- Email: [email protected]
Annealing effects on the structural and electrical properties of sputtered tungsten thin films
Annealing effects on the structural and electrical properties of sputtered tungsten thin films Andreas Kaidatzis, Vassilios Psycharis, Konstantina Mergia, Dimitrios Niarchos PII: DOI: Reference:
S0040-6090(16)30615-0 doi: 10.1016/j.tsf.2016.10.027 TSF 35551
To appear in:
Thin Solid Films
Received date: Revised date: Accepted date:
26 February 2016 22 September 2016 11 October 2016
Please cite this article as: Andreas Kaidatzis, Vassilios Psycharis, Konstantina Mergia, Dimitrios Niarchos, Annealing effects on the structural and electrical properties of sputtered tungsten thin films, Thin Solid Films (2016), doi: 10.1016/j.tsf.2016.10.027
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.
RI P
T
ACCEPTED MANUSCRIPT
NU
SC
Annealing effects on the structural and electrical properties of sputtered tungsten thin films
MA
October 11, 2016
2
PT
ED
Andreas Kaidatzis1 , Vassilios Psycharis1 , Konstantina Mergia2 , and Dimitrios Niarchos1 1 Institute of Nanoscience and Nanotechnology, NCSR Demokritos, Aghia Paraskevi, Greece
AC CE
Institute of Nuclear and Radiological Science and Technology, Energy and Safety, NCSR Demokritos, Aghia Paraskevi, Greece Corresponding author: [email protected]
Abstract
We report on the structural and electrical characterization of sputterdeposited tungsten (W) films, having thicknesses between 1.5 and 100 nm, before and after annealing in the temperature range between 200 and 800o C. In the as-deposited the β-W phase prevails, for all the thicknesses studied. A β-W to α-W transition occurs upon annealing at a temperature that depends on film thickness and it is accompanied by a corresponding resistivity drop. Films with thickness lower or equal to 8 nm are composed predominately of β-W phase after annealing at 300o C, while the α-W phase prevails after annealing at 450o C. Films with thickness higher or equal to 10 nm remain at the
1
ACCEPTED MANUSCRIPT
1
SC
RI P
T
β-W phase after annealing at 200o C, but the α-W phase prevails after annealing at 300o C. The resistivity behavior as a function of film thickness and annealing temperature are discussed. The minimum film resistivity is obtained for the 100 nm thick film after annealing at 800o C and it is 17.0 µΩ·cm.
INTRODUCTION
AC CE
PT
ED
MA
NU
Metallic tungsten (W) thin films are studied since some years as candidate materials for metallization in integrated circuits applications [1, 2, 3]. Although W has a room temperature bulk resistivity of 5.3 µΩ·cm, more than three times higher than that of Cu (1.7 µΩ·cm), the currently used material for interconnects, an increase of Cu films resistivity occurs as microelectronics interconnects dimensions become comparable to the Cu electron mean-free path (45.0 nm at room temperature), due to the influence of size effects [4]. However, W has much lower electron mean-free path (19.1 nm at room temperature [5]) and the resistivity of W lines having line-widths less than 10 nm is predicted to be lower than that of Cu lines [6]. Moreover, W has a melting point (3695 K) almost three times higher than Cu (1357 K), suffers much lower thermal degradation, is bio-degradable [2], has high resistance to electromigration [3], low thermal expansion, and does not react with most semiconductor materials. The above make W thin films promising for microelectronics metallization applications. Recently, W has also attracted interest for use in magnetic memories applications, as its high spin-orbit coupling could allow for the fabrication of electrically addressable magnetic bits of information [7]. However, although the low resistivity equilibrium α-W phase is suitable for interconnect metallization, the high spin-orbit coupling of the considerably more resistive metastable β-W phase is necessary for magnetic memories applications [7]. Furthermore, as it is necessary to anneal the multilayer stacks used for magnetic memories to temperatures up to 400o C, for improving their magnetic properties [8], the β-W to α-W transition occurring upon annealing [9] could deter the use of W for this technology. In this context, we study sputtered W thin films with thicknesses in the 2
ACCEPTED MANUSCRIPT
EXPERIMENTAL DETAILS
SC
2
RI P
T
range between 1.5 and 100 nm and the effect of vacuum annealing in the temperature range between 200o C and 800o C on their crystal structure and electrical resistivity.
AC CE
PT
ED
MA
NU
Tungsten thin films were deposited using an ultra-high vacuum (base pressure 1.5·10−9 Torr) magnetron sputtering system (ATC 2200-V, AJA Inc., USA) in diode configuration. The magnetron surface normal was at an angle of 28o with respect to the substrate surface normal and the magnetron-substrate distance was 240 mm. The source material was metallic W (50 mm diameter cylindrical target, 99.95% purity) and the working gas was high purity Ar at 3 mTorr pressure, resulting in a mean-free path for the sputtered W atoms of 44.7 mm [10]. Direct Current (DC) power was applied to the target (20 W or 1 W/cm2 ), leading to a growth rate (thickness/time) of 0.016 nm/s and a normalized growth rate (thickness/(time·power)) of 0.048 nm/(min·W). A pre-sputtering at the same conditions, for 1 min, with the shutter closed, was performed prior each deposition. The substrates used were single-side polished Si(100) wafers covered by an amorphous 500 nm thick thermal SiO2 with surface roughness approximately 0.1 nm; they were in-situ cleaned prior deposition by RF Ar plasma. During deposition the substrates were rotated at 80 revolutions per minute in order to avoid deposition inhomogeneities. Substrates remained at roomtemperature for the whole duration of the deposition runs. The abovementioned parameters were strictly controlled and maintained identical for all samples, as they have been shown to critically affect film microstructure [11, 12, 13]. All films were capped with a 2 nm thick alumina layer for protection against oxidation, deposited in-situ by RF magnetron sputtering, immediately after W deposition. The W and alumina deposition rate was determined from X-ray reflectivity (XRR) measurements of test samples. W films of 1.5, 3, 5.5, 10, 15, 20 and 100 nm thickness were fabricated. Samples were annealed in the temperature range between 200 and 800o C for 30 min, using a resistive heater mounted inside an independent high 3
ACCEPTED MANUSCRIPT
AC CE
PT
ED
MA
NU
SC
RI P
T
vacuum chamber (base pressure 2·10−7 Torr). The maximum pressure during annealing was 2·10−5 Torr, for the samples annealed at 800o C. X-ray diffraction (XRD) structural analysis was performed using a Siemens D500 diffractometer with Cu-Kα radiation, in steps of 0.03o and counting time 12 s/step. θ-2θ scans were obtained in the 30o to 50o range, in order to avoid the Bragg peak from the Si substrate at 2θ 69.2o . For the thinner films, Grazing Incidence XRD (GIXRD) has been employed, using a Rigaku SmartLab diffractometer and an incidence angle of 0.8o . XRR measurements were performed on a Bruker D8 instrument using Cu-Kα radiation and a parallel beam stemming from a Goebel mirror and on a Rigaku SmartLab diffractometer using a cross beam optics unit. Atomic force microscopy (AFM) was employed to obtain the film surface morphology, using a NT-MDT scanning probe microscope in non-contact mode and commercial SPM probes (Nanosensors PPP-FMR). All AFM images were processed using the WSxM software [14]. Film resistivity was determined by Van der Pauw (VdP) measurements performed on square samples using a programmable current source (Keithley 220) and a multimeter (Keithley 2000). Electrical contacts were made at the sample edges using silver paint and the resistivity was derived using the formalism described in reference [15]. The W films are “sandwiched” between two highly dielectric materials, silicon dioxide and alumina, thus this setup measures their actual resistivity. All measurements have been performed at room temperature. Typical currents were 0.1 to 1 mA and measured voltages ranged between 0.1 to 10 V. The studied samples are tagged as W1.5 to W100, indicating the corresponding W layer thickness; after annealing the tag A200 to A800 is added, indicating the corresponding annealing temperature.
3 3.1
Results As-deposited films
No film delamination is observed at any W layer thickness. The as-deposited films crystal structure immediately after deposition is obtained by the XRD 4
ACCEPTED MANUSCRIPT
a-W (210)
b-W (211)
(a)
SC
b-W (210)
Intensity (a.u.)
b-W (200)
RI P
T
diagrams shown in figure 1.
W (100 nm)
W (15 nm) W (10 nm)
NU
W (20 nm)
2q (degrees)
MA
30 32 34 36 38 40 42 44 46 48 50 38.5
39.0
(b) W (100 nm)
W (20 nm) W (15 nm) W (10 nm)
39.5
40.0
40.5
41.0
41.5
2q (degrees)
PT
ED
Figure 1: XRD diagrams of as-deposited samples. Figure (b) is a zoom of figure (a). Points represent experimental data and continuous lines fits to the data. The diagrams are shifted in the vertical axis for clarity purposes.
AC CE
Sputtered W films may have either the thermodynamically stable bcc structure, called α-W, or the metastable alotrope β-W with A − 15 crystal structure [16]. It has been shown [9, 17] that as deposited W films have predominately the highly resistivity β-phase below a critical thickness, which is a few tens of nm, while for larger thickness the less resistive α-phase prevails. The as-deposited structure and critical thickness depend strongly on the deposition parameters [13, 18], while annealing of the films results in a β-phase to α-phase transition [9, 16]; it has been shown that a temperature of 850o C is necessary for complete transformation [9]. The β-W to α-W phase transition onset temperature depends strongly on film thickness [9] and even a room-temperature transformation has been observed [17]. Contrary to previous reports [9, 17], our films show the presence of predominately the β-phase even up to 100 nm thickness, as it is indicated by the β-W Bragg peaks at 2θ 35.6o and 43.4o , corresponding to the (200) and (211) planes, respectively (see figure 1). However, it should be noted that all films studied in this work have been sputtered at a deposition rate of 0.016 nm/s, much lower than in the case of reference [9] (0.14 nm/s) and [17] (1.22 nm/s). 5
ACCEPTED MANUSCRIPT
AC CE
PT
ED
MA
NU
SC
RI P
T
Recent detailed studies of the deposition parameters effect on the properties of sputtered W films, show that Ar pressure and applied power may greatly affect film structure [11, 13]. High Ar pressure (≥5 mTorr) and low applied power (≤40 W) result in β-W because of the reduced energy of the deposited W species. In our case, the presence of β-phase, even up to 100 nm thickness, is in agreement with these findings due to the low power (20 W) applied to the magnetron source. Moreover, although the sputtering pressure is low, the high source-to-substrate distance should be taken into account (almost two times higher than in the case of reference [13]), which results in considerable energy-loss of the sputtered W atoms due to enhanced scattering from Ar. The wide peak at around 40o indicates the presence of both β-W and α-W, as it can be deconvoluted to the β-W (210) and α-W (110) Bragg peaks (39.88o and 40.26o , respectively). Figure 1b is a zoom-in of figure 1a from 38.5o to 41.5o . It can be seen that there is a slight shift of the peak towards the α-W (110) position upon W film thickness increase. The β-W content of the films is determined by Gaussian peak fitting at the positions that correspond to the β-W (210) and α-W (110) peaks. Before fitting, an end points weighted baseline is calculated and subtracted from the data. The calculated area of the β-W (210) peak to the total area gives approximately the β-W content in each film. The calculated values of the β-W content are shown in table 1, along with the corresponding resistivity values determined from VdP measurements. The resistivity values for films with thickness varying between 10 and 20 nm are, within errors, similar and this is due to the similar content of the β-W phase. Although there is a small decrease of the β-W content as film thickness increases from 20 nm to 100 nm, the β-W phase prevails in all films. Consequently, the corresponding film resistivity is high, as it is dominated by the strong β-W presence; the values vary from 163.0 µΩ·cm for the thinner film, to 127.1 µΩ·cm for the thicker one. It is interesting to note that for a 13% decrease of β-W content from W10 to W100, a 22% resistivity decrease is observed. This is attributed to the decreased influence of interface scattering as films become thicker.
6
ACCEPTED MANUSCRIPT
3.2
RI P
resistivity (µΩ·cm) 163.0±6.5 150.7±6.0 156.6±6.3 127.1±5.1
SC
β-W content (%) 78.8±0.4 81.7±0.3 77.9±0.4 68.9±0.6
NU
W thickness (nm) 10 15 20 100
T
Table 1: β-W content and resistivity values of the as-deposited samples.
Temporal evolution upon room-temperature stor-
MA
age
b-W (210)
a-W (210)
AC CE
Intensity (a.u.)
b-W (200)
PT
ED
Test samples have been stored in a vacuum desiccator at room-temperature and periodically measured for determining their structure and resistivity. Figure 2 shows XRD diagrams of the as-deposited W100 sample, measured within a period of 5 months after deposition.
b-W (211)
(a)
(b) 132 days
132 days 55 days
55 days 0 days
0 days
30 32 34 36 38 40 42 44 46 48 50 38.5
39.0
39.5
40.0
40.5
41.0
41.5
2q (degrees)
2q (degrees)
Figure 2: XRD diagrams of the 100 nm thick W layer during storage in roomtemperature conditions. Figure (b) is a zoom of figure (a). Points represent experimental data and continuous lines fits to the data. The diagrams are shifted in the vertical axis for clarity purposes.
7
ACCEPTED MANUSCRIPT
AC CE
PT
ED
MA
NU
SC
RI P
T
As it can be seen, with the increase of time there is only a small enhancement of the α-W (110) Bragg peak at the expense of β-W (210) one. Curve fitting of the double peak at 40o reveals a small degree of β-W to α-W phase transition upon room-temperature storage of the W100 sample. The β-W content of the sample is 74.4% immediately after deposition, while it decreases to 68.6% after 55 days and to 66.7% after 132 days, which corresponds to a 10% decrease of the β-W phase; a 11% decrease of resistivity is observed during the same period (from 127.1 µΩ·cm to 112.6 µΩ·cm), in accordance to the β-W content decrease. Thus, after several months of room temperature storage we only observe a small structural variation of the W100 sample and no measurable variation of the samples with thinner W layers (data not shown). The above findings are contrary to the room-temperature β-W to α-W phase transition reported in reference [17] at mixed-phase W films with thickness in the range between 7 and 30 nm; following the discussion of section 3.1, the enhanced tendency for obtaining α-W reported in reference [17] can be attributed to the extremely high DC power used (3 W/cm2 ), the low deposition pressure (1 mTorr), and the low source-to-substrate distance (100 mm), which result in reduced scattering of the W atoms and to an extremely high deposition rate (1.2 nm/s). However, mixed α-W and β-W phase films deposited using DC diode sputtering retain their initial properties even after 3 months of room-temperature storage [19], due to the moderate power used (1.5 to 2 W/cm2 ) and the high sputtering pressure (4 to 10 mTorr), resulting in lower deposition rate (0.13 to 0.17 nm/s) and low deposited W atoms energy, which enhance the formation of β-W phase, in agreement to our findings.
3.3
The effect of annealing
Film morphology before and after annealing is studied by means of AFM and XRR measurements. AFM was employed to explore the surface morphology of the films, yielding the surface roughness of the top 2 nm thick alumina layer. However, as the alumina layer thickness does not changes across the various samples, any discrepancies in the obtained surface roughness values 8
ACCEPTED MANUSCRIPT
MA
NU
SC
RI P
T
reflect variations of the W films morphology. All the studied films have extremely low surface roughness before and after annealing: the root meansquare surface roughness of the as-deposited samples decreases from 0.38 nm at W100 nm, to 0.11 nm at W1.5 nm (data not shown). After annealing, even at the highest temperature (800o C), surface roughness remains unchanged, within the resolution limits of the microscope; as shown in figure 3, samples W100A800 and W1.5A800 have surface roughness equal to 0.36 nm and 0.15 nm respectively. This indicates that the underlying W film morphology is not significantly affected by annealing and no W agglomeration occurs, even for the 1.5 nm thick film, an effect that would be manifested by a steep surface roughness increase due to the formation of W islands. This finding is corroborated by the XRR results discussed below.
(c) 7
AFM profiles after annealing at 800oC
6 W (1.5 nm)
Height (nm)
(b)
AC CE
PT
ED
(a)
5 4 W (100 nm)
3 2 1 0 0
200
400
600
800
1000
Distance (nm)
Figure 3: AFM images of samples after annealing at 800o C. (a) 1.5 nm thick W film. (b) 100 nm thick W film. The color scale (from dark to bright) is 2.4 nm and the white scale bar is 400 nm, in both images. (c) Line profiles obtained along the indicated directions at each image.
The XRR diagrams from W1.5, W3 and W5.5 samples before and after 9
ACCEPTED MANUSCRIPT
X-ray reflectivity (a.u.)
MA
NU
SC
RI P
T
annealing at 800o C are presented in figure 4. Least square fitting to the data was performed using the LEPTOS 6.02 software by Bruker AXS GmbH. In order to fit the experimental data, we have to assume a structural model, i.e., number of layers and interlayer profiles or interlayer roughness. Each layer is characterized by its thickness and density which are adjustable parameters to be determined by the least squares fitting. The optimum parameters calculated by the least squares minimization procedure are subsequently used for the simulation of the theoretical reflectivity curve (continuous lines in figure 4). Significant deviations between the simulated curve and the experimental data suggest that the assumed structural model is erroneous and a new one has to be envisaged. The thickness, density and roughness of the layered structure are determined.
(a)
annealed at 800oC
AC CE
0
PT
annealed at 800oC
as-deposited
1
2
3
4
5
2q (degrees)
6
7
W (5.5 nm)
(b)
W (3 nm)
ED
W (1.5 nm)
annealed o at 800 C
as-deposited
as-deposited 0
1
2
3
4
(c)
5
2q (degrees)
6
7
0
1
2
3
4
5
6
7
2q (degrees)
Figure 4: X-ray reflectivity versus scattering angle, before and after annealing at 800o C. The black open dots represent experimental data and the red lines are least-square fits to the data. W film thickness is (a) 1.5 nm, (b) 3 nm, and (c) 5.5 nm.
In order to fit the experimental XRR data initially, a single W layer on the top of the Si/SiO2 substrate was assumed followed by an Al2 O3 cover layer. However, for W1.5 and W3 samples this structural model failed to give a good agreement between the simulated curves and the experimental data. Next, a structural model in which the W layer is divided into two sublayers, W1 and W2, was assumed with the thickness, density, and roughness of both W1 and 10
ACCEPTED MANUSCRIPT
W1.5A800
W3
W3A800
AC CE
W5.5 W5.5A800
interface roughness (nm) alumina/W W/substrate
SC
NU
Al2 O3 W1 W2 Al2 O3 W1 W2 Al2 O3 W1 W2 Al2 O3 W1 W2 Al2 O3 W Al2 O3 W
MA
W1.5
thickness layer density (nm) (g/cm3 ) 1.77 3.66 1.64 3.90-17.90 0.72 19.25 1.84 2.00 1.02 5.40-14.30 1.07 14.30 1.70 3.70 0.88 7.20-18.50 2.05 18.50 1.83 3.00 0.82 7.20-19.25 1.94 19.25 2.14 2.38 4.73 19.37 2.71 1.83 4.49 19.37
ED
layer
PT
sample
RI P
T
Table 2: XRR fitting values for the 1.5, 3, and 5.5 nm thick W films, asdeposited and after annealing at 800o C.
0.50 0.36
0.35 0.32 0.44 0.42 0.40 0.41 0.29
0.40
0.27
0.47
W2 layers as parameters to be least squares fitted. The least squares fitted curves for this model and the experimental data for both the as deposited and annealed samples were found to be in reasonable agreement. The least squares fitted curves to the data are presented in figure 4 as continuous lines. The results from the least square fittings to the XRR data are shown in table 2. Least square fitting to the data from the as deposited films show that the two thinnest films are composed of two sublayers; the top layer represents a region with a density gradient, becoming less dense at its upper interface and its average density is smaller than that of the bottom layer, which has 11
ACCEPTED MANUSCRIPT
AC CE
PT
ED
MA
NU
SC
RI P
T
a density close to the tungsten bulk value (19.25 g/cm3 ). Annealing of the 1.5 nm thick film results in a decrease of the bottom layer density, which may be attributed to interdiffusion from the top and bottom interfaces; the opposite occurs at the 3 nm thick film, where a densification of the bottom sublayer is observed. The 5.5 nm thick film, both before and after annealing, is shown to be composed of a sole W layer, with density close to that of bulk W. This thickness dependent behavior is assumed to stem from the morphological evolution of the films as their thickness increases during the initial stages of film growth. In all cases, XRR shows that the films are continuous after annealing, even the 1.5 nm thick one, without any indication of W agglomeration and there is a strong presence of high-density W. The upper (W/alumina) and lower (substrate/W) W layer interfaces are shown to have roughness between 0.30 and 0.50 nm, before and after annealing, in agreement to the surface roughness values measured by AFM. The effect of annealing on the crystal structure of the W films has been studied by GIXRD and XRD measurements. In figure 5a it can be seen that after annealing of the W5.5 sample at 300o C a broad peak appears at around 40o , which is accompanied by two smaller Bragg peaks at around 35o and 44o , indicating the presence of both β-W and α-W phases. As the annealing temperature increases, the intensity of the α-W (110) Bragg peak increases, while the β-W phase is suppressed. Figures 5b and c show that for thicker films the β-W phase is preserved after annealing at 200o C. However, after annealing at 300o C, or higher, only the α-W (110) Bragg peak is present indicating the β-W to α-W phase transition. Annealing at higher temperatures does not result in any observed change: indicatively the XRD diagrams obtained after annealing at 800o C are shown. It should be noted that similar behavior is also observed for the W15 and W20 samples. The above mentioned pronounced phase transition is also manifested by a corresponding steep resistivity drop. Figure 6a shows the evolution of resistivity upon annealing, for all the W thicknesses studied. As it was mentioned in Section 3.1, the as-deposited film resistivity increases as the film thickness decreases, since interface scattering becomes increasingly important for thinner films: for 100 nm thickness the resistivity is 127.1 µΩ·cm, while it reaches 2453.9 µΩ·cm for the 1.5 nm thick film, indicating a total domination 12
ACCEPTED MANUSCRIPT
(a)
RI P
a-W b-W (210) (211)
T
b-W (210)
Intensity (a.u.)
b-W (200)
800oC
SC
450oC
300oC
as-deposited
NU
30 32 34 36 38 40 42 44 46 48 50
2q (degrees) b-W (210)
a-W b-W (210) (211)
(b)
b-W (200)
b-W (210)
a-W b-W (210) (211)
Intensity (a.u.)
MA
b-W (200)
(c) 800oC
800oC
300oC
300oC
200oC
as-deposited
as-deposited
ED
200oC
PT
30 32 34 36 38 40 42 44 46 48 50 30 32 34 36 38 40 42 44 46 48 50
2q (degrees)
2q (degrees)
AC CE
Figure 5: GIXRD diagrams of the 5.5 nm thick sample before and after annealing (a). XRD diagrams of samples before and after annealing. W thickness is (b) 10 nm and (c) 100 nm. The diagrams are shifted in the vertical axis for clarity purposes.
of interface scattering for the thinner films [20]. After annealing at 200o C there is minor resistivity change for all the films, except for the two thinner ones (W1.5 and W3) where a significant increase is observed. Annealing at 300o C results in a steep resistivity drop for the films having thickness between 10 and 100 nm. At this annealing temperature, only a minor change is observed for W5.5 and W8, indicating that their structure continues to be predominately β-phase, in conjunction to the GIXRD measurements. After annealing at 450o C a steep resistivity decrease is also observed for the 5.5 and 8 nm thick films. The two thinnest films 13
ACCEPTED MANUSCRIPT
t = 5.5 nm t = 12.5 nm
450
6000
250
4000
200 150
3000
100 50
RI P
300
2000
0 0
100 200 300 400 500 600 700 800 o
1
10
100
10
3
10
2
10
1
W thickness (nm)
MA
Annealing temperature ( C)
(b)
SC
5000
Resistivity (mWcm)
350
NU
Resistivity (mWcm)
as-deposited (b-W) A800 (a-W)
(a)
400
Resistivity (mWcm)
t = 3 nm t = 10 nm t = 100 nm
T
t = 1.5 nm t = 8 nm t = 20.5 nm
ED
Figure 6: Sample resistivity before and after annealing. (a) Resistivity versus annealing temperature. (b) Resistivity versus W thickness. Error bars are omitted for clarity purposes. The measurement error is approximately 4%.
AC CE
PT
(W1.5 and W3) present a non monotonous variation of the resistivity as a function of the annealing temperature, showing a maximum at 450o C (figure 6a). This dependence of the β-W to α-W phase transition temperature on the thickness of the film has been also reported elsewhere [9]; it is attributed to the unintended annealing of the films during sputter deposition, which results in the onset of the β-W to α-W phase transition: as film thickness increases, the deposition time increases as well, resulting in higher extent of the phase transition. The peculiar resistivity variation with annealing temperature of the two thinnest samples (W1.5 and W3) may be attributed to the competing action of bulk and interface electron scattering. According to the XRR results shown in table 2, the roughness of the bottom (SiO2 /W) and top (W/AlOx ) interfaces is not considerably affected during annealing; thus, interface electron scattering may be altered only due to the variation of the interface electronic structure and not due to the variation of interface defects. Simultaneously, there is a gradual β-W to α-W transition as annealing temperature increases. Moreover, it was previously shown [20] that the grain size to film thickness ratio of sputtered W films increases as film thickness decreases; grain size 14
ACCEPTED MANUSCRIPT
4
AC CE
PT
ED
MA
NU
SC
RI P
T
reaches a value of several times the film thickness at very thin W films. This suggests that for the W1.5 and W3 samples studied here, the top and bottom film interfaces always belong to a single grain. In this context, it is assumed that the transition from the low electron mean-free path (mfp) β-W grains to the high electron mfp α-W grains determines the electrical resistivity of W1.5 and W3: during this gradual transition to the high electron mfp α-W grains (which results in higher interface scattering), the competing interplay between the increasing interface scattering and decreasing bulk scattering causes the observed resistivity variation. Figure 6b shows the variation of the resistivity versus W film thickness. The two extreme cases are depicted: as-deposited and after annealing at 800o C; in the former case films are predominately β-phase while in the later case α-W prevails. As there is negligible change of the top and bottom W film interface roughness, any resistivity variation is attributed to changes of the film electronic structure. The slower resistivity decrease of the as-deposited films, as film thickness decreases, is explained due to the lower electron mfp in β-W, as compared to the one in α-W. A significant resistivity increase is observed between 8 nm and 10 nm thickness, indicating a β-W electron mfp close to these values. On the contrary, in the case of the predominately α-W films, a significant resistivity increase is observed between 10 nm and 20 nm thickness, indicating the higher electron mfp in α-W.
Conclusions
Thin W films were deposited by magnetron sputtering on Si/SiO2 substrates at thicknesses varying between 1.5 and 100 nm. As deposited films present both the β- and α-phase, with the beta phase content prevailing. The high content of the metastable beta phase is attributed to the low deposition rate used, compared to other studies, indicating the strong dependence of the film structure on the deposition parameters. Upon annealing a transformation from β- to α-phase occurs with the transition temperature depending on the film thickness. Films with thickness equal or higher than 10 nm are β-phase up to 200o C, while thinner films are composed of β-phase up to 300o C. A steep resistivity decrease is observed when the β- to α-phase transition occurs. 15
ACCEPTED MANUSCRIPT
NU
Acknowledgements
SC
RI P
T
This behavior is not followed by the two thinnest films (1.5 and 3 nm) which show a resistivity peak at 450o C; this is attributed to the competing action between bulk and interface electron scattering, which is manifested only at the two thinnest films due to the high contribution of interface scattering to the film’s resistivity. The lowest resistivity value is 17.0 µΩ·cm and it is obtained for the 100 nm thick film, after annealing at 800o C.
MA
Funding from the E.C. through a FP7-ICT project (GA No. 318144) is acknowledged.
ED
References
PT
[1] F. Papadatos, K. Wong, V. Arunachalam, C. H. Shin, Z. Li, M. Chudzik, W.-H. Lee, and Aimin Xing, Low resistivity tungsten for 32 nm node MOL contacts and beyond, Microelectron. Eng. 92 (2012) 123-125.
AC CE
[2] L. Yin, C. Bozler, D. V. Harburg, F. Omenetto, and J. A. Rogers, Materials and fabrication sequences for water soluble silicon integrated circuits at the 90nm node, Appl. Phys. Lett. 106 (2015) 014105. [3] S. Beyne, K. Croes, I. De Wolf, and Z. Tokei, 1/f noise measurements for faster evaluation of electromigration in advanced microelectronics interconnections, J. Appl. Phys. 119 (2016) 184302. [4] W. Steinh¨ogl, G. Schindler, G. Steinlesberger, M. Traving, and M. Engelhardt, Comprehensive study of the resistivity of copper wires with lateral dimensions of 100 nm and smaller, Appl. Phys. Lett. 97 (2005) 023706. [5] D. Choi, C. S. Kim, D. Naveh, S. Chung, A. P. Warren, N. T. Nuhfer, M. F. Toney, K. R. Coffey, K. Barmak, Electron mean free path of tungsten and the electrical resistivity of epitaxial (110) tungsten films, Phys. Rev. B, 86 (2012) 045432. 16
ACCEPTED MANUSCRIPT
RI P
T
[6] D. Choi, The electron scattering at grain boundaries in tungsten films, Microelectron. Eng. 122 (2014) 5.
SC
[7] C.-F. Pai, L. Liu, Y. Li, H.W. Tseng, D.C. Ralph, R.A. Buhrman, Spin transfer torque devices utilizing the giant spin Hall effect of tungsten, Appl. Phys. Lett. 101 (2012) 122404.
NU
[8] S. Ikeda, K. Miura, H. Yamamoto, K. Mizunuma, H. D. Gan, M. Endo, S. Kanai, J. Hayakawa, F. Matsukura, H. Ohno, A perpendicularanisotropy CoFeB/MgO magnetic tunnel junction, Nat. Mat. 9 (2010) 721.
ED
MA
[9] D. Choi, B. Wang, S. Chung, X. Liu, A. Darbal, A. Wise, N.T. Nuhfer, K. Barmak, A.P. Warren, K.R. Coffey, M.F. Toney, Phase, grain structure, stress, and resistivity of sputter-deposited tungsten films, J. Vac. Sci. Technol. A 29 (2011) 051512.
PT
[10] J. Lu, C.G. Lee, Numerical estimates for energy of sputtered target atoms and reflected Ar neutrals in sputter processes, Vacuum, 86 (2012) 1134.
AC CE
[11] K. Salamon, O. Milat, N. Radic, P. Dubcek, M. Jercinovic, S. Bernstorff, Structure and morphology of magnetron sputtered W films studied by x-ray methods, J. Phys. D: Appl. Phys. 46 (2013) 095304. [12] E. Vassallo, R. Caniello, M. Canetti, D. Dellasega, M. Passoni, Microstructural characterisation of tungsten coatings deposited using plasma sputtering on Si substrates, Thin Solid Films, 558 (2014) 189. [13] F.T.N. V¨ ullers, R. Spolenak, Alpha- vs. beta-W nanocrystalline thin films: A comprehensive study of sputter parameters and resulting materials’ properties, Thin Solid Films, 577 (2014) 26. [14] I. Horcas, R. Fern`andez, J. M. G´omez-Rodriguez, J. Colchero, J. G´omezHerrero, A. M. Baro, WSXM: A software for scanning probe microscopy and a tool for nanotechnology, Rev. Sci. Instrum. 78 (2007) 013705.
17
ACCEPTED MANUSCRIPT
RI P
T
[15] D. K. Schroder, Semiconductor Material and Device Characterization, third ed., Wiley, New Jersey, 2006. [16] M. J. O’Keefe, J. T. Grant, Phase transformation of sputter deposited tungsten thin films with A-15 structure, J. Appl. Phys. 79 (1995) 9134.
NU
SC
[17] S. M. Rossnagel, I. C. Noyan, C. Cabral, Phase transformation of thin sputter-deposited tungsten films at room temperature, J. Vac. Sci. Technol. B 20 (2002) 2047.
MA
[18] G. S. Chen, L. C. Yang, H. S. Tian, C. S. Hsu, Evaluating substrate bias on the phase-forming behavior of tungsten thin films deposited by diode and ionized magnetron sputtering, Thin Solid Films, 484 (2005) 83.
PT
ED
[19] G. S. Chen, H. S. Tian, C. K. Lin, G.-S. Chen, H. Y. Lee, Phase transformation of tungsten films deposited by diode and inductively coupled plasma magnetron sputtering, J. Vac. Sci. Technol. A 22 (2004) 281.
AC CE
[20] D. Choi, X. Liu, P. K. Schelling, K. R. Coffey, K. Barmak, Failure of semiclassical models to describe resistivity of nanometric, polycrystalline tungsten films, J. Appl. Phys. 115 (2014) 104308.
18
ACCEPTED MANUSCRIPT
T
Highlights
RI P
• The deposition and annealing parameters for obtaining α-W or β-W films are presented.
SC
• β-W films may be obtained up to 100 nm thickness and after annealing at 200◦ C.
AC CE
PT
ED
MA
NU
• High deposition rates are required for obtaining low resistivity α-W films.
19
S0040-6090(16)30615-0 doi: 10.1016/j.tsf.2016.10.027 TSF 35551
To appear in:
Thin Solid Films
Received date: Revised date: Accepted date:
26 February 2016 22 September 2016 11 October 2016
Please cite this article as: Andreas Kaidatzis, Vassilios Psycharis, Konstantina Mergia, Dimitrios Niarchos, Annealing effects on the structural and electrical properties of sputtered tungsten thin films, Thin Solid Films (2016), doi: 10.1016/j.tsf.2016.10.027
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.
RI P
T
ACCEPTED MANUSCRIPT
NU
SC
Annealing effects on the structural and electrical properties of sputtered tungsten thin films
MA
October 11, 2016
2
PT
ED
Andreas Kaidatzis1 , Vassilios Psycharis1 , Konstantina Mergia2 , and Dimitrios Niarchos1 1 Institute of Nanoscience and Nanotechnology, NCSR Demokritos, Aghia Paraskevi, Greece
AC CE
Institute of Nuclear and Radiological Science and Technology, Energy and Safety, NCSR Demokritos, Aghia Paraskevi, Greece Corresponding author: [email protected]
Abstract
We report on the structural and electrical characterization of sputterdeposited tungsten (W) films, having thicknesses between 1.5 and 100 nm, before and after annealing in the temperature range between 200 and 800o C. In the as-deposited the β-W phase prevails, for all the thicknesses studied. A β-W to α-W transition occurs upon annealing at a temperature that depends on film thickness and it is accompanied by a corresponding resistivity drop. Films with thickness lower or equal to 8 nm are composed predominately of β-W phase after annealing at 300o C, while the α-W phase prevails after annealing at 450o C. Films with thickness higher or equal to 10 nm remain at the
1
ACCEPTED MANUSCRIPT
1
SC
RI P
T
β-W phase after annealing at 200o C, but the α-W phase prevails after annealing at 300o C. The resistivity behavior as a function of film thickness and annealing temperature are discussed. The minimum film resistivity is obtained for the 100 nm thick film after annealing at 800o C and it is 17.0 µΩ·cm.
INTRODUCTION
AC CE
PT
ED
MA
NU
Metallic tungsten (W) thin films are studied since some years as candidate materials for metallization in integrated circuits applications [1, 2, 3]. Although W has a room temperature bulk resistivity of 5.3 µΩ·cm, more than three times higher than that of Cu (1.7 µΩ·cm), the currently used material for interconnects, an increase of Cu films resistivity occurs as microelectronics interconnects dimensions become comparable to the Cu electron mean-free path (45.0 nm at room temperature), due to the influence of size effects [4]. However, W has much lower electron mean-free path (19.1 nm at room temperature [5]) and the resistivity of W lines having line-widths less than 10 nm is predicted to be lower than that of Cu lines [6]. Moreover, W has a melting point (3695 K) almost three times higher than Cu (1357 K), suffers much lower thermal degradation, is bio-degradable [2], has high resistance to electromigration [3], low thermal expansion, and does not react with most semiconductor materials. The above make W thin films promising for microelectronics metallization applications. Recently, W has also attracted interest for use in magnetic memories applications, as its high spin-orbit coupling could allow for the fabrication of electrically addressable magnetic bits of information [7]. However, although the low resistivity equilibrium α-W phase is suitable for interconnect metallization, the high spin-orbit coupling of the considerably more resistive metastable β-W phase is necessary for magnetic memories applications [7]. Furthermore, as it is necessary to anneal the multilayer stacks used for magnetic memories to temperatures up to 400o C, for improving their magnetic properties [8], the β-W to α-W transition occurring upon annealing [9] could deter the use of W for this technology. In this context, we study sputtered W thin films with thicknesses in the 2
ACCEPTED MANUSCRIPT
EXPERIMENTAL DETAILS
SC
2
RI P
T
range between 1.5 and 100 nm and the effect of vacuum annealing in the temperature range between 200o C and 800o C on their crystal structure and electrical resistivity.
AC CE
PT
ED
MA
NU
Tungsten thin films were deposited using an ultra-high vacuum (base pressure 1.5·10−9 Torr) magnetron sputtering system (ATC 2200-V, AJA Inc., USA) in diode configuration. The magnetron surface normal was at an angle of 28o with respect to the substrate surface normal and the magnetron-substrate distance was 240 mm. The source material was metallic W (50 mm diameter cylindrical target, 99.95% purity) and the working gas was high purity Ar at 3 mTorr pressure, resulting in a mean-free path for the sputtered W atoms of 44.7 mm [10]. Direct Current (DC) power was applied to the target (20 W or 1 W/cm2 ), leading to a growth rate (thickness/time) of 0.016 nm/s and a normalized growth rate (thickness/(time·power)) of 0.048 nm/(min·W). A pre-sputtering at the same conditions, for 1 min, with the shutter closed, was performed prior each deposition. The substrates used were single-side polished Si(100) wafers covered by an amorphous 500 nm thick thermal SiO2 with surface roughness approximately 0.1 nm; they were in-situ cleaned prior deposition by RF Ar plasma. During deposition the substrates were rotated at 80 revolutions per minute in order to avoid deposition inhomogeneities. Substrates remained at roomtemperature for the whole duration of the deposition runs. The abovementioned parameters were strictly controlled and maintained identical for all samples, as they have been shown to critically affect film microstructure [11, 12, 13]. All films were capped with a 2 nm thick alumina layer for protection against oxidation, deposited in-situ by RF magnetron sputtering, immediately after W deposition. The W and alumina deposition rate was determined from X-ray reflectivity (XRR) measurements of test samples. W films of 1.5, 3, 5.5, 10, 15, 20 and 100 nm thickness were fabricated. Samples were annealed in the temperature range between 200 and 800o C for 30 min, using a resistive heater mounted inside an independent high 3
ACCEPTED MANUSCRIPT
AC CE
PT
ED
MA
NU
SC
RI P
T
vacuum chamber (base pressure 2·10−7 Torr). The maximum pressure during annealing was 2·10−5 Torr, for the samples annealed at 800o C. X-ray diffraction (XRD) structural analysis was performed using a Siemens D500 diffractometer with Cu-Kα radiation, in steps of 0.03o and counting time 12 s/step. θ-2θ scans were obtained in the 30o to 50o range, in order to avoid the Bragg peak from the Si substrate at 2θ 69.2o . For the thinner films, Grazing Incidence XRD (GIXRD) has been employed, using a Rigaku SmartLab diffractometer and an incidence angle of 0.8o . XRR measurements were performed on a Bruker D8 instrument using Cu-Kα radiation and a parallel beam stemming from a Goebel mirror and on a Rigaku SmartLab diffractometer using a cross beam optics unit. Atomic force microscopy (AFM) was employed to obtain the film surface morphology, using a NT-MDT scanning probe microscope in non-contact mode and commercial SPM probes (Nanosensors PPP-FMR). All AFM images were processed using the WSxM software [14]. Film resistivity was determined by Van der Pauw (VdP) measurements performed on square samples using a programmable current source (Keithley 220) and a multimeter (Keithley 2000). Electrical contacts were made at the sample edges using silver paint and the resistivity was derived using the formalism described in reference [15]. The W films are “sandwiched” between two highly dielectric materials, silicon dioxide and alumina, thus this setup measures their actual resistivity. All measurements have been performed at room temperature. Typical currents were 0.1 to 1 mA and measured voltages ranged between 0.1 to 10 V. The studied samples are tagged as W1.5 to W100, indicating the corresponding W layer thickness; after annealing the tag A200 to A800 is added, indicating the corresponding annealing temperature.
3 3.1
Results As-deposited films
No film delamination is observed at any W layer thickness. The as-deposited films crystal structure immediately after deposition is obtained by the XRD 4
ACCEPTED MANUSCRIPT
a-W (210)
b-W (211)
(a)
SC
b-W (210)
Intensity (a.u.)
b-W (200)
RI P
T
diagrams shown in figure 1.
W (100 nm)
W (15 nm) W (10 nm)
NU
W (20 nm)
2q (degrees)
MA
30 32 34 36 38 40 42 44 46 48 50 38.5
39.0
(b) W (100 nm)
W (20 nm) W (15 nm) W (10 nm)
39.5
40.0
40.5
41.0
41.5
2q (degrees)
PT
ED
Figure 1: XRD diagrams of as-deposited samples. Figure (b) is a zoom of figure (a). Points represent experimental data and continuous lines fits to the data. The diagrams are shifted in the vertical axis for clarity purposes.
AC CE
Sputtered W films may have either the thermodynamically stable bcc structure, called α-W, or the metastable alotrope β-W with A − 15 crystal structure [16]. It has been shown [9, 17] that as deposited W films have predominately the highly resistivity β-phase below a critical thickness, which is a few tens of nm, while for larger thickness the less resistive α-phase prevails. The as-deposited structure and critical thickness depend strongly on the deposition parameters [13, 18], while annealing of the films results in a β-phase to α-phase transition [9, 16]; it has been shown that a temperature of 850o C is necessary for complete transformation [9]. The β-W to α-W phase transition onset temperature depends strongly on film thickness [9] and even a room-temperature transformation has been observed [17]. Contrary to previous reports [9, 17], our films show the presence of predominately the β-phase even up to 100 nm thickness, as it is indicated by the β-W Bragg peaks at 2θ 35.6o and 43.4o , corresponding to the (200) and (211) planes, respectively (see figure 1). However, it should be noted that all films studied in this work have been sputtered at a deposition rate of 0.016 nm/s, much lower than in the case of reference [9] (0.14 nm/s) and [17] (1.22 nm/s). 5
ACCEPTED MANUSCRIPT
AC CE
PT
ED
MA
NU
SC
RI P
T
Recent detailed studies of the deposition parameters effect on the properties of sputtered W films, show that Ar pressure and applied power may greatly affect film structure [11, 13]. High Ar pressure (≥5 mTorr) and low applied power (≤40 W) result in β-W because of the reduced energy of the deposited W species. In our case, the presence of β-phase, even up to 100 nm thickness, is in agreement with these findings due to the low power (20 W) applied to the magnetron source. Moreover, although the sputtering pressure is low, the high source-to-substrate distance should be taken into account (almost two times higher than in the case of reference [13]), which results in considerable energy-loss of the sputtered W atoms due to enhanced scattering from Ar. The wide peak at around 40o indicates the presence of both β-W and α-W, as it can be deconvoluted to the β-W (210) and α-W (110) Bragg peaks (39.88o and 40.26o , respectively). Figure 1b is a zoom-in of figure 1a from 38.5o to 41.5o . It can be seen that there is a slight shift of the peak towards the α-W (110) position upon W film thickness increase. The β-W content of the films is determined by Gaussian peak fitting at the positions that correspond to the β-W (210) and α-W (110) peaks. Before fitting, an end points weighted baseline is calculated and subtracted from the data. The calculated area of the β-W (210) peak to the total area gives approximately the β-W content in each film. The calculated values of the β-W content are shown in table 1, along with the corresponding resistivity values determined from VdP measurements. The resistivity values for films with thickness varying between 10 and 20 nm are, within errors, similar and this is due to the similar content of the β-W phase. Although there is a small decrease of the β-W content as film thickness increases from 20 nm to 100 nm, the β-W phase prevails in all films. Consequently, the corresponding film resistivity is high, as it is dominated by the strong β-W presence; the values vary from 163.0 µΩ·cm for the thinner film, to 127.1 µΩ·cm for the thicker one. It is interesting to note that for a 13% decrease of β-W content from W10 to W100, a 22% resistivity decrease is observed. This is attributed to the decreased influence of interface scattering as films become thicker.
6
ACCEPTED MANUSCRIPT
3.2
RI P
resistivity (µΩ·cm) 163.0±6.5 150.7±6.0 156.6±6.3 127.1±5.1
SC
β-W content (%) 78.8±0.4 81.7±0.3 77.9±0.4 68.9±0.6
NU
W thickness (nm) 10 15 20 100
T
Table 1: β-W content and resistivity values of the as-deposited samples.
Temporal evolution upon room-temperature stor-
MA
age
b-W (210)
a-W (210)
AC CE
Intensity (a.u.)
b-W (200)
PT
ED
Test samples have been stored in a vacuum desiccator at room-temperature and periodically measured for determining their structure and resistivity. Figure 2 shows XRD diagrams of the as-deposited W100 sample, measured within a period of 5 months after deposition.
b-W (211)
(a)
(b) 132 days
132 days 55 days
55 days 0 days
0 days
30 32 34 36 38 40 42 44 46 48 50 38.5
39.0
39.5
40.0
40.5
41.0
41.5
2q (degrees)
2q (degrees)
Figure 2: XRD diagrams of the 100 nm thick W layer during storage in roomtemperature conditions. Figure (b) is a zoom of figure (a). Points represent experimental data and continuous lines fits to the data. The diagrams are shifted in the vertical axis for clarity purposes.
7
ACCEPTED MANUSCRIPT
AC CE
PT
ED
MA
NU
SC
RI P
T
As it can be seen, with the increase of time there is only a small enhancement of the α-W (110) Bragg peak at the expense of β-W (210) one. Curve fitting of the double peak at 40o reveals a small degree of β-W to α-W phase transition upon room-temperature storage of the W100 sample. The β-W content of the sample is 74.4% immediately after deposition, while it decreases to 68.6% after 55 days and to 66.7% after 132 days, which corresponds to a 10% decrease of the β-W phase; a 11% decrease of resistivity is observed during the same period (from 127.1 µΩ·cm to 112.6 µΩ·cm), in accordance to the β-W content decrease. Thus, after several months of room temperature storage we only observe a small structural variation of the W100 sample and no measurable variation of the samples with thinner W layers (data not shown). The above findings are contrary to the room-temperature β-W to α-W phase transition reported in reference [17] at mixed-phase W films with thickness in the range between 7 and 30 nm; following the discussion of section 3.1, the enhanced tendency for obtaining α-W reported in reference [17] can be attributed to the extremely high DC power used (3 W/cm2 ), the low deposition pressure (1 mTorr), and the low source-to-substrate distance (100 mm), which result in reduced scattering of the W atoms and to an extremely high deposition rate (1.2 nm/s). However, mixed α-W and β-W phase films deposited using DC diode sputtering retain their initial properties even after 3 months of room-temperature storage [19], due to the moderate power used (1.5 to 2 W/cm2 ) and the high sputtering pressure (4 to 10 mTorr), resulting in lower deposition rate (0.13 to 0.17 nm/s) and low deposited W atoms energy, which enhance the formation of β-W phase, in agreement to our findings.
3.3
The effect of annealing
Film morphology before and after annealing is studied by means of AFM and XRR measurements. AFM was employed to explore the surface morphology of the films, yielding the surface roughness of the top 2 nm thick alumina layer. However, as the alumina layer thickness does not changes across the various samples, any discrepancies in the obtained surface roughness values 8
ACCEPTED MANUSCRIPT
MA
NU
SC
RI P
T
reflect variations of the W films morphology. All the studied films have extremely low surface roughness before and after annealing: the root meansquare surface roughness of the as-deposited samples decreases from 0.38 nm at W100 nm, to 0.11 nm at W1.5 nm (data not shown). After annealing, even at the highest temperature (800o C), surface roughness remains unchanged, within the resolution limits of the microscope; as shown in figure 3, samples W100A800 and W1.5A800 have surface roughness equal to 0.36 nm and 0.15 nm respectively. This indicates that the underlying W film morphology is not significantly affected by annealing and no W agglomeration occurs, even for the 1.5 nm thick film, an effect that would be manifested by a steep surface roughness increase due to the formation of W islands. This finding is corroborated by the XRR results discussed below.
(c) 7
AFM profiles after annealing at 800oC
6 W (1.5 nm)
Height (nm)
(b)
AC CE
PT
ED
(a)
5 4 W (100 nm)
3 2 1 0 0
200
400
600
800
1000
Distance (nm)
Figure 3: AFM images of samples after annealing at 800o C. (a) 1.5 nm thick W film. (b) 100 nm thick W film. The color scale (from dark to bright) is 2.4 nm and the white scale bar is 400 nm, in both images. (c) Line profiles obtained along the indicated directions at each image.
The XRR diagrams from W1.5, W3 and W5.5 samples before and after 9
ACCEPTED MANUSCRIPT
X-ray reflectivity (a.u.)
MA
NU
SC
RI P
T
annealing at 800o C are presented in figure 4. Least square fitting to the data was performed using the LEPTOS 6.02 software by Bruker AXS GmbH. In order to fit the experimental data, we have to assume a structural model, i.e., number of layers and interlayer profiles or interlayer roughness. Each layer is characterized by its thickness and density which are adjustable parameters to be determined by the least squares fitting. The optimum parameters calculated by the least squares minimization procedure are subsequently used for the simulation of the theoretical reflectivity curve (continuous lines in figure 4). Significant deviations between the simulated curve and the experimental data suggest that the assumed structural model is erroneous and a new one has to be envisaged. The thickness, density and roughness of the layered structure are determined.
(a)
annealed at 800oC
AC CE
0
PT
annealed at 800oC
as-deposited
1
2
3
4
5
2q (degrees)
6
7
W (5.5 nm)
(b)
W (3 nm)
ED
W (1.5 nm)
annealed o at 800 C
as-deposited
as-deposited 0
1
2
3
4
(c)
5
2q (degrees)
6
7
0
1
2
3
4
5
6
7
2q (degrees)
Figure 4: X-ray reflectivity versus scattering angle, before and after annealing at 800o C. The black open dots represent experimental data and the red lines are least-square fits to the data. W film thickness is (a) 1.5 nm, (b) 3 nm, and (c) 5.5 nm.
In order to fit the experimental XRR data initially, a single W layer on the top of the Si/SiO2 substrate was assumed followed by an Al2 O3 cover layer. However, for W1.5 and W3 samples this structural model failed to give a good agreement between the simulated curves and the experimental data. Next, a structural model in which the W layer is divided into two sublayers, W1 and W2, was assumed with the thickness, density, and roughness of both W1 and 10
ACCEPTED MANUSCRIPT
W1.5A800
W3
W3A800
AC CE
W5.5 W5.5A800
interface roughness (nm) alumina/W W/substrate
SC
NU
Al2 O3 W1 W2 Al2 O3 W1 W2 Al2 O3 W1 W2 Al2 O3 W1 W2 Al2 O3 W Al2 O3 W
MA
W1.5
thickness layer density (nm) (g/cm3 ) 1.77 3.66 1.64 3.90-17.90 0.72 19.25 1.84 2.00 1.02 5.40-14.30 1.07 14.30 1.70 3.70 0.88 7.20-18.50 2.05 18.50 1.83 3.00 0.82 7.20-19.25 1.94 19.25 2.14 2.38 4.73 19.37 2.71 1.83 4.49 19.37
ED
layer
PT
sample
RI P
T
Table 2: XRR fitting values for the 1.5, 3, and 5.5 nm thick W films, asdeposited and after annealing at 800o C.
0.50 0.36
0.35 0.32 0.44 0.42 0.40 0.41 0.29
0.40
0.27
0.47
W2 layers as parameters to be least squares fitted. The least squares fitted curves for this model and the experimental data for both the as deposited and annealed samples were found to be in reasonable agreement. The least squares fitted curves to the data are presented in figure 4 as continuous lines. The results from the least square fittings to the XRR data are shown in table 2. Least square fitting to the data from the as deposited films show that the two thinnest films are composed of two sublayers; the top layer represents a region with a density gradient, becoming less dense at its upper interface and its average density is smaller than that of the bottom layer, which has 11
ACCEPTED MANUSCRIPT
AC CE
PT
ED
MA
NU
SC
RI P
T
a density close to the tungsten bulk value (19.25 g/cm3 ). Annealing of the 1.5 nm thick film results in a decrease of the bottom layer density, which may be attributed to interdiffusion from the top and bottom interfaces; the opposite occurs at the 3 nm thick film, where a densification of the bottom sublayer is observed. The 5.5 nm thick film, both before and after annealing, is shown to be composed of a sole W layer, with density close to that of bulk W. This thickness dependent behavior is assumed to stem from the morphological evolution of the films as their thickness increases during the initial stages of film growth. In all cases, XRR shows that the films are continuous after annealing, even the 1.5 nm thick one, without any indication of W agglomeration and there is a strong presence of high-density W. The upper (W/alumina) and lower (substrate/W) W layer interfaces are shown to have roughness between 0.30 and 0.50 nm, before and after annealing, in agreement to the surface roughness values measured by AFM. The effect of annealing on the crystal structure of the W films has been studied by GIXRD and XRD measurements. In figure 5a it can be seen that after annealing of the W5.5 sample at 300o C a broad peak appears at around 40o , which is accompanied by two smaller Bragg peaks at around 35o and 44o , indicating the presence of both β-W and α-W phases. As the annealing temperature increases, the intensity of the α-W (110) Bragg peak increases, while the β-W phase is suppressed. Figures 5b and c show that for thicker films the β-W phase is preserved after annealing at 200o C. However, after annealing at 300o C, or higher, only the α-W (110) Bragg peak is present indicating the β-W to α-W phase transition. Annealing at higher temperatures does not result in any observed change: indicatively the XRD diagrams obtained after annealing at 800o C are shown. It should be noted that similar behavior is also observed for the W15 and W20 samples. The above mentioned pronounced phase transition is also manifested by a corresponding steep resistivity drop. Figure 6a shows the evolution of resistivity upon annealing, for all the W thicknesses studied. As it was mentioned in Section 3.1, the as-deposited film resistivity increases as the film thickness decreases, since interface scattering becomes increasingly important for thinner films: for 100 nm thickness the resistivity is 127.1 µΩ·cm, while it reaches 2453.9 µΩ·cm for the 1.5 nm thick film, indicating a total domination 12
ACCEPTED MANUSCRIPT
(a)
RI P
a-W b-W (210) (211)
T
b-W (210)
Intensity (a.u.)
b-W (200)
800oC
SC
450oC
300oC
as-deposited
NU
30 32 34 36 38 40 42 44 46 48 50
2q (degrees) b-W (210)
a-W b-W (210) (211)
(b)
b-W (200)
b-W (210)
a-W b-W (210) (211)
Intensity (a.u.)
MA
b-W (200)
(c) 800oC
800oC
300oC
300oC
200oC
as-deposited
as-deposited
ED
200oC
PT
30 32 34 36 38 40 42 44 46 48 50 30 32 34 36 38 40 42 44 46 48 50
2q (degrees)
2q (degrees)
AC CE
Figure 5: GIXRD diagrams of the 5.5 nm thick sample before and after annealing (a). XRD diagrams of samples before and after annealing. W thickness is (b) 10 nm and (c) 100 nm. The diagrams are shifted in the vertical axis for clarity purposes.
of interface scattering for the thinner films [20]. After annealing at 200o C there is minor resistivity change for all the films, except for the two thinner ones (W1.5 and W3) where a significant increase is observed. Annealing at 300o C results in a steep resistivity drop for the films having thickness between 10 and 100 nm. At this annealing temperature, only a minor change is observed for W5.5 and W8, indicating that their structure continues to be predominately β-phase, in conjunction to the GIXRD measurements. After annealing at 450o C a steep resistivity decrease is also observed for the 5.5 and 8 nm thick films. The two thinnest films 13
ACCEPTED MANUSCRIPT
t = 5.5 nm t = 12.5 nm
450
6000
250
4000
200 150
3000
100 50
RI P
300
2000
0 0
100 200 300 400 500 600 700 800 o
1
10
100
10
3
10
2
10
1
W thickness (nm)
MA
Annealing temperature ( C)
(b)
SC
5000
Resistivity (mWcm)
350
NU
Resistivity (mWcm)
as-deposited (b-W) A800 (a-W)
(a)
400
Resistivity (mWcm)
t = 3 nm t = 10 nm t = 100 nm
T
t = 1.5 nm t = 8 nm t = 20.5 nm
ED
Figure 6: Sample resistivity before and after annealing. (a) Resistivity versus annealing temperature. (b) Resistivity versus W thickness. Error bars are omitted for clarity purposes. The measurement error is approximately 4%.
AC CE
PT
(W1.5 and W3) present a non monotonous variation of the resistivity as a function of the annealing temperature, showing a maximum at 450o C (figure 6a). This dependence of the β-W to α-W phase transition temperature on the thickness of the film has been also reported elsewhere [9]; it is attributed to the unintended annealing of the films during sputter deposition, which results in the onset of the β-W to α-W phase transition: as film thickness increases, the deposition time increases as well, resulting in higher extent of the phase transition. The peculiar resistivity variation with annealing temperature of the two thinnest samples (W1.5 and W3) may be attributed to the competing action of bulk and interface electron scattering. According to the XRR results shown in table 2, the roughness of the bottom (SiO2 /W) and top (W/AlOx ) interfaces is not considerably affected during annealing; thus, interface electron scattering may be altered only due to the variation of the interface electronic structure and not due to the variation of interface defects. Simultaneously, there is a gradual β-W to α-W transition as annealing temperature increases. Moreover, it was previously shown [20] that the grain size to film thickness ratio of sputtered W films increases as film thickness decreases; grain size 14
ACCEPTED MANUSCRIPT
4
AC CE
PT
ED
MA
NU
SC
RI P
T
reaches a value of several times the film thickness at very thin W films. This suggests that for the W1.5 and W3 samples studied here, the top and bottom film interfaces always belong to a single grain. In this context, it is assumed that the transition from the low electron mean-free path (mfp) β-W grains to the high electron mfp α-W grains determines the electrical resistivity of W1.5 and W3: during this gradual transition to the high electron mfp α-W grains (which results in higher interface scattering), the competing interplay between the increasing interface scattering and decreasing bulk scattering causes the observed resistivity variation. Figure 6b shows the variation of the resistivity versus W film thickness. The two extreme cases are depicted: as-deposited and after annealing at 800o C; in the former case films are predominately β-phase while in the later case α-W prevails. As there is negligible change of the top and bottom W film interface roughness, any resistivity variation is attributed to changes of the film electronic structure. The slower resistivity decrease of the as-deposited films, as film thickness decreases, is explained due to the lower electron mfp in β-W, as compared to the one in α-W. A significant resistivity increase is observed between 8 nm and 10 nm thickness, indicating a β-W electron mfp close to these values. On the contrary, in the case of the predominately α-W films, a significant resistivity increase is observed between 10 nm and 20 nm thickness, indicating the higher electron mfp in α-W.
Conclusions
Thin W films were deposited by magnetron sputtering on Si/SiO2 substrates at thicknesses varying between 1.5 and 100 nm. As deposited films present both the β- and α-phase, with the beta phase content prevailing. The high content of the metastable beta phase is attributed to the low deposition rate used, compared to other studies, indicating the strong dependence of the film structure on the deposition parameters. Upon annealing a transformation from β- to α-phase occurs with the transition temperature depending on the film thickness. Films with thickness equal or higher than 10 nm are β-phase up to 200o C, while thinner films are composed of β-phase up to 300o C. A steep resistivity decrease is observed when the β- to α-phase transition occurs. 15
ACCEPTED MANUSCRIPT
NU
Acknowledgements
SC
RI P
T
This behavior is not followed by the two thinnest films (1.5 and 3 nm) which show a resistivity peak at 450o C; this is attributed to the competing action between bulk and interface electron scattering, which is manifested only at the two thinnest films due to the high contribution of interface scattering to the film’s resistivity. The lowest resistivity value is 17.0 µΩ·cm and it is obtained for the 100 nm thick film, after annealing at 800o C.
MA
Funding from the E.C. through a FP7-ICT project (GA No. 318144) is acknowledged.
ED
References
PT
[1] F. Papadatos, K. Wong, V. Arunachalam, C. H. Shin, Z. Li, M. Chudzik, W.-H. Lee, and Aimin Xing, Low resistivity tungsten for 32 nm node MOL contacts and beyond, Microelectron. Eng. 92 (2012) 123-125.
AC CE
[2] L. Yin, C. Bozler, D. V. Harburg, F. Omenetto, and J. A. Rogers, Materials and fabrication sequences for water soluble silicon integrated circuits at the 90nm node, Appl. Phys. Lett. 106 (2015) 014105. [3] S. Beyne, K. Croes, I. De Wolf, and Z. Tokei, 1/f noise measurements for faster evaluation of electromigration in advanced microelectronics interconnections, J. Appl. Phys. 119 (2016) 184302. [4] W. Steinh¨ogl, G. Schindler, G. Steinlesberger, M. Traving, and M. Engelhardt, Comprehensive study of the resistivity of copper wires with lateral dimensions of 100 nm and smaller, Appl. Phys. Lett. 97 (2005) 023706. [5] D. Choi, C. S. Kim, D. Naveh, S. Chung, A. P. Warren, N. T. Nuhfer, M. F. Toney, K. R. Coffey, K. Barmak, Electron mean free path of tungsten and the electrical resistivity of epitaxial (110) tungsten films, Phys. Rev. B, 86 (2012) 045432. 16
ACCEPTED MANUSCRIPT
RI P
T
[6] D. Choi, The electron scattering at grain boundaries in tungsten films, Microelectron. Eng. 122 (2014) 5.
SC
[7] C.-F. Pai, L. Liu, Y. Li, H.W. Tseng, D.C. Ralph, R.A. Buhrman, Spin transfer torque devices utilizing the giant spin Hall effect of tungsten, Appl. Phys. Lett. 101 (2012) 122404.
NU
[8] S. Ikeda, K. Miura, H. Yamamoto, K. Mizunuma, H. D. Gan, M. Endo, S. Kanai, J. Hayakawa, F. Matsukura, H. Ohno, A perpendicularanisotropy CoFeB/MgO magnetic tunnel junction, Nat. Mat. 9 (2010) 721.
ED
MA
[9] D. Choi, B. Wang, S. Chung, X. Liu, A. Darbal, A. Wise, N.T. Nuhfer, K. Barmak, A.P. Warren, K.R. Coffey, M.F. Toney, Phase, grain structure, stress, and resistivity of sputter-deposited tungsten films, J. Vac. Sci. Technol. A 29 (2011) 051512.
PT
[10] J. Lu, C.G. Lee, Numerical estimates for energy of sputtered target atoms and reflected Ar neutrals in sputter processes, Vacuum, 86 (2012) 1134.
AC CE
[11] K. Salamon, O. Milat, N. Radic, P. Dubcek, M. Jercinovic, S. Bernstorff, Structure and morphology of magnetron sputtered W films studied by x-ray methods, J. Phys. D: Appl. Phys. 46 (2013) 095304. [12] E. Vassallo, R. Caniello, M. Canetti, D. Dellasega, M. Passoni, Microstructural characterisation of tungsten coatings deposited using plasma sputtering on Si substrates, Thin Solid Films, 558 (2014) 189. [13] F.T.N. V¨ ullers, R. Spolenak, Alpha- vs. beta-W nanocrystalline thin films: A comprehensive study of sputter parameters and resulting materials’ properties, Thin Solid Films, 577 (2014) 26. [14] I. Horcas, R. Fern`andez, J. M. G´omez-Rodriguez, J. Colchero, J. G´omezHerrero, A. M. Baro, WSXM: A software for scanning probe microscopy and a tool for nanotechnology, Rev. Sci. Instrum. 78 (2007) 013705.
17
ACCEPTED MANUSCRIPT
RI P
T
[15] D. K. Schroder, Semiconductor Material and Device Characterization, third ed., Wiley, New Jersey, 2006. [16] M. J. O’Keefe, J. T. Grant, Phase transformation of sputter deposited tungsten thin films with A-15 structure, J. Appl. Phys. 79 (1995) 9134.
NU
SC
[17] S. M. Rossnagel, I. C. Noyan, C. Cabral, Phase transformation of thin sputter-deposited tungsten films at room temperature, J. Vac. Sci. Technol. B 20 (2002) 2047.
MA
[18] G. S. Chen, L. C. Yang, H. S. Tian, C. S. Hsu, Evaluating substrate bias on the phase-forming behavior of tungsten thin films deposited by diode and ionized magnetron sputtering, Thin Solid Films, 484 (2005) 83.
PT
ED
[19] G. S. Chen, H. S. Tian, C. K. Lin, G.-S. Chen, H. Y. Lee, Phase transformation of tungsten films deposited by diode and inductively coupled plasma magnetron sputtering, J. Vac. Sci. Technol. A 22 (2004) 281.
AC CE
[20] D. Choi, X. Liu, P. K. Schelling, K. R. Coffey, K. Barmak, Failure of semiclassical models to describe resistivity of nanometric, polycrystalline tungsten films, J. Appl. Phys. 115 (2014) 104308.
18
ACCEPTED MANUSCRIPT
T
Highlights
RI P
• The deposition and annealing parameters for obtaining α-W or β-W films are presented.
SC
• β-W films may be obtained up to 100 nm thickness and after annealing at 200◦ C.
AC CE
PT
ED
MA
NU
• High deposition rates are required for obtaining low resistivity α-W films.
19