Effective work function shift induced by TiN sacrificial metal gates as a function of their thickness and composition in 14nm NMOS devices C. Suarez-Segovia 1,2 , P. Caubet 1 , V. Joseph 1 , O. Gourhant 1 , G. Romano 1 , F. Domengie 1 , G. Ghibaudo 2 1 STMicroelectronics, 850 rue Jean Monnet, 38926, Crolles cedex, France. Phone Number: +33-7-88-22-38-64 Email : carlosaugusto.suarezsegovia@st.com 2 IMEP-LAHC, Minatec/INPG, BP 257, 38016 Grenoble, France Abstract For the first time, we investigated the effects of compo- sition and thickness of TiN on the effective work func- tion (EWF) shift induced by sacrificial gates in 14nm MOS devices with HfON-based dielectrics. No impact of sacrificial TiN composition was observed. Instead, WF was modulated of 35meV by tuning sacrificial TiN thickness. This shift is attributed to the presence of di- poles and/or fixed charges at the IL/HK interface. 1. Introduction The continuous scaling of MOS transistors makes it imperative to replace the conventional SiO 2 /Poly-Si with HKMG stack [1]. A typical high-k metal gate (HKMG) stack structure contains a silicon oxide based interfacial layer (IL), a high-k dielectric, followed by a final metal gate electrode. Generally, band edge metal gates are needed to obtain low threshold voltage (Vth). However, measure- ments performed on the processed gate stacks show unde- sirable flat band voltage (Vfb) shifts. This can be attributed to one or several potential drops within the gate stack. As already reported by Jha et al [2], the voltage drops in the insulator at the flat band condition has two origins. The first one is related to the fixed charges, which induces voltage drops increasing linearly with the distance from the gate. The second origin is related to the dipoles at the Si/IL, IL/HK and HK/Metal interfaces. Indeed, according to Iwamoto, Vfb shift in HKMG stacks is mainly determined by the dipole layer at the high-k/SiO 2 interface [3]. Compo- sitional and thickness tuning of the TiN final metal gate film and inclusion of capping layers have recently been studied in order to achieve the desired effective work func- tion in gate-first integration. Nevertheless, HKMG stack engineering can lead to etching issues and serious chal- lenges for gate patterning [4]. To solve this problem, sacri- ficial gate technique has been recently introduced in ad- vanced 14nm CMOS gate stacks. Thus, the aim of this work is to determine the effective work function shift in- duced by sacrificial TiN (sac-TiN) metal gates. 2. Extraction of effective work function shift induced by sacrificial metal gates The CV analysis of MOS devices is widely used for its ability to determine the equivalent oxide thickness (EOT) and the flat band voltage (Vfb) by fitting experimental measurements with quantum simulations [5]. According to Eqs. (1) and (2), EWF depends on metal vacuum work function (WFm) but also on interfacial drops (V) and on the fixed charges at the IL/HK and Si/IL interfaces. Vfb = EWF - Si (1) EWF = WFm + V – EOT HK (Q SiOx/HK ) – EOT(Q Si/SiOx ) (2) The shift in EWF induced by fixed charges depends on the EOT. Therefore, we use the IL-bevel method in order to get rid of this shift. The extrapolation of the curve EWF (EOT) to EOT=0 leads to a typical effective work function (EWF 0 ), which is characteristic of each HKMG stack without the effect of fixed charges at the IL/Si interface (Q Si/SiOx ). Charbonnier demonstrated that the y-intercept is invariable irrespective of the HfO 2 thickness [6]. We can then neglect the contribution of the fixed charges at the IL/HK interface (Q SiOx/HK ) to the EWF 0 . In order to extract the EWF shift induced by sacrificial metal gates, we will apply the IL-bevel method on two types of samples. In the first one, the electrode consists only of the final TiN metal that will serve as a reference. In the other one, we deposit the sacrificial metal gate to be studied, followed by a ther- mal treatment under N 2 atmosphere at 900 ºC. The purpose of the stage anneal is to activate the diffusion of elements tuning the work function into the high-k. Next, the metal gate is removed by wet etching, and we deposit the same final TiN metal used as reference. Fig. 1: a) EWF shift induced by sacrificial gates, b) Sample with only Final TiN, c) Sample with also sacrificial TiN metal gates. In this way, the EWF 0 obtained with IL-bevel method for samples b) (only final TiN) and samples c) (with sac-TiN) can be expressed as written in Eqs. (3) and (4), respectively. The workfunction shift induced by the sacrificial TiN is therefore determinated using Eq. (5). EWF sample 1 = WF Final + Final (3) Extended Abstracts of the 2014 International Conference on Solid State Devices and Materials, Tsukuba, 2014, - 838 - J-2-3 pp838-839