Recent Development of Ligand Chemistries for Next Generation Conformal PEALD/ALD of Metal & Oxides

Transcription

1 Recent Development of Ligand Chemistries for Next Generation Conformal PEALD/ALD of Metal & Oxides ALD Workshop, Semicon Europa 2014, Grenoble, October 7 th 2014 N.Blasco and JM.Girard, Air Liquide Electronics

2 Outlines Need for new precursors and ligand systems Interest of the combinatorial heteroleptic chemistry for ALD New classes of heteroleptic precursors: Cyclopentadienyl-amidinate precursors: examples of Ln (Y 2 O 3 ), Co, Ni Allyl-amidinate precursors (Ni metal PEALD) Conclusions 2 October, 7 th 2014

3 Why New Precursors? New materials Emerging memories (ReRAM, SST-RAM, PCRAM) : HfO2, NiO, Nb2O5, Ta2O5, ZrO2, TiO2, SbTe, etc.. Extending DRAM/eDRAM stacks: ZMZ, M = Al, Nb, Ta, etc Sub 20nm CMOS scaling: HKMG stack, WF control and capping BEOL metals cap & liners (Co), barriers & self-formed barriers (Ru, Mn) Hardmasks Generational increase in new materials 180nm130nm 90nm 65nm 45nm 32nm 22nm 14nm Source: T.Henry, SMC Oct

4 Elemental Coverage ALD precursor availability vs Periodic table High volume manufacturing scale Study scale IA 4 H IIA Li Be Na Mg IIIA IVA VIIIB K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te Cs Ba Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po He F Ne Cl Ar Br Kr I Xe At Rn Fr Ra Rf Db Sg Bh Hs Mt Uun Er Tm Yb October, 7th, 2014 IIIB IVB VA VIA VIIA [ VIIIA ] IB La Ce Pr Nd Ac Th Pa U VB VIB VIIB B C N O IIB Al Si P S Pm Sm Eu Gd Tb Np Pu Am Cm Bk Uuq Dy Ho Cf Es Fm Md No Lr Lu

5 Why New Tailored Precursors? Increasing structures complexity DRAM 60:1 aspect ratio, 3D-NAND for >256Gb Emerging memories: ReRAM, PCRAM, STT-RAM Avanced logics FinFET Gate + back-end Metal liners most architectures are moving towards 3D and require new deposition processes capables of depositing conformal thin ALD layers in challenging conditions: low/high temperature, high aspect ratio, halide-free processes 3D FinFET transistor 5 Toshiba s P-BICS Samsung TCAT

6 ALD Precursor Metallic center: TARGET Ligand: VECTOR - Enabling metal supply - Enabling metal vaporization - Enabling ALD reaction Ligand choice is directly impacting facilitization and process performance. Heteroleptic chemistry allows to optimize properties - «take the best of each» 6

7 Interest of the Heteroleptic Approach Combining the advantageous properties of ligands from different families Functionalities to optimize: facilitization and ALD process performance aspects - Type of M-L bond - Steric hindrance - Polarity - By-products Heteroleptic Precursors Facilitization - Melting Point - Reactivity - Volatility - Thermal stability Process performance - Temperature window - Growth rate / throughput - Crystal., impurities) - 3D HAR conformality Ligand «toolbox» Effect / industrial viability 7

8 Ligand Chemistries in ALD - Examples X Halogen R Alkyl OR NR2 NR SiR3 CO Alkoxide Alkylamido Imide Alkylsylil Carbonyl Guanidinato Diene R N N Acac b-diketonato AcNac (Ketiminato) DAD Diazadiene Allyl PR3 Arene Phosphine NacNac (Diketiminato) R R Pyrrole-imine Amidinate Cyclopentadienyl Heterolpeptic chemistry: Combining ligands of different families in the same precursor 8

9 Heteroleptic Chemistry ZrCp(NMe2)3 Example Well-known AL monocyclopentadienyl improvement Bulky cyclopentadienyl ligand stabilization effect wider ALD window at the expense of GPC due to higher steric hindrance >50ºC increase in ALD window, 100% 250ºC in 60:1 HAR structures Zr(NEtMe)4 vs. ZrCp(NMe2)3 ZyALD TEMAZ 9 Precursor GPC w/ O3 (A/cycle) ALD window ( C) Zr(NEtMe) ZrCp(NMe2)

10 Heteroleptic Chemistry TiCp*(OMe)3 Clear to yellow liquid M.W.: Torr at 95 C d=1.081 g cm-3 White solid M.W.: M.P.: 192 C 1Torr at 142 C CVD Non uniform Reduction of the melting point, volatility imrpovement ( monomer) Improvement of ALD window: 275 C 350 C GPC: 350 C 10

11 Heteroleptic Chemistry TiCp*(OMe)3 Step coverage improvement TiO2 280C w/ O3 1:10 patterned wafer 1:40 patterned wafer S/C: ~94% S/C: ~72% 1:10 patterned wafer S/C: ~63% 1:40 patterned wafer No film Strong improvement of step coverage, both at 280 C and 350 C 11

12 Outlines Introduction approaches for new ALD processes Interest of the combinatorial heteroleptic chemistry for ALD New classes of heteroleptic precursors: Cyclopentadienyl-amidinate precursors: examples of Ln (Y2O3), Co, Ni Allyl-amidinate precursors (Ni metal PEALD) Conclusions 12

13 Lanthanides Homoleptic Precursors State-of-the-Art ReRAM resistive switch HKMG capping layers (La2O3) Ln-doped MOx (Y-doped HfO2) Lack of suitable precursor sources Reactive, liquid, volatile compounds desired Applications deployment limited by availability b-diketonate O3 required Low GPC Solid form Cyclopentadienyl R R Ln High reactivity Thermally stable Solid form 50 nm 1 :2 Y doped H fo 2 o m ( ) Intensity (arb. units) Ln series can be used for variety of applications: C N 2 a n n e a lin g d u rin g 1 0 m in. m (1 1 0 ) c(2 0 0 ) 50 nm 1:8 Y doped H fo 2 o C N 2 a n n e a lin g d u rin g 1 0 m in Amidinate High volatility High reactivity. Solid form Melting point reduction 45 2 T h e ta (2 ) R 13 c(2 2 0 )

14 Novel Lanthanide Cp-amidinates Concept Liquid 120 C Yttrium Thermal stability Volatility Reactivity Melting point Cyclopentadienyl Amidinate Combination Y(iPrCp)2(iPr-amd) is the only liquid Y precursor reported to date beyond unvolatile Y(nBuCp)3 14

15 Y(iPrCp)2(iPr-amd) ALD Evaluation Y2O3 ALD w/ H2O (University of Helsinki) Niinistö ALD 2013 GPC: 1.2Å/cycle GPC: 1.2Å/cycle Precursor Volatility Melting point GPC ALD window Ref. Y(MeCp)3 (H2O) 130 C 124 C 1.2 Å/cycle 200~400 C Niinistö et al. Y(iPr-amd)3 (H2O) Sub. 90 C >220 C 0.8 Å/cycle 270~295 C De Rouffignac et al. Y(iPrCp)2(iPr-amd) (H2O) 130 C <.R.T. 1.2 A/cycle >350 C Niinistö et al. Physical properties: Y(RCp)2(iPr-amd) outperforms homoleptic equivalents ALD performance: offers wider ALD window and equivalent/higher GPC 15

16 Extension to Other Lanthanides: Ce and Dy Same melting point issue exists for most lanthanides: Er, Dy, Ce, Gd, Pr New series of Liquid (or low mp) and volatile precursors for Ln2O3 ALD, opening new applications 16

17 Extension to Other Lanthanides: Gd and Pr Same melting point issue exists for most lanthanides: Er, Dy, Ce, Gd, Pr New series of Liquid (or low mp) and volatile precursors for Ln2O3 ALD, opening new applications 17

18 Outlines Introduction approaches for new ALD processes Interest of the combinatorial heteroleptic chemistry for ALD New classes of heteroleptic precursors: Cyclopentadienyl-amidinate precursors: examples of Ln (Y2O3), Co Allyl-amidinate precursors (Ni metal PEALD) Conclusions 18

19 State-of-the-art Ni ALD/PEALD Limited availability of high throughput Ni ALD/PEALD processes Ni ALD is challenging because of Ni reduction potential: -0.25V Few reported precursors and processes Reduction of Ni precursors requires NH3 (or more reducing species) CVD/ALD/P EALD GPC Å/cy Compound VP MP Temp. Concern NiCp2 2.4Torr 100 C C Ni(tBu-amd)2 1Torr 120 C 69 C ALD H C Reactivity Lim et al. Ni(dmamp)2 Sub. 0.01Torr 50 ºC 118 C PEALD H C O content Lee et al. Ni(tBu-DAD)2 115 C C ALD H2N-NMe C Volatility Knisley et al. Ni(Me-allyl) (ipr-amd) 1Torr 75 C Liquid PEALD NH C This work This work Ni(Me-allyl) (PCAl) 1Torr 137 C 58 C PEALD NH C This work This work No Ni metal ALD reported. NiO ALD + reduction. Most processes require NH3 as co-reactant. Need for highly reactive precursors. 19 Ref.

20 Novel Metal Allyl-Amidinates Concept Nickel Thermal stability Volatility Reactivity Melting point No O content Ni Allyl Ni Amidinate Combination Liquid 75 C Ni(ŋ3-2-methylallyl)(N,N -diisopropylacetamidinate) Allyl ligand: weak anionic bond, low molecular weight Heteroleptic allyl high volatility, high reactivity, liquids metal precursor source 20

21 Ni(Me-allyl)(iPr-amd) Comparison Thermal properties vs. homoleptic compounds Strong volatility enhancement Best-in-class stability vs. volatility compromise Vapor Pressure Melting Point Ni(EtCp)2 1.2Torr 100 C Liquid Ni(tBu-amd)2 1Torr 120 C 95 C Ni(iPr-DAD) C Liquid Ni(Me-allyl) (ipr-amd) 1Torr 75 C Liquid Precursor MW More volatile than homoleptic Cp and amidinates, and heteroleptic Cp-amidinates 21

22 Ni(Me-allyl)(iPr-amd) PEALD Performance Temp-act (oc) Reactant Pfurnace(Torr) Cycle R.F. (W) Prec. (sccm) NH3 (sccm) Tcan( C) NH * Annealing T Furnace - actual( C) Atmosphere H2. F.R.*(sccm) Pfurnace (Torr) Time (min) 280 H Res (on SiO2) 19µΩ cm Pure Ni films obtained at 240 C Carbon removed at ALD step, not at annealing step (carbidic) Self-saturation demonstrated at A/cycle 22

23 Ni(Me-allyl)(iPr-amd) PEALD Performance Temp-act (oc) Reactant Pfurnace(Torr) Cycle R.F. (W) Prec. (sccm) NH3 (sccm) Tcan( C) NH * Annealing T Furnace - actual( C) Atmosphere H2. F.R.*(sccm) Pfurnace (Torr) Time (min) 280 H As dep Annealed Smooth films obtained up to 230 C, Rugosity increases from 240 C w/ likely nucleation effect, 23

24 Ni(Me-allyl)(iPr-amd) PEALD Performance PEALD Reactant Pfurnace (Torr) Plasma (W) Prec.(sccm) NH3 (sccm) Tcan ( C) NH Time (min) Atmosphere F.R. (sccm) Temp actual (oc) Pressure (Torr) 60 H Annealing SEM: after annealing Ar 12s Prec 5s Ar 4s Ar 5s NH3 5s Temp. AR Diameter SC Res (SiO2) Prec 10s Ar 5s 240 C 1:10 155nm 83% 42μΩ.cm NH3 10s Temp. AR SC Res (SiO2) ~230 C 1: % 9μΩ.cm Conductive and conformal films achievable in 1:10 AR. Higher AR structures: limited by plasma radicals lifetime. 24

25 Outlines Introduction approaches for new ALD processes Interest of the combinatorial heteroleptic chemistry for ALD New classes of heteroleptic precursors: Cyclopentadienyl-amidinate precursors: examples of Ln (Y2O3), Co, Ni Allyl-amidinate precursors (Ni metal PEALD) Conclusions 25

26 Conclusions Novel heteroleptic chemistry with known ligands can provide drastic improvement of precursor properties and ALD performance. New classes of heteroleptic precursors released: Heteroleptic chemistry will continue to enable new ALD/PEALD processes and will remain key, for instance for selective ALD on dielectrics, New classes of more reactive ligand families are continuously searched, together with more reactive co-reactants with higher reducing potential. 26

27 Heteroleptic Chemistry ZrCp(NMe2)3 Example Defectivity improvement Better precursor stability can decrease adders & improve uniformity 300mm O2 PEALD comparison w/ TEMAZ Courtesy of Dr. M. Gros-Jean - ST Microelectronics TEMAZ 450Å ZyALD ZyALD 450Å NU: 2.7% Immediate reduction of particle contamination w/ ZyALD Growth rate: 1 Å/cycle - uniformity obtained ~2.7% 27