This site uses cookies. By continuing to use this site you agree to our use of cookies. To find out more, see our Privacy and Cookies policy.
Brought to you by:
Paper

Traceable thickness measurement of ultra-thin HfO2 films by medium-energy ion scattering spectroscopy

, , , , and

Published 12 February 2020 © 2020 BIPM & IOP Publishing Ltd
, , Citation Kyung Joong Kim et al 2020 Metrologia 57 025001 DOI 10.1088/1681-7575/ab57f0

0026-1394/57/2/025001

Abstract

The thicknesses of a series of ultra-thin HfO2 films were precisely determined by mutual calibration by x-ray photoelectron spectroscopy (XPS) and x-ray reflectometry (XRR) in the recent Consultative Committee for Amount of Substance (CCQM) pilot study P-190. From these well-defined reference film thicknesses, the measurement capability of medium-energy ion scattering spectroscopy (MEIS) for the thickness of HfO2 films was investigated. The film thicknesses determined by MEIS showed a small difference, within 2%, from the reference thicknesses and an offset value of 0.017 nm. The MEIS thicknesses can also be determined by mutual calibration between the transmission electron microscopy (TEM) thicknesses and the MEIS intensity ratios in the region of the substrate and HfO2 film. From linear fitting with the reference thicknesses, the MEIS thicknesses determined by mutual calibration showed a slope value of 1.011 and an offset value of 0.015 nm. As a result, MEIS can be a traceable method to determine the absolute thickness of ultra-thin HfO2 films, and a zero-offset method for application of the mutual calibration method.

Export citation and abstract BibTeX RIS

1. Introduction

The thickness measurement of ultra-thin oxide films is very important for the development of semiconductor devices. For this reason, the thickness measurement of nanometer SiO2 films was chosen as the first subject of the Consultative Committee for Amount of Substance (CCQM) Surface Analysis Working Group (SAWG) [15]. In the first pilot study, P-38, many kinds of thickness measurement methods were compared for thickness measurements of nanometer SiO2 films on Si substrates. There were large offset values in a range from 0.5 nm to 1.0 nm in spectroscopic ellipsometry (SE), transmission electron microscopy (TEM), Rutherford backscattering spectroscopy (RBS), nuclear reaction analysis (NRA), medium-energy ion scattering spectroscopy (MEIS) and x-ray reflectometry (XRR) [2]. These offset values are too large to accept because the thickness measurement range is only a few nanometers. As a result, it was found that the absolute thickness of ultra-thin oxide films is very difficult to determine using an individual surface analysis method.

Fortunately, however, the offset value of x-ray photoelectron spectroscopy (XPS) was found to be zero [6, 7]. This is because thickness measurement by XPS is determined from the relative ratio of the signal intensities of the constituent elements in the film and the substrate materials. That is, if the oxide thickness is close to 0, the peak area of the oxide component should be 0. The fact that the offset value of XPS is zero provided an opportunity to design a mutual calibration method to determine the absolute thickness of nanometer oxide films [8]. The mutual calibration method combines the strong points of a zero-offset method (such as XPS) and a length-unit traceable method (such as TEM). The mutual calibration method using XPS and TEM was applied to thickness measurements of SiO2 [8], Al2O3 [9], and HfO2 films [10].

HfO2 is a dielectric material that can be used as an alternative to SiO2. Therefore, a traceable thickness measurement of ultra-thin HfO2 films, using physical or chemical methods, is required for advanced semiconductor industries. Thickness measurement of nanometer HfO2 films was investigated as a recent subject of the pilot study P-190 by the CCQM SAWG. In P-190, the certified reference thicknesses of six HfO2 films were determined by mutual calibration between the average thicknesses by XPS and XRR [11].

MEIS is known to be an excellent method to determine the amount of substance of ultra-thin films [1214]. In this study, MEIS was investigated as a zero-offset method for application of the mutual calibration method, and suggested as a traceable method to determine the absolute thickness of ultra-thin HfO2 films.

2. Experimental

2.1. Fabrication of samples

A series of HfO2/SiO2/Si(1 0 0) films were fabricated for certification of thickness by mutual calibration with a length-unit traceable method and a zero-offset method. Six HfO2 films with nominal thicknesses of 1.0 nm, 1.5 nm, 2.0 nm, 2.5 nm, 3.0 nm, and 4.0 nm were grown on the polished side of Si(1 0 0) substrates by atomic layer deposition. Before the growth of the HfO2 films, to prevent the diffusion of oxygen atoms from the HfO2 films to Si(100) substrate, 2 nm SiO2 layers were grown on the Si(1 0 0) substrates by thermal oxidation, as shown in figure 3. After growth of the HfO2 films, the wafers were cut into small specimens with sizes of 10 mm  ×  10 mm.

2.2. Determination of reference thickness

To investigate the traceability of thickness measurement, the determination of well-defined film thicknesses is an important prerequisite. The reference thicknesses of the HfO2 layers in a series of HfO2/SiO2/Si(1 0 0) films were determined by mutual calibration between the average thicknesses by XPS and XRR reported in the pilot study P-190 by 11 national metrology institutes. The linear fitting line showed the good linear relationship between the two average thicknesses of the XPS and XRR. The reference thicknesses were determined from the average values of the XPS thicknesses corrected by the slope, and the XRR thicknesses corrected by the offset value of the linear fitting line. These reference thicknesses can be regarded as traceable values because they are based on the thickness obtained by XRR, where the thickness scale is based on the wavelength of x-rays.

2.3. TEM analysis

Thicknesses of HfO2 films were determined from high-resolution (HR) TEM micrographs collected using FEI-F30 microscopes operating at 300 kV. The film thicknesses of the samples were determined from the lattice constant of the Si(1 0 0) substrate. The HfO2/SiO2 interface was determined from the point with half of the average contrast of the SiO2 layer and that of the HfO2 layer. More than ten TEM images at different locations were obtained.

2.4. MEIS analysis

Thickness of the HfO2 films was measured using an MEIS system (K-120, K-MAC, Korea), which consists of an ion source, an accelerator, an energy analyser, and a detector, as schematically shown in figure 1. The incident and scattering angles are determined by the geometric arrangement between the ion source, the sample, and the detector. In MEIS analysis, ions generated from an ion source are accelerated to impinge on the sample surface, and the energy of the ions scattered by the nuclei of the constituent atoms is measured by an energy analyser. From the energy distribution of scattered ions, the in-depth locations of the constituent atoms can be determined from kinematic factors, and the quantity of atoms can also be determined from the scattering cross section of the atoms. An electrostatic analyser, magnetic sector analyser, and time of flight (TOF) analyser are widely used for MEIS analysis. Of these, the TOF analyser, because it is the most quantitative, is robust for thickness measurement of ultra-thin oxide films. The TOF analyser acquires scattered particles, both charged and neutralized, and so neutralization correction is not needed in the analysis; therefore, there is no potential error in the charge correction. In addition to this, a full energy spectrum is always obtained, which provides a perfect internal reference.

Figure 1.

Figure 1. A schematic diagram of a typical MEIS system. The incident angle is α, and the scattering angle is θ.

Standard image High-resolution image

3. Results and discussion

In this study, to investigate the measurement capability of the MEIS system, the thicknesses of a series of HfO2/SiO2/Si(1 0 0) films used in the pilot study P-190 were measured by MEIS and HR-TEM, as shown in figure 2. In the measurement, He+ ions with a voltage of 100 keV were used. The incident angle was 45°, and the scattering angle was 130°. The pulse width of the ion beam was 350 ps. The ion beam current was about 100 pA, as monitored in situ during the experiment using a Faraday cup installed in the middle of the ion beam column. The ion dose on the sample was kept at 500 nC by controlling the irradiation time, which was calculated based on the ion beam current. The sample was continuously rotated around the normal axis of the surface to ensure random spectrum measurement. The detector open area for the data acquisition was confined to Δθ  =  5° and Δ$\phi $   =  5°.

Figure 2.

Figure 2. MEIS spectra of a series of HfO2/SiO2/Si(1 0 0) films used in the pilot study P-190.

Standard image High-resolution image

The signal intensity of crystalline Si is very reproducible within the relative standard deviation of 0.54%. On the other hand, Hf peaks show a gradual increase proportional to the reference thicknesses described in figure 2.

3.1. Thickness measurement by HR-TEM

Because its scale is based on the lattice constant, HR-TEM is a length-unit traceable thickness measurement method. In particular, in thin films grown on Si(1 0 0) wafers, the crystalline lattice planes of Si can be directly used as an internal standard to measure the absolute film thickness. The lattice distance between Si(1 1 0) planes in the cross-sectional TEM image is 0.543 nm.

An HR-TEM image of an HfO2/SiO2/Si(1 0 0) film can be simply converted to an intensity line profile image using the average contrast of the region of interest (ROI), as shown in figure 3. For precise measurement, the aspect ratio of the lattice line should be maximized by aligning the lines parallel to the lattice direction, and as such, parallel to the interface and film surface.

Figure 3.

Figure 3. An HR-TEM image (a) and an intensity profile image (b) of HfO2 (1.5 nm)/SiO2(2.0 nm)/Si(1 0 0) film.

Standard image High-resolution image

The locations of the interface and the surface at which the film thickness is measured can be determined from the contrast profile. The location of the SiO2/HfO2 interface can be determined from the point of half contrast between the average contrast of the SiO2 (${{I}_{{\rm Si}{{{\rm O}}_{{\rm 2}}}}}$ ) and HfO2 (${{I}_{{\rm Hf}{{{\rm O}}_{{\rm 2}}}}}$ ) layers, (${{I}_{{\rm Si}{{{\rm O}}_{{\rm 2}}}}}+{{I}_{{\rm Hf}{{{\rm O}}_{{\rm 2}}}}}$ )/2. In the same manner, the location of the film's surface can be determined from the average contrast of the HfO2 (${{I}_{{\rm Hf}{{{\rm O}}_{{\rm 2}}}}}$ ) and the glue (Iglue) layers using the relation (${{I}_{{\rm glue}}}+{{I}_{{\rm Hf}{{{\rm O}}_{{\rm 2}}}}}$ )/2.

The film thickness can be measured from the distance between the SiO2/HfO2 interface and the surface of the HfO2 layer. The thickness of the HfO2 layer can be simply determined from the ratio of the line width of the HfO2 layer and the width of 20 lines of Si(1 0 0) planes corresponding to the value of 5.431 nm for ten Si lattice constants [1517]. The average TEM thicknesses of the HfO2 films (TTEM), derived from more than ten TEM images at different locations, are shown in table 1.

Table 1. Thicknesses of HfO2 films determined by HR-TEM (TTEM).

Number of specimen Measured thickness (nm) Uncertainty parameters (nm) Expanded uncertainty U (nm)
um ud ur ul
1 1.25 0.03 0.100 0.01 0.00 0.21
2 1.60 0.02 0.100 0.01 0.00 0.20
3 2.11 0.02 0.100 0.01 0.00 0.21
4 2.57 0.05 0.100 0.01 0.00 0.22
5 3.03 0.03 0.100 0.01 0.00 0.21
6 3.76 0.02 0.100 0.01 0.00 0.21

The combined standard uncertainty (uc) is calculated from the equation $u_{{\rm c}}^{2}=u_{{\rm m}}^{2}+u_{{\rm d}}^{2}+u_{{\rm r}}^{2}+ u_{{\rm l}}^{2}$ . Here, um is the standard uncertainty in the measurement of film thickness, and ud is the standard uncertainty in the definition of the interface and surface, which is related to the offset value of the TEM. It was reported in the range of (0.1–0.2) nm [810]. In this study, a small value of 0.1 nm was assigned as the value of ud. Here, ur is the standard uncertainty in the measurement of the line width of the periodic Si(1 0 0) lattice planes. It is the uncertainty in the measurement of the interval of the periodic 20 Si(1 0 0) lattice planes, as shown in figure 3(b). The ur value was measured to be small to be 0.01 nm. Here, ul is the standard uncertainty of the variation of the Si lattice constant; ul is negligibly small to be 0.89  ×  10−8 nm [15]. The expanded uncertainty (U) was determined from the equation U  =  Kuc at a 95% confidence level.

3.2. Thickness measurement by MEIS

In MEIS analysis, the thickness of the HfO2 films can be determined from the measured number of constituent Hf atoms within a unit area and the number density of the bulk HfO2. The number of the constituent Hf atoms is measured by simulation of the obtained MEIS spectrum. MEIS is a program developed at K-MAC and used as a simulation tool for MEIS spectra analysis. In the analysis, the Andersen cross section [18] and Chu straggling [19] were used. For the stopping power, fitted values based on the experimental data [20] were used following recent MEIS round robin test results [21]. The simulations took into account the solid angle of the detector, and the kinematic broadening of each element was reliably applied. Multiple scattering effects were included in the calculation.

The thickness of the HfO2 films (TMEIS) can be determined by dividing the measured number of Hf atoms in the unit area (atoms cm−2) by the number density of the bulk HfO2 (=8.32  ×  1022 atoms cm−3). The number of Hf atoms was measured three times by MEIS. The number of Hf atoms and the thicknesses of HfO2/SiO2/Si(1 0 0) films derived from the number of Hf atoms are shown in table 2. The linear fitting results of the measured thickness compared to the reference thicknesses show a slope value of 0.981 and an offset value of 0.017 nm. The slope value of 0.981 means that the measured MEIS thicknesses of the HfO2 films are close to the reference thicknesses, within 2%.

Table 2. The number of Hf atoms and thicknesses of HfO2/SiO2/Si(1 0 0) films measured by MEIS.

Number of Hf atoms (1  ×  1015 atoms cm−2) MEIS thickness (nm) TMEIS (nm) Stdev (nm) RSD (%)
1 2 3 1 2 3
2.14 2.11 2.12 0.773 0.762 0.764 0.766 0.006 0.765
3.35 3.38 3.38 1.208 1.221 1.221 1.217 0.008 0.617
4.60 4.54 4.54 1.660 1.640 1.640 1.647 0.012 0.701
5.79 5.81 5.78 2.092 2.099 2.086 2.092 0.007 0.311
7.04 7.01 6.98 2.541 2.531 2.520 2.531 0.011 0.415
9.44 9.48 9.55 3.410 3.425 3.450 3.428 0.020 0.589

3.3. Thickness measurement by mutual calibration

Mutual calibration using a combination of a zero-offset method and a length-unit traceable method was suggested as a useful method to determine the absolute thickness of ultra-thin oxide films. MEIS can be used as a zero-offset method in the mutual calibration method because the film thickness depends on the number of atoms in the oxide film according to the basis of the number density of c-Si. In this study, the thickness of nanometer HfO2 films was directly measured by the mutual calibration method from the MEIS spectra and the thicknesses measured by HR-TEM shown in table 1.

In MEIS, the number of scattered ions is related to the scattering cross section of the constituent atoms. In the thickness measurement of an HfO2 thin film grown on crystalline Si substrate, the signal intensity of the substrate can be a basis for thickness measurement because the number density of crystalline Si is a constant. Figure 4 shows the MEIS spectra of Si(1 0 0) substrate (magenta) without film and HfO2/SiO2/Si(1 0 0) film (blue). The fact that the intensities of Si are identical in the substrate without film and in the ultra-thin HfO2/SiO2/Si(1 0 0) film is the basis for the thickness measurement.

Figure 4.

Figure 4. MEIS spectra of (a) Si(1 0 0) substrate without film (magenta) and (b) HfO2/SiO2/Si(1 0 0) film (blue).

Standard image High-resolution image

For this reason, the relative intensity ratio (RMEIS  =  IB/IA) of film material (B) to substrate (A) can be converted to the thickness (TMEIS) of the HfO2 layer. The proportional factor can be determined from the slope, derived from mutual calibration by MEIS, a zero-offset method, and HR-TEM, a length-unit traceable method. The intensities of the crystalline Si substrate (IA) were determined to be in the energy range from 46 keV to 56 keV from the MEIS spectra shown in figure 4. The intensities of Hf in the six HfO2 films (IB) were also determined to be in the energy range from 85 keV to 95 keV. Table 3 shows the MEIS intensities of the substrate (IA) and of the HfO2/SiO2/Si(1 0 0) films (IB), and their ratios (R  =  IB/IA) determined from the three MEIS spectra.

Table 3. MEIS intensities of substrate (IA) and HfO2 HfO2/SiO2/Si(1 0 0) films (IB) and their ratios (R  =  IB/IA) determined from three MEIS spectra.

Measurement 1 2 3 4 5 6
1 IA 456 211 456 813 457 875 458 174 459 707 462 033
  IB 111 006 175 448 239 460 303 839 369 497 500 151
  R  =  IB/IA 0.2433 0.3841 0.5230 0.6632 0.8038 1.0825
2 IA 456 968 455 821 457 874 458 486 458 420 461 614
  IB 111 259 175 679 238 932 303 589 368 062 500 280
  R  =  IB/IA 0.2435 0.3854 0.5218 0.6622 0.8029 1.0838
3 IA 457 262 456 454 458 058 457 814 459 414 461 386
  IB 111 185 176 238 238 702 303 155 367 120 502 713
  R  =  IB/IA 0.2432 0.3861 0.5211 0.6622 0.7991 1.0896
Average of R 0.2433 0.3852 0.5220 0.6625 0.8019 1.0853  
Stdev of R 0.0002 0.0010 0.0009 0.0006 0.0025 0.0038  
RSD of R (%) 0.07 0.27 0.18 0.09 0.31 0.35  

The variations in the values of the signal intensities of the crystalline Si substrate show very small relative standard deviations of 0.46%, 0.43%, and 0.38%. The signal intensities of the Hf peaks show a gradual increase proportional to the reference thicknesses. The variations in the values of the signal intensity ratios (R  =  IB/IA) in the three MEIS spectra of the HfO2/SiO2/Si(1 0 0) films are also very small at 0.07%, 0.27%, 0.18%, 0.09%, 0.31%, and 0.35%. These reproducible results can be a basis for traceable thickness measurement of ultra-thin oxide films.

In this study, the relative intensity ratios (RMEIS  =  IB/IA) of the film material (B) and substrate (A) were converted to thicknesses. The MEIS thicknesses (TMEIS) of the HfO2 films can be determined by mutual calibration from the average relative intensity ratio (RMEIS  =  IB/IA), shown in table 3, and the traceable thicknesses by TEM (TTEM), shown in table 1. Figure 5 shows the mutual calibration results of the MEIS intensity ratio (RMEIS) and the average TEM thicknesses (TTEM). The MEIS intensity ratios are very linearly proportional to the average TEM thicknesses.

Figure 5.

Figure 5. The mutual calibration result of the MEIS intensity ratio (RMEIS) and average TEM thicknesses (TTEM).

Standard image High-resolution image

The slope m and offset values c are key parameters for the determination of the film thickness by mutual calibration. The measured value of m is 3.065  ±  0.097; the measured value of c is 0.497 nm  ±  0.066 nm. From the slope and offset values, the certified MEIS thicknesses ($T_{{\rm MEIS}}^{{\rm C}}$ ) and TEM thicknesses ($T_{{\rm TEM}}^{{\rm C}}$ ) can be determined as follows:

Table 4 shows the certified MEIS thicknesses ($T_{{\rm MEIS}}^{{\rm C}}$ ) and TEM thicknesses ($T_{{\rm TEM}}^{{\rm C}}$ ) together with the measured raw thicknesses and the reference thicknesses. The slope and the offset values were derived from linear fitting of the measured and certified thicknesses by MEIS and TEM to the reference thicknesses. The MEIS thicknesses can be used as certified values because the length scale was calibrated from TEM based on the lattice constant of crystalline Si substrate.

Table 4. Reference thicknesses and certified thicknesses of HfO2 films determined by mutual calibration with MEIS and TEM.

Reference thickness (nm) Thickness (nm) Certified thickness (nm)
TRef RMEIS TTEM $T_{{\rm MEIS}}^{{\rm C}}$ $T_{{\rm TEM}}^{{\rm C}}$
0.76 0.24 1.25 0.74 0.75
1.20 0.39 1.60 1.18 1.10
1.64 0.52 2.11 1.60 1.61
2.08 0.66 2.57 2.03 2.07
2.52 0.80 3.03 2.46 2.53
3.36 1.09 3.76 3.33 3.26
slope 3.099 1.008 1.011 1.008
offset 0.015 −0.479 0.015 0.025
R 1.000 0.999 1.000 0.999

Figure 6 shows the linear fitting results of the MEIS thicknesses (TMEIS) and the reference thicknesses (TRef). The slope value (1.011) close to unity means the MEIS thicknesses are identical to the reference thicknesses, within 1.1%. The small offset value of 0.015 nm also means that the MEIS thicknesses are close to the reference thicknesses from P-190 performed by national metrology institutes (NMIs). It means that the role of MEIS as a zero-offset method can be said to be comparable to the role of average XPS thickness in P-190 within the uncertainty level.

Figure 6.

Figure 6. Linear fitting of MEIS thicknesses (TMEIS) and reference thicknesses (TRef).

Standard image High-resolution image

4. Conclusion

To investigate the measurement capability of the MEIS system, the thickness of HfO2 layers in a series of HfO2/SiO2/Si(1 0 0) films was measured by MEIS. The MEIS thicknesses determined by simulation showed a difference of about 2% from the reference thicknesses and the offset value of 0.017 nm. The MEIS thicknesses can be determined without simulation by mutual calibration between the TEM thicknesses and the MEIS intensity ratios in the region of the substrate and HfO2 film. The MEIS thicknesses obtained by mutual calibration showed a slope value of 1.011 and offset value of 0.015 nm in linear fitting with the reference thicknesses. MEIS can be used as a traceable method to determine the absolute thickness of ultra-thin oxide films and a zero-offset method for application of the mutual calibration method.

Acknowledgments

This work was supported by a National Research Council of Science and Technology (NST) grant from the Korean Government (MSIT) (No. CAP-18-04-KRISS).

Please wait… references are loading.
10.1088/1681-7575/ab57f0