Enhanced HVPMOS

[0001] This application is a continuation of U.S. patent application Ser. No. 12/851,256, filed Aug. 5, 2010, U.S. Pat. No. 8,217,452, entitled “ENHANCED HVPMOS”, the contents of which are incorporated herein by reference in its entirety.

[0002] This subject matter is generally related to semiconductor devices, and more particularly to p-channel lateral double diffused metal-oxide-semiconductor (LDMOS) for high voltage applications.

[0003] P-channel LDMOS is widely used as high side power devices due to reduced gate drive circuitry. When combined with n-channel LDMOS, p-channel LDMOS can be employed in level shifters in many applications such as motor drivers or display panels. A good specific on-state resistance/breakdown voltage (Rdson/BV) trade-off of n-channel LDMOS is possible due to the Reduced SURface Field (RESURF) principal since the handle wafer is grounded and drain forward biased. However, for p-channel LDMOS, the RESURF principle is inhibited because the drain and handle wafer are commonly biased to the same potential.

[0004] Some conventional designs have been developed to overcome this issue. In both Bulk and thick film silicon-on-insulator (SOI) technology, the vertical depletion can also be assured with a source metallization or gate polysilicon which acts as a field plate. The presence of the n-type floating layer associated with the field plate defines a double-RESURF effect which leads to a competitive Rdson/BV tradeoff. In a thin film SOI substrate, however, the small active silicon area reduces the possibility to define an n-type floating region without degrading the device Rdson. Consequently, only the effect of the field plate is possible and the doping concentration of the drift region which sustains the voltage has to be lowered, leading to an increase of Rdson.

[0005] A p-channel LDMOS device with a controlled n-type buried layer (NBL) is disclosed. Shallow trench isolation (STI) oxidation is defined, partially or totally covering the drift region length. The NBL layer, which can be defined with a p-well mask, connects to the n-well diffusion, thus providing an evacuation path for electrons generated by impact ionization. High immunity to the Kirk effect is achieved, resulting in a significantly improved safe-operating-area (SOA). The addition of the NBL deep inside the drift region supports a space-charge depletion region which increases the RESURF effectiveness, thus improving BV. An optimum NBL implanted dose can be set to ensure fully compensated charge balance among n and p doping in the drift region (charge balance conditions). Since the drift depletion action is enhanced with the addition of the NBL layer, the p-well implanted dose can be further increased to maintain a charge balance, which leads to an Rdson reduction.

[0006] The p-channel LDMOS with additional NBL improves the BV versus Rdson trade-off in p-channel LDMOS transistors and consumes less surface area, resulting in lower cost transistors.

[0012] FIG. 1 is a cross sectional view showing the structure of a conventional p-channel LDMOS 100. LDMOS 100, which is an SOI LDMOS, includes substrate 122, buried oxide layer (BOX) 120, n-well region 102, p-well region 118, n-doped active region 106, p-doped active regions 104, 116, source 108, gate 110 and drain 114.

[0013] BOX 120 (e.g., SiO2) is formed in substrate 122 using, for example, a Separation by IMplantation of OXygen process (SIMOX). Substrate 122 can include a semiconductor material such as silicon (Si) or silicon-germanium (SiGe). In this example, substrate 122 is doped with n-type impurities.

[0014] P-channel LDMOS 100 is fabricated in n-type substrate 122. An n-doped layer is formed in substrate 122 as n-well region 102, and a p-doped layer is formed inside n-well region 102 forming drift region 118 (hereinafter also referred to as p-well region 118). N-well region 102 and drift region 118 can be formed in substrate 122 overlying BOX 120 using implantation and photoresist masks. Some examples of n-type impurities are Phosphorous and/or Arsenic. Likewise, drift region 118 can be implanted with a p-type impurity using a photoresist mask. Some examples of p-type impurities are Boron and/or Indium. Drift region 118 is laterally adjoining n-well region 102.

[0015] N-doped active region 106, which forms an n-well contact region, can be formed in n-well region 102 using implantation and a photoresist mask. N-well 102 is connected to source 108 terminal through n-doped active region 106 to avoid the activation of inherent parasitic bipolar transistor activation by short-circuiting n-well region 102.

[0016] P-doped active region 104, which forms a contact region for source 108, is formed in n-well region 102 using implantation and a photoresist mask. N-doped active region 106 and p-doped region 104 can be connected together at the same potential. P-doped active region 116, which forms a contact region for drain 114, can be formed in p-well region 118 using implantation and a photoresist mask.

[0017] Source 108 at least partially overlies n-doped active region 106 and p-doped active region 104. Gate 110 at least partially overlies p-doped active region 104 and n-well region 102. Gate 110 can include doped polysilicon disposed in one or more layers of dielectric material (e.g., silicon oxide) or other known conductive materials. Drain 114 at least partially overlies p-doped region 116. A shallow trench isolation (STI) oxidation 124 spaces drain 114 and gate 110 apart, so that a high drain-to-gate voltage can be applied. Electrodes of conductive material can be disposed on source 108, drain 114 and gate 110.

[0018] Gate 110 can be used to induce a field-enhanced depletion region between source 108 and drain 114, and thereby creating a “channel.” Channel current can be controlled by a vertical electric field induced by gate 110 and a lateral field that exists between source 108 and drain 114.

[0019] P-channel LDMOS 100 has a low optimal drift doping concentration which leads to a high Rdson, as compared to an n-channel LDMOS. To optimize the BV and improve the Rdson/BV trade-off of the p-channel LDMOS 100 structure, a new drift region design is needed that includes an NBL, as described in reference to FIG. 2.

[0020] FIG. 2 is a cross sectional view showing the structure of a p-channel LDMOS device 200, including an NBL 202. P-channel LDMOS 200 has the same structure as p-channel LDMOS 100 except for the inclusion of NBL 202.

[0021] In some implementations, NBL 202 can be included in substrate 122 using a high-energy Phosphorus and Boron multi-implantation sequence. NBL 202 can be underlying p-well region 104 and vertically adjoining BOX 120. STI 124 can be previously defined in substrate 122, partially or totally covering the length of drift region 118. STI 124 can be defined by the following process steps: stack deposition (oxide+protective nitride), lithography print, dry etch, trench fill with oxide, chemical-mechanical polishing of the oxide, removal of the protective nitride and adjusting the oxide height.

[0022] NBL 202 can be defined with a p-well mask, so that it connects to n-well diffusion, thus providing an evacuation path for electrons generated by impact ionization. NBL 202 provides high immunity to the well-known Kirk effect, thus significantly improving the safe-operating area (SOA). The addition of NBL 202 deep inside drift region 118 supports a space-charge depletion region which increases the RESURF effectiveness, thus improving BV. An optimal NBL 202 implanted dose can be set to ensure full depletion before breakdown, thus achieving good reliability conditions with a compensated charge balance among n and p doping in drift region 118 (charge balance conditions). Since the drift depletion action is enhanced with the addition of NBL 202, the p-well implanted dose can be further increased to maintain charge balance, which leads to Rdson reduction.

[0023] FIG. 3 is a flow diagram of a process 300 for fabricating the p-channel LDMOS of FIG. 2. In some implementations, process 300 can begin by preparing an SOI substrate including a BOX (302). The SOI substrate can be prepared using SIMOX, for example. STI is formed in the substrate partially or totally overlying the drift region (304). An n-well region is formed in the SOI substrate, so that the n-well region is vertically adjoining the BOX (306). A p-well region is formed in the n-well region to provide a drift region including an NBL disposed between the BOX and the drift region (308). The NBL is laterally connected to the n-well region to provide an evacuation path for electrons generated by impact ionization. An n-doped region is formed in the n-well region to form a contact region for a source (310). A p-doped region is formed in the drift region to form a contact region for a drain (312). A gate is formed in dielectric material disposed on the top surface of the SOI substrate so that the gate region at least partially overlies the n-well region and the drift region, defining a channel (314). In some implementations, electrodes formed from conductive material can be disposed on the source, gate and drain, respectively, using a metallization process.

[0024] FIG. 4 is a schematic diagram illustrating charge distribution along the Silicon/BOX/Silicon region in the proposed p-channel LDMOS structure. For example, in thin film SOI substrates with a Silicon active thickness (T_SOI) of 1.6 um, NBL 202 thickness (T_NBL) can be made as small as possible to not excessively reduce the drift current path which could penalize Rdson. As shown in FIG. 4, NBL 202 is depleted by the combined action of substrate field-effect action, and the p-well/NBL junction. Then, the optimal NBL implanted charge can be appropriately chosen to compensate both depletion effects. Hence, the total depletion process in NBL 202 can be defined by
W_NBL=W_N/P+W_ox/N,   [1]
where W_N/P is the depletion region in NBL 202 due to the p-well/NBL junction diode represented by

[0025] $\begin{array}{cc}{W}_{N/P}=\sqrt{\frac{2·{\varepsilon }_{\mathrm{si}}}{q·\mathrm{NBL}\left(1+\frac{N_{\mathrm{NBL}}}{N_{\mathrm{pwell}}}\right)}\left(V_D-V_{\mathrm{bi}}\right)}\approx \sqrt{\frac{{\varepsilon }_{\mathrm{si}}·V_D}{q·N_{\mathrm{NBL}}}},& \left[2\right]\end{array}$
where N_NBL and N_pwell are the constant doping concentration of the NBL and p-well layers, respectively, and W_ox/N is the depletion region in NBL 202 due to the SOI substrate field effect action, defined by

[0026] $\begin{array}{cc}{W}_{\mathrm{ox}/N}=\frac{{\varepsilon }_{\mathrm{si}}·{\varepsilon }_{\mathrm{ox}}·V_D}{q·N_{\mathrm{NBL}}·\left({\varepsilon }_{\mathrm{si}}·T_{\mathrm{BOX}}+{\varepsilon }_{\mathrm{ox}}·T_{\mathrm{NBL}}\right)},& \left[3\right]\end{array}$
where ε_si is the dielectric constant of Silicon, ε_ox is the dielectric constant of oxide, q is the electronic charge, V_bi is the built-in potential, V_D is the threshold voltage and T_BOX is the thickness of the BOX.

[0027] In optimal RESURF conditions, the vertical full depletion of the NBL region has to take place before the junction breakdown. The voltage capability of a planar junction in a lateral diode is given by:

[0028] $\begin{array}{cc}{V}_{\mathrm{BR}}^{\mathrm{Ld}}=\frac{{\varepsilon }_{\mathrm{si}}·E_c^2}{2·q·N_{\mathrm{NBL}}},& \left[4\right]\end{array}$
where ε_si is the dielectric constant of Silicon and E_c is the Silicon critical electric field (≅3×10⁵ V/cm). Then, if the planar breakdown expression [4] is introduced into expression [1], and the theoretical bound in 1D-RESURF theory is considered by setting N_NBL=N_pwell in expression [2], then the W_NBL can be simplified to the following expression:

[0029] $\begin{array}{cc}{W}_{\mathrm{NBL}}=\frac{2\times {10}^{12}}{N_{\mathrm{NBL}}}\left(\sqrt{\frac{1}{2}}+\frac{1\times {10}^{12}}{N_{\mathrm{NBL}}}·\frac{{\varepsilon }_{\mathrm{ox}}}{\left({\varepsilon }_{\mathrm{si}}·T_{\mathrm{BOX}}+{\varepsilon }_{\mathrm{ox}}·T_{\mathrm{NBL}}\right)}\right).& \left[5\right]\end{array}$

[0030] FIG. 5 is a graph of calculated constant doping concentration in the p-well and NBL layers as a function of p-well (T_pwell) and NBL (T_NBL) thickness. The theoretical drift resistances (R_LDD) can be extracted from

[0031] $\begin{array}{cc}{R}_{\mathrm{LDD}}=\frac{1}{q·N_{\mathrm{pwell}}·{\mu }_p}·\left(\frac{L_{\mathrm{LDD}}}{T_{\mathrm{Active}}·W}\right),& \left[6\right]\end{array}$
where L_LDD is the drift region and T_Active=T_SOI−(T_NBL−W_P/N) in the proposed p-channel LDMOS, and the depletion region in the p-well laver, W_P/N, expressed as:

[0032] $\begin{array}{cc}{W}_{P/N}=\sqrt{\frac{{\varepsilon }_{\mathrm{si}}·V_{\mathrm{bi}}}{q·N_{\mathrm{pwell}}}}.& \left[7\right]\end{array}$

[0033] While this document contains many specific implementation details, these should not be construed as limitations on the scope what may be claimed, but rather as descriptions of features that may be specific to particular embodiments. Certain features that are described in this specification in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable sub combination. Moreover, although features may be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a sub combination or variation of a sub combination.