Method and Apparatus for Characterizing the Quality of Electrically Thin Semiconductor Films

The present invention generally relates to quality control of electrically thin semiconductor films (herein also referred to either as semiconductor thin films or thin films), and more particularly, to a method and apparatus for characterizing the quality of electrically thin semiconductor films and their interfaces. The phrase "electrically thin semiconductor film" is defined herein to be any semiconductor film for which there exists some surface potential such that a change in this potential results in a change in the potential at the semiconductor/substrate interface. In the semiconductor manufacturing industry, an effective way to characterize electrically thick semiconductor films is to make an electrical device on the material and then perform measurements using the device. The phrase "electrically thick semiconductor film" is defined herein to be any semiconductor film for which the potential at the semiconductor/substrate interface is independent of the surface potential for all values of the surface potential. This electrically thick film characterization technique is more sensitive than any other material diagnostic technique. The electrical device typically made on the semiconductor material for this purpose is a capacitor. Capacitance versus voltage, or C-V, measurements are made using the capacitor to characterize the quality of the electrically thick semiconductor films.

In the past, the capacitor has been made in an electrically thick semiconductor film and the C-V measurement was made by providing electrical contacts on opposite sides of the insulating substrate supporting the semiconductor layer, thus using the insulating substrate as the gate material One electrical contact was provided by the semiconductor layer itself available on the topside of the substrate. The other electrical contact was provided on the opposite side of the insulating substrate, i e. on the backside.

For the case where the insulating substrate is very thick (such as quartz or sapphire) the steps performed to provide such backside electrical contact were, first thinning the substrate (to a thickness of approximately 200 to 400 microns) and, then, evaporating metal on the backside of the thinned substrate to form the contact. The thinning step had to reduce the substrate to a thickness which would give interpretable capacities. Several problems are presented with this manner of making electrical contacts with a capacitor on the semiconductor material when taking the required C-V measurements. One problem is that the substrate thinning process is time consuming. Another problem is that the thinning process itself may affect the interface between the insulating substrate and semiconductor layer so as to change its electrical properties and thereby produce distorted results. A further problem is that even after thinning, thousands of volts of bias are required to accomplish the measurements. An additional problem is that the signal-to-noise ratio of the data is small because the thinned substrate is still many times thicker than the semiconductor film. Finally, as a consequence of the small signal-to-noise ratio this measurement has been restricted to doped, electrically thick semiconductor films.

For the case where the substrate is thin enough initially such that thinning is not required (such as Self-Implanted-OXide or SIMOX) the backside electrical contact is made either directly to a conducting substrate if one is used such as, for example silicon in the case of SIMOX, that is supporting the insulator or to an evaporated conductor such as aluminum. The problem with this type of technique is that it has been restricted to electrically thick semiconductor films. To make such films usually requires special processing thus increasing the complexity and cost of the measurement.

Consequently, there has been a long-felt need to devise a more reliable, quicker and simpler measurement technique for characterizing the quality of electrically thin semiconductor films.

The present invention satisfies the above described long-felt need and relates to a method and apparatus for characterizing the quality of electrically thin semiconductor films and their interfaces, including interfaces with gate materials, substrate materials and semiconductor compound materials such as epilayer oxides. The method and apparatus of the present invention employ a pair of electrical contacts on the same side of the film, i.e. the topside, to permit C-V measurements of electrically thin semiconductor films and their interfaces with the substrates such as insulating substrates such that derived C-V data is meaningful By eliminating the use of an electrical contact on the opposite side of the substrate, the method and apparatus of the present invention obviate the need for performance of the time-consuming substrate thinning process and/or the need for an electrically thick film. The primary advantage of the present invention is that it allows quick sampling and inspection of electrically thin semiconductor films, such as silicon films, in an integrated circuit manufacturing environment which up to the present has not been possible.

The C-V measurements provide pertinent information concerning the electrical characteristics of the electrically thin semiconductor films and the interfaces with substrates and other interfaces. These measurements are interpreted to reach a judgement as to the quality of the electrically thin semiconductor films and interfaces. Also, the method of the present invention has a large signal-to-noise figure of merit.

Accordingly, it is the primary object of the present invention to disclose a simple, quick and reliable method and apparatus for characterizing electrically thin semiconductor films and their interfaces.

Another object of the present invention is to disclose a method and apparatus for characterizing electrically thin semiconductor films and their interfaces by interpretation of low frequency C-V measurements.

Still another object of the present invention is to disclose a method and apparatus for characterizing an electrically thin semiconductor film by employing a pair of electrical contacts on the same side of the film, one contacting a capacitor made on the film and the other contacting the film itself.

Other objects, advantages and novel features of the invention will become apparent from the following detailed description of the invention when considered in conjunction with the accompanying drawings.

FIG. 1 is a flow diagram depicting a method of characterizing semiconductor material in accordance with the present invention.

FIG. 2 is a general diagram of an apparatus for characterizing semiconductor material in accordance with the present invention.

FIG. 3 is an enlarged cross-sectional view taken along line 3--3 of FIG. 2 of a semiconductor device of the characterizing apparatus.

FIG. 4 is a simplified band diagram illustrating the relationship of conduction and valence bands to Fermi level in an electrically thin semiconductor film.

FIG. 5 is a symbolic diagram of a biased capacitor.

FIG. 6 is a general diagram of a biased capacitor with a resistor in series and parallel.

FIG. 7 is a diagrammatic sectional view of the sample capacitor and substrate contact.

FIG. 8 is schematic diagram of a basic lock-in amplifier.

FIG. 9 is a general diagram of a circuit using the lock-in amplifier to measure capacitance.

FIG. 10 is a detailed diagram of the apparatus used in the present invention to measure capacitance.

FIG. 11 is a cross section of a circular capacitor illustrated by way of example.

Referring now to the drawings. FIG. 1 illustrates a flow diagram 10 depicting the basic steps of the method for characterizing the quality of an electrically thin semiconductor film in accordance with the present invention.

The characterizing method as illustrated by the flow diagram 10 is performed with respect to a sample of semiconductor material composed of semiconductor and substrate layers. The method basically includes a first step, as represented by block 12 of flow diagram 10, of constructing a capacitor and a pair of electrical contacts on the same side of the semiconductor layer of the material, and a second step, as represented by block 14 of flow diagram 10, of taking capacitance and voltage, C-V, measurements which provide the electrical transfer characteristics of the capacitor.

The first step (block 12) of the method utilizes a number of additional steps which are similar to those involved in processes used conventionally in constructing layered semiconductor devices and thus need only be briefly described. First, the semiconductor is cleaned. Next, a gate oxide to provide the insulator layer of the capacitor is grown or deposited on the semiconductor layer. Then, through a sequence of steps which may, for example, include depositing a conductive metal layer, then depositing, curing and removal of a photoresist mask and unmasked metal, and sintering the remaining masked metal, the capacitor and electrical contacts are formed on the semiconductor layer and the substrate. Likewise, other methods could be used to form the capacitor and contacts within the scope of the present invention. The second step (block 14) of the method is carried out by operation of a C-V measuring system which will be described below.

FIG. 2 generally illustrates apparatus 16 for use in characterizing the quality of semiconductor material. FIG. 3 shows a cross-section of the semiconductor/insulator structure 18 for which C-V measurements are to be made by characterizing apparatus 16 and by the characterizing method described herein.

More particularly, apparatus 16 includes low frequency C-V measuring system 20 connected to capacitor 22 and electrically thin semiconductor film 24 by capacitor contact 26 and semiconductor film contact 28. As seen in FIG. 3, semiconductor device 18 includes capacitor 22, such as a Metal Oxide Semiconductor (MOS) capacitor, constructed of upper layer 26 of any suitable type of electrical conductor material, such as aluminum or doped polysilicon, middle layer 30 of an insulator material, such as silicon dioxide or silicon nitride, and lower layer 24 of an electrically thin semiconductor material, such as a thin film of silicon, germanium or gallium arsenide. Semiconductor structure 18 also includes insulating substrate 32, composed of a material such as sapphire (Al.sub.2 O.sub.3) or quartz (SiO.sub.2), which mounts layers capacitor 22. Insulating substrate 32 is usually much thicker than upper, middle and lower layers 26, 30 and 24 making up capacitor 22 such that for purposes of analysis provided below substrate 32 is assumed to be infinitely thick.

As seen in FIG. 2, electrical contact 28 of electrically thin semiconductor film 24 is grounded and thus for low frequencies holds the Fermi level of the semiconductor material constant everywhere in capacitor 22. FIG. 4 is a simplified band diagram illustrating the relationship of conduction and valence bands to the above-mentioned Fermi level in a semiconductor material layer of an electrically thin semiconductor film on an insulator. As is well-known, the conduction band is the level of energy required for electrons to be free in the crystalline structure of and thus available for conducting electricity. The valence band is the level of energy required to retain electrons within the crystalline structure of semiconductor in non-conducting valence bonds about the nucleus where they are not free to conduct electricity. There is a gap G between these two energy levels. The Fermi level, E.sub.F, is a derived number which tells how many electrons are in the conduction band compared to the valence band. If the Fermi level is close to the conduction band, then there are numerous electrons available for conducting electricity. Conversely, if the Fermi level is close to the valence band, then more electrons are trapped at the valence sites and are not available for conducting electricity.

Given the construction of semiconductor structure 18 of FIGS. 2 and 3, coupled with the fact that it is possible to hold the Fermi level at zero with respect to substrate contact 28, the following condensed analysis is possible. First, the definitions of variables used herein are as follows:

C.sub.m =measured capacity

C.sub.ox =capacitor insulator (oxide) capacity

C.sub.t =series capacity of capacitor semiconductor and insulator layers

.phi..sub.c =potential edge of conduction band states

.phi..sub.v =potential edge of valence band states

Q.sub.sb =fixed charge of interface of substrate layer and capacitor semiconductor layer

Q.sub.t =total charge in capacitor, including its interfaces

N.sub.sb =density of states at interface of substrate layer and capacitor semiconductor layer

q=charge on electron

.phi..sub.b =potential at interface of substrate layer and capacitor semiconductor layer

.phi..sub.s =potential at interface of capacitor semiconductor and insulator layers

.epsilon..sub.si =dielectric constant of semiconductor (silicon) layer

E(x)=electric field at a point x

E.sub.B =electric field at interface of substrate layer and capacitor semiconductor layer

.beta.=(1/(26.times.10.sup.-3) volts=q/kT

t.sub.epi =thickness of capacitor semiconductor layer

Z.sub.s =e.sup..beta..phi.s

Z.sub.b =e.sup..beta..phi.b

N.sub.c =density of conduction band states

N.sub.v =density of valence band states

a=.epsilon..sub.si .beta.E.sub.b .sup.2 /2/q/N.sub.c /exp(-.beta..phi..sub.c)-exp(.beta..phi..sub.b)

Using Kirchoff's law, the definition of capacitance and a total differential, the following is true:

Assuming that voltage is measured from mid-gap to the Fermi level, the following is true from Gauss's law: ##EQU1## Assuming that the capacitor semiconductor layer is lightly P-type, Poisson's equation in the semiconductor layer reduces to

It can be easily shown that ##EQU2##

Equation 4 is now solved for E(.phi.) and is found to be

In order to find the complete solution the boundary condition that .phi.(x=0)=.phi..sub.s is imposed. This is done by using the fact that the relationship

can be used to rewrite equation 7 into the form ##EQU3## Equation 9 states that the integral of Equation 8 from the top of the epi-film to the back of the film must be equal to the film thickness. The integral can solved by making the following substitutions;

By substituting Equation 10 and 11 into Equation 9 one arrives at ##EQU4## which has two solutions depending one whether a is positive or negative. These solutions are: ##EQU5## Q.sub.t can now be found from Gauss' Law and is

Using equations 1, 2, 11, 13 and 14 complete, closed formed solutions, showing the relationship of capacity and voltage of a lightly doped semiconductor layer have been found. Two very important limiting cases follow immediately. Whether or not the semiconductor is lightly doped, by using Gauss's law it can be shown that when the semiconductor film is completely depleted out

The critical parameters that are to be measured are Q.sub.sb and N.sub.sb. Q.sub.sb is the fixed charge at the interface of the substrate layer and the capacitor semiconductor layer. N.sub.sb s is the interface state density at the interface of the substrate layer and the capacitor semiconductor layer.

The characterization of the electrically thin film semiconductor by the C-V technique of the present invention has a distinct advantage over the prior art C-V technique in that the analysis covers the cases where there is no depletion edge in the film. Also this analysis results in solutions from strong inversion to strong accumulation with very simple relationships in the depletion case. These solutions allow quick inspection of thin semiconductor films in an integrated circuit foundry environment which up to now has not been possible.

The capacitor used in the characterizing method of the present invention is preferably a two-terminal device. FIG. 5 shows a simple diagram of a capacitor C.sub.m (V.sub.g) which is a MOS capacitor with its gate biased by a power supply at voltage V.sub.g. To determine the capacity of such a device the definition

was used. In Equation (17) it is emphasized that C.sub.m is a function of V.sub.g and Q is the total charge in the capacitor. Typically the gate voltage is modulated by a small sinusoidal voltage such as

where t is time, and f is the frequency and V.sub.o is the amplitude of the a.c. signal. Thus ##EQU6## where I.sub.c is the instantaneous current in the capacitor. An important assumption now made is that the capacity is constant in the sampling range

Since the current in a fixed capacitor must be .pi./2 radians out of phase with the driving signal, Equation (19) reduces to

where I.sub.90 is the out-of-phase current and is a function of V.sub.g.

A more realistic perspective of the thin film capacitor is shown in FIG. 6. In this diagram, a voltage controlled capacitor C.sub.m (V.sub.g) has resistor R.sub.p in parallel and resistor R in series with a power supply and a function generator. Typical oxides have a net resistance greater than 10.sup.15 ohms for a 5 picofarad capacitor. The time constant associated with that value is greater than 10.sup.3 seconds. Thus for sampling frequencies greater than 0.1 hertz, parallel resistor R.sub.p can be ignored. However, for thin film devices and/or very high frequencies, series resistor R can force the current in Equation (19) to some phase other than .pi./2, thus making a correction necessary

Assuming the a.c. amplitude, V.sub.0, is small, a simple a.c. analysis can be used to solve for R and C in FIG. 6. These calculated values for R and C are

I.sub.0 and I.sub.90 are the in-phase and out-of-phase amplitudes of the peak current. The effect of R on the measurement of C can now be evaluated. Inspection of Equation (22) shows that for 2.pi.fRC<or=0.1 only a 1% or less error on the measurement of C will result from the presence of R in using Equation (21).

A still more realistic perspective on a typical thin film capacitor C.sub.m (V.sub.g)s which may be used is shown by way of example in FIG. 7. As shown in FIG. 7, the resistance is distributed and thus Equations (21) and (22) may not apply. The validity of using the criteria 2.pi.fRC <0.1 to allow the use of Equations 21 and 22 to calculate the capacitance for a circular capacitor is now addressed.

By way of example, the cross section of a circular capacitor is shown in FIG. 11. The thickness of the film shown in FIG. 11 is exaggerated. The periphery of the device may be held at ground potential and the gate electrode at V.sub.g. A small a.c. signal, V.sub.ac, is then added to V.sub.g and is expressed by

where V.sub.0 is the amplitude of the signal, j=(-1).sup.1/2, f is the frequency and t is time. For example V.sub.0 =10-25 mV.

Since the resistance and capacitance of this structure are distributed, a differential equation must be derived to describe the current flow caused by the applied alternating voltage. FIG. 11 shows a small element under the gate oxide of length dr. The current and the voltage in this element are related by the following equations

where i is the current in the element, r is the distance along the device, c is the capacitance per unit area, .rho. is the sheet resistance, assumed nearly constant, V is the voltage at the element, and t is time. Substituting Equation (25) into Equation (24) gives the diffusion equation which is written as

The boundary condition associated with Equation (26) is

where r.sub.o is the radius of the circular capacitor.

The solution to Equation (26) with the boundary condition stated in Equations (27) is

The measured current, i.sub.m (t) can now be found from Equations (29), (28), and (24); and is ##EQU7## Integrating Equation (30) gives

The Bessel function terms in Equation (31) are now expanded about .lambda.r.sub.o with terms containing (.lambda.L).sup.6 or higher being left out. After this expansion and some algebra, Equation (31) reduces to

Thus equating .rho./(8.pi.) and .pi.r.sub.o .sup.2 c with R and C in Equations 21 and 22 and FIG. 6, it is now clear that if

Thus, when the criterion in Equation 32 is met, Equation 32 can be written to an accuracy of 1.3% as

thus establishing that the series resistance and capacitance can be considered lumped for the device used in this technique, provided that 2.pi.fRC<0.1.

In the previous section it was shown that the capability of measuring in-phase and out-of-phase components of a capacitor's response to a small sinusoidal driving signal is very important in determining the capacity. This measurement can be accomplished using lock-in amplifier 38 schematically illustrated in FIG. 8. The basic components of lock-in amplifier 38 are mixer 40 and low pass filter 42. Mixer 40 mixes a reference signal from a local oscillator (not shown) and an input signal to produce an intermediate output signal which includes the sum and difference of the frequencies of the reference and input signals. The intermediate output signal then passes through low pass filter 42 which filters out nearly all a.c. components. Thus the d.c. level at the filter 42 output will be proportional to the product of the reference and input signal amplitudes at frequency f.

Lock-in amplifier 38 also includes a phase shifter (not shown) so that the phase of the reference signal relative to the input signal can be changed. It is to be understood that it is within the scope of the present invention that components other than lock-in amplifier 38 could be utilized to implement its function such as any phase sensitive amplifier. This allows the measurement of the in-phase and out-of-phase components of the input signal. By the appropriate filter choice, a.c. noise in the system can be reduced or eliminated by using the lock-in amplifier.

Referring to FIG. 9, lock-in amplifier 38 is shown being used to measure resistor R and capacitor C.sub.m in series. Function generator 44 is connected to power supply 46 and drives the portion of the circuit containing capacitor C.sub.m and resistor R with a small amplitude sinusoidal signal. This same function generator produces the reference signal at one input (e.g. Ref.) to lock-in amplifier 38 which is exactly in-phase with the driving signal. Sense resistor R.sub.m is placed in series with C.sub.m and R. The value of R.sub.s is <<1/(2.pi.fC). The voltage drop across R.sub.s is the other input (e.g Signal) to lock-in amplifier 38.

To measure the in-phase component, the reference signal is mixed with the input signal and then passed through the low pass filter. V.sub.out is then proportional to the in-phase component of the current in the circuit as discussed in the previous section. The reference is then shifted by .pi./2. Now V.sub.out is proportional to the out-of-phase component of the circuit. Equations (21) and (22) may then be used to measure C.sub.m and R.

FIG. 10 illustrates by way of example a system that may be used in the performance of the present invention. Semiconductor devices to be qualified are contained and probed in metal light-tight box 48. Box 48 is grounded, as is the outside of all coaxial cables connecting the various pieces of equipment. Holes and electrons are produced on the perimeter of the device by turning light 49 in box 48 "on" then "off" before each measurement. This ensures that capacitor C.sub.m is in d.c. equilibrium after each bias change. If the light is used during the experiment it must be only bright enough such that Equation (33) is satisfied during the measurement. Too bright a light can produce errors in the measurement if too much is coupled into the area under the capacitor. The d.c. bias is provided by bias function generator 56 while an a.c. modulation is provided by function generator 50 which also provides the reference signal for lock-in amplifier 38 The d.c. bias and the a.c. modulation are added together by operational amplifier 53 whose output is the stimulus for the test capacitor. This is desirable since a function generator with its own internal d.c. bias often has the phase difference between it output and trigger as a function of this bias. By summing the a.c. and d.c. components through operational amplifier 53 this problem is solved. Resistors 55, 57, 59, and 61 are for adjusting the gain and offset of operational amplifier 56. Their absolute values are preferably chosen such that they are much less than the input impedance of the operational amplifier. Their relative values are chosen to give the voltage swing the user needs for the gate voltage. The output of lock-in amplifier 38 is sensed by external voltmeter 52. Voltmeter 52 is used in the present example since lock-in amplifier 38 can not be used on the instrument bus 63 which in the present example is an IEEE-488 bus Thus the analog output of lock-in amplifier 53 is accessed by the computer via a voltmeter that can communicate with bus 63. Microcomputer 54 is used to control voltage bias function generator 56, power supply 58 for light 49, and voltmeter 52. This is done over instrument bus 63. Under the control of a software computer program commands to these three devices can be sent. Information can also be transferred to the computer over this bus from the voltmeter. Thus, all data is stored in a convenient format on a computer floppy disc, allowing easy data manipulation and retrieval. By way of example, commercially-available components that may be used are as follows: an Ithaco 393 Lock-In Amplifier as lock in amplifier 38; a Tektronix AM501 operational amplifier as operational amplifier 53; a Wavetek 182 Function Generator as function generator 50; a Keithy 619 Electrometer as voltmeter 52; an IBM Personal System/2 Model 50 Computer as microcomputer 54; a Tektronix 5010 Function Generator as voltage bias function generator 56; and a Tektronix 5010 Power Supply as power supply 58. A source code listing written in Quick Basic suitable for implementation as the source code for computer 54 is provided by way of example in the following pages.

It is thought that the present invention and many of its attendant advantages will be understood from the description herein and it will be apparent that various changes may be made in the form, construction and arrangement of the parts thereof without departing from the spirit and scope of the invention or sacrificing all of its material advantages, the forms hereinbefore described being merely exemplary embodiments thereof. ##SPC1##