Semiconductor Structure with N-type Region Codoped with Group I or II Elements

Background of the Invention.

The various embodiments of the present invention relate generally to semiconductor structures . More particularly, the various embodiments relate to a semi- conductor structure having a highly n-doped region, and to a method of fabricating such a semiconductor structure.

Advances in the miniaturization of semicon ductor structures follow Moore's law according to which the performance and density of transistors double every three years. To comply with Moore's law, in a field effect transistor (FET) the distance between source and drain must shrink by a factor of ^2 in each generation, thereby doubling the number cf transistors per chip. IIow- ever, transistor dimensions cannot be scaled down without facing technological hurdles.

One example of a semiconductor structure is a silicon (Si) -based metal oxide semiconductor FET (MOSFET) having a source, a drain, a channel, and a gate region. The source region provides a supply of mobile charge carriers enabling current to flow from the source to the drain when the transistor is turned on. The source and drain regions are electrically isolated from one another by an oppositely charged channel region. A controlling gate electrode is separated from the channel by an insulating oxide material. By applying a voltage across the insulating material, an electric field is created. If the applied voltage repels the channel charge and attracts the source and drain charge, a conducting layer is formed, and a current can flow from the source to the drain.

By way of background, such a semiconductor structure is produced using one or more doping technologies, such as diffusion and ion implantation. Doping refers to the process of intentionally introducing impuri- ties (i.e., dopants) into an intrinsic semiconductor, such as Si, in order to change its electrical properties. For Si, the most common dopants are elements from group III or group V, wherein the group number refers to the Roman numeral of the columns in the periodic table. To dope Si, Arsenic (As), boron (B), gallium (Ga) and p ^' hos- phorus (P) can be used. By doping Si with a group V element, e.g. P, extra electrons are added, which become unbonded from individual atoms and allow the compound to be electrically conductive. This results in an n-type or n- doped material, wherein the group V element becomes a donor. Doping Si with a group III element, e.g., B, creates holes in the silicon lattice that are free to move. This is an electrically conductive, p-type or p-doped material, and the group III element becomes an acceptor. Scaling down the dimensions of the above

MOSFET can be achieved, among other measures, by increasing the density of dopants, for example, in the source and drain regions. However, increasing the density. of dopants negatively influences the performance of the transistor due to negative effects, such as dopant deactivation and dopant diffusion, which distorts the abrupt and shallow doping profile required for ultra short gate lengths .

Mϋller, D.C. and Fichtner, W. describe in "Highly n-doped silicon: Deactivating defects of donors", Physical Review B 70, 245207 (2004), that doping Si with donors from the group V, such as P, As and Sb,- with active concentrations well above the respective solid solubility limit deactivates donors upon subsequent thermal processing. This deactivation can be ascribed mainly to the formation of so-called donor-vacancy (D _n V) clusters. Dopants in such clusters have been deactivated and as

such no longer contribute to the concentration of the charge carriers .

US Patent No. 6,037,640 relates to the fabrication of ultra-shallow junctions in FET devices and de- scribes that the prior art used carbon co-implants or fluorine co-implants to reduce diffusion of dopants in silicon during thermal processing. US Patent No. 6,037,640 proposes to implant a first ion, such as Si, germanium (Ge) , As, indium (In) , Ga, into the surface of a Si substrate, and then a dopant ion, such as As, P or Sb.

Summary of Certain Inventive Aspects

The main defects involved in both the diffusion and deactivation of donors, as described above, are the D _n V clusters. The vacancies act as electron sinks in highly n-doped silicon. It is believed that they form at high Fermi level positions and in the vicinity of one or more donor atoms, and that they mediate, for example, the As and Sb diffusion. Hence, it is a general aspect of the various embodiments described herein to provide a highly n-doped semiconductor structure with a shallow doping profile and a method of fabricating such a semiconductor structure with reduced donor vacancies .

Accordingly, one aspect involves a semiconductor structure with a region that is doped with a first dopant material and codoped with at least one second dopant material. As described herein, such a co-doping strategy thwarts the formation of donor-compensating complexes during post-implantation thermal processing. The codopant, i.e., a third atomic species like an alkali metal or an earth alkaline metal, is introduced into a highly n-doped region of, for example, a silicon wafer. An effective codopant material should either bind strongly to vacancies or to the donor atoms in order to forestall the donor-vacancy clustering process. Moreover,

it should not introduce additional acceptor states into the crystal of the semiconductor structure, neither as isolated impurity nor when forming complexes with other lattice imperfections. Accordingly, one aspect involves a semiconductor structure having a semiconductor substrate and a first region on the semiconductor substrate. The first region is defined by a first dopant material of an n-type conductivity. The first region includes at least one sec- ond dopant material selected from the group consisting of lithium, sodium, beryllium, magnesium, and calcium. Hence, the first region of such a semiconductor structure is not only doped with the first dopant, for example, P, As, or Sb, but codoped with one or more of the listed dopant materials .

In one embodiment, the fist region includes the first dopant material at a first concentration and the second dopant material at a second concentration, wherein the second concentration is about 10% to about 100% of the first concentration.

Another aspect involves a method of fabricating a semiconductor structure that includes defining a region on a semiconductor substrate or layer. A first dopant material of an n-type conductivity is introduced into the region. Further, at least a second dopant material selected from the group consisting of lithium, sodium, beryllium, magnesium, and calcium is introduced into the region. Subsequent temperature processing, e.g., through rapid thermal annealing, allows the atoms to re- arrange and the codopants to associate with vacancies and donor-vacancy clusters.

According to another aspect, the co-doping strategy is implemented in a field effect transistor. The source and drain regions are of a first conductive type, and the gate region is of a second conductive type. The source region, the drain region and the gate region are formed in a semiconductor substrate, wherein one of the

first and second conductive types is an n-type conductivity. The n-type conductivity is defined by a first dopant material introduced into the semiconductor substrate to provide for the n-type conductivity in the semiconductor substrate. At least one second dopant material selected from the group consisting of lithium, sodium, beryllium, magnesium, and calcium is added to the first dopant material.

Brief Description of the Drawings

These and other aspects, advantages, and novel features of the invention will become apparent upon reading the following detailed description and upon reference to the accompanying drawings. In the drawings, ^' same elements have the same reference numerals.

Figure 1 is a schematic illustration of one embodiment of a field effect transistor having a codoped region,

Figure 2 is an illustration depicting the behavior of various group I and II elements in silicon, and

Figure 3 is an illustration depicting energy gains when a codopant assumes a vacant lattice position, as a function of adjacent arsenic atoms.

Detailed Description of Certain Inventive Embodiments

The certain inventive embodiments are hereinafter described with reference to a metal oxide semiconductor field effect transistor. However, it is contemplated that a MOSFET, regardless of whether implemented as an NMOS or a PMOS transistor, is merely one example of a semiconductor structure. Other semiconductor structures, such as bipolar transistors or memory devices

(DRAM, SRAM, and SDRAM) may equally be implemented using the technology described herein with reference to a MOSFET.

Further, the technology described with refer- ence to a MOSFET is not limited to creating pnp, npn or pn junctions having highly n-doped regions. In principle, the technology may be used to create any highly n-doped region on a semiconductor substrate or layer, where the above described D _n V clusters may occur. Figure 1 shows an exemplary structure of ^" one embodiment of a MOSFET ₇ more particularly an NMOS transistor, implemented on a substrate 2. The substrate 2 is in one embodiment a Si substrate and hereinafter, without limitation, referred to as Si substrate 2. In other em- bodiments, the substrate 2 may be a SiGe substrate or a strained Si substrate. The MOSFET has a structure including a channel region 4, a drain region 6, a source region 8, a gate 12, and an oxide layer 10. In the illustrated NMOS transistor, the drain and source regions 6, δ are highly n-doped regions formed by doping the Si substrate 2 with a group V element, such as P, As, Sb. The channel region 4 is a p-doped region formed by doping the Si substrate 2 with a group III element, such as B or Al . The drain and source regions 6, 8 are coupled to an electri- cal contact E^, Eg, and the gate 12 is coupled to an electrical contact E _Q . The basic processes and methods of fabricating and operating such a MOSFET structure are known to one of ordinary skill in the art. Emphasis is therefore given to the new and inventive features of the technology described herein with reference to the MOSFET illustrated in Figure 1.

Further, Figure 1 illustrates the highly υ- doped drain and source regions 6, 8 as abutting to the channel region 4. The gate 12 and the oxide layer 10 par- tially overlap the drain and source regions 6,8. In the illustrated embodiment, the drain and source regions 6, 8 extend into the Si-substrate 2 and have a depth of about

8 - 45 run. As is known in the art, the depth is defined as the depth where the concentration of the dopant falls under about 10 ¹⁸ cm ^"3 .

In addition to being highly doped with a group V element, within the drain and source regions 6, 8 the Si-substrate 2 is further doped with one or more elements selected from group I or group II of the periodic table, i.e., alkali metals or earth alkaline metals. More particularly, the drain and source regions 6, 8 may be co-doped with lithium (Li) , sodium (Na) , beryllium (Be) , magnesium (Mg) or calcium (Ca) , or a combination of these elements. For illustrative purposes, the element Mg is shown within the n-type drain and source regions 6, 8. It has been found that, for example, the light group II metals Mg and Be are very well suited for disposing of the abundant vacancies present in highly n- doped Si, for example, in the drain and source regions 6, 8. These metals diffuse in Si through interstitial lattice positions and show a strong tendency to get trapped at vacancy sites of donor-vacancy clusters. Thereby, they convert the acceptor-like D _n V clusters into electron donors .

Consequently, the degree of donor deactivation is reduced drastically. The concentration of conduc- tion electrons in the corresponding area, for example, the drain and source regions 6, 8, increases accordingly. In addition, junctions formed by codoping P, As, or Sb, for example, with Be and/or Mg will show less donor diffusion at annealing temperatures and will therefore re- tain their desired shallow character and abrupt profile.

In order to achieve highly n-doped Si with minimal deactivation and diffusion, the respective area, for example, the drain and source regions 6, 8, is in one embodiment doped with one of a group V donor (P, As, Sb) and a codopant selected from the group of Li, Na, Be, Mg and Ca. The concentration of the codopant is in one embodiment almost as abundant as the dopant, i.e. the con-

centration of the codopant is about the same order of magnitude as the concentration of the group V dopant. As understood by one of ordinary skill in the art, the term "same order" means a certain value +/- a factor of 10. In one embodiment, the concentration of the codopant is between about 10% and about 100% of the concentration of the group V dopant.

In one embodiment, the concentration of the donor may be about 6 x 10^° cm ^~ 3 _{f an} ci the concentration of the codopant may be about 1 x 10^0 cm ^"3 . In another embodiment, the concentration of the donor may be about 1 x 10^1 cm~3 _{f anc} j the concentration of the codopant may be about 2 x 10 ²⁰ cm ^"3 .

In general, the donors P, As, Sb may be co- doped with Li, Na, Be, Mg, Ca at concentrations of the above-mentioned range, i.e., between about 10% and 100% of the donor concentration. In one embodiment with As or P as dopants, the concentration of the co-dopant is, for example, between about 30% and about 40% of the dopant concentration. In another embodiment with P as dopant, the codopant may be Na, Ca or Mg at a concentration of about 50% of the concentration of the P dopant. In yet another embodiment, As may be codoped with Li, Be or Mg at a codopant concentration of about 30% to about 40%. In an embodiment where Sb is the dopant, the codopant may be Li, Be cr Mg at a codopant concentration of above about 40%, for example, at about 50%.

Figure 2 is an illustration depicting the energy behavior (in eV) of the above-mentioned group I and II elements in Si as a function of the number of adjacent arsenic atoms. For comparative reasons, the illustration shows further the energy behavior of Si, Ge and carbon (C) . As discussed above, the mentioned codopants Li, Na, Be, Mg, Ca have the ability to neutralize acceptor levels of D _n V clusters. Further, these codopants have the tendency to be trapped at lattice vacancies.

The hard sphere radius that can be accommodated in hexagonal <111> tunnels in Si without an appre ^¬ ciable lattice dilatation is about 1.05A. This criterion for a high diffusivity in bulk Si is fulfilled in all of the codopants Li, Na, Be, Mg, Ca. For example, a Li ion has a radius of about 0.76A, a Na ion has a radius of about 1.02A, a Be ion has a radius of about 0.45A, a Mg ion has a radius of about 0.72A, and a Ca ion has a radius of about 1.00A. As shown in Figure 2, interstitial Li and Na exhibit relatively low vacancy trapping energies of -1.08 eV and -1.74 eV, respectively. These elements move along the <111> channels in a zigzag fashion, alternatively- passing through the tetrahedral and hexagonal intersti- tial sites of the Si lattice, which are their equilibrium and saddle points for diffusion, respectively.

Such a high diffusivity of a possible co- dopant may not be desirable in the manufacturing process of shallow junctions. However, the presence of 10- ¹ - cm ^" -^ carbon reduces the room temperature diffusion rate of Li to 0.1% of its value in pure Si. It is therefore very likely that in heavily n-doped samples, all of the metals discussed exhibit a dramatically reduced diffusivity, due to their tendency to bind to donors and donor-vacancy clusters.

The energy gain of this trapping process from the interstitial to the substitutional position is small for the alkali metals Li and Na because these light metals cause minimal lattice distortion when interstitial. Interstitial C, Si, and Ge, on the other hand, evoke large lattice relaxations, such that the substitutional configuration is energetically favored for any number of neighboring As atom, by more than 1 eV. The trapping energies of the earth alkaline metals Ca, Mg, and Be are smaller than those of the group IV elements in undoped Si (0 As on the x-axis), but at high As-doping, an interesting feature becomes apparent: The ionized interstitial

earth alkalines, though small, show an even stronger tendency to annihilate vacancies in the AS3-V and AS4-V complexes than the large Si interstitial. Especially Mg and Be are therefore major codopant candidates for the pre- elusion of the AS3-V cluster formation. In addition, the trapping energies of Mg range from -3.65 eV to -3.10 eV for 0 - 3 As atoms next to the vacancy.

Figure 3 is an exemplary illustration depicting energy gains when a codopant assumes a vacant lattice position as a function of the number of adjacent As atoms. More particularly, the energy is gained when a substitutional codopant assumes a formerly vacant lattice position next to 0 - 4 As atoms. For an isolated vacancy (0 As on the x-axis) , this energy corresponds to the negative vacancy formation energy in Si and is independent of the codopant species. However, when one or more donors are adjacent to the vacancy, then the discrepancy between group IV elements and the group I or II elements (Li, Na, Be, Mg, Ca) described herein becomes apparent. A dashed line for the Si atom illustrates the fact that both AS ₃ -V and AS4-V are thermodynamically stable complexes: The Asβ-Si and As4~Si complexes are energetically 0.78 eV and 2.51 eV higher than configurations with a central vacancy, respectively. This may not be the case for AS3 4-codopant complexes if the codopant is an alkali or earth alkaline metal. Hence, in the absence of a light metal codopant, the AS3-SX and As4-Si complexes will be formed given enough time and thermal energy to allow the system evolve towards the thermodynamic equilibrium. In the presence of light alkali or earth alkaline codopants, the latter can associate with the complexes with a resulting energy gain of -4.30 eV to -1.84 eV, whereupon the electrical activation of the donors involved is partially restored, as ex- plained above.

The above-described semiconductor structure can be fabricated using known semiconductor processing

technologies. For example, the drain and source regions 6, 8 on the semiconductor substrate 2 can be defined through a lithographic process. Once these regions 6, 8 are defined the first and second dopant materials P, As, Sb and Li ₇ Na, Be, Mg, Ca, respectively, can be introduced into the regions 6, 8.

The introduction of both dopant and codopant materials can be achieved either by an as-grown process, ion implantation, molecular beam epitaxy (MBE) , or chemi- cal vapor deposition (CVD) . Subsequent temperature processing allows the atoms to rearrange and the codopants to associate with vacancies and donor-vacancy clusters. The temperature processing may be performed using known annealing technologies, such as annealing using a flash lamp or a laser, rapid thermal annealing, furnace annealing, millisecond annealing, spike annealing and SPEG (solid phase epitaxial growth) annealing.

In addition to the above-described embodiments of the semiconductor structure and method, a skilled artisan will recognize that the semiconductor and method may advantageously include any, some, or all of the features and aspects discussed in the foregoing description with reference to Figures 1 to 3. Additionally, other combinations, omissions, substitutions and modifi- cations will be apparent to the skilled artisan in view of the disclosure herein. Accordingly, the present invention is not intended to be limited by the reaction of the described embodiments, but is to be defined by reference to the appended claims .