Analysis of titanium species in titanium oxynitride films prepared by plasma enhanced atomic layer deposition

Accepted Manuscript Analysis of titanium species in titanium oxynitride films prepared by plasma enhanced atomic layer deposition

Małgorzata Kot, Karsten Henkel, Chittaranjan Das, Simone Brizzi, Irina Kärkkänen, Jessica Schneidewind, Franziska Naumann, Hassan Gargouri, Dieter Schmeißer PII: DOI: Reference:

S0257-8972(16)31270-1 doi: 10.1016/j.surfcoat.2016.11.094 SCT 21840

To appear in:

Surface & Coatings Technology

Received date: Revised date: Accepted date:

31 May 2016 4 November 2016 25 November 2016

Please cite this article as: Małgorzata Kot, Karsten Henkel, Chittaranjan Das, Simone Brizzi, Irina Kärkkänen, Jessica Schneidewind, Franziska Naumann, Hassan Gargouri, Dieter Schmeißer , Analysis of titanium species in titanium oxynitride films prepared by plasma enhanced atomic layer deposition. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Sct(2016), doi: 10.1016/ j.surfcoat.2016.11.094

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Analysis of titanium species in titanium oxynitride films prepared by plasma enhanced atomic layer deposition Małgorzata Kota, Karsten Henkela,*, Chittaranjan Dasa;#, Simone Brizzia, Irina Kärkkänenb, Jessica Schneidewindb, Franziska Naumannb, Hassan Gargourib, and Dieter Schmeißera a

Brandenburg University of Technology Cottbus-Senftenberg, Applied Physics and Sensors,

PT

K.-Wachsmann-Allee 17, 03046 Cottbus, Germany [email protected]; [email protected]*; [email protected];

RI

[email protected]; [email protected]

SC

*corresponding author: phone: +49 355 694069, fax: +49 355 693931 #present address: Technische Universität Darmstadt, FG Oberflächenforschung

SENTECH Instruments GmbH, Schwarzschildstraße 2, 12489 Berlin, Germany

MA

b

NU

Jovanka-Bontschits-Str. 2, 64287 Darmstadt, Germany

[email protected]; [email protected];

PT E

D

[email protected]; [email protected]

Abstract

A comparative study of thin titanium oxynitride (TiOxNy) films prepared by plasma enhanced

CE

atomic layer deposition using tetrakis(dimethylamino)titanium (TDMAT) and N2 plasma as

AC

well as titanium(IV)isopropoxide and NH3 plasma is reported. The comparison is based on the combination of Ti2p core level and valence band spectroscopy and current-voltage measurements. The TDMAT/N2 process delivers generally higher fractions of TiN and TiON within the Ti2p spectra of the films and stronger photoemissions within the bandgap as resolved in detail by high energy resolution synchrotron-based spectroscopy. In particular, it is shown that higher TiN contributions and in-gap emission intensities correlate strongly with increased leakage currents within the films and might be modified by the process parameters and precursor selection.

ACCEPTED MANUSCRIPT Keywords: titanium oxynitride; plasma enhanced atomic layer deposition (PEALD); ALD process

AC

CE

PT E

D

MA

NU

SC

RI

PT

parameters; Ti-N contributions; in-gap defect states; leakage current

ACCEPTED MANUSCRIPT 1. Introduction Titanium oxynitride (TiOxNy) films are widely used, where the oxygen to nitrogen (O/N) ratio determines the field of application, e.g. nitrogen-rich films are used as anti-reflective coatings [1] or biomaterials [2] while oxygen-rich films are applied for thin film resistors [3] or solar selective collectors [4]. In particular, nitrogen doping of titanium oxide (TiO2) is

PT

essential for the absorption within the visible light wavelength for photo-catalytic and photovoltaic applications [5-8]. Several deposition techniques including metal-organic

RI

chemical vapor deposition (MOCVD) and reactive magnetron sputtering have been used to

SC

produce TiOxNy films [9,10]. The atomic layer deposition (ALD) allows the fabrication of high-quality ultra-thin films with high conformity and homogeneity, where in particular a

NU

low-temperature budget can be applied in the plasma enhanced ALD (PEALD) [11-13]. The

MA

ALD process allows typically the variation of the process parameters promoting the achievement of a desired functionality via the modification of the electronic, optical and/or

D

electrical properties of the deposited layers [14-16]. Ultra-thin ALD layers became very

PT E

attractive within the last years for a wide range of applications such as batteries, (electro)photo-catalysis, photovoltaics, or fuel cells (to name a few) [17-20]. Here, in particular the high controllability and low process temperatures can be beneficially applied. For example,

CE

recently a thin “leaky” TiO2 ALD layer was used to protect silicon (Si) photo-electrodes

AC

against corrosion in (electro-)photo-catalytic devices (water splitting) and performance improvements could be demonstrated [21-25]. Here, the thin layer is on the one hand inhibiting the direct contact of the small-band-gap semiconductor to the electrolyte but on the other hand still delivering large current densities despite a high difference between the relevant energetic band edges between Si and TiO2. Hence, the modification of the nitrogen content within TiOxNy layers via the ALD parameters could be important for this kind of application to further improve the energy conversion efficiency. Recently we have compared two PEALD processes of TiOxNy films [26,27] where an oxygen-free and an oxygen-

ACCEPTED MANUSCRIPT containing titanium precursor were used. For the former process tetrakis(dimethylamino)titanium (TDMAT) was used together with nitrogen (N2) plasma whereas for the later one titanium(IV)isopropoxide (TTIP) was applied in combination with ammonia (NH3) plasma. In Ref. 26, laboratory-based X-ray photoemission spectroscopy (XPS) was used to demonstrate that both processes can be applied in combination with

PT

varying ALD parameters to produce TiOxNy layers with modified O/N ratio and electrical conductivity. While the TTIP/NH3 process delivered in general oxygen-rich films, in the

RI

TDMAT/N2 process nitrogen-rich films accompanied by higher conductivity were

SC

produced. These results are based on the analysis of the N1s, O1s and Ti2p core level XPS spectra. It was further speculated that based on exemplarily performed XPS spectrum

NU

decomposition results the current density might be correlated to the higher TiN contributions

MA

within the films. In Ref. 27, we have analyzed the nitrogen species by synchrotron-based spectroscopy at the N1s edge and on the N1s core levels, respectively. The main focus was on

D

the contributions of Ti-N and Ti-ON bonds across the TiOxNy films and on other nitrogen

PT E

contributions resulting from precursor residuals. In this work, we highlight in the first part the peak decomposition of the laboratory-based Ti2p core level XPS data of the whole sample set in order to get a complete picture of the

CE

species bonded to titanium (including in advance to Ref. 27 also TiO2). Based on these results

AC

the aforementioned speculation [26] that in particular TiN species are responsible for the higher current densities within the films is confirmed in part two. In the third and fourth parts synchrotron radiation-based XPS (SR-XPS) is applied to record the Ti2p core levels and valance band (VB) spectra of selected films. Using these results; we discuss a non-destructive depth-profiling of the films and investigate in-gap states in the valence band and its correlation on the current density. Hereby, the role of the process parameters substrate temperature (Tsubstr), plasma power (PP) and plasma pulse duration (Ptime), and their

ACCEPTED MANUSCRIPT influence on the titanium composition across the TiOxNy film, on the in-gap state intensity as well as on the leakage currents of these films are discussed. 2. Materials and methods All TiOxNy layers were prepared in the SI ALD LL reactor (SENTECH Instruments GmbH) [14,15] on Si (001) n-type 4” wafers (1-30cm). One complete ALD cycle was typically

PT

performed in the sequence of titanium precursor pulse/purge/nitrogen plasma pulse/purge with the durations of 0.5s/3s to 5s/Ptime/2s and 0.5s/15s/30s/15s in the TTIP/NH3 and

RI

TDMAT/N2 processes, respectively, using nitrogen gas flow rates of 40sccm/144sccm and

SC

60sccm/265sccm during the titanium/nitrogen plasma pulses correspondingly. The complete details of the PEALD process of the TiOxNy films are given elsewhere [26,27]. The

NU

investigated samples are summarized in Table I including the process parameters Tsubstr, PP

MA

and Ptime. The name index “I-” is used for the TTIP/NH3 processed samples while “M-” is chosen for the TDMAT/N2 ones. The varied parameter is then added to the sample name (see

D

Table I).

PT E

For XPS and electrical characterization pieces of 10mm x 10mm were cut from the middle of the wafers.

Laboratory-based XPS study was performed using an Al Kα radiation source (1486.6eV,

CE

Specs GmbH) and a hemispherical electron analyzer (Omicron) where the photo-excited

AC

electrons were collected at a take-off angle () of 90°. SR-XPS was conducted at the undulator beamline U49/2-PGM2 [28] at BESSY-II in Berlin/Adlershof with the ASAM endstation [29]. Here the excitation energy was 620eV and the PHOIBOS-150 (SPECS GmbH) hemispherical electron analyzer equipped with a 1D delay line detector was employed to collect the photo-electrons at =45°. During XPS and SR-XPS the base pressure of the measurement chamber was better than 1 × 10-9 mbar. For analysis, all XPS data were background corrected [30] and their decomposition was performed with Gaussian-Lorentzian line profiles. For current-voltage (I-V) measurements an aluminum contact having a diameter

ACCEPTED MANUSCRIPT of 500µm was thermally evaporated through a shadow mask and the I-V data were recorded on the final metal-insulator-semiconductor (MIS) stack using an Agilent E3649A power supply unit and PREMA4001 and HP34401 ampere and volt meters, respectively. In the I-V loop, the voltage was started at 0V, then it was driven towards negative direction (i.e. inversion of semiconductor surface) and afterwards to its positive counterpart (i.e.

PT

accumulation) applying a step width of 0.1V and a ramp of approximately 0.09V/s.

RI

Table I: List of the investigated samples, their process parameters and the calculated ratios of

SC

the titanium species within the Ti2p core level XPS data (TiO2/Ti2p, TiON/Ti2p, TiN/Ti2p) measured with Al K excitation. The corresponding O/N and O/Ti ratios are given for

NU

completeness and if marked by a star taken from [26]. The samples selected for synchrotron

MA

measurements (Ti2p core level and VB spectra taken with excitation energy of 620eV) are checked in the column “Syn”.

TDMAT/ N2

Ptime [s] 8 8 8 6 8 15 8 8 8 30 30 30

t [nm] 6.3 6.6 6.5 7.0 6.2 7.8 6.2 6.8 7.8 6.9 7.3 21.6

D

PP [W] 100 150 200 100 100 100 100 100 100 200 200 200

PT E

I-100W I-150W I-200W I-6s I-8s I-15s I-200C I-240C I-300C M-250C M-285C M-350C

Tsubstr [°C] 200 200 200 200 200 200 200 240 300 250 285 350

CE

TTIP/ NH3

Sample

AC

Process

TiO2/Ti2p [%] 97.7 61.6 62.9 96.1 96.4 97.7 93.0 96.3 96.8 28.1 29.1 27.9

TiON/Ti2p [%] 2.3 32.0 27.6 3.9 3.6 2.3 7.0 3.7 3.2 55.0 51.1 47.6

3. Results and Discussion 3.1 Laboratory-based XPS data of the Ti2p core level

TiN/Ti2p [%] 0 6.4 9.5 0 0 0 0 0 0 16.9 19.8 24.5

O/N

O/Ti

*

2.3 2.3* 2.3* 2.3 2.4 2.5 2.5* 2.3* 2.6* 1.0* 0.9* 0.9*

4.4 3.8* 4.0* 4.0* 4.3* 5.9* 4.3* 5.3* 10.6* 0.7* 0.7* 0.8*

Syn x x

x x x

ACCEPTED MANUSCRIPT Laboratory-based XPS measurements were performed on all samples given in Table I. The Ti2p core level spectra recorded with Al K excitation are depicted in Figure 1. The spectra of the two processes, namely TTIP/NH3 and TDMAT/N2, exhibit strong differences. While the signals of the TTIP/NH3 films, excepting the samples I-150W and I-200W, are more or less TiO2-like; in the TDMAT/N2 processed samples and the aforementioned samples I-150W

PT

and I-200W stronger Ti(O)N contributions within the films affect the line shapes of the Ti2p core levels obviously. Within the TiO2-like spectra of the TTIP/NH3 sample set the spectrum

RI

of the I-300C sample is slightly shifted (~0.3eV) towards higher binding energy. The Tsubstr

SC

(300°C) of this sample is not within the ALD window of this process (190-270°C [26]) which likely leads to precursor decomposition and changed chemistry prior to the surface reaction.

NU

But also the measurement accuracy as well as the step width in the measurement of the

TTIP/NH

Ti2p

3

AlK

I-200W

MA

laboratory-based XPS (0.1eV) should be taken into consideration here.

D

Normalized Intensity

I-150W

I-100W

PT E

I-15s I-8s I-6s

I-300C I-250C I-200C

468

CE

2

AC

TDMAT/N

465

462

459

456

M-350C M-285C M-250C

453

Binding Energy [eV]

Figure 1: (Color Online) Ti2p core level spectra of the TiOxNy films investigated in this work recorded with Al K excitation. The sample names are indicated next to the corresponding spectrum.

Peak decomposition of the Ti2p core level is performed after background correction for all samples and Figure 2 shows representative examples for both PEALD processes, TTIP/NH3

ACCEPTED MANUSCRIPT (Fig. 2a) and TDMAT/N2 (Fig. 2b). The samples I-150W (Fig. 2a) and M-250C (Fig. 2b) are chosen for illustration as these samples were prepared keeping the substrate temperature within the ALD window [26] and show also, as mentioned above, reasonable different signals than pure TiO2 layers. The main differences between the two processes are still distinctly

1

TTIP/NH3

PT

visible in Fig. 2.

Ti2p AlK

I-150W

RI SC

a) 0

TDMAT/N2

NU

1

M-250C

MA

Normalized Intensity

data TiON TiN TiO2

b) 465

460

455

D

0 470

PT E

Binding Energy [eV]

Figure 2: (Color Online) Ti2p core levels (Al K excitation) of the a) I-150W sample

CE

prepared within the TTIP/NH3 process and b) M-250C sample fabricated within the

AC

TDMAT/N2 process and the corresponding peak decomposition. The peak decomposition contributions are given in the legend in part a).

Both spectra can be decomposed into three components, namely TiO2, TiON, and TiN [31], however the relative contributions of Ti(O)N components is much higher within the M-250C sample prepared by the TDMAT/N2 process. Within the Ti2p spectrum of the TTIP/NH3 process (Fig. 2a, sample I-150W) the doublet peak at the binding energies of 458.9eV (Ti 2p3/2) and 464.6eV (Ti 2p1/2) is assigned to TiO2, the doublet at 457.5eV and 463.3eV to

ACCEPTED MANUSCRIPT TiON and the doublet at 455.6eV and 461.4eV to TiN. The TDMAT/N2 sample M-250C delivers the three components inside the Ti2p spectra (Fig. 2b) as follows: TiO2 at 458.7eV and 464.4eV, TiON at 456.8eV and 462.6eV and TiN at 455.5eV and 461.2eV. All Ti2p contributions of both processes show a typical spin orbit splitting of 5.7eV to 5.8eV [31]. The positions of the binding energies of the TiO2 and TiN components are similar for the

PT

I-150W and M-250C samples taking into account the measurement accuracy and analyzer step width (0.1eV). However, in the I-150W sample the TiON component is shifted by 0.7eV

RI

towards higher binding energy in comparison to the M-250C sample. Here, the plasma

SC

reactant (NH3) may deliver hydrogen species within the film, which are not fully removed during the purging step. The N1s core level spectra of these samples confirm this fact [27]. It

NU

should be also noted that the decomposition of the Ti2p spectrum of the I-150W sample could

MA

also be performed reasonably without considering TiN contributions. Subsequently, the resulting TiON content would be higher but the decomposed TiON peaks would appear at

D

binding energies of 456.9eV and 462.7eV and hence quite similar to the M-250C sample.

PT E

However, the N1s peak decomposition of this sample [27] rather confirms the existence of TiN contributions. Despite this discussion, the sum of Ti(O)N contributions in the Ti2p core level of the TTIP/NH3 I-150W sample is 38.4% (see Table I) and only approximately the half

CE

of the value in the TDMAT/N2 processed sample M-250C (71.9%). This underlines the fact

AC

that the TDMAT/N2 process produces more nitrogen inside the films than the TTIP/NH3 process. Only the relative decomposition of the Ti2p spectra was reported here. Looking further to the O/N and O/Ti ratios determined from the total film composition [26] it gets even clearer that the oxygen containing precursor process (TTIP/NH3) produces oxygen-rich films. The O/N ratios of all TTIP/NH3 samples investigated in this work were between 4 and around 11, while the ones of the TDMAT/N2 process are below 0.8. Also the O/Ti ratios of the TTIP/NH3 samples are much higher (>2) than of the TDMAT/N2 samples (~1). These values

ACCEPTED MANUSCRIPT are repeated for completeness in Table I. The Ti2p peak decomposition of all samples is

TDMAT/N2

TTIP/NH3

1.0 0.8

TiO2

0.6

TiON 350°C

285°C

250°C

300°C

240°C

Tsubstr

SC

Tsubstr

RI

Ptime

PP

200°C

8s

15s

6s

200W

0.0

150W

0.2

PT

TiN

0.4

100W

Relative Content in Ti2p

summarized in Table I and illustrated in Figure 3.

NU

Figure 3: (Color Online) Ti2p core level decomposition of all samples investigated in this work (with Al K excitation) in dependence of the process (TTIP/NH3, TDMAT/N2) and

MA

parameter (PP, Ptime, Tsubstr) selection. The color code is given beside the diagram.

D

Beside the fact of nitrogen- (TDAMT/N2) and mostly oxygen-richness (TTIP/NH3), it’s

PT E

noteworthy that only the samples I-150W and I-200W contain higher TiON and desirable TiN contributions whereas in all other samples of this TTIP/NH3 process no TiN contributions and

CE

only small TiON signals (equal to 7% or less within the Ti2p) could be detected. Hence, as the TDMAT/N2 process samples were characterized after the TTIP/NH3 films, the focus for

AC

the TDMAT/N2 process was set to the variation of Tsubstr at higher fixed plasma power of 200W. There seems to be a small dependence of the TiON signal within the Ti2p on Tsubstr in both processes. In the TTIP/NH3 process this contribution is decreasing from 7% to 3.2% within the temperature range of 200°C to 300°C and in the TDMAT/N2 process from 55% to 47.6% for 250°C ≤ Tsubstr ≤ 350°C. In addition an increase of the TiN signals with the increase of Tsubstr can be observed for the TDMAT/N2 samples. Based on the results reported in this chapter representative samples were selected for further SR-XPS measurements (see chapter 3.3).

ACCEPTED MANUSCRIPT 3.2 I-V characterization A generally higher leakage current for the TDMAT/N2 samples in comparison to the TTIP/NH3 samples was already reported in Ref. 26. In the same way a dependence of the leakage current on Tsubstr was observed for the TDMAT/N2 processed samples while the current was more or less constant with Tsubstr in the TTIP/NH3 processed samples [26]. In this

PT

work we show the I-V measurements on selected samples (Fig. 4) supporting the aforementioned fact that the nitrogen-rich samples of the TDMAT/N2 process possess higher

RI

currents with ohmic-like behavior. Here, I-V measurements on the same samples as selected

SC

for SR-XPS characterization (chapters 3.3 and 3.4) are shown. All I-V data exhibit diode-like

12

M-350C

MA

M-285C

10 8

M-250C

D

6 4

PT E

2

Current Density [A/cm ]

(positive voltages) the current increases.

NU

character typical for MIS stacks with n-type semiconductor where in the accumulation region

2 0 0

2

4

I-150W

I-100W

6

CE

Electric Field [MV/cm]

AC

Figure 4: (Color Online) I-V characteristics of selected TiOxNy samples of the TTIP/NH3 process: I-100W (rectangles) and I-150W (open triangles) as well as of the TDMAT/N2 process: M-250C (circles), M-285C (stars) and M-350C (reversed triangles).

The current density data of the TDMAT/N2 samples at a fixed electric field (0.93MV/cm) are plotted versus Tsubstr in Fig. 5a. In addition the results of chapter 3.1 are used to plot the TiON and TiN contributions within the Ti2p core level data for the TDMAT/N2 processed samples versus Tsubstr (Fig. 5b) in order to investigate the probable influence of these components onto

ACCEPTED MANUSCRIPT the I-V characteristics of these films. Here it should be noted that the TiN and TiON contributions within the Ti2p core level data are related to the total film composition to link them to the I-V characteristic, which reflects the bulk behavior of the complete film, and are

TDMAT/N2

2

a)

PT

10

0.1

RI

1

14

SC

E=0.93 MV/cm

b)

12

8 6 TiN (of Ti2p)

4 240

260

280

300

320

NU

TiON (of Ti2p)

10

MA

Rel. total amaount [%] Current dens. [A/cm ]

plotted as total relative amount in Fig. 5b.

340

360

D

Substrate Temperature [°C]

PT E

Figure 5: (Color Online) Dependence of a) the current density (rectangles, at electric field of 0.93MV/cm) and b) TiN (open triangles) and TiON (open circles) contents within the Ti2p

CE

core level spectra (recorded with Al K excitation) on the substrate temperature of the

AC

PEALD process for the TDMAT/N2 processed samples. The TiN and TiON contents are referred to the total film composition (given in relative total amount).

It’s obvious that the TiN signal strength follows the same trend with Tsubstr as the leakage current density whereas the TiON content stays relatively constant with Tsubstr. A similar trend can be seen when the TiN and TiON signals within the N1s core levels [27] are analyzed in the same way (data not shown). Therefore, it is conclusive that the resistivity of the films depends strongly on the nitrogen directly bonded to titanium within the film which depends on the Tsubstr of the TDMAT/N2 PEALD process. The reader should note that the leakage

ACCEPTED MANUSCRIPT current data are shown in logarithmic scale whereas the TiN contributions linearly, that’s why the decrease of the resistivity with the TiN content increase should follow an exponential relation. For films prepared by MOCVD using the TDMAT precursor decreased resistivity with increasing process temperature is also reported [32,33]. In the films of the Tsubstr dependent series of the TTIP/NH3 process (samples I-200C, I-240C

PT

and I-300C) no TiN contributions were detected. Hence, the independence of the current density on Tsubstr for these samples reported in [26] might be cross-linked to that fact. In

RI

contrast, the small fluctuation of the TiON content with Tsubstr in these samples seems to have

SC

no effect on the current density. Comparing the I-V characteristics of the samples I-100W and I-150W in Fig. 4, a higher current density is observed for the sample I-150W. For this

NU

TTIP/NH3 sample a reasonable TiN signal appeared within the Ti2p core level while no TiN

MA

was found in the I-100W sample (refer to Fig. 3). However, the TiN contribution of the I150W sample is smaller than in all TDMAT/N2 samples (see Fig. 3 and Table I). These two

D

facts and the appearance of the I-V characteristic of the I-150W sample in between the ones

PT E

of the TDMAT/N2 and the I-100W samples support the correlation of the current density with the amount of Ti-N chemical bonds.

CE

3.3 SR-XPS data of the Ti2p core levels Based on the results reported in chapter 3.1 the following samples were selected for SR-XPS

AC

measurements: I-100W, I-150W to further investigate the influence of the plasma power in the TTIP/NH3 process and all three samples of the TDMAT/N2 process to examine the Tsubstr effect. In particular these samples were also chosen to record the VB spectra with high resolution (see chapter 3.4). Figure 6 depicts the SR-XPS core level spectra of the mentioned samples recorded with an excitation energy of 620eV. Also here a clear difference between both processes is evident. The shoulder at lower binding energies of the main peak (~455eV to 458eV) due to Ti(O)N contributions is much more pronounced in the TDMAT/N2 films

ACCEPTED MANUSCRIPT than in the TTIP/NH3 samples. Furthermore, the signal intensity of this shoulder is higher in the I-150W sample than in the I-100W sample. a)

TTIP/NH3

Ti2p I-150W I-100W

TDMAT/N2

PT

b)

M-350C

RI

Normalized Intensity

620eV

M-285C

470

465

460

SC

M-250C

455

NU

Binding Energy [eV]

Figure 6: (Color Online) Ti2p core levels recorded with SR-XPS (excitation energy of 620eV)

MA

of a) the I-100W and I-150W samples prepared within the TTIP/NH3 process and b) M-250C, M-285C and M-350C samples fabricated within the TDMAT/N2 process. The sample names

PT E

D

are indicated next to the corresponding spectrum.

The same samples as used in Fig. 2 are selected for the illustration of the peak decomposition

CE

and to discuss the film composition along the depth of the film. Accordingly, the peak decompositions of the Ti2p SR-XPS core level spectra are shown for the samples I-150W and

AC

M-250C in Fig. 7. Again, these spectra are decomposed into the three components TiO2, TiON, and TiN [31]. Within the Ti2p spectrum of the TTIP/NH3 sample I-150W (Fig. 7a) the TiO2 doublet peak appears at binding energies of 459.2eV (Ti 2p3/2) and 464.9eV (Ti 2p1/2), the TiON doublet at 457.6eV and 463.4eV and the TiN doublet at 455.6eV and 461.4eV, respectively. The TDMAT/N2 sample M-250C delivers the three components inside the Ti2p spectra (Fig. 1b) as follows: TiO2 at 458.8eV and 464.5eV, TiON at 456.9eV and 462.7eV and TiN at 455.5eV and 461.2eV. As in the laboratory-based XPS data (Fig. 2), also in the

ACCEPTED MANUSCRIPT SR-XPS data (Fig. 7) all Ti2p contributions show typical spin orbit splits of 5.7eV to 5.8eV [31]. TTIP/NH3

Ti2p

I-150W

620eV

data TiON TiN TiO2

a) 1

PT

0

TDMAT/N2

RI

M-250C

SC

Normalized Intensity

1

b) 0 470

465

460

455

MA

NU

Binding Energy [eV]

Figure 7: (Color Online) SR-XPS Ti2p core levels (excitation energy of 620 eV) of a) the I150W sample prepared within the TTIP/NH3 process and b) M-250C sample fabricated within

PT E

D

the TDMAT/N2 process and the corresponding peak decomposition. The peak decomposition contributions are given in the legend in part a).

CE

It is obvious that in particular the TiN contribution is higher in the TDMAT/N2 sample

AC

compared to the TTIP/NH3 sample. The positions of the binding energies of the single components of the Ti2p spectra recorded with an excitation energy of 620eV in the SR-XPS (Fig. 7) are very similar (within the step width of measurement) to the ones recorded with Al Ka excitation in the laboratory-based XPS (Fig. 2) for both samples. Only the TiO2 contributions of the I-150W sample show a slight shift of 0.3eV towards higher binding energy in the 620eV data in respect to the Al K data. Here some contribution of OH groups might be reasonable as the measurements were done ex-situ as well as the wafers were stored in normal environmental conditions between processing and characterization. For excitation

ACCEPTED MANUSCRIPT energy of 620eV and =45°, the inelastic mean free path (IMFP) of the photo-excited electrons is 6.6Å corresponding to an information depth of ~1.4nm [34]. Hence, the surface of the layer has a higher impact on the measurement and adsorbed OH might contribute to the signals. In contrast, with Al K excitation (=90°) the IMFP is 21.6Å and an information depth of 6.4nm can be deduced [34] delivering more bulk information of the film. The peak

PT

decomposition data of the samples I-150W and M-250C in dependence of the excitation

RI

I-150W (TTIP/NH3)

M-250C (TDMAT/N2)

SC

1.0 0.8

0.4 0.2 0.0 620 eV

Al K

620 eV TiN

Al K

TiON

D

TiO2

NU

0.6

MA

Relative Content in Ti2p

energy and hence on the information depth are compared in Fig. 8.

PT E

Figure 8: (Color Online) Ti2p core level decomposition of the samples I-150W (TTIP/NH3 process) and M-250C (TDMAT/N2 process) performed on spectra recorded with excitation

CE

energies of 620eV and Al K, respectively, as marked in the diagram. The color code is given

AC

below the diagram.

It shows that in both films the Ti(O)N contributions are reduced at the surface of the films. The effect is stronger in the M-250C sample where surface oxidation seems to be occurring during wafer storage. Nevertheless the TiN content is always stronger within the TDMAT/N2 film compared to the TTIP/NH3 sample. Just the TiON content at the surface near region seems to be comparable for both samples. The data of the I-150W sample support the conclusion that in the TTIP/NH3 process a higher plasma power should be used in order to increase the nitrogen content within the film.

ACCEPTED MANUSCRIPT Summarizing this chapter, surface oxidation is clearly visible in particular for the TDMAT/N2 film. This might open the hypothesis that the initial film of this process without any oxygen content in the precursors is in fact TiN and the contact to air promotes the surface oxidation. Typically XPS measurements are not necessarily bulk sensitive. However, in our case, the information depth of the XPS recorded with Al K excitation (6.4nm) is similar to the film

PT

thickness of the M-250C sample (6.9nm). Even though an oxygen-free titanium precursor is used in the TDMAT/N2 process, still strong TiO2 and TiON contributions are observed deep

RI

in the bulk. Therefore, it seems to be rather reasonable that beside the surface oxidation

SC

residual oxygen and water in the chamber contribute strongly to the final film composition due to the thermodynamically favorable oxidation in comparison to the nitridation [35]. Here

NU

in-situ spectroscopic experiments directly performed after the film deposition [36] might be

MA

helpful to spotlight this issue. However, we would like to emphasize that all samples were produced in the wafer scale reactor at which XPS cannot be performed with our facilities.

D

3.4 In-gap defect states in the valence band spectra

PT E

In the laboratory-based XPS investigations we found hints for states within the electronic bandgap. Therefore, VB spectra were recorded using SR-XPS with high resolution. Here the

CE

focus of sample selection (see chapter 3.3) was set to monitor a wide range of the aforementioned differences within the sample sets. Figure 9 shows the VB spectra of the

AC

corresponding samples recorded with excitation energy of 620eV. In fact, strong electron emissions within the energetic region of the bandgap located at around 0.5eV to 0.9eV below the Fermi energy can be observed for the TDMAT/N2 samples. These emissions are highlighted by peak area filling within the spectra in the corresponding energy region in Fig. 9. These emissions are also but less pronounced present in the TTIP/NH3 films, where the intensity is higher for the sample I-150W than for the sample I-100W.

ACCEPTED MANUSCRIPT a)

TTIP/NH3

VB

Normalized Intensity

620eV

I-150W I-100W

TDMAT/N2

b)

PT

M-350C M-285C

14

12

10

8

6

4

2

0

RI

M-250C

VBM

-2

SC

Binding Energy [eV]

NU

Figure 9: (Color Online) VB spectra recorded with SR-XPS (excitation energy of 620 eV) of a) the I-100W and I-150W samples prepared within the TTIP/NH3 process and b) M-250C,

MA

M-285C and M-350C samples fabricated within the TDMAT/N2 process. The sample names are indicated next to the corresponding spectrum. The in-gap state peaks are illustrated by

PT E

D

filling the area under the peaks.

The origin of these signals can be attributed to Ti+3 (3d) defect states [37,38]. In TiO2 ALD

CE

films this kind of defects is also observed, where a higher photoemission intensity is found in case of resonant excitation (at the Ti-L edge) [16,39]. Recently, for the TiO2 rutile surface the

AC

presence of emissions ~1eV below the conduction band minimum was discussed in terms of polaronic states formed by localized excess electrons from shallow donor states which might be caused by bridging oxygen vacancy and adsorbed or substitutional hydrogen [40]. Nitrogen doping might promote this effect too. In the present work the excitation in the VB measurement was done off-resonant. As the in-gap state related signals in the off-resonant data of the TiOxNy films of this work are higher than in the off-resonant data of ALD TiO2 films [16,39] we suspect that the in-gap state density is higher when nitrogen doping is done. This is also evident when comparing the Ti2p core level decomposition details of the TiOxNy

ACCEPTED MANUSCRIPT samples (Fig. 3, Table I) with the peak intensity of the in-gap states. It’s obvious that the samples with higher nitrogen content have also higher in-gap state density. Finally, in Fig. 10 the leakage current values taken from Fig. 4 at different electric fields are plotted versus the in-gap state intensity (corresponding to its peak area as illustrated in Fig. 9) in a normalized manner (the values are normalized to the data of the sample I-100W). The

PT

TDMAT/N2

RI

TTIP/NH3

SC

100

TSubstr

10

@ E [MV/cm] 0.5 0.75 0.93

PP

1 1.0

1.5

2.0

2.5

MA

Normalized In-Gap State Intensity

NU

Normalized Current Density

current density is scaled logarithmical and the in-gap state intensity linearly.

Figure 10: (Color Online) Normalized current density versus normalized in-gap state

D

intensity. Samples used for this diagram are those investigated at the synchrotron (see Table

PT E

1). The values are deduced from the currents (Fig. 4) at different fixed electric fields of 0.5MV/cm (open rectangles), 0.75MV/cm (circles) and 0.93MV/cm (triangles) as well as

CE

from the area under the in-gap state line shapes (illustrated in Fig. 9) and normalized in each case to the lowest value corresponding to the sample I-100W. The arrows indicate the

AC

influence of rising plasma power and substrate temperature.

A clear indication of an exponential dependence of the current density (and hence the conductivity) of the films on the in-gap state intensity (and hence the in-gap state density) can be observed and is indicated by the dashed line in Fig. 10. This trend is similar to the current behavior in respect to the TiN content within the Ti2p core level discussed in dependence of Tsubstr in Fig. 5 (see chapter 3.2). Here the trend is more generalized by the normalization done in Fig. 10 and the fact that the data of both PEALD processes, TTIP/NH3 and TDMAT/N2,

ACCEPTED MANUSCRIPT are combined. Hence we can argue that the nitrogen doping and in particular the nitrogen bonded in the form of Ti-N promote the appearance of in-gap states which are causing higher leakage currents. The in-gap states are directly related to the existence of Ti3+ states [37,38]. In that respect the Ti-N bonds help to provide a stable doping of the TiO2 films whereas oxygen vacancies and the Ti-ON bonds are not stable but are depending on the oxygen partial

PT

pressure and cause long term instabilities. It should be mentioned that beside the aforementioned doping process via Ti3+ formation there is also the possibility that the Ti-N

RI

bonds cause a softening of the TiO2 lattice and thereby enable the existence of small polarons

SC

[39]. Small polarons also contribute to the intensity of the in-gap states [40].

NU

4. Conclusion

A comparison of thin TiOxNy films prepared within two different PEALD processes was

MA

reported, where the oxygen-free and oxygen containing precursors TDMAT and TTIP, respectively, were used in combination with N2 and NH3 plasma sources. The comparison is

D

based on the combination of spectroscopic and electrical characterization, where Ti2p core

PT E

level and valence band spectra were recorded with laboratory-based and synchrotron-based photoemission spectroscopy as well as current-voltage measurements were carried out. The

CE

influence of process parameters on the film composition and its subsequent influence on the electrical characteristic were investigated with a special focus on the contributions within the

AC

Ti2p core level data as well as on photoemissions within the bandgap region of the VB spectra. We like to emphasize that these data represent one of the rare systems in which a direct comparison of spectroscopic and electric properties is demonstrated. We found that the TDMAT/N2 process delivers generally higher contributions of TiN and TiON within the Ti2p spectra of the films and stronger photoemissions within the bandgap. Strong surface oxidation in particular of the TDMAT/N2 films was observed by nondestructive depth profiling using different excitation energies in XPS measurements. The applicability of this method for the TiOxNy films of this work was discussed as XPS is

ACCEPTED MANUSCRIPT typically not bulk sensitive. The surface oxidation and the strong oxygen contributions within the films (even though an oxygen-free titanium precursor is used in combination with N2 plasma within the TDMAT/N2 process) were attributed to the sample storage in air, on the one hand, and to oxygen and water residuals in the ALD reactor, on the other hand. In particular, it was shown that higher TiN contributions and in-gap emission intensities

PT

correlate strongly with increased leakage currents within the films. The in-gap states were directly related to the existence of the Ti3+ states where the Ti-N bonds assist the stable

RI

nitrogen doping [37,38]. The Ti-N bonds can further cause the existence of small polarons

SC

contributing to the in-gap states intensity as well [39,40].

The TiN contribution and in-gap state intensities are varying strongly with the plasma power

NU

in the TTIP/NH3 process and with the substrate temperature in the TDMAT/N2 process. As

MA

these properties depend sensitively on the mentioned process parameters the conductivity and finally the functionality of the TiOxNy layers can be modified by the process parameters and

D

the precursor selection of the PEALD process within the film fabrication. For the materials

PT E

science community our data enable a careful adjustment of the preparation parameters to obtain the desired electrical and optical properties of the films. Our results are helpful to limit

strategies.

CE

the choice and variation of the parameters and avoid long-lasting empirical optimization

AC

Furthermore the results might be important for (photo-)electro-chemical and photovoltaic energy conversion systems. For example, ultra-thin ALD passivation layers with adjusted conductivity may contribute to improved long-term stability of Si photo-electrodes [21-25]. Applications of photo-absorbers (where typically thicker layers are necessary for efficient absorption) within the visible light range might benefit from our results as well when combined with the ongoing development in the field of spatial ALD [13,41] with much higher growth rates.

ACCEPTED MANUSCRIPT Acknowledgements This work was partially supported by the German Federal Ministry of Education and Research (BMBF, grant numbers 03IN2V4A, 03IN2V4B) and the German Research Foundation (DFG, SCHM 745/31-1). We acknowledge the support of G. Beuckert, C. Schwiertz, and F.

AC

CE

PT E

D

MA

NU

SC

RI

PT

Reichmann (BTU Cottbus-Senftenberg) in XPS measurements and data handling.

ACCEPTED MANUSCRIPT References [1] F. Maury, F.-D. Duminica, TiOxNy coatings grown by atmospheric pressure metal organic chemical vapor deposition, Surf. Coat. Technol. 205 (2010) 1287-1293. [2] J. Koerner, L.A. Butterworth, I.V. Mayer, R. Dasbach, H.J. Busscher, Bacterial adhesion to titanium-oxynitride (TiNOX) coatings with different resistivities: a novel approach for the development of biomaterials, Biomaterials 23 (2002) 2835-2840.

PT

[3] N.D. Cuong, D.-J. Kim, B.-D. Kang, C.S. Kim, S.-G. Yoon, Characterizations of high resistivity TiNxOy thin films for applications in thin film resistors, Microelectron. Reliab. 47 (2007) 752-754.

RI

[4] C. Nunes, V. Teixeira, M.L. Prates, N.P. Barradas, A.D. Sequeira, Graded selective coatings based on chromium and titanium oxynitride, Thin Solid Films 442 (2003) 173-178.

SC

[5] R. Asahi, T. Morikawa, T. Ohwaki, K. Aoki, Y. Taga, Visible-light photocatalysis in nitrogen-doped titanium oxides, Science 293 (2001) 269-271.

MA

NU

[6] S.-I. In, P.C.K. Vesborg, B.L. Abrams, Y. Hou, I. Chorkendorff, A comparative study of two techniques for determining photocatalytic activity of nitrogen doped TiO2 nanotubes under visible light irradiation: Photocatalytic reduction of dye and photocatalytic oxidation of organic molecules, J. Photochem. Photobiol. A: Chem. 222 (2011) 258-262.

D

[7] S. Zhang, F. Peng, H. Wang, H. Yu, S. Zhang, J. Yang, H. Zhao, Electrodeposition preparation of Ag loaded N-doped TiO2 nanotube arrays with enhanced visible light photocatalytic performance, Catal. Commun. 12 (2011) 689-693.

PT E

[8] K. Siuzdak, M. Szkoda, M. Sawczak, A. Lisowska-Oleksiak, Novel nitrogen precursors for electrochemically driven doping of titania nanotubes exhibiting enhanced photoactivity, New J. Chem. 39 (2015) 2741-2751.

CE

[9] F. Fabreguette, L. Imhoff, J. Guillot, B. Domenichini, M.C. Marco de Lucas, P. Sibillot, S. Bourgeois, M. Sacilotti, Temperature and substrate influence on the structure of TiNxOy thin films grown by low pressure metal organic chemical vapour deposition, Surf. Coat. Technol. 125 (2000) 396-399.

AC

[10] S.K. Rawal, A.K. Chawla, R. Jayaganthan, R. Chandra, Structural, Wettability and Optical Investigation of Titanium Oxynitride Coatings: Effect of Various Sputtering Parameters, J. Mater. Sci. Technol. 28 (2012) 512-523. [11] S.M. George, Atomic Layer Deposition: An Overview, Chem. Rev. 110 (2010) 111-131. [12] H.B. Profijt, S.E. Potts, M.C.M. van de Sanden, W.M.M. Kessels, Plasma-Assisted Atomic Layer Deposition: Basics, Opportunities, and Challenges, J. Vac. Sci. Technol. A 29 (2011) 050801. [13] V. Miikkulainen, M. Leskelä, M. Ritala, R.L. Puurunen, Crystallinity of inorganic films grown by atomic layer deposition: Overview and general trends, J. Appl. Phys. 113 (2013) 021301.

ACCEPTED MANUSCRIPT [14] J. Haeberle, K. Henkel, H. Gargouri, F. Naumann, B. Gruska, M. Arens, M. Tallarida, D. Schmeißer, Ellipsometry and XPS comparative studies of thermal and plasma enhanced atomic layer deposited Al2O3-films, Beilstein J. Nanotechnol. 4 (2013) 732-742. [15] K. Henkel, H. Gargouri, B. Gruska, M. Arens, M. Tallarida, D. Schmeißer, Capacitance and conductance versus voltage characterization of Al2O3 layers prepared by PEALD at 25°C ≤ T ≤ 200°C. J. Vac. Sci. Technol. A 32 (2014) 01A107.

PT

[16] C. Das, K. Henkel, M. Tallarida, D. Schmeißer, H. Gargouri, I. Kärkkänen, J. Schneidewind, B. Gruska, M. Arens, Thermal and plasma enhanced atomic layer deposition of TiO2: Comparison of spectroscopic and electric properties, J. Vac. Sci. Technol. A 33 (2015) 01A144.

RI

[17] J. Bachmann, Atomic layer deposition, a unique method for the preparation of energy conversion devices, Beilstein J. Nanotechnol. 5 (2014) 245-248.

SC

[18] Q. Peng, J.S. Lewis, P.G. Hoertz, J.T. Glass, G.N. Parsons, Atomic layer deposition for electrochemical energy generation and storage systems, J. Vac. Sci. Technol. A 30 (2012) 010803.

MA

NU

[19] H. Si, Q. Liao, Z. Zhang, Y. Li, X. Yang, G. Zhang, Z. Kang, Y. Zhang, An innovative design of perovskite solar cells with Al2O3 inserting at ZnO/perovskite interface for improving the performance and stability, Nano Energy 22 (2016) 223-231.

D

[20]: X. Dong, X. Fang, M. Lv, B. Lin, S. Zhang, J. Ding, N. Yuan, Improvement of the humidity stability of organic–inorganic perovskite solar cells using ultrathin Al2O3 layers prepared by atomic layer deposition, J. Mater. Chem. A 3 (2015) 5360-5367.

PT E

[21] S. Hu, M.H. Richter, M.F. Lichterman, J. Beardslee, T. Mayer, B. S. Brunschwig, N.S. Lewis, Electrical, Photoelectrochemical, and Photoelectron Spectroscopic Investigation of the Interfacial Transport and Energetics of Amorphous TiO2/Si Heterojunctions, J. Phys. Chem. C 120 (2016) 3117-3129.

CE

[22] M.F. Lichterman et al., Protection of inorganic semiconductors for sustained, efficient photoelectrochemical water oxidation, Catal. Today 262 (2016) 11-23.

AC

[23] M. T. McDowell, M. F. Lichterman, A. I. Carim, R. Liu, S. Hu, B. S. Brunschwig, N. S. Lewis, The Influence of Structure and Processing on the Behavior of TiO2 Protective Layers for Stabilization of n-Si/TiO2/Ni Photoanodes for Water Oxidation, ACS Appl. Mater. Interfaces 7 (2015) 15189-15199. [24] C. Das, M. Kot, K. Henkel, D. Schmeisser, Engineering of Sub-Nanometer SiOx Thickness in Si Photocathodes for Optimized Open Circuit Potential, ChemSusChem 9 (2016) 2332-2336. [25] C. Das, M. Tallarida, D. Schmeißer, Si microstructures laminated with a nanolayer of TiO2 as long-term stable and effective photocathodes in PEC devices, Nanoscale 7 (2015) 7726-7733. [26] M. Sowińska, K. Henkel, D. Schmeißer, I. Kärkkänen, J. Schneidewind, F. Naumann, B. Gruska, H. Gargouri, Plasma-enhanced atomic layer deposition of titanium oxynitrides films: A comparative spectroscopic and electrical study, J. Vac. Sci. Technol. A 34 (2016) 01A127.

ACCEPTED MANUSCRIPT [27] M. Sowinska, S. Brizzi, C. Das, I. Kärkkänen, J. Schneidewind, F. Naumann, H. Gargouri, K. Henkel, D. Schmeißer, Analysis of nitrogen species in titanium oxynitride ALD films, Appl. Surf. Sci. 381 (2016) 42-47. [28] D.R. Batchelor, R. Follath, D. Schmeißer, Commissioning results of the BTUC-PGM beamline, Nucl. Instrum. Methods Phys. Res. Sect. A 467-468 (2001) 470-473.

PT

[29] D. Schmeißer, P. Hoffmann, G. Beuckert, Electronic Properties of the Interface formed by Pr2O3 growth on Si(001), Si(111) and SiC(0001) surfaces, in E. Zschech, C. Whelan, T. Mikolajick (Eds.), Materials for Information Technology, Springer, New York, 2005, pp 449460.

SC

[31] http://www.lasurface.com/database/elementxps.php

RI

[30] M. Repoux, Comparison of background removal methods for XPS, Surf. Interface Anal. 18 (1992) 567-570.

NU

[32] I. J. Raaijmakers, Low temperature metal-organic chemical vapor deposition of advanced barrier layers for the microelectronics industry, Thin Solid Films 247 (1994) 85-93.

MA

[33] S. Panda, J. Kim, B.H. Weiller, D.J. Economou, D.M. Hoffman, Low temperature chemical vapor deposition of titanium nitride films from tetrakis(ethylmethylamido)titanium and ammonia, Thin Solid Films 357 (1999) 125-131.

D

[34] S. Tanuma, C.J. Powell, D.R. Penn, Calculations of electron inelastic mean free paths. V. Data for 14 organic compounds over the 50-2000 eV range, Surf. Interface Anal. 21 (1994) 165-176.

PT E

[35] X. Song, D. Gopireddy, C. G. Takoudis,Characterization of titanium oxynitride films deposited by low pressure chemical vapor deposition using amide Ti precursor, Thin Solid Films 516 (2008) 6330-6335.

CE

[36] M. Tallarida, D. Schmeisser, In-situ ALD experiments with synchrotron radiation photoelectron spectroscopy, Semicond. Sci. Technol. 27 (2012) 074010.

AC

[37] Z. Zhang, S.-P. Jeng, V.E. Henrich, Cation-ligand hybridization for stoichiometric and reduced TiO2 (110) surfaces determined by resonant photoemission, Phys. Rev. B 43 (1991) 12004-12011. [38] R. Sanjinés, H. Tang, H. Berger, F. Gozzo, G. Margaritondo, F. Lévy, Electronic structure of anatase TiO2 oxide, J. Appl. Phys. 75 (1994) 2945-2951. [39] C. Das, M. Richter, M. Tallarida, D. Schmeißer, Electronic properties of atomic layer deposited films, anatase and rutile TiO2 studied by resonant photoemission spectroscopy. J. Phys. D: Appl. Phys. 49 (2016) 275304. [40] P.G. Moses, A. Janotti, C. Franchini, G. Kresse, C.G. Van de Walle, Donor defects and small polarons on the TiO2(110) surface, J. Appl. Phys. 119 (2016) 181503. [41] P. Poodt et al., Spatial atomic layer deposition: A route towards further industrialization of atomic layer deposition, J. Vac. Sci. Technol. 30 (2012) 010802.

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT Table I: List of the investigated samples, their process parameters and the calculated ratios of the titanium species within the Ti2p core level XPS data (TiO2/Ti2p, TiON/Ti2p, TiN/Ti2p) measured with Al K excitation. The corresponding O/N and O/Ti ratios are given for completeness and if marked by a star taken from [26]. The samples selected for synchrotron measurements (Ti2p core level and VB spectra taken with excitation energy of 620eV) are

RI

TiON/Ti2p [%] 2.3 32.0 27.6 3.9 3.6 2.3 7.0 3.7 3.2 55.0 51.1 47.6

SC

TiO2/Ti2p [%] 97.7 61.6 62.9 96.1 96.4 97.7 93.0 96.3 96.8 28.1 29.1 27.9

NU

t [nm] 6.3 6.6 6.5 7.0 6.2 7.8 6.2 6.8 7.8 6.9 7.3 21.6

MA

Ptime [s] 8 8 8 6 8 15 8 8 8 30 30 30

D

PP [W] 100 150 200 100 100 100 100 100 100 200 200 200

PT E

TDMAT/ N2

I-100W I-150W I-200W I-6s I-8s I-15s I-200C I-240C I-300C M-250C M-285C M-350C

Tsubstr [°C] 200 200 200 200 200 200 200 240 300 250 285 350

CE

TTIP/ NH3

Sample

AC

Process

PT

checked in the column “Syn”. TiN/Ti2p [%] 0 6.4 9.5 0 0 0 0 0 0 16.9 19.8 24.5

O/N

O/Ti

4.4* 3.8* 4.0* 4.0* 4.3* 5.9* 4.3* 5.3* 10.6* 0.7* 0.7* 0.8*

2.3 2.3* 2.3* 2.3 2.4 2.5 2.5* 2.3* 2.6* 1.0* 0.9* 0.9*

Syn x x

x x x

ACCEPTED MANUSCRIPT Figure Captions Figure 1: (Color Online) Ti2p core level spectra of the TiOxNy films investigated in this work recorded with Al K excitation. The sample names are indicated next to the corresponding spectrum. Figure 2: (Color Online) Ti2p core levels (Al K excitation) of the a) I-150W sample

PT

prepared within the TTIP/NH3 process and b) M-250C sample fabricated within the

RI

TDMAT/N2 process and the corresponding peak decomposition. The peak decomposition

SC

contributions are given in the legend in part a).

Figure 3: (Color Online) Ti2p core level decomposition of all samples investigated in this

NU

work (with Al K excitation) in dependence of the process (TTIP/NH3, TDMAT/N2) and

MA

parameter (PP, Ptime, Tsubstr) selection. The color code is given beside the diagram. Figure 4: (Color Online) I-V characteristics of selected TiOxNy samples of the TTIP/NH3

D

process: I-100W (rectangles) and I-150W (open triangles) as well as of the TDMAT/N2

PT E

process: M-250C (circles), M-285C (stars) and M-350C (reversed triangles). Figure 5: (Color Online) Dependence of a) the current density (rectangles, at electric field of

CE

0.93MV/cm) and b) TiN (open triangles) and TiON (open circles) contents within the Ti2p

AC

core level spectra (recorded with Al K excitation) on the substrate temperature of the PEALD process for the TDMAT/N2 processed samples. The TiN and TiON contents are referred to the total film composition (given in relative total amount). Figure 6: (Color Online) Ti2p core levels recorded with SR-XPS (excitation energy of 620eV) of a) the I-100W and I-150W samples prepared within the TTIP/NH3 process and b) M-250C, M-285C and M-350C samples fabricated within the TDMAT/N2 process. The sample names are indicated next to the corresponding spectrum.

ACCEPTED MANUSCRIPT Figure 7: (Color Online) SR-XPS Ti2p core levels (excitation energy of 620eV) of a) the I150W sample prepared within the TTIP/NH3 process and b) M-250C sample fabricated within the TDMAT/N2 process and the corresponding peak decomposition. The peak decomposition contributions are given in the legend in part a). Figure 8: (Color Online) Ti2p core level decomposition of the samples I-150W (TTIP/NH3

PT

process) and M-250C (TDMAT/N2 process) performed on spectra recorded with excitation

RI

energies of 620eV and Al K, respectively, as marked in the diagram. The color code is given

SC

below the diagram.

Figure 9: (Color Online) VB spectra recorded with SR-XPS (excitation energy of 620eV) of

NU

a) the I-100W and I-150W samples prepared within the TTIP/NH3 process and b) M-250C,

MA

M-285C and M-350C samples fabricated within the TDMAT/N2 process. The sample names are indicated next to the corresponding spectrum. The in-gap state peaks are illustrated by

D

filling the area under the peaks.

PT E

Figure 10: (Color Online) Normalized current density versus normalized in-gap state intensity. Samples used for this diagram are those investigated at the synchrotron (see Table

CE

1). The values are deduced from the currents (Fig. 4) at different fixed electric fields of 0.5MV/cm (open rectangles), 0.75MV/cm (circles) and 0.93MV/cm (triangles) as well as

AC

from the area under the in-gap state line shapes (illustrated in Fig. 9) and normalized in each case to the lowest value corresponding to the sample I-100W. The arrows indicate the influence of rising plasma power and substrate temperature.

ACCEPTED MANUSCRIPT Figures

TTIP/NH

Ti2p

3

AlK

I-200W

PT

Normalized Intensity

I-150W

I-100W

RI

I-15s

2

468

465

PT E

D

TDMAT/N

MA

NU

SC

I-8s I-6s

462

459

CE AC

I-200C

M-350C M-285C M-250C

456

Binding Energy [eV]

Figure 1

I-300C I-250C

453

ACCEPTED MANUSCRIPT

1

TTIP/NH3

Ti2p AlK

I-150W

PT RI SC

a)

0

TDMAT/N2

NU

1

MA

M-250C

0 470

PT E

D

Normalized Intensity

data TiON TiN TiO2

465

b) 460

455

AC

CE

Binding Energy [eV]

Figure 2

RI

SC

0.8

Tsubstr

PT E CE AC Figure 3

Tsubstr

TiO2 TiN TiON

350°C

285°C

300°C

240°C

MA

200°C

15s

8s

Ptime

D

PP

6s

200W

150W

0.2

250°C

0.4

NU

0.6

0.0

PT

TDMAT/N2

TTIP/NH3

1.0

100W

Relative Content in Ti2p

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

2

Current Density [A/cm ]

12

M-350C

M-285C

8

M-250C

6

RI

I-150W

2

SC

4

I-100W

0 2

4

NU

0

AC

CE

PT E

D

MA

Electric Field [MV/cm]

Figure 4

PT

10

6

TDMAT/N2

2

a)

PT

10

SC

RI

1

0.1

E=0.93 MV/cm

NU

14 12

MA

8

D

6 4 240

260

280

300

TiN (of Ti2p)

320

340

AC

CE

Substrate Temperature [°C]

Figure 5

b)

TiON (of Ti2p)

10

PT E

Rel. total amaount [%] Current dens. [A/cm ]

ACCEPTED MANUSCRIPT

360

ACCEPTED MANUSCRIPT

Ti2p

RI

SC

I-150W

NU

I-100W

MA

TDMAT/N2

465

CE

470

PT E

D

Normalized Intensity

620eV

PT

a)

TTIP/NH3

460

AC

Binding Energy [eV]

Figure 6

b)

M-350C M-285C M-250C

455

ACCEPTED MANUSCRIPT

TTIP/NH3

Ti2p

I-150W

620eV

PT

data TiON TiN TiO2

RI

a)

1

SC

0

TDMAT/N2

NU

M-250C

MA

Normalized Intensity

1

D PT E

0 470

b)

465

460

455

AC

CE

Binding Energy [eV]

Figure 7

M-250C (TDMAT/N2)

I-150W (TTIP/NH3)

PT

1.0 0.8

RI

0.6

0.2 0.0 620 eV

Al K

620 eV TiN

AC

CE

PT E

D

MA

TiO2

Figure 8

SC

0.4

Al K

NU

Relative Content in Ti2p

ACCEPTED MANUSCRIPT

TiON

ACCEPTED MANUSCRIPT

VB

RI

SC

I-150W

NU

I-100W

MA

TDMAT/N2

12

10

CE

14

PT E

D

Normalized Intensity

620eV

PT

a)

TTIP/NH3

8

6

M-250C

2

AC

Binding Energy [eV]

Figure 9

M-350C M-285C

VBM

4

b)

0

-2

100

RI

TSubstr

10

SC

@ E [MV/cm] 0.5 0.75 0.93

PP

1 1.0

1.5

2.0

2.5

AC

CE

PT E

D

MA

Normalized In-Gap State Intensity

Figure 10

PT

TDMAT/N2

TTIP/NH3

NU

Normalized Current Density

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT Highlights  Titanium oxynitride PEALD processes using TDMAT/N2 or TTIP/NH3 are compared  Influence of ALD parameters on film composition and leakage current is shown  Strong photo-emissions within the bandgap are observed in the valence band spectra  Ti-N bonded nitrogen and in-gap states correlate strongly with leakage current

AC

CE

PT E

D

MA

NU

SC

RI

PT

 TDMAT/N2 process delivers higher TiN and in-gap state contributions