Effect of Growth Temperature during the Atomic Layer Deposition of

Subscriber access provided by UNIV OF LOUISIANA

Functional Inorganic Materials and Devices

Effect of growth temperature during the atomic layer deposition of the SrTiO3 seed layer on the properties of RuO2/SrTiO3/Ru capacitors for dynamic random access memory applications Sang Hyeon Kim, Woongkyu Lee, Cheol Hyun An, Dae Seon Kwon, DongGun Kim, Soon Hyung Cha, Seong Tak Cho, and Cheol Seong Hwang ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.8b17366 • Publication Date (Web): 12 Nov 2018 Downloaded from http://pubs.acs.org on November 15, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Effect of Growth Temperature during the Atomic Layer Deposition of the SrTiO3 Seed Layer on the Properties of RuO2/SrTiO3/Ru Capacitors for Dynamic Random Access Memory Applications Sang Hyeon Kim,†, § Woongkyu Lee,†,‡, § Cheol Hyun An,† Dae Seon Kwon,† DongGun Kim,† Soon Hyung Cha,# Seong Tak Cho,† and Cheol Seong Hwang†,*

†Department

of Materials Science and Engineering and Inter-University

Semiconductor Research Center, Seoul National University, Seoul 08826, Republic of Korea

‡Department

of Materials Science and Engineering, Northwestern University,

Evanston, IL 60208, United States

#Department

of Engineering Practice, Seoul National University, Seoul 08826,

Republic of Korea

1 ACS Paragon Plus Environment

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 52

KEYWORDS: atomic layer deposition, SrTiO3, growth temperature, crystallization, DRAM, capacitor, seed layer

ABSTRACT The atomic layer deposition process of SrTiO3 (STO) films at 230 °C were studied with Sr(iPr3Cp)2 and Ti(CpMe5)(OMe)3 (Pr, Cp, and Me are propyl, cyclopentadienyl, and methyl groups, respectively) on Ru substrates. The growth behavior and property of STO films grown at 230 ° C were compared with those deposited at 370 ° C. With the limited over-reaction of the Sr precursor during the initial growth stage at a lower temperature, the cation composition was more controllable, and the surface morphology after crystallization annealing at 650 ° C had more uniform grains with fewer defects. Here, the excess reaction of the Sr precursor means the chemical-vapor-deposition-like growth of the SrO component mediated through the thermal decomposition of the adsorbed Sr precursor molecules via the reaction with the oxygen supplied from the partly oxidized Ru substrate. The second STO was


Page 3 of 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


grown at 370 ° C (main layer) on the annealed first STO layer (crystallized seed layer) to lead to the in-situ crystallization of the main layer. Due to the improved microstructure of STO films induced by the seed layer deposited at 230 °C, the bulk dielectric constant of 167 was obtained for the main layer, which was higher than the value of 101 where the seed layer was deposited at 370 ° C, even though the crystallization annealing condition of the seed layer and the deposition condition of the main layer were consistent. The seed layer grown at 230 ° C, however, had a lower dielectric constant of only ~49, whereas the high-temperature seed layer had a dielectric constant of ~106. Therefore, the low-temperature seed layer posed a severe limitation in acquiring an advanced capacitor property with the involvement of a low-dielectric interfacial layer.

Introduction



Page 4 of 52

SrTiO3 (STO) thin films have been studied for about two decades as one of the promising candidates for the next-generation dielectric of the dynamic random access memory (DRAM) capacitors.1-10 With the aid of the atomic layer deposition (ALD) technique, dielectric films with a permittivity of over 100 can be obtained in a three-dimensional structure, which can mitigate the charge loss problem of aggressively shrunken DRAM capacitors.11-17 Leskelä's group first accomplished the ALD growth of STO films with a permittivity of 100.16-17 Thereafter; several other groups joined this research field with various strategies, such as intermixing Sr- and Ti-rich STO or film growth with the aid of plasma energy.5,18-23 The authors’ group has also focused on the thermal ALD and performance improvement of the STO films since 2005.2,7,10,20,24-32 Thermal ALD is a preferred method of growing such dielectric films over the plasma-based processes because of its almost unlimited capability to grow a conformal film in a severely three-dimensional structure. The DRAM capacitor node, whose minimum lateral dimension is as small as 20 nm and whose height is ~1,000 nm, is an example structure, making the aspect ratio as high as ~50. Optimizing the multi-cation STO film, however, was a quite complicated task


Page 5 of 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


due to the involvement of various metal precursors, which have different degrees of reactivity towards the common oxygen sources, such as O3 and H2O.10 Another complication originated from the difficult crystallization to achieve a high dielectric constant. When the film was grown at a low temperature (~600 ° C), which causes a large leakage current increase, was not required after the main layer deposition by this method.25 Annealing process was performed only before STO main layer growth for crystallization of STO seed layer. A growth temperature higher than 350 ° C, however, was necessary to achieve a sufficiently high dielectric constant of the main STO films. This high process temperature caused an unexpected problem in STO growth when the commercially



Page 6 of 52

viable Sr precursor Sr(iPr3Cp)2 (Pr: propyl group; Cp: cyclopentadienyl group) was used. Sr(iPr3Cp)2 does not thermally decompose up to a temperature of ~400 ° C, and can thus be used in high-temperature ALD. In these ALD processes, highdensity O3 (~250 g/m3) is usually adopted as the oxygen source, which also oxidizes the Ru substrate to RuOx (x99.999%). STO main layer was grown on the crystallized STO seed layer by combining two TiO2 cycles and one SrO cycle at 370 °C. The layer densities of the Sr and Ti elements were measured via X-ray fluorescent spectroscopy (XRF, Thermoscientific, ARL Quant’X). The physical thicknesses of the STO films were measured with an ellipsometer (Gaertner Scientific Corporation, L115B). The surface morphologies of the STO films were confirmed via fieldemission scanning electron microscopy (SEM, Hitachi, S-4800) and atomic force microscopy (AFM, JEOL, JSPM-5200). The crystal structures were examined via



Page 10 of 52

glancing angle X-ray diffraction (GAXRD, PANalytical, X’pert Pro) with a 2 °incident angle, and the densities of the films were measured via X-ray reflectivity (XRR), using Cu Kα radiation. The depth profiles were investigated through the time-of-flight secondary ion mass spectroscopy (ToF-SIMS, ION-TOF, SIMS-5), which detected negative ion sputtering with a Cs+ gun. To measure the electrical characteristics, 20 nm RuO2 and 50 nm Pt were sequentially deposited through a sputtering method using a shadow mask with a 300 μm inner diameter on the thin film as a top electrode, and a Ru substrate layer was used as a bottom electrode. The area of the top electrode was measured using an optical microscope. The capacitance-voltage (C-V) electrical characteristics were measured using an HP4194A impedance analyzer at 10 kHz, and the leakage current density-voltage (I-V) values were measured using an HP4140 pA meter at room temperature.

Results and Discussion


Page 11 of 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In Figure 1(a) and (b), the self-limited growth behaviors of the SrO and TiO2 ALD process, respectively, at a temperature as low as 230 ° C, were explained. All the precursor pulse/purge and oxygen source pulse/purge steps showed well-saturated growth behavior, which indicates that 230 ° C is well within the ALD temperature window for both the Sr(iPr3Cp)2/H2O and Ti(CpMe5)(OMe)3/O3 combinations. The deposited amounts of Sr and Ti hardly increased when the Sr and Ti precursor injection times were over 3 seconds. 3 seconds was also a sufficient precursor feeding time for SrO and TiO2 ALD at 370 ° C in the previous studies.20,28 In this study, 6s-5s-6s-5s and 3s-10s-2s-10s were set as the ALD step times (precursor injection-Ar purge-oxygen source injection-Ar purge) for SrO and TiO2, respectively. Also, the established ALD cycles for the SrO and TiO2 layers were termed “subcycles”. Furthermore, STO films were grown by combining the SrO and TiO2 ALD processes, and the cation composition was controlled by changing the SrO/TiO2 subcycle ratio, as shown in Figure 1(c). The stoichiometric cation composition (Sr:Ti=1:1) was achieved by the TiO2:SrO=2:1 subcycle ratio.



Page 12 of 52

Once the ALD process condition of STO was obtained, the growth behavior of the STO film as a function of the ALD cycle was investigated, and the results are summarized in Figure 2(a). The variations of the film thickness of the STO films at both 230 and 370 °C (hereafter referred to as “lower temperature (LT)” and “higher temperature (HT)”, respectively) were studied with an identical subcycle ratio of TiO2:SrO=2:1. Note that this subcycle ratio is not the process condition for the stoichiometric STO film growth at HT. While the aforementioned excess growth of the STO film during the initial ALD stage (less than 20 cycles) existed in HT deposition, the growth at LT had higher linearity with suppressed initial enhanced growth. It is believed that the overgrowth of the Sr element due to the excess reaction between the substrate and the Sr precursor was retarded due to the lower thermal energy in the LT case. Figure 2(b) shows the Sr composition ratio (Sr/(Sr+Ti)) vs. thickness of the deposited STO films at HT and LT. In the HT case, the STO film had a ~70% Sr/(Sr+Ti) ratio in the first few cycles. The Sr composition gradually decreased and was not saturated until the STO film thickness became 17 nm. Interestingly, even the LT deposition process showed a ~70% Sr/(Sr+Ti) ratio at


Page 13 of 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


the initial stage, although the excess growth was significantly suppressed, as shown in Figure 2(a). The almost identical ~70% Sr/(Sr+Ti) ratio at the very thin film (~1 nm) suggested that the lowered deposition temperature could not retard the unwanted CVD-like deposition when the film was grown immediately on the Ru (RuOx) surface. The LT process, however, showed a much faster decrease in the Sr/(Sr+Ti) ratio and provided the saturated Sr composition of ~52-53% at a ~4-5 nm thickness, which was achieved at over 17 nm for the HT process with Sr composition of ~43-44%. This implies that excessive Sr incorporation for the HT process proceeded via the diffusion of the oxygen atoms liberated from the RuOx underlayer through the previously grown STO layer. Such diffusion must be a thermal activation process so that the LT process could suppress such an adverse effect, and so that the film could acquire the desired composition at a much lower thickness. As mentioned in the Introduction section, the 3-nm-thick STO films grown via the HT and LT processes were annealed by the RTA process to crystallize them so that they could serve as the crystalline seed layer for the subsequent main-layer STO growth at HT. The appropriately crystallized seed layer could induce the in-situ



Page 14 of 52

crystallization of the STO films (main layer) during the ALD process at HT, which could provide a highly desirable capacitance performance. Hereafter, the TiO2:SrO subcycle ratio was adequately adjusted to obtain stoichiometric STO films for the target thickness (2:1 for the LT seed, 5:1 for the HT seed [the HT seed tends to have excessive Sr incorporation, so the TiO2 subcycle number should be increased to make the 3-nm-thick film cation stoichiometric], and 2:1 for the HT main layer). Figures 2(c) and (d) show the planar SEM and AFM images of the STO seed layer grown at HT and LT, respectively, after the RTA crystallization annealing. Consistent with the previous reports, severe micro- or nano- voids and cracks were formed in the STO films grown at HT.32 The surface morphology, however, notably improved in the LT sample, with fewer defects, and the root mean squared (RMS) roughness, estimated from the AFM images, decreased from 0.81 nm to 0.33 nm. This could have been due to the better film property of the STO seed layer with a more optimal growth behavior and a constant cation composition during film growth, as shown in Figure 2(a) and (b). Moreover, due to the less initial excess growth and the reduced growth amount per cycle and per time, the total deposition cycle and process time


Page 15 of 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


required for STO seed layer growth at LT were 54 cycles and 1,296 seconds, respectively, which are far larger than those at HT (18 cycles, 270 seconds). Therefore, the atoms in the STO seed layer grown at LT might have a better chance to be stabilized with a more uniform configuration. Due to the amorphous phase of the as-deposited seed layer, the surface morphology was very smooth with a RMS roughness value of lower than ~0.5 nm in both HT and LT cases. No distinctive defects were observed in as-deposited amorphous films and the voids and cracks might be produced by the densification during the annealing process. (See Supporting Information Figure S1) It was previously reported that the surface morphology of the STO seed layer affected the morphology of the following STO main layer32, which influenced the growth rate of the main STO layer, although the growth conditions for the main layer (HT) remained invariant. In the inset of Figure 2(a), the growth rate of the main layer on the HT seed layer was 0.15 nm/cycle, slightly higher than that on the LT seed layer (0.12 nm/cycle), probably due to the higher roughness of the HT STO seed layer surface. (The roughness of the films is also shown in Figure 2.) In both cases, no initial excess growth was observed during



Page 16 of 52

the main layer deposition. This indicates that the ratio of exposed Ru substrate region by the defect formation in the seed layer during annealing was not large enough to show a significant effect in the thickness and Sr-composition measurement. The overgrowth effect was not evidently shown in this case because the ellipsometry and XRF measured the entire (seed and main) STO films. The growth rate of the main layer was higher for both cases compared to the saturated growth rate of the seed layer at HT shown in Figure 2(a) (0.07 nm/cycle), which suggests the higher growth rate of the in-situ crystalline phase than that of the amorphous phase (the seed layer directly on top of the Ru substrate). GAXRD confirmed the crystallinity of the STO films, as shown in Figure 3(a). The GAXRD peak of STO was hardly observed in the annealed STO seed layer grown at LT due to the low thickness of 3 nm, while sharp peaks of the Ru (100), (002), and (101) planes from the substrate existed (blue line). This was also the case for the HT seed.20 When the STO main layer was deposited at HT on the annealed STO seed layer (total STO thickness: ~24 nm), however, distinct peaks of the STO (110), (111), and (200) planes appeared, although there was no annealing after the main-layer


Page 17 of 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


growth. This implies that the method of in-situ crystallization of STO films with the aid of a pre-crystallized STO seed layer also worked well with the newly developed LT process of the STO seed layer (black line), similar to the previous studies’ reports on the HT seed layer process.20 The similar in-situ crystallization of the main STO on the crystallized HT seed was again confirmed in this work (red line). The peak intensity and full-width half-maximum of the STO peaks from the two samples were comparable (inset of Figure 3(a)). As clearly shown in the inset of Figure 3(a), however, the position of the STO (110) and (200) peaks from the LT seed sample shifted towards the higher 2 theta region compared with the peaks of the HT seed sample (32.2° 32.5°; 46.3°  46.5°). In JCPDS #350734, the bulk STO films have the peaks of the (110) and (200) planes at 32.5 and 46.5 °, respectively. This suggests that the main STO film with the HT seed had a certain level of stress (whose precise nature can hardly be determined using the GAXRD pattern), which was well relieved when the LT seed was adopted. XRR analyses were conducted to understand the physical structure of the STO films further, as shown in Figure 3(b) and (c). The simulation of the spectra (red line) using the parameters related with the



Page 18 of 52

density, roughness, and thickness reproduced the experiment data (black line) precisely. The fitting revealed the following: HT seed and HT main - 4.40 g/cm3 density, 0.7 nm interface roughness, and 22 nm thickness; and LT seed and HT main - 4.26 g/cm3 density, 0.8 nm interface roughness, and 22 nm thickness. Therefore, the density decrease of the film through the adoption of the LT seed layer was not very high, and the interface roughness was similar. The roughness of the main layer surface, however, was obviously improved from 3.7 nm for the HT seed to 1.8 nm for the LT seed, due to the decreased voids and cracks of the STO film. This explains the growth rate difference of the main layer shown in the inset of Figure 2(a). The electrical properties of RuO2/STO/Ru-structured capacitors were investigated and are shown in Figure 4, where STO films were grown through the two-step processes (main layer growth at HT after crystallization of the seed layer grown at LT and HT). Figure 4(a) shows the variation of tox as a function of the physical oxide thickness (tphy) of the STO films (summation of the 3-nm-thick seed layer and the main layer). The tox of a certain sample can be calculated from the estimated capacitance density (C/A, where A is the electrode area) through tox=3.9 × 0/(C/A),


Page 19 of 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


where 0 is the vacuum permittivity. For the comparison, the data with the HT seed layer process from the previous report20 were included in the same graph. In Figure 4(a), tox could be estimated using the equation below.

tox = (3.9/kbulk (main)) × tphy (main) + (3.9/kseed) × tphy (seed) + toxi(intrinsic) (1)

Here, kbulk(main), tphy(main), kseed, tphy(seed), and toxi(intrinsic) are the bulk dielectric constant of the main layer, the physical oxide thickness of the main layer, the dielectric constant of the seed layer, the physical oxide thickness of the seed layer, and the tox values by intrinsic interfacial property, such as dead layer and low-k interfacial layer. All the parameters, except for tphy(main), were considered constant; tphy(main) was the experimental variable. In the HT seed case (main(HT)/seed(HT)), kseed was close to

kbulk(main) (shown in Figure 5(d)), and equation (1) can be simplified into equation (2) below.



Page 20 of 52

tox = (3.9/kbulk) × tphy + toxi(intrinsic)

(2)

Here, kbulk is the bulk dielectric constant of the STO film, which is similar to kbulk(main) (also, to kseed). The best-linear-fitted line (solid red line) in Figure 4(a) for the HT seed case produced a kbulk of 101 from the inverse slope, and a 0.02 nm toxi(intrinsic) was obtained at the Y-axis intercept. In the LT seed case (main (HT) on crystalline seed (LT)), however, kseed could be different from kbulk

(main)

(kseed must be much

smaller than kbulk (main) by Figure 5(d)), and equation (1) was used to interpret the LT seed data in Figure 4(a). As the interface roughness between STO and Ru was identical (Figure 3(b) and (c)), and as the main layer surface roughness was even decreased in the LT seed layer case, it is believed that the capacitance loss due to the intrinsic interface property of the LT seed case is comparable to that of the HT seed case. Therefore, 0.02 nm was also adopted for toxi(instrinsic) for the LT seed case. Now, the seed and the main layer regions could be separately considered in the LT seed case shown in Figure 4(a). When tphy is over 3 nm (main-layer region), the inverse slope of the best-linear-fitted line (solid black line) from the measured data


Page 21 of 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


gave a 167 kbulk(main). It is notable that the main layer showed such a large difference in kbulk depending on the seed layer type, although the cases were grown under identical conditions. In the previous study, it was identified that the nanovoid regions of the STO seed layer induced the amorphous-phase regions of the STO main layer due to the absence of a crystallized seed matrix, which is essential for the in-situ crystallization of the main layer.32 The density of micro/nano-defects in the seed layer is directly related to the portion of the low-permittivity amorphous phase in the main layer, which must have contacted the Ru substrate film without the intervening crystallized seed layer.32 Therefore, kbulk(main) was believed to have been enhanced by the LT seed process, due to the improved film morphology of the seed layer (the SEM images in Figure 2) associated with the decrease in the initial overgrowth and the longer deposition time. The lower stress in the STO film with the LT seed compared to that with the HT seed, as shown by the GAXRD results in Figure 3(a), might have also contributed to the improved dielectric property. In the tphy range of lower than 3 nm (seed layer region), another blue-dotted linear line was plotted from the end of the main-layer data at a 3 nm tphy to a 0.02 nm Y-intercept (toxi(intrinsic)).



Page 22 of 52

From the inverse slope of this plot, kbulk(seed) was estimated to be 23, which is much lower than the expected value in crystallized STO thin films. The possible reasons for such low-k value of the thin seed layer will be discussed later (Figure 5). Figure 4(b) shows a summary of the electrical properties of capacitors with an HT main layer on HT20 or that underwent the LT seed layer process. Since the bottom electrode was grounded and the bias was applied to the top electrode, the leakage current is from the electrons injected from the Ru bottom electrode into the STO layer. As the thickness of the dielectric layer decreased, the tox also decreased, but the leakage current density increased in both the HT and LT seed cases. For the LT seed layer case, 0.70 nm tox was achieved with a low leakage current density (9.2× 10-8 A/cm2 @ 0.8 V), which is acceptable in DRAM capacitor application at a 10-nm-

tphy sample. The tphy value that satisfies the low leakage current level (

Effect of Growth Temperature during the Atomic Layer Deposition of

Recommend Documents