the experimental and simulated particle ux distribu-tions of sputtered Ge particles are shown for varying primary ionenergy Eion and incidence angle a.The inuence of the primary ion incidence angle a on the parti-cle ux distribution for both inert gases is shown in Fig. 2(a and b).For all parameter sets, the Ge particle ux is higher for sputteringwith Xe than for sputtering with Ar and the particle ux increaseswith increasing incidence angle for both ion species, because thetotal sputter yield is increased [15,16].In Fig. 2(c and d) the inuence of the primary ion energy Eion isshown for an incidence angle of a = 30 for sputtering with Ar andXe ions, respectively. The Ge particle ux increases with increasingprimary ion energy and is again higher for sputtering with Xe ionsthan for sputtering with Ar ions.Allparticleuxdistributionsshowacosine-likeshapethatistiltedinforwarddirectiondependingontheprimaryionenergyand the incidence angle. The data can be tted byU U cosnb b 1where U is the maximum value of the particle ux, n is the expo-nent (under-cosine for n < 1 and over-cosine for n > 1) and b is theemission angle of the maximum value of the particle ux. Table 1gives an overview of the best-t parameters. There is no tilting ofthe cosine distribution for normal incidence. Additionally, the parti-cle ux distribution is nearly perfectly cosine-like for Ar ion bom-bardmentata = 0andunder-cosineforXeionbombardmentata = 0. Forotherincidenceangles, theparticleuxdistributionisover-cosineandnincreaseswithincreasingincidenceangleanddecreasingprimaryionenergy. ForsputteringwithXeions, nishigher than for sputtering with Ar ions. The tilting of the particle uxdistributions b, also increases with decreasing primary ion energyand is higher for Xe bombardment than for Ar bombardment. This Fig. 1. Schematic sketch of the ion beam sputter setup.R. Feder et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 8895 89tiltingof thecosine-distributionandtheunder-cosineandover-cosinebehaviorarecausedbyananisotropicdistributionof therecoil ux in the target. For the lower primary ion energy and theheavier primary ions, this anisotropy is closer to the target surfaceand therefore, the inuence on the angular distribution of the sput-tered particles is higher.Theparticleuxescalculatedfromsimulationresultsareinexcellentagreementwiththeexperimentalvaluesfor a = 0, 30and for all ion energies. However, the simulation results are about50% higher for an incidence angle of 60 Fig. 2(a and b), for sputter-ing with Ar and Xe ions. These deviations can have different ori-gins. Therstpossibilityisthatnotthetotal primaryionbeamhits the target due to the high incidence angle and the beam diam-eter and divergence. This effect can be excluded, because the angu-lar distribution of the primary ions was measured and the effectiveuxofprimaryionsattheeffectiveareaofthesputtertargetistakenintoaccount. Otherpossibleoriginsof thedeviationsarelikely related to the targets surface roughness and the mixing ofdifferent incidence angles due to the beam divergence. It is known,that structuringandrougheningeffects at the targets surfaceincrease with increasing incidence angle of the primary ions [28].Measurements on Ag targets have shown a similar behavior [6].In Refs. [13,14], the over-cosine behavior of the angular distri-bution of sputtered Ge particles was also studied for Ar bombard-ment at normal incidence. The values of n for ion energies between0.6 keVand1.25 keVarebetween1.25and1.51andagreewellwith the values presented in this work.3.2. Energy of sputtered Ge ionsFig. 3 demonstrates the energy distribution of Ge ions sputteredfromaGetargetwithAr(a)andXe(b)ionswithaprimaryionenergyof Eion = 1.5 keVunder anincidenceangleof a = 30 atselected emission angles b. Due to the different mass ratio betweenprimary ion and target atom, the maximum energy of sputtered Geions is higher for sputtering with Xe than for sputtering with Ar.Theenergydistributionof theGeionsshowstheexpectedEbbehavior accordingtotheThompsonformula[17] for emissionangles b 6 70 for both primary ion species, with b indicating theexponent inthe power functionwitha value of about 2. ForFig. 2. Experimental (symbols) and simulated (lines) Ge particle ux distributions under variation of process parameters (ion species (a, c vs. b, d), incidence angle (a, b),energy (c, d)).Table 1Best-t parameters according to Eq. (1) for the experimental and simulated particle ux distributions for different combinations of ion species, primary ion energy and incidenceangle.Conditions Experimental distributions Simulated distributionsIon species Eion (eV) a () U (1013at cm2s1) n b () U (1013at cm2s1) n b ()Ar 500 30 8.4 0.3 2.1 0.3 30.4 1.5 9.6 0.3 1.7 0.1 16.1 1.01000 0 8.8 0.3 1.1 0.3 0.8 6.0 8.7 0.1 1.3 0.1 0.1 0.21000 30 13.4 0.2 1.8 0.1 16.7 0.8 13.3 0.3 1.5 0.1 11.2 0.71000 60 15.2 0.8 1.8 0.3 15.6 2.4 23.6 0.9 1.9 0.1 16.4 1.21500 30 15.8 0.2 1.7 0.2 12.5 1.7 15.8 0.3 1.5 0.1 9.2 0.6Xe 500 30 11.6 0.5 2.8 0.4 42.6 1.7 12.7 0.5 2.1 0.2 24.1 1.31000 0 9.4 0.1 0.7 0.1 1.9 2.4 10.7 0.1 1.2 0.1 0.1 0.21000 30 19.2 0.4 2.0 0.2 27.8 1.1 19.5 0.6 1.7 0.1 15.7 0.91000 60 24.9 1.6 2.2 0.4 23.0 2.6 45.7 1.7 2.5 0.2 22.8 1.21500 30 25.8 0.3 2.1 0.1 22.7 0.6 24.5 0.6 1.6 0.1 12.9 0.890 R. Feder et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 8895b = 80theshapeofthecurveforXeionbombardmentchangesduetothefact that thegenerationof direct recoilsispossible.These direct recoil particles have higher energies than the particlessputteredinacollisioncascadeandcanonlyoccurif a + b > 90[7,27].The energy distributions are in good agreement with the energydistributionsofsputteredSi observedbyPelletetal. [8,9]. Theyalso observed a decrease of the number of sputtered particles withincreasingenergyaccordingtoThompsonwithadditional highenergetic contributions fromdirectlysputteredparticles. Thesecontributions also increase with increasing incidence and emissionangle. In [10,11], metal targets were studied. They also show thesame behavior, but the inuence of directly sputtered particles ismorepronounced. ThesamewasfoundinformerstudiesonAg[6]. The simulated energy distributions in [12] also show the samebehavior.InFig. 4, theaverageenergyof thesputteredGeionsisshown as a function of the emission angle for different incidenceangles a(0, 30, 60) andprimaryionenergies Eion(1.0 keV,1.5 keV). The values are calculated from experimental energy dis-tributions, seeforinstanceFig. 3. Theaverageenergyofthesputtered Ge ions increases only slightly with increasing emissionanglebforbothprimaryionspecies, lessforsputteringwithArions than for sputtering with Xe ions. Also, increases slightlywith increasing primary ion incidence angle. The inuence of theprimaryparametersismuchlessthanforsputteringaAgtarget[6] and the average energy distribution of the sputtered particlesis closer tothe isotropic distribution predicted by the Thompsonmodel [17]. Thiscanbecorrelatedtotheamorphizationof theGe surface under ion bombardment that leads to a more isotropicenergy distribution [5].In Fig. 5 the energy distributions of sputtered Ge atoms, calcu-lated from simulation results for sputtering with Ar (a) and Xe (b)(Eion = 1.5 keV; a = 30) at selected emission angles are shown. Thegeneral behavior of all curves agrees with the experimental curvesof the ESMS signals from Fig. 3. All curves follow the power func-tionbehavior predictedby Thompson[17]. Incontrast totheexperimental curvesinFig. 3, thechangeof thecurvesduetodirectly sputtered particles for large emission angles is not visibleandthedirectsputteringprocessesseemtobeneglectedinthesimulation. Instead, the Ge atoms sputtered with Ar show a slightlyhigher maximum energy than the Ge atoms sputtered with Xe.In Fig. 6 the average energies of sputtered Ge particles as afunctionof the emissionangle calculated fromthe simulatedenergydistributions(examplegiveinFig. 5)forsputteringwithAr (a)and Xe(b) for differentprimary ion energiesEion (0.5 keV,1.0 keV, 1.5 keV) and incidence angles a (0, 30, 60) are outlined.The average energy increases with increasing emission angle b uptoa maximumbetween60 and80, followedby a decreasetowards an emission angle of 90. Additionally the average particleFig. 4. Average energy of Ge ions sputtered by Ar (a) and Xe (b) ions as a function of the emission angle b, under variation of incidence angle a and primary ion energy Eion.Data are calculated from experimental energy distributions as shown in Fig. 3.Fig. 3. Experimental energy distribution of Ge ions sputtered by Ar (a) and Xe (b) ions (Eion = 1.5 keV; a = 30) at selected emission angles b.R. Feder et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 8895 91energy is signicantly higher for an incidence angle of 60 in com-parison to the other incidence angles. Increasing the primary ionenergy also leads to an increase of for both primary ion species.TheshapeofthecurvesissimilarforsputteringwithArandXeions. For an incidence angle of 60, the increase of the average par-ticle energyofthesputtered particlesismuchmore pronouncedthan for the other incidence angles. This effect is due to the inu-enceofdirectlysputteredparticles[6,27]. Thesimulatedcurves,withexceptionofthecurvesfora = 60, reproducetheshapeofthe experimental curves well and the absolute values of the aver-age particle energies are comparable. For an incidence angle of 60,there is a large deviation between the data calculated from theexperimental and the simulated values. A possible explanation forthis deviation might be a correlation between the surface rough-ness of the target and the emission of directly sputtered particlesat large emission angles. It has been shown that ion bombardmentcan cause smoothing, roughing or structuring of the target surfacedepending on ion species, incidence angle and energy of the ionsand that the roughening increases with increasing incidence angle[28]. A rough surface may cause many more interactions of theseparticles with other target atoms. Another possible explanation isthat for Ge, in contrast to sputtering a Ag target, these directly scat-tered particles are not emitted as charged particles and thereforethey cannot be detected with the ESMS.3.3. Energy of scattered Ar and Xe ionsTheenergydistributionofArandXeionsbackscatteredfromtheGetarget fordifferent emissionanglesat Eion = 1.5 keVanda = 30aredemonstratedinFig. 7. Themaximumenergyofthebackscatteredparticlesincreaseswithincreasingemissionanglefor both ion species and is higher for Ar ions than for Xe ions dueto the different mass ratio between primary ion and target atom.Intheenergydistributionsof scatteredAr ionsadditional highenergetic maxima occur between 200 eV and 400 eV,which shiftwith increasing emission angle to higher energies. These maximaoriginate from a direct scattering process near or at the target sur-face and can be calculated as described elsewhere [7].Additionally, thereisachangeintheshapeof thecurveforb = 80 for both ion species. This behavior can be explained witha scattering of primary ions at implanted projectiles in the Ge tar-get during the irradiation. Fig. 8 illustrates this for the energy of ArandXe ions backscatteredfromthe Ge target for Eion = 1 keV,a = 60 and b = 50. The parameters are chosen for a clearer dem-onstration of the two scattering processes. Two high energetic con-tributions canbe clearly seenfor scatteredAr ions. The highenergeticcontributionsareassignedtoAr/ArscatteringandAr/Ge scattering. Incontrast,there isonly one highenergetic maxi-mumin the energy distribution of scattered Xe ions. This isFig. 5. Energy distributions of Ge atoms sputtered by Ar (a) and Xe (b) (Eion = 1.5 keV; a = 30) at selected emission angles, calculated from simulation results.Fig. 6. Average energies of Ge atoms sputtered by Ar (a) and Xe (b) as a function of the emission angle b for different combinations of primary ion energy Eion and incidenceangle a, calculated from simulation results.92 R. Feder et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 8895explained by the condition for the detection of directly scatteredparticles: the mass ratio mtarget/mion must be larger than the sineof the scattering angle [7]. In the geometry chosen here, the scat-tering angle is c = 180ab. Due to that relation, the maximumscattering angle for Xe ions at a Ge atom is about 33 and thereforeonly a contribution to the energy spectra for Xe/Xe scattering andnot for Xe/Ge scattering can be observed.In Fig. 9 the energy distributions of scattered Ar (a) and Xe (b)primary ions (Eion = 1.5 keV; a = 30) reected from a Ge target cal-culatedfromsimulationresultsat selectedemissionanglesareshown. High energetic maxima can be observed in the energy dis-tributions of backscattered Ar, but not in the energy distribution ofbackscattered Xe. The difference between experimental and simu-lated data for Xe ion bombardment originates from the fact that inthe simulation, implantation of primary ions is neglected.In Fig. 10 the average energies of scattered Ar (a) and Xe (b)primary ions reected from a Ge target are outlined as a function ofthe emission angle calculated from simulation results for differentcombinations of Eion and a. For both primary ions and all combina-tionsof Eionandatheaverageenergyofthescatteredparticlesincreaseswithincreasingemissionangle.alsoincreaseswithincreasingprimaryionenergyandprimaryionincidenceanglefor both species. The average energy of scattered Ar particles is sig-nicantly higher than for Xe particles in every case. This is an effectof the contributions of the high energetic, directly scattered parti-cles to the average energy of all scattered particles. Only the pri-mary ions directly scattered at Ge atoms contribute to theaverage energy, due to the fact that no contamination of the targetby primary ions is considered in the simulation, and therefore noscattering between Ar/Ar or Xe/Xe is calculated. The contributionof directly scattered Xe ions at Ge atoms can be neglected due tothe reasons given above.Fig. 7. Experimental energy distribution of Ar (a) and Xe (b) ions (Eion = 1.5 keV; a = 30) reected from Ge at selected emission angles.Fig. 9. Energy distributions of Ar (a) and Xe (b) ions (Eion = 1.5 keV; a = 30) reected from Ge selected emission angles, calculated from simulation results.Fig. 8. Experimental energydistributionof ArandXeionsreectedfromaGetarget. Primary ion energy, incidence and emission angle were chosen to show bothdirect scattering maxima (Eion = 1.0 keV; a = 60; b = 50).R. Feder et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 8895 93The average energy of the scattered particles could not be calcu-lated from experimental data, because for most of the Ar ion spec-trathemaximumenergyof thescatteredparticlesexceedstheenergy range of the ESMS.3.4. Total energies of sputtered and scattered particlesWedenethenormalizedtotal energyEtotof thesputteredparticlesasthesumof theenergyof all sputteredGeparticlesFig. 10. Average energies of Ar (a) and Xe (b) ions reected from Ge as a function of the emission angle b for different combinations of primary ions energy Eion and incidenceangle a, calculated from simulation results.Fig. 11. Normalized total energies of sputtered Ge particles (a, b) and scattered primary ions (c, d) as a function of the emission angle b for different combinations of ionspecies (Ar (a, c), Xe (b, d)), primary ions energy Eion and incidence angle a, calculated from simulation results.94 R. Feder et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 8895calculated from simulation results, divided by the number of pri-mary ions. The normalized total energy of the scattered particlesis calculated in an analogous manner. Fig. 11 shows the normalizedtotal energies of sputtered (a, b) and scattered (c, d) particles as afunction of the emission angle for sputtering with Ar (a, c) and Xe(b, d)ionsundervariationof primaryionenergyandincidenceangle.For the total energy of the sputtered particles, all curves are ofsimilar shape for both ion species. Like the average energy of thesputtered particles, the normalized total energy increases signi-cantlywithincreasingionincidenceangleandslightlywiththeprimary ion energy. Additionally, the normalized total energies ofthesputteredparticlesareslightlyhigherforsputteringwithXeionsthanforsputteringwithArions, especiallyforthehighestinvestigated primary ion energies, high incidence angles and highemission angles, due to the different mass ratio.The normalizedtotal energy of the scatteredparticles alsoincreases with increasing incidence angle and slightly withincreasing primary ion energy. For sputtering with Ar ions, the nor-malizedtotal energiesof thescatteredparticlesarecomparablewiththenormalizedtotalenergiesofthesputteredGeparticles.Although the average energy of the sputtered particles is much lessthan the average energy of the scattered ions, the large differencebetween the sputter yield and the backscatter yield lead to compa-rable total energies. For sputtering with Xe ions, a large inuenceof the primary ion incidence angle is obvious. For a = 0, the nor-malizedtotal energies of thesputteredparticles aretoosmalland therefore out of the scale. For a = 30 and the three differentprimary ion energies the normalized total energies of the scatteredXe ions are about two orders of magnitude lower than the normal-ized total energies of the scattered Ar ions, and also more than twoordersofmagnitudelower thanthenormalizedtotalenergiesofthesputteredGeparticles. For a = 60, thecurvegetsadifferentshape, becauseof additional energycontributionsfromdirectlyscattered particles like shown in Fig. 10(b).4. SummaryThe properties of secondary (sputtered and scattered) particleswere analyzed for sputtering Ge with Ar and Xe ions under system-aticvariationof theionbeamandgeometrical parameters:pri-mary ion energy, incidence angle, emission angle and ion species.The measured particle ux distributions for sputtered Ge particlesare ina goodagreement withthe values calculatedfromtheTRIM.SPsimulations. Forincidenceanglesdifferentfrom0, theangular distributions of the sputtered particles are of over-cosinetype and tilted in forward direction with respect to the target nor-mal. The tilting of the cosine-distribution for a 0 is caused by ananisotropic distribution ofthe recoil ux in thetarget,especiallyfor the lower and heavier primary ion energies. In contrast to mea-surements on a Ag target [6], the energy distributions of sputteredGe particles show only a small dependence on the emission angle.The values of the average energy of the sputtered particles there-fore onlyslightlyincrease withincreasingemissionangle andincreasingprimaryionincidenceangle. Theaverageenergiesofthesputteredparticlescalculatedfromsimulationresultsshowthe same behavior and are of comparable value, except the simula-tion results for an incidence angle of 60. The experimental energydistributions of scattered Ar and Xe ions differ signicantly. In theenergy distributions of the Ar ions, additional maxima from directscattering processes between primary ion and target atoms as wellas between primary ion and implanted primary ions occur. For Xe,thereareonlyadditionalmaximarelatedtoscatteringprocessesbetweenprimaryionsandimplantedprimaryions. Theaverageenergy of the Ar particles scattered from the Ge target are muchhigher than those of the Xe particles as a result of the contributionofthedirectlyscatteredparticles. Simulationsandexperimentaldata show a considerable difference in the ratio of the normalizedtotal energies of sputtered and scattered particles betweensputtering with Ar or Xe ions, originating from the different energydistributions of backscattered particles.The results presented in this paper should inuence the Ge thinlm properties. It is expected that the properties of the Ge lms arenot as sensitive to the geometrical and primary particle propertiesas the properties of the Ag lms studied before [3,6,7], because thepropertiesof thelmformingparticlesarelessaffectedbythegeometrical and primary ion parameters.AcknowledgementsTheauthorswanttothanktheDeutscheForschungsgemeins-chaft (DFG) for nancial support (Project BU2625/1-1), F. Scholze,I. Herold, P. Hertel, M. Mller, R. Woyciechowski andtheIOMWorkshop for technical support as well as R. Wunderlich for helpwith the RBS measurements.References[1] C. Bundesmann, I.-M. Eichentopf, S. Mndl, H. Neumann, Thin Solid Films 516(2008) 8604.[2] C.A. Davis, Thin Solid Films 226 (1993) 30.[3] C. Bundesmann, R. Feder, J.W. Gerlach, H. Neumann, ThinSolidFilms551(2014) 46.[4] J.P. Biersack, W. Eckstein, Appl. Phys. A: Mater. Sci. Process. 34 (1984) 73.[5] H. Oechsner, Appl. Phys. 8 (1975) 185.[6] R. Feder, C. Bundesmann, H. Neumann, B. Rauschenbach, Nucl. Instrum.Methods B 316 (2013) 198.[7] R. Feder, F. Frost, H. Neumann, C. Bundesmann, B. Rauschenbach, Nucl.Instrum. Methods B 317 (2013) 137.[8] C. Pellet, C. Desgranges, C. Schwebel, J. Aubert, Appl. Phys. A 55 (1992) 359.[9] C. Pellet, C. Schwebel, C. Resseguier, Nucl. Instrum. Methods B 78 (1993) 294.[10] A. Goehlich, N. Niemller, H.F. Dbele, Phys. Rev. B 62 (2000) 9349.[11] A. Goehlich, D. Gillmann, H.F. Dbele, Nucl. Instrum. MethodsB179(2001)351.[12] M. Stepanova, S.K. Dew, Nucl. Instrum. Methods B 215 (2004) 357.[13] H.H. Andersen, B. Stenum, T. Sorensen, H.J. Whitlow, Nucl. Instrum. Methods B6 (1985) 459.[14] T.K. Chini, M. Tanemura, F. Okuyama, Nucl. Instrum. MethodsB119(1996)387.[15] D. Rosenberg, G.K. Wehner, J. Appl. Phys. 33 (1962) 1842.[16] N. Laegreid, G.K. Wehner, J. Appl. Phys. 32 (1961) 365.[17] M.W. Thompson, Philos. Mag. 18 (1968) 377.[18] R. Behrisch, Sputtering by Particle Bombardment, vol. 1, Springer, Berlin, 1981.[19] G.K. Wehner, D. Rosenberg, J. Appl. Phys. 31 (1960) 177.[20] Y. Yamamura, Nucl. Instrum. Methods 194 (1982) 515.[21] J. Roth, J. Bohdansky, W. Eckstein, Nucl. Instrum. Methods 218 (1983) 751.[22] M. Zeuner, F. Scholze, B. Dathe, H. Neumann, Surf. Coat. Technol. 142144(2001) 39.[23] F. Jona, H.R. Wendt, Surf. Sci. 24 (1971) 343.[24] M. Zeuner, H. Neumann, J. Meichsner, J. Appl. Phys. 81 (1997) 2985.[25] M.L. Yu, in: R. Behrisch, K. Wittmaack (Eds.), Sputtering by ParticleBombardment, vol. 3, Springer, Berlin, 1991.[26] W. Eckstein, IPP Rep. 9 (132) (2002) 162.[27] W. Eckstein, Nucl. Instrum. Methods B 27 (1987) 78.[28] F. Frost, B. Ziberi, A. Schindler, B. Rauschenbach, Appl. Phys. A 91 (2008) 551.R. Feder et al. / Nuclear Instruments and Methods in Physics Research B 334 (2014) 8895 95