Effect of Ion Bombardment on the Chemical Properties of Crystalline Tantalum Pentoxide Films
Israel Perez, Victor Sosa, Fidel Gamboa Perera, Jose Trinidad Elizalde Galindo, Jose Luis Enriquez-Carrejo, Pierre Giovanni Mani Gonzalez
EEffect of Ion Bombardment on the Chemical Properties of Crystalline TantalumPentoxide Films
Israel Perez ∗ National Council of Science and Technology (CONACYT)-Department of Physics and Mathematics,Institute of Engineering and Technology, Universidad Aut´onoma de Ciudad Ju´arez,Av. del Charro 450 Col. Romero Partido, C.P. 32310, Ju´arez, Chihuahua, M´exico
V´ıctor Sosa
Applied Physics Department, CINVESTAV Unidad M´erida,km 6 Ant. Carretera a Progreso, A.P. 73, C.P. 97310 M´erida, Yucat´an, M´exico
Fidel Gamboa Perera
Applied Physics Department, CINVESTAV Unidad M´erida,km 6 Ant. Carretera a Progreso, A.P. 73, C.P. 97310 M´erida, Yucat´an, M´exico
Jos´e Trinidad Elizalde Galindo
Institute of Engineering and Technology, Universidad Aut´onoma de Ciudad Ju´arez,Av. del Charro 450 Col. Romero Partido, C.P. 32310, Ju´arez, Chihuahua, M´exico
Jos´e Luis Enr´ıquez-Carrejo
Institute of Engineering and Technology, Universidad Aut´onoma de Ciudad Ju´arez,Av. del Charro 450 Col. Romero Partido, C.P. 32310, Ju´arez, Chihuahua, M´exico
Pierre Giovani Mani Gonz´alez
Institute of Engineering and Technology, Universidad Aut´onoma de Ciudad Ju´arez,Av. del Charro 450 Col. Romero Partido, C.P. 32310, Ju´arez, Chihuahua, M´exico (Dated: May 7, 2019)The effect of argon ion bombardment on the chemical properties of crystalline Ta O films grownon Si(100) substrates by radio frequency magnetron sputtering was investigated by X-ray photoelec-tron spectroscopy. All samples were irradiated for several time intervals [(0.5, 3, 6, 9) min] and theTa 4 f and O 1 s core levels were measured each time. Upon analysis at the surface of the films, weobserve the Ta 4 f spectrum characteristic of Ta O . Irradiated samples exhibit the formation of Tasuboxides with oxidation states Ta , Ta , Ta , Ta , and Ta . Exposing the films, after ionbombardment, to ambient for some days stimulates the amorphous phase of Ta O at the surfacesuggesting that the suboxides of Ta are unstable. Using a sputtering simulation we discuss thatthese suboxides are largely generated during ion bombardment that greatly reduces the oxygen totantalum ratio as the irradiation time increases. The computer simulation indicates that this is dueto the high sputtering yield of oxygen. I. INTRODUCTION
Investigations in high dielectric constant materialspoint to tantalum pentoxide (Ta O ) as one of the mostpromising candidates to deal with several modern tech-nological challenges. Due to its low leakage current andhigh dielectric constant, tantalum pentoxide has beenused in storage capacitors, insulators, catalysts, gas de-tectors, and memory devices . In addition to its elec-trical properties, Ta O possesses high refraction index( n = 2 .
18 at λ = 550 nm) and a wide band gap of ∼ .Regarding its atomic structure, Ta O solidifies in ei-ther amorphous or crystalline structure; the latter show-ing two phases below 1500 K . The study of thephysical and chemical properties of Ta O films is ofgreat importance not only from the technological per- spective but also from the scientific one. In previousresearch the existence of several polymorphs were es-tablished and although most researchers agree that thesystem crystallizes in either hexagonal ( δ phase) or or-thorhombic ( β phase) structures, the ground-state crys-tal structure is still under investigation . The atomicstructure and the properties of Ta O strongly depend onthe fabrication methods. In the last two decades experi-mental techniques such as pulsed laser deposition (PLD),low pressure chemical vapour deposition (LPCVD), di-rect current (DC) and radio frequency (RF) sputtering,ion assisted deposition (IAD), and electron beam evap-oration (EBE) have been used . Moreover, to en-sure the full oxidation of samples most techniques involvepost-deposition heat treatments under oxygen flow andtemperatures above 473 K . If the temperature ishigh enough, crystalline phases are favoured resulting in atransition of the chemical and physical properties . a r X i v : . [ c ond - m a t . m t r l - s c i ] M a y In an earlier work we investigated the effect of the an-nealing temperature on the morphological, crystal, vibra-tional, and optical properties of crystalline Ta O films.Here we wish to extend the study in Ta O films to findout how their chemical properties are modified by ionbombardment. In the past, several researchers have usedTa O films (in some cases a standard BCR-261T) inXPS depth profile experiments to study several aspectsof the samples such as film composition, chemical prop-erties, and etch rate for different ion guns, ion energiesand/or irradiation fluences . From these studies, itis generally believed that, as the ion bombardment pro-gresses, there is a preferential sputtering of oxygen to becaused by the low atomic number of oxygen that leadsto the generation of several oxidation states of tantalum.However, very little is discussed on the processes takingplace during irradiation. For instance, no information isknown on the magnitude of the sputtering yields of bothoxygen and tantalum; i.e., how many oxygen/tantalumatoms are sputtered during irradiation and how muchmaterial is displaced to other lattice sites? Also, it isnot well understood how the atomic lattice is modifiedduring ion irradiation and in turn how this leads to thegeneration of the different oxidation states of tantalum.In this sense, the aim of this research is to shed somelight on these issues that so far have not been completelyelucidated.For this purpose we use an ion argon gun (with a fixedacceleration energy) from an XPS system, and sputterthe sample surface of some crystalline Ta O films forseveral time intervals. Keeping in mind that argon isinert we then assess the effect of ion bombardment onthe sample surface and obtain in situ the changes in thechemical states of Ta. The advantage of using the ion gunof the XPS system is that, samples remain in the samechamber under ultra high vacuum ( ∼ − mbar) con-ditions, avoiding in this way any contamination or anyundesired oxidation process that otherwise would arisewhen using an external ion gun. Furthermore, the atomicprocesses involved in ion bombardment at the film sur-face are analyzed with the aid of a computer simulationthat throws very important quantitative and qualitativeinformation. The simulation turns out to be very usefulnot only to build a wider picture of such processes butalso to enrich the interpretation of XPS measurementswhen depth profiles are performed.To achieve our goal, three Ta films were grown onSi(100) substrates by RF magnetron sputtering and thecrystalline phase, i.e., Ta O , was induced by exposingthe films to post-deposition heat treatments at 1273 K inambient atmosphere. The crystalline structure and mi-crostructure were determined by X-ray diffraction (XRD)and scanning electron microscopy (SEM), respectively.XPS in combination with an ion gun were used to evalu-ate the chemical properties for different sputtering timeintervals. Finally, using the actual data from the ion ir-radiation we carried out the simulation and discuss thebombardment process in detail. II. EXPERIMENTALA. Film growth and annealing
Three amorphous Ta films were grown at different ses-sions at room temperature on 5 mm × . × − mbar. To activate the plasma, argongas was flushed into the chamber to a partial pressure of(2 . ± . × − mbar. The native oxide layer on thetarget surface was removed by a 5 min pre-sputteringprocess at 60 W. Immediately the power was raised to120 W and the rotary base was activated with a speedof 0.2 rpm. During deposition there was no intentionalsubstrate heating or cooling and no oxygen was injectedinto the chamber. To oxidize the samples, two films wereexposed to post-deposition heat treatments in air for onehour at 1273 K using a Thermo Scientific Thermolynecylindrical furnace (model F21135). These films were la-beled as F05 and F33. The as-deposited film, labeled F0,was kept for reference and was not exposed to any heattreatment. The deposition rate was determined using thedeposition time and the sample thickness which was pre-viously measured in the SEM. The film thicknesses were(2.4, 0.5 and 3.3) µ m for F0, F05, and F33, respectively.The deposition rates were 2.8 ˚A/s for F0 and F33, and2.4 ˚A/s for F05. B. Crystal structure and morphology
After exposing the samples to heat treatments, theircrystalline structure was studied by a Siemens diffrac-tometer model D-5000 with Cu K α radiation ( λ = 1 . ◦ with a time perstep of 3 s and operating parameters of 34 kV and 25 µ A. A field emission microscope JSM7000F was used toevaluate the morphology and grain size of the samples.
C. Ion bombardment and XPS characterization
To study the effect of ion bombardment on the chem-ical properties of Ta O films, the samples were placedinside the vacuum chamber of a Thermo Scientific K-Alpha XPS spectrometer at a base pressure of 5 × − mbar. As a sputtering agent we used argon with a purityof 99.998%. Argon ions were accelerated with a voltage of3 kV by an ion gun EX06 Thermo Scientific, generatingan electric current of 10 µ A. The incidence angle betweenthe sample and the ion gun was 90 ◦ . The etched area was
20 25 30 35 40 45 50 55 60 65 70 75F33 ( )( ) ( ) I n t e n s i t y / a . u . ( ) ( ) ( ) ( )( )( )( ) ( )( )( )( ) PDF 025-0922
F05
FIG. 1. X-ray diffraction patterns for the crystalline films,F05 and F33. The lower pattern corresponds to the referencePDF 025-0922 and the samples were sputtered at different timeintervals ( t sput ), namely: (0.5, 3.0, 6.0, 9.0) min. To eval-uate the effect of the ion bombardment on the film sur-face, simulations were carried out with the SRIM/TRIMsoftware . This software gives, in real time, informa-tion on the total and partial sputtering yield, damagearea, number of sputtered atoms, among other magni-tudes. Particular settings for the simulations are givenin the corresponding section below.After each sputtering session, the spectra of the Ta 4 f and O 1 s core levels were measured in the same instru-ment using an Al K α X-ray source with photon energyof 1486.7 eV, and set to 12 kV and 40 W. The X-raybeam spot has a diameter of 400 µ m and makes an an-gle relative to the sample of 30 ◦ . For the XPS scans weused steps of 0.1 eV. Both the stoichiometry of the filmsand the oxidation states of Ta were determined by peakdeconvolution of the XPS spectra using Voigt functionsas implemented in the AAnalizer software (see below formore details) . III. RESULTS AND DISCUSSIONA. Crystalline structure
As expected, the as-deposited film F0 displays no peaksin the XRD pattern indicating an amorphous structure;for this reason its pattern is not shown here. On thecontrary, the diffraction patterns for F05 and F33 implya crystalline structure (see Figure 1). These patterns canbe indexed to the orthorhombic phase β − Ta O (PDF 00-025-0922 with lattice parameters a = 6 . b =40 . c = 3 . α = β = γ = 90 ◦ for thespatial group P O .It is worth noting, however, that the patterns show notraces of other Ta oxides. According to the literature ,Ta O is the most stable oxide of Ta exhibiting amor-phous and crystalline phases; the other oxides such asTaO, Ta O, TaO , Ta O (except for TaO x ) have shownto be difficult to obtain as pure phases. Moreover, ex-cept for Ta O and TaO x , all suboxides are crystalline.We will come back to this in our discussion on chemicalstates.The crystallite size D for the samples F05 and F33 wasestimated using the celebrated Scherrer equation D = Kλ Γ cos θ , (1)where K = 0 . λ = 1 . θ = 28 . ◦ for both films, we found that Γ=(0.22,0.27) ◦ and hence D =(36, 30) nm, respectively. B. Morphology and grain size
In Figure 2 the SEM images for the three films aredisplayed. F05 and F33 are shown after annealing. F0(Figure 2a) shows a smooth and cloudy-like pattern andthis same pattern was exhibited by the films F05 (Fig-ure 2b) and F33 (Figure 2c) before annealing. As thefilms were exposed to the heat treatment, large grainsstarted to show up, both samples resembling the pow-der microstructure of Ta O . We note that both sam-ples exhibit some black spots. To elucidate the nature ofthese spots we took backscattering electron images (notshown). The results suggest that the spots may belongto another oxide phase of Ta, most probably the δ phaseof Ta O or a suboxide. We performed an EDS (Energydispersive x-ray spectroscopy) analysis on these spots butthe atomic concentration of O and Ta did not change sig-nificantly and therefore conclusions could not be drawn.As for the grain size (not to be confused with the crystal-lite size above), according to these images, we estimated adiameter varying from 100 nm to 500 nm with a mode of300 nm for F05 and F33 which is a relatively large size fora grain and this allows us to consider our films as granu-lar bulk materials. Despite that both films have differentthickness, the grain size and morphology are quite similaras well as the manifestation of the black spots, indicatingthat the fabrication process is quite reproducible. a) b) c) FIG. 2. SEM images of the films: F0 (a), F05 (b), and F33 (c).F0 exhibits a cloudy-like microstructure whereas the othersshow randomly oriented grains with black spots
C. Chemical states by ion bombardment
1. Ta f core level To investigate the chemical states of the films beforeand after ion bombardment, we carried out XPS mea-surements on all samples and deconvoluted the spectrafor the Ta 4 f and O 1 s core levels. The binding ener-gies of our spectra were calibrated with respect to theO 1 s peak at 532 eV. After calibration we checked forsurface contamination from the C 1 s core level. The car-bon spectra (not shown) revealed the presence of carboncontamination which is quite ubiquitous in most samples(adsorbed carbon). We stress that we have used the O 1 s binding energy as reference because oxygen, in our sam-ples, has just two contributions (see below). The first onecan be associated to O-Ta bonding and the second oneto O-C bonding; and the O 1 s core level is well known toappear at 532 eV. This energy is a good candidate for ref- erence because we are interested in analyzing variationsin the Ta 4 f binding energy and they can be detectedwith respect to our set of samples, and due to native car-bon, the C 1s binding energy would not be a good choicefor reference.The spectra for Ta 4 f core level as a function of sput-tering time are given in the left column of Figure 3. Start-ing with the Ta 4 f core level for the three films withoutsputtering (0 min, black spectra) we observe that thereis a spin-orbit doublet corresponding to the levels 4 f / and 4 f / . The binding energy for the doublets is around(27.8 ± ± in stoichiomet-ric amorphous Ta O films , suggesting that at thesurface Ta O forms. This is expected because duringannealing the film surface is in an oxygen rich environ-ment that favours surface oxidation much better than indeeper layers. We checked this by sputtering the samplessurface for 30 s and observed the red spectra where newfeatures appear. We then exposed the samples to am-
32 30 28 26 24 22 20 538 536 534 532 530 528
Ta4f F0 F0 O1sTa4f I n t e n s i t y / a . u . F05 F05
O1sTa4f
F33 F33
Binding Energy/eV
O1s
FIG. 3. Normalized to height XPS spectra for Ta 4 f (left)and O 1 s (right) for the three films bient for seven days and observed again the same blackspectra without sputtering; an indicative that in the sur-face the oxidation state +5 is recovered. A closer lookat the spectrum for zero etching of F0 and the spectraof F05 and F33 for 30 s etching shows the presence of ashoulder in the low energy region (around 24 . ± . d electrons in amorphous TaO x ( x = 1 . , .
00) suggestingthe presence of TaO x in our films .
2. Ion bombardment analysis
Before going further, it is useful to have a wide pic-ture of the ion bombardment process as a function ofsputtering time. This will help us to acquire a deeperunderstanding of the atomic configuration involved andhow it affects the chemical states of the samples. Forthis goal we carried out a sputtering simulation using thetransport of ions in matter (TRIM) package in the sur-face sputtering/monolayer collision steps mode . Thismode gives a full treatment of sputtering for full damagecascades based on Monte Carlo methods. The simulationassumes a stoichiometric target of Ta O with a thick-ness of 100 nm and bulk density of 8.2 g/cm . Althoughour films are thicker than 100 nm, this thickness was se-lected to make calculations much faster since simulationsfor thicker targets give the same results but in longertimes. During the simulation the film is bombarded byargon ions, impinging normal to the target surface. Theion energy was set to the actual energy of 3 keV; andthe ion direction is assumed to be parallel to the X -axiswith the target surface parallel to the Y Z plane, so thepenetration depth ( P ) is measured along the − X axis.The number of incident ions ( N ) was computed us-ing the actual ion gun current of 10 µ A. Accordingly, wedetermined that for a surface of 1 nm , about 32 ionsarrive per second. We just then multiplied this numberby the corresponding sputtering time intervals, that is: t sput =(0.5, 3, 6, 9) min. Table I gives the parameters ineach case.During the bombardment several processes occur.When an ion arrives at the surface, surface atoms maybe either knocked out of the material (sputtered) or dis-placed deeper into the lattice (implanted). When an ionhas a head-on collision with a surface or lattice atom, TABLE I. Sputtering parameters for the simulation: Sputter-ing time ( t sput ), number of incident ions N in a nm , damagearea ( A ), penetration depth ( P ), sputtering yields ( Y ) for Taand O, and number of sputtered atoms ( N sput ) t sput min N A nm P nm Y Ta Y O N sput very often the collision is elastic, otherwise is inelasticand the ion keeps moving and interacting with other lat-tice atoms. The recoiling atoms (atoms displaced aftera collision with either ions or displaced atoms), in turncollide with other lattice atoms, losing energy as theypropagate and causing an atomic cascade. This cascadeis depicted in Figure 4 (a-d) in blue dots for O atomsand in green dots for Ta atoms for the different sputter-ing times (or equivalently, for different N ). It is evidentthat most recoiling atoms are O atoms. As the recoilingatoms collide they lose energy and finally stop at someplace within the lattice (red dots). When atoms finallystop they may reside either at an interstitial site or atany other lattice site different from the initial one. Inthis sense, it is likely that a recoiling atom may replaceanother lattice atom of the same species. If the impingingatom has an energy less than either the lattice bindingenergy ( E bin ) or the surface binding energy ( E sbe ), theatom gives off energy to the lattice/surface atom forcingit to vibrate and release the excess energy as phonons. Inour case we use E bin = 3 eV for both Ta and O and E sbe is 2 eV for O and 8.1 eV for Ta. So a 3 keV ion wouldeasily displace target atoms.The zone affected by the ion bombardment where mostatoms recoil is called the damage zone. The images (a-d)show such zone along the XZ plane. The effects alongthe XY plane are similar and, on average, they are thesame as those on the XZ plane and for this reason theyare not shown here. In a three dimensional view, thedamage zone would resemble a half ellipsoid (the dam-age ellipsoid). The red semi-ellipse delimits an average damage area A defined for quantification purposes. Thisarea is a measured of the displaced material during ionbombardment. As summarized in Table I, this quantityincreases as t sput increases. Roughly speaking, the dam-age area doubles its size from 0.5 min to 9 min. Thepenetration depth in turn (delimited by the red verticalline), varies little for it increases about 1 nm from 0.5 minto 3 min and just 1.7 nm from 3 min to 9 min. Simula-tions realized by us (not shown) with higher ion energiesindicate that the penetration depth largely depends, asexpected, on the ion energy and, as shown here, little onthe sputtering time. The simulation also gives the sput-tering yield ( Y ) per incident ion; the values are 0.64 forTa and 4.34 for O. For this reason, the number of atomsthat leave the sample surface ( N sput ) runs from 4800 for30 s to 86 400 for 9 min; and so roughly 88% of the sput-tered atoms are oxygen atoms and the rest are Ta atoms.According to our simulation the preferential sputtering ofoxygen is not only a consequence of the low atomic num-ber of oxygen, as generally believed , but also a resultof the low surface binding energy of oxygen which causesoxygen atoms to be more prone, than tantalum atoms,to undergo either displacement or sputtering. Therefore,one would expect the depletion of O atoms and both thetemporary generation of O vacancies on the surface andthe formation of Frenkel defects in the lattice.With this information in mind, we can figure out the FIG. 4. Simulation of damage areas for the different sputtering times along the XZ plane: a) 0.5 min, b) 3 min, c) 6 min, andd) 9 min. Blue dots represent recoiling oxygen atoms and green dots represent recoiling tantalum atoms. Red dots representstopped Ta and O atoms. The red vertical line is a projection to the horizontal used to determine the penetration depth andthe half ellipsis are used to delimit an average damage area possible effect on the chemical properties. As we can seethe effect of sputtering does not extend more than 20 nmdeep; so what happen to the chemical environment of thedamage zone after the ion bombardment stops? Unfor-tunately, the simulation does not tell us anything on thismatter. It just informs us that many atoms are sput-tered from the grains and many others are displaced toother locations within the lattice, leaving a great amountof structural defects and in particular low coordinationnumbers. So, this scenario strongly suggests that theatomic structure in that zone is highly and energeticallyunstable.To recover stability, several processes must occur tomaintain not only the structural stability at the grainssurface but also electroneutrality. The atoms in the dam-age zone must seek for the most favourable energetic sitesand, due to oxygen depletion and the large amount ofpunctual defects, one would expect, in first instance, newcoordination geometries leading to the transition of Tafrom Ta to other oxidation states; and, consequently,the formation of metastable phases of Ta oxides suchas TaO, Ta O, TaO , TaO x , or Ta O . Later, as timegoes by (it could be some minutes or days depending on both the oxygen concentration around the surface andthe physi- and chemisorption processes taking place ),the surface tends to readsorb O atoms from the environ-ment and, as we have just discussed above, the surfacetends to oxidize in the most stable phase of Ta oxide,i.e., Ta O . In this scenario one realizes that after sput-tering, there exists amorphous Ta O at the very sur-face of the grains; and for the rest of the bulk, there ismainly TaO x for F0 and the β -Ta O phase for F05 andF33. Hence, as one sputters away the topmost atoms,one destroys the crystalline Ta O phase and the chem-ical states of Ta change for some time until amorphousTa O is recovered . The rich additional features thatappear in the spectra as the sputtering time increaseslends support for this line of thought (0.5 min to 9 minspectra in Figure 3).
3. Deconvolution analysis of Ta f core level To shed some light on such features, we conductedan analysis of the Ta 4 f spectra with the softwareAAnalizer. Because of spin-orbit coupling, the decon-
32 30 28 26 24 22 20 32 30 28 26 24 22 20 32 30 28 26 24 22 20 F0 F05
F33 I n t e n s i t y / a . u . Ta Ta Ta Ta Ta Binding Energy/eV Ta FIG. 5. Deconvolution of Ta 4 f XPS spectra for the three films. Left column for F0, central column for F05, and right columnfor F33 volution process was done with doublets. The latterwere fitted using Voigt functions and a Shirley-Sherwoodbackground . The deconvolution analysis reveals the ex-istence of six sets of peak splitting (see Figure 5) indicat-ing the presence of the six oxidations states of Ta, fromTa to Ta . The fitting parameters for the doubletswere the following: Gaussian=1.43 eV for the oxidationstates +2, +3, +4, +5 and Gaussian=0.9 eV for the ox-idation states 0 and +1; and Lorentzian=0.02 eV for alldoublets. This kind of features has been reported in ab-initio calculations for the δ phase of Ta O films and have been described by four doublets with the 4 f / binding energies of (22.0, 23.2, 24.6, 26.1) eV and spin-orbit splitting of 1.9 eV. They were attributed to Ta ,Ta , Ta /Ta , and Ta , respectively, observed inamorphous Ta O films. In our case the six doublets canbe associated to the states Ta , Ta , Ta , Ta , Ta ,and Ta with binding energies at (21.7 ± ± ± ± ± ± ± f spectra, the percentage of oxidation state of Ta(see Figure 6). Due to the complexity of the spectra, thepeak fitting may not be entirely accurate and over inter-pretation of the results should be avoided. Nonetheless,some general trends can be indeed summarized. In Fig-ure 6 we can observe that, at the surface of all samples,Ta in Ta has the major contribution; with more than70% in all cases. As expected the content of Ta in F0is much less than that one in F05 or F33. As the irradia-tion time increases, the state +5 decreases between 15%and 35% whereas the other states increase, suggestingthe presence of Ta suboxides. Ta in metallic state (Ta )is barely present in all cases with less than 1% and thepresence of suboxides varies from 10% to 35%. These
5+ 4+ 3+ 2+ 1+ 0 F0 F05 % o f o x i d a t i o n s t a t e F33
Sputtering time/min
FIG. 6. Percentage of oxidation state of Ta as a function ofsputtering time for the three films. Lines are just a guide tothe eye values are reasonable according to previous reports foramorphous samples .We also estimated the stoichiometry of our samplesusing the typical expression for composition C x of anatomic species x : C x = I x SF x (cid:80) Ni =1 I i SF i , (2)where I i are the i peak intensities for each atomic species,that is, Ta and O; and SF i is the sensitivity factor for theatomic species i . The sensitivity factors are given by theXPS instrument ( SF O = 2 .
93 and SF Ta = 8 .
62) and thepeak intensities were obtained during the deconvolutionprocess. In stoichiometric Ta O the O to Ta ratio is 2.5.We have plotted the atomic percentage ratio for Ta infigure 7. There we can see that all samples are non-stoichiometric. The ratio of F0 ranges from 0.80 to 0.95as sputtering time increases which seems reasonable sincethis sample was not exposed to a heat treatment andone would expect the same atomic concentration all overthe sample (except at the surface where there is slightlymore oxygen available). This finding suggests that the O % / T a % o f T a Sputtering time/min F0 F05 F33
FIG. 7. O at% to Ta at% ratio for Ta (Ta O ) as a functionof sputtering time for the three coatings. Lines are just a guideto the eye composition of this film is mainly TaO x , reaffirming ourdiscussion above. The ratio for F05 goes from 1.50 to0.90 as the sputtering time increases and that one for F33goes from 1.75 to 1.20; these values indicate an overalllow oxygen content. The non-stoichiometry is likely to beconsequence of two important factors, namely: the lackof an oxygen rich atmosphere during the manufacturingprocess of the samples; and the oxygen depletion takingplace during ion bombardment. It is evident from theplot that the latter factor has a major impact on theoxygen depletion.We remark that, according to our analysis, most Tasuboxides only appear after sputtering the film surface(even for sputtering times as short as 30 s). If they werepresent all over the bulk of the film, they would show upin the XRD patterns; however the XRD patterns showno signs of impurity phases of Ta oxide (of course, ex-cluding Ta O and TaO x which are amorphous and thuscannot be detected by XRD). To further check these as-sertions, after sputtering F33 for 9 min, we exposed it toair for three days and obtained again its Ta 4 f spectrumat the surface; the results showed anew the spectrum cor-responding to amorphous Ta O . This spectrum is quitesimilar to the one displayed in Figure 3 and therefore itis not shown here.
4. O s core level We can extract even more information on the chem-ical states analyzing the oxygen core level. The resultsfor the O 1 s core level for all films are displayed in theright column of Figure 3. Starting with the black spectra(no etching) we observe a hump at high energies, around(533 . ± .
2) eV. We also note a prominent signal at 532eV which is typical from Ta-O binding . The situationfor the sputtered films is different for the hump seems tovanish. The deconvolution unveils the presence of twocomponents, indicating that even for bottom layers thissignal is still present although its intensity is diminished(see figure 8). This feature suggests C-O binding and hasbeen observed before by some researchers . On onehand, O. Kerrec et al. grew amorphous Ta O films andexposed them to distilled water. They attributed thefeature to hydroxylic groups and/or water adsorbed atthe surface. On the other hand, Atanassova et al. grewalso amorphous Ta O thin films (6 nm to 13 nm) on Sisubstrates. They explained that the feature was due toSi-O binding rather than contamination. In a more re-cent investigation, R. Simpson et al. studied the effecton the chemical composition of Ar + sputtered amorphousTa O films (30 nm thick) grown on Ta foil and foundthat the C 1 s core level disappears after 3 nm depth.Thus since our coatings are thick and the intensity of thefeature diminishes considerably as the sputtering timeincreases, we attribute it to surface contamination, mustprobably a carbon compound (CO) due to the presenceof carbon on the surface. We rule out the formation ofTa-C binding because is well known that tantalum car-bide crystallizes at temperatures larger than 1600 ◦ C . IV. CONCLUSIONS
The results of SEM showed a granular structure forthe crystalline samples resembling the powder morphol-ogy. Using XRD, the samples were indexed to the or-thorhombic phase of Ta O for the crystalline films and,accordingly, we observed no traces of crystalline Ta sub- oxides in the bulk of all samples. Meanwhile F0 wasfound to be amorphous. In this sense, the crystallineanalysis indicates, after the heat treatment, a structuraltransition of F05 and F33 from Ta amorphous to crys-talline Ta O . We carried out XPS analysis in order tostudy the effect of ion irradiation on the chemical prop-erties of our samples for several time intervals. From thedeconvolution of the Ta 4 f spectra it can be concludedthat the chemical states of Ta vary from Ta to Ta with a nil contribution of the oxidation state Ta whenthe samples are irradiated. There is an overall reductionof the oxygen to tantalum ratio as the sputtering timeincreases which is caused by the high oxygen sputteringyield. Our findings strongly suggest the manifestationof Ta suboxides generated at the surface derived by thecreation of Frenkel defects and atom displacement mainlyduring ion bombardment as indicated by the simulation.We found evidence that the suboxides are unstable andthe surface composition can be restored to the amorphousphase of Ta O after several days of air exposure. Weconclude that all films, after the ion bombardment, ex-hibit an amorphous Ta O phase at the surface althoughsome degree of non-stoichiometry is present. We mainlyattribute this to oxygen depletion during irradiation andto the lack of an oxygen rich environment during bothgrowth and annealing. We have demonstrated that ionirradiation induces the formation of several unstable ox-idation states of tantalum. ACKNOWLEDGEMENTS
We are grateful to Wilian Cauich, Daniel Aguilar, andJorge Ivan Betance for their technical support during theXPS, XRD, and SEM sessions. We are grateful to the ed-itor of this journal and the anonymous reviewers for rais-ing several comments that greatly improved the qualityof this research. Prof. I. Perez is thankful to Dr. Al-berto Herrera for helpful comments and discussions onthe XPS analysis. Funding: This work was partiallysupported by the National Council of Science and Tech-nology (CONACYT) Mexico and the program C´atedrasCONACYT through project 3035.
Declaration of interest: none ∗ Contact Author: [email protected] T. Kaga, H. Shinriki, F. Murai, Y. Kawamoto, Y. Nak-agone, F. Takeda, K. Itoh, DRAM Manufacturing in the’90s -Part 3: A Case Study, Semicond. Int. 6 (1991) 98-101 K.W. Kwon, C.S. Kang, S.O. Park, H.K. Kang, S.T.Ann, Thermally robust Ta O capacitor for the 256-MbitDRAM, IEEE Trans. Electron. Devices. 43 (1996) 919-923 C. Chaneliere, J.L. Autran, R.A.B. Devine, B. Baland,Tantalum pentoxide (Ta O ) thin films for advanced di-electric applications, Mater. Sci. Eng. R-rep 22 (1998) 269-322 H. Chen, Electrical and material characterization of tan-talum pentoxide (Ta2O5) charge trapping layer memory,Appl. Surf. Sci. 257 (2011) 7481 C. Chaneliere, S. Four, J.L. Autran, R.A.B. Devine, N.P.Sandler, Properties of amorphous and crystalline Ta O thin films deposited on Si from a Ta(OC H ) precursor,Electrochem. J. Appl. Phys. 83 (1998) 4823-4829 R.H. Dennard, F.H. Gaensslen, H. Yu, V.L. Rideout, E.Bassous, A.R. LeBlanc, Design of ion-implanted MOS-FET’s with very small physical dimensions, IEEE J. Solid-State Circ. 9 (1974) 256-268
538 536 534 532 530 528 538 536 534 532 530 528 538 536 534 532 530 528 F0 Ta-O
Ta-O
F05
F33
Ta-O I n t e n s i t y / a . u . Binding Energy/eV
FIG. 8. Deconvolution of O 1 s XPS spectra for the films. Left column for F0, central column for F05, and right column forF33 S. Shibata, Dielectric constants of Ta O thin films de-posited by r.f. sputtering, Thin Solid Films 277 (1996) 1-4 E. Atanassova, Thin RF sputtered and thermal Ta O onSi for high density DRAM application, Microelectron. Re-liab. 39 (1999) 1185-1217 G. Atak, ¨O. Duyar, D. Coskun, Annealing effects of NiOthin films for all-solid-state electrochromic devices, SolidState Ionics 305 (2017) 43-51 Q. Liu, G. Dong, Q. Chen, J. Guo, Y. Xiao, M.-P.Delplancke-Ogletree, F. Reniers, and X. Diao, Charge-transfer kinetics and cyclic properties of inorganic all-solid-state electrochromic device with remarkably improved op-tical memory, Sol. Energy Mater. Sol. Cells 174 (2018) 545-553 Q. Liu, G. Dong, Q. Chen, J. Guo, Y. Xiao, M.-P.Delplancke-Ogletree, F. Reniers, and X. Diao, Dynamicbehaviors of inorganic all-solid-state electrochromic device:Role of potential, Electrochim. Acta 269 (2018) 617-623 D. Dong, W. Wang, A. Barnab´e, L. Presmanes, A.Rougier, G. Dong, F. Zhang, H. Yu, Y. He, X.Diao, Enhanced electrochromism in short wavelengths for NiO:(Li, Mg) films in full inorganic device ITO/NiO:(Li,Mg)/Ta O /WO /ITO, Electrochim. Acta 263 (2018) 277 J.D.T. Kruschwitz, W.T. Pawlewicz, Optical and dura-bility properties of infrared transmitting thin films, Appl.Opt. 36 (1997) 2157-2159 C. Chaneliere, S. Foura, J.L. Autran, R.A.B. Devine, Com-parison between the properties of amorphous and crys-talline Ta O thin films deposited on Si, Microelectron.Reliab. 39 (1999) 261-268 S.P. Garg, N. Krishnamurthy, A. Awashi, M. Venkatra-man, The O-Ta (Oxygen-Tantalum) system, J. PhaseEquilib. 17 (1996) 63-77 K.T. Jacob, C. Shekhar, Y. Waseda, An update onthe thermodynamics of Ta O , J. Chem. Thermodyn. 41(2009) 748753 S. P´erez-Walton, C. Valencia-Balv´ın, A.C.M. Padilha,G.M. Dalpian, J.M. Osorio-Guill´en, A search for theground state structure and the phase stability of tantalumpentoxide, J. Phys. Condens. Matt. 28 (2016) 035801-11 Y. Yang, Y. Kawazoe, Prediction of a new ground-statecrystal structure of Ta O , Phys. Rev. Mat. 2, 034602 (2018) E. Atanassova, T. Dimitrova, J. Koprinarova, AES andXPS study of thin RF-sputtered Ta O layers, Appl. Surf.Sci. 84 (1995) 193-202 H. Shinriki, M. Nakata, UV-O and Dry-O : Two-step An-nealed Chemical Vapor-Deposited Ta O Films for StorageDielectrics of 64-Mb DRAMs, IEEE Trans. Electron De-vices ED. 38 (1991) 455-462 S. Kamiyama, P.-Y. Lesaicherre, H. Suzuki, A. Sakai, I.Nishiyama, A. Ishitani, J. Electrochem. Soc. 140 (1993)1617 G.Q. Lo, D.L. Kwong, S. Lee, Appl. Phys. Lett. 62 (1993)973 Y. Kuo, Reactive Ion Etching of Sputtered Deposited Tan-talum Oxide and its Etch Selectivity to Tantalum, J. Elec-trochem. Soc. 139 (1992) 579-583 N. Donkov, A. Zykova, V. Safonov, R. Rogowska, J. Smo-lik, Tantalum Pentoxide Ceramic Coatings Deposition onTi4A16V Substrates for Biomedical Applications, Prob-lems At. Sci. Technol. Plasma Physics Series 17 (2011)131-133 T. Dimitrova, U. K. Arshak, E. Atanassova, Crystalliza-tion effects in oxygen annealed Ta O thin films on Si,Thin Solid Films 381 (2001) 31-38 S.-J. J. Wu, B. Houng, B. S. Huang, Effect of growthand annealing temperatures on crystallization of tantalumpentoxide thin film prepared by RF magnetron sputteringmethod, J. Alloys Compd.475 (2009) 488-493 D. Cristea, D. Constantin, A. Crisan, C.S. Abreu, J.R.Gomes, N.P. Barradas, E. Alves, C. Moura, F. Vaz, L.Cunha, Properties of tantalum oxynitride thin films pro-duced by magnetron sputtering: The influence of process-ing parameters, Vacuum 98 (2013) 63-69 T. Tsuchiya, H. Imai, S. Miyoshi, P.A. Glans, J. Guo,S. Yamaguchi, X-Ray absorption, photoemission spec-troscopy, and Raman scattering analysis of amorphous tan-talum oxide with a large extent of oxygen nonstoichiome-try, Phys. Chem. Chem. Phys. 13 (2011) 17013-17018 S.-C. Wang, K.-Y. Liu, J.-L. Huang, Tantalum oxide filmprepared by reactive magnetron sputtering deposition forall-solid-state electrochromic device, Thin Solid Films 520(2011) 1454-1459 S.V. Jagadeesh-Chandra, C.-J. Choi, S. Uthanna, G.Mohan-Rao, Structural and electrical properties of radiofrequency magnetron sputtered tantalum oxide films: In-fluence of post-deposition annealing, Mater. Sci. Semicond.Process. 13 (2010) 245-251 I. Perez, J.L. Enr´ıquez-Carrejo, V. Sosa, F. Gamboa, J.R.Farias-Mancillas, J.T. Elizalde-Galindo, C.I. Rodriguez-Rodriguez, Evidence for structural transition in crystallinetantalum pentoxide films grown by RF magnetron sputter-ing, J. Alloys Compd. 712 (2017) 303-310 E. Atanassova, D. Spassov, X-ray photoelectron spec-troscopy of thermal thin Ta O films on Si, Appl. Surf.Sci. 135 (1998) 71-82 V.R.R. Medicherla, S. Majumder, D. Paramanik, S.Varma, Formation ofself-organized Ta nano-structures byargon ion sputtering of Ta foil: XPS and AFM study, J.Electron Spectrosc. Relat. Phenom. 180 (2010) 1-5 D.R. Baer, M.H. Engelhard, A.S. Lea, P. Nachimuthu,Comparison of the sputter rates of oxide films relative tothe sputter rate of SiO , J. Vac. Sci. Technol. A 28 (2010)1060-1072 N. Benito, C. Palacio, Nanostructuring of Ta O surfacesby low energy Ar + bombardment, Appl. Surf. Sci. 351(2015) 753-759. R. Simpson, R. G. White, J.F. Watts, M. A. Baker, XPSinvestigation of monatomic and cluster argon ion sputter-ing of tantalum pentoxide, Appl. Surf. Sci. 405 (2017) 79-87 J. P. Biersack and L. Haggmark, A Monte Carlo computerprogram for the transport of energetic ions in amorphoustargets, Nucl. Instr. and Meth. 174 (1980) 257-269 J. F. Ziegler and J. M. Manoyan, The stopping of ions incompounds, Nuclear Inst. and Meth. B35 (1989) 215-228 A. Herrera-Gomez, M. Bravo-Sanchez, O. Ceballos-Sanchez, M.O. Vazquez-Lepe, Practical Methods for Back-ground Subtraction in Photoemission Spectra, Surf. Inter-face Anal. 46, (2014) 897-905 J.Y. Kim, B. Magyari-K¨ope, K.-J. Lee, H.-S. Kim, S.-H.Lee, Y. Nishi, Electronic structure and stability of lowsymmetry Ta O polymorphs, Phys. Status Solidi RRL 8(2014) 560-565 S.-H. Lee, J. Kim, S.-J. Kim, S. Kim, G.-S. Park, HiddenStructural Order in Orthorhombic Ta O , Phys. Rev. Lett.110 (2013) 235502-5 J. Lee, W. Lu, E. Kioupakis, Electronic properties of tan-talum pentoxide polymorphs from first-principles calcula-tions, Appl. Phys. Lett. 105 (2014) 202108-5 Z. Helali, M. Calatayud, C. Minot, Novel Delta-Ta O Structure Obtained from DFT Calculations, J. Phys.Chem. C. 118 (2014)13652-13658 Y. Guo, J. Robertson, Comparison of oxygen vacancy de-fects in crystalline and amorphous Ta O , Microelectron.Engin. 147 (2015) 254-259 J.-Y. Kim, B. Magyari-K¨ope, Y. Nishi, J.-H. Ahn, First-principles study of carbon impurity effects in the pseudo-hexagonal Ta O , Current Appl. Phys. 16 (2016) 638-643 K. Chen, G.R. Yang, M. Nielsen, T.M. Lu, E.J. Ry-maszewski, X-ray photoelectron spectroscopy study ofAl/Ta O and Ta O /Al buried interfaces, Appl. Phys.Lett. 70 (1997) 399-401 A. Muto, F. Yano, Y. Sugawara, S. Iijima, The Study ofUltrathin Tantalum Oxide Films Before and After Anneal-ing with X-Ray Photoelectron Spectroscopy, Jpn. J. Appl.Phys. 33 (1994) 2699-2702 E. Atanassova, M. Kalitzova, G. Zollo, A. Paskaleva,A. Peeva, M. Georgieva, G. Vitalib, High temperature-induced crystallization in tantalum pentoxide layers andits influence on the electrical properties, Thin Solid Films426 (2003) 191-199 M.V. Ivanov, T.V. Perevalov, V.S. Aliev, V.A. Grit-senko,V.V. Kaichev, Electronic structure of δ − Ta O withoxygen vacancy: ab initio calculations and comparisonwith experiment, J. Appl. Phys. 110 (2011) 024115 M.V. Ivanov, T.V. Perevalov, V.Sh. Aliev, V.A. Gritsenko,V.V. Kaichev, Ab Initio Simulation of the Electronic Struc-ture of δ − Ta O with Oxygen Vacancy and Comparisonwith Experiment, J. Exp. Theo. Phys. 112 (2011) 1035-1041 H. Szymanowski, O. Zabeida, J. E. Klemberg-Sapieha, L.Martinu, Optical properties and microstructure of plasmadeposited Ta O and Nb O films, J. Vac. Sci. Technol. A23 (2005) 241-247 O. Kerrec, D. Devilliers H. Groult, P. Marcus, Study of dryand electrogenerated Ta O and Ta/Ta O /Pt structuresby XPS, Mat. Sci. Eng. B55 (1998) 134-14253