 |
Description  |
|
|
BACKGROUND AND SUMMARY OF THE INVENTION
This invention relates generally to thin film transistor (TFT) processes and fabrication, and more particularly, to a polycrystalline film, and method of forming the polycrystalline film, from a microcrystalline film.
Low-temperature polysilicon formation is a process that has been intensively investigated for more than a decade now. These investigations have become more important due to the potential of this material in flat panel display related technology,
particularly, in the area of active matrix liquid crystal displays (AMLCDs). Polysilicon TFTs offer the advantages of: (a) smaller device dimensions, thus higher aperture ratio for the display; (b) higher TFT "on" current, thus less sensitivity to gate
and bus line signal delays; and, (c) elimination of external drivers and interconnects, thus possible yield and cost improvement (especially in small size displays).
Earlier efforts focused on the formation of polysilicon by SPC methods (i.e. low temperature furnace anneal). Recently, significant emphasis has been placed on excimer laser anneal as a more suitable process for the formation of high quality
polysilicon material from an amorphous-silicon precursor film. ELA technology has significantly matured over the past decade. A number of commercial tools are currently available to flat panel display manufacturers. However, problems still plague this
process. Control of grain size is difficult, especially under operation close to the super-lateral-growth (SLG) regime, where small variations in laser energy density result in large variations in grain size. Uniformity of grain size is another major
issue, especially in the overlap region between successive laser beam shots. Moreover, damage of the underlying substrate is another problem, typically encountered as the energy density of the laser increases to accommodate a larger grain size in the
annealed polysilicon layer.
In light of these problems, it is desirable to develop a process that can compensate for some of the intrinsic shortcomings of ELA. More specifically, it is desired to obtain high quality polysilicon at the minimum laser energy "load", to avoid
hardware instabilities and substrate damage. This issue becomes even more important as technology moves to cheaper substrates, such as soda-lime glasses or organic substrates. Furthermore, it is desirable to reduce the variability of the polysilicon
material characteristics by appropriate engineering of the structure of the silicon film precursor. The method of depositing amorphous silicon on the transparent substrate is also crucial in the fabrication of polycrystalline films having large crystal
grains. Previous work in the area of solid phase crystallization of silicon has indicated that certain control of the structural characteristics of polysilicon is possible through materials engineering of the phase of the as-deposited film. Recent
evidence has found that the starting phase of the film affects the crystalline characteristics of the post-ELA (excimer laser annealed) polysilicon.
Typically, an LCD is made by mounting a transparent substrate on a heated susceptor. The transparent substrate is exposed to gases which include elements of silicon and hydrogen. The gases decompose to leave solid phased silicon on the
substrate. In a plasma-enhanced chemical vapor deposition (PECVD) system, the decomposition of source gases is assisted with the use of radio frequency (RF) energy. A low-pressure (LPCVD), or ultra-high vacuum (UHV-CVD), system pyrolytically decomposes
the source gases at low pressures. In a photo-CVD system the decomposition of source gases is assisted with photon energy. In a high-density plasma CVD system high-density plasma sources, such as inductively coupled plasma and helicon sources are used. In a hot wire CVD system the production of activated hydrogen atoms leads to the decomposition of the source gases.
It would be advantageous if a high quality polycrystalline film could be obtained by excimer laser crystallization, at low levels of laser energy density.
It would be advantageous if crystallization process steps could be reduced by optimization of the deposition and annealing procedures. It would also be advantageous if good crystal structural characteristics could be obtained with a low
variability against disturbance factors (i.e. variation in laser energy density and/or other process parameters).
It would be advantageous if a polycrystalline film could be fabricated on a transparent substrate having an electron mobility of 150 cm.sup.2 /Vs, or more.
Accordingly, a method of forming a polycrystalline film having high electron mobility and low threshold voltage is provided. The method comprises the steps of:
a) depositing a microcrystallite film having a microcrystallite density and a microcrystallite size;
b) annealing the microcrystallite film; and
c) in response to the microcrystallite density and size and the annealing, forming a polycrystalline film having a polycrystalline grain size and a polycrystalline grain size uniformity. The embedded microcrystallite seed crystals surviving
annealment form nucleation sites in the polycrystalline film. That is, the crystal grains are directly responsive to quantifiable variables such as microcrystallite size, density, and annealing energies.
Specifically, the step of annealing involving heating the microcrystalline film, melting the amorphous matter, selectively melting microcrystallites, and cooling the microcrystalline film. A second density of unmelted microcrystallites are left
embedded in the molten amorphous matter. Then, Step c) includes crystallizing the amorphous matter melted in Step b), using the unmelted microcrystallites as nucleation sites. The polycrystalline film grain size is responsive to the size and density of
unmelted microcrystallites.
Generally, the microcrystalline is formed on a first region adjacent, and overlying the transparent substrate. Step b) includes selectively melting the microcrystalline film so that the second density of microcrystallites is primarily in the
first region adjacent the transparent substrate.
Step a) includes forming a first microcrystallite crystalline fraction which is a product of the first microcrystallite size and the first microcrystallite density, and in which the first crystalline fraction in the range of approximately 0.01 to
80%. In some aspects of the invention, where larger sized microcrystallites are used, the pre-melt (as-deposited) crystalline fraction is in the range 0.01 to 25%. The typical as-deposited microcrystallite size is between 150 to 300 .ANG., although
microcrystallites as large as approximately 1000 .ANG. are useful.
Step b) includes forming microcrystallites having a post-melt microcrystallite size and post-melt microcrystallite density. The post-melt size and density is the result of the annihilation of microcrystallites smaller than a microcrystallite
critical size and the partial melting of microcrystallites having a pre-melt size larger than the critical size. Then, Step b) includes forming a post-melt microcrystalline film.
Because the melting process annihilates microcrystallites smaller than the critical size, the percentage of microcrystallites having a size less than the average, post-melt size, is smaller than the percentage of smaller than average sized
pre-melt microcrystallites. Thus, the post-melt microcrystallites are more uniform in size. The more uniformly sized post-melt microcrystallites result better control of the post-melt microcrystallite density. The polycrystalline grains are more
directly responsive to the post-melt crystalline fraction than most of the other process variables. A post-melt density of at least approximately 1.times.10.sup.8 microcrystallites per cm.sup.2 has been found to be effective, with approximately a 1
micron average separation between microcrystallites.
Preferably, Step b) includes uses an excimer laser crystallization (ELC) process, having a wavelength of approximately 308 nm, or less, to melt the amorphous matter and selectively melt the microcrystallites. The post-melt crystalline fraction
is responsive to the ELC energy density as well as the pre-melt crystalline fraction.
In one aspect of the invention, Step a) includes depositing the microcrystalline film by a PECVD process using a SiH.sub.4 and H.sub.2 gas mixture. The deposition process includes using a power level of approximately 0.16783 W per cm.sup.2, at a
temperature of approximately 320.degree. C., a total pressure of approximately 1.2 Torr, and a flow rate of SiH.sub.4 to H.sub.2 of approximately 100:1.
FIG. 1a shows deposition rate v. the percentage of crystalline fraction. The deposition rate is critical to the pre-melt crystalline fraction. When the microcrystalline film deposition rate is less than 2 .ANG. per second (.ANG./s), a
crystalline fraction as high as 80% can be formed. Higher deposition rates can be used to form a crystalline fraction in the range between 0.01% and 50%. Preferably, the microcrystalline film is deposited with the PECVD method at a deposition rate of
less than 10 .ANG./s and a deposition temperature of approximately 380 degrees C.
A liquid crystal display (LCD) is also provided comprising a transparent substrate and a TFT polycrystalline semiconductor film, overlying the transparent substrate, which has an electron mobility of greater than 150 cm.sup.2 /Vs, a threshold
voltage less than 2 volts, a grain size larger than 0.5 microns, and grain size uniformity of less than 10%. The polycrystalline film is formed from the above-described method. Namely, depositing amorphous matter in deposition conditions which result
in a microcrystalline film having a pre-melt crystalline fraction; annealing the microcrystalline film, with the annealing process selectively melting of microcrystallites to from a post-melt crystalline fraction; and, forming the polycrystalline TFT
film with an electron mobility responsive to the post-melt crystalline fraction.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1a shows deposition rate v. the percentage of crystalline fraction.
FIG. 1b illustrates the concept of grain size uniformity.
FIG. 2 is a flowchart illustrating the present invention method for forming a polycrystalline film having high electron mobility and a low threshold voltage.
FIG. 3 illustrates the relationship between SiH.sub.4 flow rate and amorphous film deposition rate.
FIG. 4 summarizes the average grain size of polysilicon as a function of the laser energy density and the crystalline content of the film.
FIG. 5 illustrates the placement of microcrystallites in amorphous matter, in a single dimension.
FIG. 6 illustrates the relationship between microcrystallite size and deposition rate, of silicon films deposited by PECVD.
FIGS. 7a-7c illustrate examples of the relationship between post-melt density and pre-melt crystallite size.
FIG. 8 illustrates the relationship between pre-melt and post-melt density as a function of the applied laser energy density.
FIGS. 9 and 10 illustrate the relationship between crystalline fraction, energy density, and either electron mobility or threshold voltage.
FIG. 11 illustrates the relationship between microcrystallite size and microcrystallite density in PECVD and LPCVD silicon films.
FIG. 12 illustrates the results of the deposition of double-layers.
FIGS. 13-16 illustrate steps in a method of forming an LCD.
FIG. 17 depicts the LCD of FIG. 13, further comprising a second, amorphous matter film overlying the microcrystalline film.
FIG. 18 depicts the LCD of FIG. 17 where the second film is specifically a microcrystalline film.
FIG. 19 depicts the LCD of FIG. 18 following the melting phase of annealment.
DETAIL DESCRIPTION OF THE PREFERRED EMBODIMENT
In the present invention, excimer laser anneal (ELA) processes are described for the phase transformation of amorphous to polycrystalline silicon (polysilicon). Polysilicon films, obtained from laser annealed microcrystalline silicon,
demonstrate a larger grain size than that of laser annealed amorphous silicon films. This grain size enhancement is attributed to the embedded nucleation seeds in the microcrystalline film that survive the melting process during laser annealing. Proper
engineering of the as-deposited film structure, via optimization of deposition conditions, yields at least a twofold increase in the grain size of the post ELA polysilicon films.
The measurement of grain attributes is considered with respect to digitized versions of TEM micrographs. Some grain attributes are: grain area, grain perimeter, grain major axis, grain minor axis, and grain shape factor. The last attribute is a
measure of how circular the grain is and is calculated as:
S.sub.f =4.pi..multidot.(area)/(perimeter).sup.2 (1)
According to this definition, a perfect circle has a shape factor of 1.0 and a line has a shape factor that approaches zero. The average shape factor of the grains of the laser annealed polysilicon films of the present invention are in the range
of 0.72-0.74, indicating quite regularly shaped (equiaxial) grains. An equivalent grain size is defined to be based on the measured major and minor axes. The equivalent grain size is given by the geometric mean of the major and minor axes as:
where r is the equivalent grain size, a is the major axis of the grain, and b is the minor axis of the grain.
FIG. 1b illustrates the concept of grain size uniformity. The uniformity of the grain size is defined as the ratio of the standard deviation (.sigma.) to the average grain size (.mu.). FIG. 1b illustrates typical grain size distributions
obtained from polycrystalline films resulting from excimer laser crystallization of amorphous silicon (dotted line curve) and microcrystalline (solid line curve) film precursors. The statistics of mean (.mu.) and standard deviation (.sigma.) for each
distribution are displayed. The uniformity of grain size is calculated by dividing the standard deviation by average. To express as a percentage, the above-calculated uniformity is multiplied by 100. The microcrystalline precursor of FIG. 1b has a
uniformity of 9.8%, while the amorphous silicon precursor has a uniformity of 31.5%.
FIG. 2 is a flowchart illustrating the present invention method for forming a polycrystalline film having high electron mobility and a low threshold voltage. Step 10 provides amorphous matter. Step 12 deposits amorphous matter embedded with
microcrystallites having a first predetermined microcrystallite density and a first predetermined microcrystallite size. Typically, the amorphous matter and microcrystallites deposited in Step 12 are silicon. Alternately, the amorphous matter and
microcrystallites deposited in Step 12 are a silicon-germanium compound. In this manner, a microcrystalline film is formed. Typically, Step 12 includes depositing a microcrystalline film having a thickness of less than approximately 1000 .ANG., whereby
the polycrystalline film is well suited to the manufacture of thin film transistors. The preferred thickness is less than approximately 400 .ANG..
Step 14 anneals the microcrystallite film deposited in Step 12. Step 16, in response to the microcrystallite first density and microcrystallite first size formed in Step 12, and the annealing performed in Step 14, forms a polycrystalline film
having a first polycrystalline grain size of greater than 0.5 microns and a first polycrystalline grain size uniformity of less than 10%. Step 18 is a product where the embedded microcrystallite seed crystals surviving annealment form nucleation sites
in the polycrystalline film.
Preferably, a transparent substrate selected from the group consisting of glass, quartz, and plastic is provided and Step 12 includes depositing the microcrystalline film on the transparent substrate. In this manner, a polycrystalline film is
formed suitable for a TFT on an LCD.
Specifically, Step 14 includes sub-steps. Step 14a heats the microcrystalline film deposited in Step 12 to a critical temperature. At a critical temperature the amorphous matter and microcrystallites begin to melt. The critical temperature for
amorphous matter is lower than it is for microcrystallites (single crystal material). Further, as is discussed below, the melting rates of amorphous matter and microcrystallites are different. Therefore, the calculus of melting is complex, involving
material, temperature, and melting rates.
Step 14b is the melting phase of annealing, which is responsive to reaching the critical temperature in Step 14a and the first microcrystallite size deposited in Step 12. Step 14b melts the amorphous matter and selectively melts
microcrystallites, leaving unmelted microcrystallites embedded in the molten amorphous matter. Typically, the matter continues to melt as long as the matter is above the critical temperature. Step 14c is the cooling phase of the annealing process,
where the matter made molten in Step 14b is permitted to cool to the critical temperature. When a laser is used, the laser shot typically occurs during Step 14a, the energy imparted into the microcrystalline film melts the microcrystalline film in Step
14b. Eventually, the microcrystalline film cools to the critical temperature. In a single region of microcrystalline film, Step 16 follows Step 14c. However, if the film as a whole is considered, the timing of Steps 14c and 16 can be coincident, or
Step 16 may even precede Step 14c, as the film has different temperatures and temperature gradients in different film regions.
Regardless, Step 16 includes crystallizing the amorphous matter melted in Step 14b using the unmelted microcrystallites as nucleation sites, whereby the polycrystalline film grain size is responsive to the size and density of unmelted
microcrystallites. Step 16 is distinct from Step 14c in that matter does not necessarily form a crystal structure at the change of temperature. The matter is crystallized along a lateral growth velocity, discussed in more detail below. Specifically,
the present invention method permits a polycrystalline product in Step 18 with a first grain size of at least approximately 1 micron, a first uniformity, or difference of grain sizes, of less than approximately 10%, a first electron mobility of greater
than approximately 150 cm.sup.2 /Vs.
FIG. 3 illustrates the relationship between SiH.sub.4 flow rate and amorphous film deposition rate. In one aspect of the present invention, silicon films are deposited by PECVD over a wide range of conditions. The deposition rate of the film is
a critical factor in determining the microstructure of the deposited layer. The key factors affecting the rate are the flows of reacting gases (SiH.sub.4 and H.sub.2), and the process pressure. The plasma power has a minor effect (within the range of
0.15 W/cm.sup.2 to 0.30 W/cm.sup.2). There is little sensitivity to the deposition temperature (in the range of 320-390.degree. C.).
During plasma decomposition of silane a complicated network of chemical reactions typically occurs. However, the net reaction can be expressed by:
The forward reaction represents film deposition, while the reverse reaction film "etching". The exact rate of growth of the film depends upon the balance between these two reactions. Under standard deposition conditions, the reaction is far
from its equilibrium point, leading to film deposition. It should be noted, however, that even in the case of undiluted source gas, the reverse ("etching") reaction does occur. Increasing silane flow rate and/or the total pressure results in increasing
the rate of the forward reaction (deposition rate increases). Adding hydrogen to the source gas, pushes the reaction described in equation (3) in the reverse direction. Thus, hydrogen effectively decreases the deposition rate of the film by promoting
film erosion. This "erosion" mechanism results in removal of energetically unfavorable configurations, which typically tend to be weakly bonded silicon atoms. Since crystalline phase is the lowest energy configuration, it often is the survived
structure. Unlike processing at high temperatures, the low temperatures typical of PECVD process result in limited grain growth, and the material obtained is typically viewed as microcrystalline silicon. Fine polycrystalline silicon structures,
however, are obtainable.
Increasing plasma power is another way to increase the rate of the "etching" reaction. However, the rate of the forward reaction is also boosted and, as a result, the net effect on the structural characteristics of the as-deposited film is not
as significant.
Returning to FIG. 2, in some aspects of the invention Step 12 includes depositing the microcrystalline film by a PECVD process using a SiH.sub.4 and H.sub.2 gas mixture. Step 12 also includes depositing the microcrystalline film at a power level
of approximately 0.16783 W per cm.sup.2, at a temperature of approximately 320.degree. C., a total pressure of approximately 1.2 Torr, a flow rate of SiH.sub.4 to H.sub.2 of approximately 100:1.
Alternately, Step 12 includes depositing microcrystalline film through a process selected from the group consisting of low pressure chemical vapor deposition (LPCVD), ultra-high vacuum CVD, and hotwire CVD, although PECVD is the preferred method. Regardless of the deposition method, Step 12 includes depositing microcrystalline film through chemistries selected from the group consisting of disilane (Si.sub.2 H.sub.6), higher silanes represented by the formula Si.sub.N H.sub.2N+2, where N is
greater than 2, and combinations of silane/fluorosilane chemistries represented by the structural formula Si.sub.N H.sub.2N+2 /Si.sub.M F.sub.2M+2, where N and M are greater than, or equal to, 1. M and N need not be equal.
As noted above, favorable results occur when Step 12 includes a microcrystalline film deposition rate of less than 2 .ANG. per second (.ANG./s). More specifically, Step 12 includes depositing the microcrystalline film with the PECVD method.
The deposition conditions include a deposition rate of less than 20 .ANG./s and a deposition temperature in the range of approximately 100 to 400 degrees C. A deposition rate of approximately 4 .ANG./s and a deposition temperature of about 320 degrees C.
has also been found effective. Alternately, Step 12 includes depositing the microcrystalline film with the LPCVD method, and the deposition conditions include a deposition rate of less than 20 .ANG./s and a deposition temperature of approximately 560
degrees C.
Based on the established relationship between microstructure and deposition conditions, one foundation of the present invention is the correlation between as-deposited microstructure and post-ELA-process materials characteristics (i.e. grain
size). FIG. 4 summarizes the average grain size of polysilicon as a function of the laser energy density and the crystalline content of the film. The crystalline content of the film is controlled by the deposition rate, as shown in FIG. 1a. Laser
annealing is assumed to be performed in vacuum at a substrate temperature of 450.degree. C. FIG. 4 depicts a strong correlation between grain size and deposition rate. Specifically, two dependency regimes can be distinguished: at low energy density
values (i.e., <240 mJ/cm.sup.2) no significant dependency of the grain size upon the crystalline content is observed. However, at mid-to-high energy density values, a strong correlation between grain size and crystalline content is observed, with the
average grain size becoming increasingly larger as the crystalline content increases (i.e., when the film is deposited at a lower deposition rate).
The following explanation is relevant to the results of FIG. 4. When as-deposited film has no embedded crystallites (c=0%), the maximum achievable grain size is about 460 nm at a laser energy density of about 320 mJ/cm.sup.2. Depositions of the
film in microcrystalline phase (c=5%) improves the grain size of the polysilicon film at a much lower energy density (about 280-290 mJ/cm.sup.2). At the same time, the variability of the grain size to changes in the input laser energy is far superior
for microcrystalline film, due to the controlled nucleation process inherent with the use of embedded seeds. Additional increases in grain size can be achieved by optimizing the pre-anneal crystalline content in the film (c=20%). Depending upon the
microcrystalline content in the pre-annealed film and the selected energy density, the grain size of the post-annealed film can increase by at least a factor of 2-3 when compared to a amorphous silicon precursor.
One important prediction of this model relates to the optimum laser energy density as a function of the as-deposited microstructure. At the same energy density, microcrystalline material tends to develop a larger grain size than amorphous
material. Alternatively, this can be thought of as analogous to a smaller laser energy requirement to achieve the same quality characteristic (for example, a grain size of about 340 nm requires 310 mJ/cm.sup.2 when the film is deposited at 10 .ANG./s,
but only 250 mJ/cm.sup.2 if the film is deposited at 2 .ANG./s). Therefore, the optimum energy density level (maximizing the grain size) is smaller for microcrystalline films than for amorphous films. The result is important, since reduction of the
operating laser energ | | |
