Silicon Oxidation Studies Using Ellipsometry

Table of Contents

1. Introduction 1

2. Theory 1

3. Facility/Equipment

3.1 Oxidation Facilities 6

3.2 Thickness Measurement by Ellipsometry 7

4. Experiment 9

5. References 13

6. Appendices 14

A. Expectations

B. Health & Safety Notes

C. Equipment/Facility Care

D. Silicon Sample Preparation and Handling

E. Thermometry Details

F. Discussion of Experimental Techniques

G. Details of Water Source Pneumatics

Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

Silicon Oxidation Studies Using Ellipsometry

1. Introduction

An important step in silicon microelectronics processing is the formation of an oxide layer to act variously
as a diffusion barrier, an insulator, or an interface to a metallization [Refs. 1, 2]. Oxide layers can also be
used as the dielectric for on-chip capacitors.

Thermal oxidation of silicon in dry or wet oxygen is a widely used process in VLSI. The purpose of this
experiment is to evaluate the growth kinetics of the oxidation process as a function of processing
parameters such as temperature or atmosphere [Refs.3, 4]. You will also be exposed to ellipsometry, an
important optical instrument in thin film technology [Ref. 5]. An interesting result in this experiment is the
empirical observation of the color of the oxidized silicon samples.

Some background reading [e.g. Refs. 3,4,7] is strongly recommended.

2. Theory

The formation of the oxide of silicon (SiO2) on a silicon surface is termed oxidation. The ability to form
this oxide (SiO2) is the cornerstone of the planar (in a plane, i.e. two dimensional) processing of silicon
integrated circuits. Although there are many different ways in which to form this oxide, the process most
often employed is thermal oxidation [Refs. 6, 7]. It is this method that will be explored in this experiment.

The method of thermal oxidation enables the user to produce SiO2 films of desired thickness and beneficial
Si/SiO2 interface properties. It is due to this control that thermal oxidation is the most common growth
technique found in silicon processing. Thermally grown SiO2 is used in layers or in thicknesses ranging
from 6 nm to 103 nm typically. Some of the functions of these films include: a) masking against ion
implantation and diffusion; b) passivation of the silicon surface; c) isolation of individual devices (e.g. local
oxidation of silicon, LOCOS); d) use as a gate oxide and capacitor dielectric in MOS devices; and e) use as
a tunneling oxide in electrically alterable ROMs.

Freshly cleaved silicon rapidly forms a very thin native oxide layer when exposed to an oxidizing ambient
(e.g. oxygen, water, water vapour) at room temperature. The oxidation is self-limiting as surface silicon is
covered by the native oxide of about 1 to 2 nm thickness. At elevated temperatures more rapid growth and
thicker oxides result.

The reactions governing the formation of SiO2 are given as:

Si(solid ) + O2 (vapor) SiO2 (solid ) Dry Oxidation (1)

Si(solid ) + H 2O(vapor) SiO2 (solid ) + 2H 2 Wet Oxidation (2)

The oxidation reaction occurs initially at the ambient-silicon interface. As the oxide grows, the reaction
continues at the SiO2/Si interface while consuming fresh substrate silicon there. Based on the relative
densities and molecular weights of Si and SiO2, it is found that the amount of silicon consumed is 44% of
the final oxide thickness, as shown in Figure 1. This relationship is technologically important for
controlling step heights that form at the SiO2/Si interface between areas subjected to varying degrees or
duration of exposure to oxidizing ambients as a result of masking (patterning) processes [Ref. 2]. Such step
heights govern the topographical evolution of the silicon device structure.

2
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

SiO2 Surface x0
0.44xo
SiO2 Original Si interface

Silicon Substrate

Figure 1 - Thermal oxidation of silicon

The Linear-Parabolic model proposed by Deal and Grove assumes that the oxidation of SiO2 on Si occurs
as a result of two fluxes that sequentially transport the oxidizing species from the gas to the Si/SiO2
interface, where a third flux is involved in the consumption of the oxidant by the reaction with silicon.
The Deal-Grove model is a very accurate representation of the thermal oxidation process for most
oxides thicker than 50 nm [Refs. 6, 7]. It is a classic mass transport model.

Flux is defined as the number of atoms (or molecules) crossing a unit area per unit time with dimensions of
number /cm2sec. The three fluxes involved in this process are identified as: a) F1, flux of the oxidizing
species moving from the bulk of the gas phase to the gas/oxide interface, b) F2, the flux of the oxidizing
species as it diffuses through the existing oxide to the SiO2/Si interface, and c) F3, the flux of the oxidizing
species as it is consumed by reaction at the SiO2/Si interface.

These three quantities are defined in the following manner:

F1 = h(C * C 0 ) (3)

where C* is the oxidant concentration in the oxide in equilibrium with the partial pressure of the oxidant in
the bulk of the gas, C0 is the oxidant concentration at the outer surface of the oxide and h is defined as the
gas phase mass-transfer coefficient:

h = hg/HkT (4)

where hg is a mass-transfer coefficient, H is Henrys law constant, k is Boltzmanns constant and T is the
gas temperature. C* (=HpG ) is directly proportional to the equilibrium oxidant partial pressure pG.

dC (C0 Ci )
F2 = D = D (5)
dx 0 x0

where D is the diffusion coefficient of the oxidizing species in the oxide, Ci and C0 are the concentrations
of the oxidizing species at the growth interface and the oxide surface, respectively, and x0 is the thickness
of the growing oxide.

F3 = k s C i (6)

where ks is the chemical reaction rate constant of the reaction taking place at the interface and Ci , again, is
the concentration of the oxidizing species at the growth interface.

3
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

Equations (3), (5) and (6) are fully developed in more detail in the listed references [Refs. 6, 7]. Under
steady-state condition, that is when F1 = F2 = F3 = F, two important limiting cases arise. The first, the
diffusion-controlled case, arises when the diffusion constant D is very small such that the oxidant
concentration is minimal at the SiO2/Si interface (Ci 0) but approaches the gas bulk value at the oxide
surface (C0 C*). The second limiting case occurs when D is very large, and the process is therefore
reaction- controlled at the Si/SiO2 interface where the supply of oxidant (concentration Ci C*) sustains
the oxidation at a rate proportional to ks and C* (and therefore pG ).

In the steady state, Equations (3), (5) and (6) therefore yield a set of solutions for Ci and Co:

C*
Ci =
k k x (7)
1 + s + s o
h D
and

k x
C * 1 + s o
D
C0 = (8)
k k x
1 + s + s o
h D
In order to calculate the rate of growth (dx0/dt) we need to introduce a new variable N1, which is the
number of oxidant molecules incorporated into a unit volume of oxide. There are 2.2 x 1022 molecules of
SiO2 per cm3 in the oxide, and since two oxygen atoms are incorporated per SiO2 molecule, N1 for dry
oxidation is 2.2 x 1022 cm-3, and 4.4 X 1022 cm-3 for wet oxidation.

To calculate the rate, the differential equation:

dx 0 ksC *
N1 = F3 =
dt k k x (9)
1+ s + s 0
h D
is solved by applying initial boundary conditions, that is x0 = xi at t = 0, where xi includes the native oxide
that had grown on Si in room ambient, or the oxide from a previous process. The solution becomes:

x02 + Ax 0 = B(t + ) (10)

where;

1 1
A = 2 D + (11)
k s h

2 DC *
B= (12)
N1

4
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

xi2 + Axi
= (13)
B
where represents the time displacement needed to account for the initial oxide layer xi.

Since diffusion is a thermally activated process, the rate constant B in Eq.12 can be expected to follow an
Arrhenius relation with temperature as D does:
EaD
kT
B = B0 e (14)

where B0 is a pre-exponential constant that depends on C* and N1, EaD is the activation energy (of the order
of 1 eV) specific to a given diffusing oxidant, k is Boltzmanns constant and T is the temperature in oK. B
(in m2/hr) is called the Parabolic Rate Constant.

Equations (12, 13) suggest B/A is independent of D but should follow an Arrhenius relation through kS:

E
B C * kTaS C *
= ks = k0 e (15)
A N1 N1
where EaS is the activation energy (of the order of 2 eV, approximating that for Si-Si bond breaking, which
is necessary for oxidation) and k0 is a constant for a given surface crystallographic direction. B/A (in
m/hr) is called the Linear Rate Constant. Note that deduction of the activation energy from B/A data
assumes a constant C*, which is not true a priori under all experimental conditions. In particular, the
equilibrium partial pressure pG of the oxidant is expected to be temperature dependent, likely governed by
its own activation energy.

The linear dependence of both rate constants on C* suggests that their measured values can be very specific
to apparatus and operation conditions, especially regarding humidification-dehumidification effects
between the water source and the furnace. It should be noted that the description of the original apparatus in
the Deal-Grove publications may not be detailed enough to suggest experimental equivalence. Only under
excess humidification condition, achievable in a humidification apparatus of large enough scale, would C*
be effectively forced to be a constant by saturating the surface of the oxide with water molecules. C* would
then be the same value at any temperature as long as the apparatus can maintain the excess.

The Deal-Grove model works well for wet oxidation but not for thin oxides (<30 nm) grown in dry oxygen.
See Refs 6, 7 for more information.

5
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

3. Facility/Equipment (Please request a demonstration before operation.)

Oxidation Facilities (See Appendix C and follow facility/equipment care instructions carefully.)

Figure 2 shows a (quartz) tube furnace in which a crucible carrying multiple silicon chips is centrally
positioned. Ambient gas flows through either a dry channel (99.999% pure nitrogen, <1ppm water, <1ppm
oxygen) or a wet channel (99.999% oxygen humidified by bubblers) with appropriate pneumatics. A
spouted, 300ml tall-form Pyrex beaker serves as endcap to limit back-diffusion of air at the tube exit and
minimize the steady state flowrate of process gases.

For wet oxidation, high purity oxygen controlled by needle valves at two flowmeters in series (Matheson
tubes 602, 604) is fed through a wet channel comprising a 2-stage bubbler. Each stage is a 500 mL flask of
distilled water heated to the 94 - 98oC range by immersion in a water bath in turn temperature-controlled by
a digital dial for repeatable setting. Thermometers monitor both flask and bath temperatures (See Appendix
E for thermometry details). Boiled distilled water is used to fill both flask and bath to minimize start-up
time upon siphoning out cooled water from the previous lab. Stage 2 flask has a glass diffuser and glass
beads to stabilize temperature. Stage 1 has a perforated Teflon tubing as diffuser. Plastic beads limits
evaporation and stabilize temperature at both baths. Temperature drop in Stage 2 water is limited to 2
degrees C during bubbling cycles. The water vapour pressure in the wet oxygen stream is thus maximized
by a high flowrate without boiling or causing liquid injection into the furnace. The incoming water vapour
pressure at the furnace tube would be close to atmospheric as the above-mentioned endcap effectively vents
the flask to atmosphere through the furnace tube.

When selected for nitrogen purge, the dry channel allows to chemically arrest wet oxidation of samples
while in the furnace and in transit at the tube exit where the endcap is necessarily off. Given an appropriate
N2 flowrate, as controlled by fine (Matheson tube 601) and coarse (10 LPM air equivalent) flowmeters in
series, N2 purge on/off controls the oxidation time.

Quartz coupling tube

Thermocouple assembly with
wire mesh shield
Silicon chips

Crucible(s) with
perforated Flowmeter Dry
(Coarse) Channel
handle Flowmeters
Quartz tube Vent 1

Pyrex
Vent 2
endcap Dry
N2 65 psi

Thermometers Thermometers

Furnace Wet
Wet Channel
Flowmeters

Heated bath with

Porcelain Temp plastic floats
end table Contr
O2 40 psi
ol STAGE 2 STAGE 1
O
94-96 C Distilled
water with 94-98OC
glass beads Distilled water

Figure 2 Oxidation apparatus (pneumatics abbreviated - see Appendix for details)

6
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

Vent 1 relieves pressure in Stage 1 and should never be completely closed. Vent2 is under pneumatic
control, opening during dry N2 purge to keep moisture from the furnace tube and closed otherwise. Other
details of the pneumatic system is given in Appendix G.

The ceramic (Alundum, i.e. vitrified aluminum oxide) crucible has a perforated handle to facilitate
manipulation by a hooked (quartz) push-rod. Another jig constructed from a pair of quartz tubes provides a
protective rest for the tip of the push-rod when idle but hot. Tongs are provided to handle hot crucibles and
the endcap during sample loading/unloading cycles. A porcelain crucible (not shown in Fig.2, carries
silicon chips and sits in the Alundum crucible as a secondary vessel. A porcelain end table provides
support for the endcap as well as a working surface when transferring silicon samples from a hot porcelain
crucible to a cooling surface before their further transfer to a plastic waffle pack (also called bare die
tray).

The push-rod is never handled by bare hands to eliminate finger grease contamination. Although it can be
safely handled in Nitrile gloves up to 1/3 of the cool end, thermal mitts are recommended instead for those
prone to forget this and at the expense of dexterity. The very end of the push-rod is at room temperature
where contact or proximity of hand/face should be avoided as burns can result from infrared strongly
guided through the rod.

The furnace temperature is controlled by an Omega CN2100 Controller. The temperature sensor is a
thermocouple mounted on the heating jacket at approximately the geometric centre of the furnace but
outside the quartz tube. Silicon sample or crucible temperature is therefore not directly measured but the
thermocouple temperature is assumed to be representative (See Appendix E. for thermometry details). To
set the process sequence, use the following steps:

a. Check power connections (230 VAC and 115VAC).

b. On the Controller Keypad, press TUNE, LAST, YES in sequence
c. Repeatedly press PARAMETER to cycle through the parameter values
d. Set Ramp and Soak: YES gives a linear warm up and a subsequent hold; NO allows you to set a
continuous temperature
e. Press END OF TUNE
f. To start the process press MANUAL/AUTO
g. To shut down, set to room temperture under MANUAL mode.

Due to thermal loading in the laboratory, the room temperature may rise in time, necessitating careful
calibration of thermocouple-based thermometers for repeatable control of water source temperatures during
and between lab. periods. Calibration against a bath of crushed ice is recommended.

Thickness Measurement by Ellipsometer

A classic industrial tool, the Gaertner Rotating Analyzer Ellipsometer measures oxide thicknesses ranging
potentially from a few tens to about 1000 nanometers, depending on experimental details and variations.
(Model LS104SA with LS104B controller, S/N 1533-AK.. Software: NI-DAQ Driver and GEMP v.1.2,
1998. PCI interface card. Gaertner Scientific Corp., Skokie, Illinois. 1-847-673-5006. See manuals at the
workstation).

Data can be downloaded in a text file for exporting to spreadsheets.

NIST traceable calibration samples at 100C and 900C (nominal) thicknesses, SiO2 on 100mm Si Wafers are
provided.

Especially for thin oxides, comparison between results from various ellipsometer measurement modes
(algorithms) should be made to discern systematic uncertainties and to decide on preferred modes.

7
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

Figure 2 Schematic diagram of the ellipsometer setup

The ellipsometer software is a semi-automated program that measures four quantities (refractive index N,
the extinction coefficient K, the ellipsometry coefficients and ) by a fitting routine. Default angle of
incidence is Phi = 70o. Operate the instrument as follows:

a. Turn on controller by the ON switch on the front panel and turn up the light intensity of the
microscope by an adjacent knob.
b. Clean the stage with a cleanroom towel wetted with IPA (isopropyl alcohol).
c. Verify these instrument settings: Polarizer Drum (45O), Beam Attenuator Lever (In, Closed),
Photodetector Mode Toggle Switch (A).
d. Start up the computer and enter the program via the GEMP icon on the desktop. Confirm default
settings for PHI (70O), POLARIZER (45O), AMBIENT (1 for air).
e. Place an uncoated substrate (of same thickness as the sample to be analyzed) at centre stage and
therefore in beam path.
f. Adjust the two screws under the stage to superimpose the two crosshairs in the microscope eyepiece.
g. Adjust the stage vertical position upon loosening the setscrew and aiming the reflected beam at the
aperture of the anaylzer. Turn the analyzer drum to an orientation corresponding to maximum
(bargraph) signal on the front panel of the controller. Further maximize the signal by fine-adjusting the
stage vertical position before fastening the setscrew again.
h. Re-align the two crosshairs in the microscope to align the sample normal to the optical symmetry axis.
i. Select CALCULATION MODE Substrate. Enter SAMPLE ID for substrate.
j. Push MEASURE & CALCULATE. Push STORE MEASURED DATA to record the N and K values
for the substrate (Ns, Ks) in a text file.
k. Remove the substrate sample and substitute with the sample to be analyzed. Verify that the crosshair
alignment is preserved which should be the case if the two samples are of identical thicknesses and
parallelism and no dust particle has affected their seating.
l. Select CALCULATION MODE Thickness 1 & Nf1. Enter Exp(ected) thickness and Nf1. Enter
substrate values (Ns, Ks) just measured.
m. Push MEASURE & CALCULATE. Push STORE MEASURED DATA to record the measured
thickness and Nf1for Film 1 in a text file.
n. If necessitated by lack of solution or unreasonable results, iterate by adjusting input parameters and by
fixing Nf1 (AutoFix Nf1 under CALCULATION MODE).
o. For oxide thicknesses less than 300A, compare results from CALCULATION MODE Thin Oxide.
p. Save measurement session record in the text file before program exit.
q. Shut off controller. Leave no sample on the stage.

8
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

4. Experiment

Obtain silicon samples from the teaching staff. Note sample Lot No. on the container in case they prove
relevant to experiment understanding. Request a demonstration of proper sample handling protocols and
review written instructions (Appendix D).

Observe all health and safety as well as facility/equipment care instructions (Appendices).

Ve nt 1 &
N ee d le Va lve

A d a ptor
Ve nt 2 & w ith glas s
Pn e um a tic Va lve rod ha n dle

D ry pu rg e TC 1
in le t

TC2 S ta ge 1

C a rrier g a s in le t

B ath T C
Sta ge 2

Fig.3 Picture of water source

9
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

Set Up

At the beginning of each lab. period, check that these default and SHUTDOWN conditions apply before
changing them to IDLE and other operating conditions:

Pneumatics

Vent 1 (Needle valve on post): Default is Full CCW (maximum vent)

N2/O2 (wall-mounted valve): Default is O2 (wet channel carrier gas)

N2 Bottle O2 Bottle DRY N2 DRY Channel WET Channel

CONDITION Valve Valve N2/O2 PURGE Flowmeter Flowmeter
[Regulator] [Regulator] Select ON/OFF Valve Valve
Closed Closed
SHUTDOWN DRY N2 OFF Closed Closed
[CCW] [CCW]
Open Open 6 LPM air
IDLE DRY N2 ON Closed
[65 psi] [10 psi] equivalent

Electrical

Stage 2 Water Stage 1 Bath

115VAC 230VAC
Power Bath Controller Controller Furnace
CONDITION Panel Panel
Cables Digit.Dial @503 Dial (Limit) Controller
Switches Switches
(Limit is preset) @ maximum
Preset(Preset)
SHUTDOWN Plugged OFF OFF Power OUT OFF
Power OUT
Preset(Preset) ON-
IDLE Plugged ON ON Power IN
Power IN Room Temp

Water Source Set Up

At the beginning of each lab. period, drain room temperature water left over from previous lab. period in
both flask and bath of each stage by thoroughly siphoning each with the corresponding squeeze bottle-
tubing kit provided. For temporary access, use the glass thermocouple (TC2) adaptor on Stage 2 and the
adaptor with a glass rod handle on Stage 1. As for all others, both adaptors are Teflon sleeved for ease of
removal when cold. Replace when hot kettle water has been introduced (see below), applying minimal
finger pressure to seal it will self-seal by thermal expansion. (Note: this practice of draining cooled water
and adding measured volumes of boiled distilled water saves lab. time, removes particulate contaminants in
the flasks, and ensures accurate flask water charge. Alternatively, to skip the use of a kettle, one can start
heating hours beforehand.)

Wet oxidation at 800o C - 900oC, short durations

This is to establish the wet oxidation rate at a typical process temperature by measuring oxide thickness as
a function of time to look for evidence that the rate limiting mechanism is indeed surface reaction (i.e.
linear regime) at short oxidation times. You will need multiple runs with multiple samples per run to
optimize statistics and to detect systematic errors. Oxidation time for each run has to be short enough to be
well in the linear regime yet long enough to have enough data points collected in an affordable lab. period.
For good planning, consult with teaching staff and see Appendix F for suggested techniques.

Data is to be collected at nominally 800, 825, 850, 875 and 900OC following these generic instructions:

a. With furnace still cold, plan push-rod position so as to place crucible(s) as close to the center of the
tube furnace as possible.
b. Set furnace controller to lowest temperature setpoint used for a given lab. session, and monitor warm-
up to ensure proper operation and no excessive temperature transients.

10
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

c. Allow at least 15 min. of thermal equilibration at each new furnace temperature setting before loading
samples.
d. Fetch distilled water in 1 litre bottle and 1 gallon jug provided (supply in 4th floor core in Stirling Hall)
e. Measure 350 ml of fresh distilled water in a cleaned beaker and boil in clean kettle.
f. Temporarily remove the (Teflon sleeved) glass thermocouple adaptor and TC2 on the Stage 2 flask and
empty the 350 ml of boiled distilled water into it through a clean glass funnel. Secure TC2 and adaptor
again with a very gentle twisting action to start seal.
g. Start monitoring Stage 2 flask water temperature against time.
h. Fill 2nd stage water bath from a full 2-litre kettle of boiling distilled water until the plastic floats rise to
the point their upper layer is level with the rim of the bath.
i. Start monitoring Stage 2 bath temperature.
j. Turn power switch on the Stage 2 bath to IN. Upon equilibration (in 20 30 min.), this should yield a
bath temperature of 98 + 1OC. This is the highest possible setting short of vigorous boiling in this bath,
and correspondingly sets a quiescent flask temperature of 96 + 1OC without carrier gas bubbling.
k. Similarly fill the Stage 1 flask with 500 ml of boiling distilled water. Similarly fill the bath. Turn bath
power switch to IN and start monitoring flask and bath temperatures. Verify flask temperature control
at 94 - 98OC, especially to avoid vigorous boiling; fine-adjust where necessary.

Note: If wet oxygen injection is done for the first time, it is recommended that a preliminary short
run be done under supervision as suggested in Appendix F to familiarize with its controls and
operating range.

l. With a clean porcelain crucible on a cleanroom towel, transfer silicon chips (polished side up) as
planned and identify them by location in a diagram.
m. When ready to introduce samples into furnace, remove the endcap with tongs. Position the Alundum
crucible at the mouth of the furnace tube, again using tongs.
n. Transfer the loaded porcelain crucible by tongs to the Alundum crucible, and push the assembly so it
does not protrude from the furnace tube.

Fig.4 Crucibles assembly being inserted (chips not shown)

o. Re-apply endcap and leave for 1 min. to allow crucibles to warm up (to reduce thermal shock).
p. Remove endcap. With above-mentioned precaution against burn hazards, insert crucible assembly at
about 1 cm/sec by the push-rod to the planned position. (Keep hands and face away from the furnace-
heated section and the end of the push-rod!)

11
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

DRY Wet Channel

DRY N2 DRY
N2 and O2 Channel Flowmeter
N2/O2 PURGE Coarse
CONDITION BottleValve Flowmeter Valve
Select ON/OFF Flowmeter
Valve Tube 602 (mm)
DRY N2 6 LPM
DRY N2 ON Open Closed
PURGE (Air Equiv.)
Open Open
WET O2 DRY N2 OFF Closed NA 110mm max.
( S.S. bead)

q. Following the pneumatic settings appropriate to each condition as shown in the above table, execute
each oxidation cycle as follows: From DRY N2 PURGE condition, switch to WET O2 condition
(default: max. wet channel setting) and start stopwatch when the 2nd stage flask water temperature
reaches the prescribed steady-state value to mark the beginning of wet oxidation. Upon reaching the
planned duration, switch to DRY N2 PURGE condition again while promptly stopping the stopwatch.
(The datalogged TC2 trace gives an independent record of WET O2/DRY N2 switching events by
virtue of the slight cooling effect of carrier gas on Stage 2 water.)
r. Maintain a 2-min. DRY N2 purge. Remove the endcap temporarily and withdraw crucible(s) at 1
cm/sec. to mouth of tube. Wait 1 min. before unloading crucible onto porcelain platform.
s. Unload silicon chip(s) in the porcelain vessel. Let cool for 1 min. before transfer to (plastic) waffle
packet.
t. For each oxidized chip sample, discern the oxide colour to help identify [colour chart, Table 4.2 of
Ref. 7] the most likely expected thickness as input to the ellipsometry measurement. A thickness
reading which is also consistent with the colour chart is the most likely the correct value.
u. By division of labour, measure oxide thickness promptly while running the experiment for timely
feedback and to minimize cumulative dust particle contamination, which can compromise
measurements given the lack of cleanroom environment. Plot oxide thickness vs. oxidation time data
while in the lab. and consult with teaching staff about data quality.
v. Adjust experimental plans at various temperatures to optimize data quality for determination of the rate
constants. Attempt to determine a characteristic activation energy for wet oxidation where data trend
suggests it is appropriate. Compare with published values.

12
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

Additional Exercises (Optional)

For detailed investigation or preventive maintenance. As there is usually barely enough time to complete
the basic runs, these should be attempted only when there is lab. time left or by investing extra lab. time.

Wet oxidation at 1000oC at larger thicknesses

This is to explore the wet oxidation rate at a process temperature and at oxide thickness commonly
applicable to silicon chip manufacture to look for evidence that the rate limiting mechanism is indeed
diffusion (i.e. parabolic regime) at long oxidation times. One approach is to use pre-oxidized chips with
known thicknesses and grow additional oxide of modest thickness and process duration.

w. Measure oxide thickness on pre-oxidized chips and use colour charts to ascertain range. Minimize
handling to avoid excessive dust particulate contamination before furnace entry. Alternatively,
sacrificially measure sample chips from the same lot while mounting other clean chips without
handling.
x. Repeat Steps l though v above but run at 1000oC furnace temperature. Analyse oxide thickness vs. time
data and if possible determine parabolic rate constant.

Water source operating parameters dependence

y. Repeat any wet oxidation experiment done above but at a different Stage 2 (quiescent) flask water
temperature (e.g. 93, 90 + 1OC, Step j, adjust digital dial setting) to explore temperature dependence
effects in the water source. Compare and interpret results at the two water bath temperatures.
z. Repeat any wet oxidation experiment done above but at different carrier gas flowrates to explore the
presence or absence of excess moisture conditions at a given furnace temperature.

Mass flow and temperature effects

aa. Repeat any wet oxidation experiment done above by loading crucibles and chips in different ways or in
combinations to explore mass flow and/or temperature effects due to geometry and position. Such
effects may be furnace temperature and carrier gas flowrate dependent so finding a meaningful
comparison could be most informative concerning this apparatus.

13
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

5. References

* Indicates availability of photocopy at experiment station.

1. Silicon Processing for the VLSI Era, Volume 1: Process Technology, Chapter 7, S. Wolf, R. N.
Tauber, Lattice Press, (PO Box 340-W, Sunset Beach, CA, 90742, USA),1986, ISBN 0-9616721-3-7.
[Queens Engineering/Science Library TK7874 .W635 1986 v.1.]

2. Fundamentals of Microfabrication, Marc Madou, CRC Press, Baton Rouge 1997.

3. *Deal, B.E. The Oxidation of Silicon in Dry Oxygen, Wet Oxygen, and Steam. J. Electrochem. Soc.
Vol.110, No.6, pp.527-533 (1963) [Online]

4. *Deal, B.E., A.S. Grove General Relationship for the Thermal Oxidation of Silicon J. of App.Phys.
vol.36, No.12, pp. 3770-3778 (1965). [Online]

5. Ellipsometry and Polarized Light , R.M.A. Azzan and N.M. Bashara, North-Holland 1977. [Queens
Engineering/Science Library QC443 A96]

6. *Physics and Technology of Semiconductor Devices, Chapter 2 Thermal Oxidation and Fig.4.14,
Andrew S. Grove, John Wiley & Sons, 1967.

7. *The Science and Engineering of Microelectronic Fabrication, Chapter 4, Stephen A. Campbell,

Oxford University Press, 1996, ISBN 0-19-510508-7 [Queens Engineering/Science Library
TK7871.85 C251996]

8. *Safety Infogram, Canadian Centre for Occupational Health and Safety, Laboratories; I09,
Compressed Gas Regulator Use.

9. Flowmeter Calibration Data for Matheson tubes 601 (air), 602 (oxygen), BOC (formerly Matheson
Gas Products Canada, Inc., 530 Watson Street East, Box 89, Whitby, Ontario L1N 5R9 1(905) 668-
3397, 1(800) 263-2620.)

10. *Certification of Measurement (Traceable to NIST), Standard 100C and 900C SiO2 on 100mm Si
Wafers, Gaertner Scientific Corporation Models L118SW-100 and L118SW-900.

14
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

APPENDIX A. EXPECTATIONS

Evaluation and grading will generally be based on the following factors:

1. Conformance to Health & Safety and Equipment/Facility Care instructions (see below).
2. Good documentation discipline as evident in lab. notebook and raw data records.
3. Observation by teaching staff of in-lab performance and behaviour.
4. Coverage of scope specified under the Experiment Section.
5. Grasp of principles and sound laboratory practices as evident in report and presentation.
6. Extra credit from evidence of relevant innovation or extra effort to:
6.1 Enhance own learning experience, or
6.2 Offset any unforeseen equipment/facility related problems, or
6.3 Suggest improvements to the Experiment.

APPENDIX B HEALTH & SAFETY NOTES

Compressed gases As long as regulators are properly attached and operated at low pressures (=<20 psi),
there should be minimal direct hazard. See Ref. 8 for safety instructions.

1. High temperatures The furnace is operable up to 1100oC such that all precaution must be taken to
avoid burns. Do not open the furnace casing while it is heated. Note the following hot items (and
proper handling tools): endcap (tongs for beakers), crucible (tongs), push-rod (padded mits, quartz tube
stand), silicon chips (metal tweezers, porcelain cooling dish). Avoid burn from guided infrared at the
tip of the push-rod when handling it there (see Section 3).

2. Steam Water near boiling point has a vapour pressure close to 1 atmosphere. When the wet channel
is selected, the much increased steam pressure is relieved through the furnace tube to atmosphere.
Boiling in the water flask is to be avoided to preclude uncontrolled moisture content and pressure
hazard. When moist oxygen is further heated in the furnace, it behaves as a rapidly expanding gas with
unsaturated steam. The effluent steam is bled into the lab. through the beak of the endcap and should
be kept in mind as a burn hazard.

3. Hot water up to 2 litres of boiled distilled water will be handled in each lab. session.

4. Vacuum The coupling tube assembly has a vacuum jacket over which a wire mesh implosion shield
is always applied.

5. Lasers The ellipsometer laser beam, although low power (about 1 mW), can be reflected off silicon
samples and shiny tools. Wear safety eyewear when aligning beam.

6. Glass/Quartzware Handle gently and report any breakage and chipping for proactive replacement.

15
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

APPENDIX C. EQUIPMENT/FACILITY CARE

1. Generic care (daily):

1.1 Observe all cleanliness protocols and leave no litter.
1.2 Keep all items (including documentation) at original workstation.
Return all mobile items to toolboxes and storage facility provided.
Exit all programs to idle at operating system level.
Promptly report all malfunction, damages, losses, and material depletion to the faculty member in charge of
the experiment.

2. Experiment-specific care:
2.1 Follow proper compressed gas turn-on and shut-down procedure to avoid a sudden high pressure
surge. Release regulator pressure (handle fully CCW) before opening and closing the shut-off valve on
the cylinder. Upon shut-down, always bleed off the regulator pressure by temporarily operating the
control valves of the pneumatic system.
2.2 Always wear clean Nitrile gloves when handling cold furnace parts (crucibles and quartz tube) to avoid
figure grease contamination and its subsequent burning.
2.3 Always wear clean Nitrile gloves when handling glassware. Clean glassware regularly by rinsing in
distill water and then IPA. Invert over cleanroom cloth or paper for storage.
2.4 Move crucible at about 1 cm/sec inside the furnace to avoid thermal shock of crucibles.
2.5 A reference sample (silicon and silicon dioxide) is available to check ellipsometry basic function.

16
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

APPENDIX D. SILICON SAMPLE PREPARATION AND HANDLING

1. Silicon source
1.1 Wafer specifications

Diameter Thickness Polish Crystal orient. Resistivity Dopant

100 mm 475 - 575 m Single-side (111) 1 - 5 ohm-cm Phosphorus

1.2 Protective oxide: A thermal oxide of 0.15 to 1.2 m thickness is to be grown on the wafer by the
commercial source to protect the silicon surface during dicing and storage.

2. Dicing
2.1 Die (chip) size: 4.5 mm x 5.0 mm is a convenient size for accommodating multiple samples in one
crucible. A 100 mm dia. wafer should yield a large number of samples.
2.2 Cleanliness: A commercial supplier should have high pressure RO (reverse osmosis) water jet to cool
and clean the samples during dicing. The samples are then air-dried before storage in waffle packs
(see Item 5. below).

3. Surface preparation: The protective oxide needs to be stripped either by the commercial supplier or
locally using methods outlined here.
3.1 Facility requirements: Lab. with a chemical fume hood, a sink with running water, chemical shower
and eyewash, and storage for inorganic chemicals.
3.2 Chemical safety tools: face shield, Nitrile gloves, heavy duty plastic gloves, chemical apron, protective
sleeves, First Aid Kit for hydrofluoric acid. Use at all times.
3.3 Process tools: Heated water bath, glass thermometer, polypropylene bottles, squeeze bottle and beaker,
500 ml Teflon (TFE) beaker, Teflon (FEP) sample wash basket kit, stopwatch or timer.
3.4 Materials: 49% reagent grade hydrofluoric acid (HF), distilled or deionized water (DW, DI),
cleanroom wipes (towels), 1:1 diluted HF (49%: DW).
3.5 Procedure (under fume hood): Preheat water bath to 50oC as measured by an immersed glass
thermometer. Adjust the bath temperature controller to maintain temperature. Then remove the
thermometer to keep it away from HF fumes. Batch-load oxidized silicon samples into sample wash
basket and apply cover and handle. Add 100 ml of diluted HF (1:1) to TFE beaker and place wash
basket with samples in acid. Place beaker in the water bath and allow about 10 minutes warming
before setting timer for 30 min. etch. To quench the etch, transfer the wash basket to another 400 ml
beaker full of DW. Decant the TFE beaker to a storage bottle (for reuse) and rinse the beaker by
squeeze bottle into a waste bottle for dilute HF. Rinse samples from the other beaker with DW
similarly and collect dilute HF waste. Unload samples onto a cleanroom cloth and cover with another
one to blot any remaining water beads. Promptly load samples into clean waffle pack for storage
identify sample lot. Rinse the outside of all bottles and beakers by tap water, dry, and then store in
clean toolbox.
3.6 Waste disposal: Dilute DW. Soak all wet towels in tap water before disposal.

4. Sample Handling
4.1 Precaution: Do not talk over exposed samples. Always work on a table over cleanroom towels.
4.2 Gloves: Always wear cleanroom gloves while handling sample boxes and tweezers.
4.3 Operation: Unlock the keepers on the waffle packs. Remove paper liner by tweezers. Hold each sample
by edges with tweezers perpendicular to sample surface. Complete all transfers and close/lock the
waffle pack immediately.

5. Sample Storage
5.1 Separate waffle packs (bare die trays) are supplied for fresh and processed samples.
5.2 A grid pattern in each pack facilitates sample identification and counting. Use a map to carefully plan
and manage processed samples, which are so small they cannot be identified by other means.

17
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

APPENDIX E. THERMOMETRY DETAILS

Thermometry is a critical factor in this experiment as the rate of wet oxidation is dependent on both sample
temperature and relative humidity in the furnace ambient, the latter being in turn dependent on control of
temperatures in the water vapour source:

1. Sample temperature
1.1. Calibration - At 800OC and above, heat transfer in the furnace is predominantly radiative, such
that the temperatures of the furnace thermocouple, the crucible and the samples, all positioned at
centre, should be very close regardless of heat and mass transfer effects in the oxidizing gas.
However, the accuracy of the furnace thermocouple calibration can only be independently
checked against melting temperatures of known metal alloys placed in the furnace while being
optically monitored this had not yet been done for this apparatus. It can be noted that an error
of 10 degrees Kelvin amounts to ~1% in absolute temperature readings.
1.2. Temperature uniformity in furnace - The dry channel, when selected for O2 (99.999% purity),
allows an experimenter to perform dry oxidation to check the furnace temperature profile apart
from flow effects. This test is most suitable when done by a lab. technician as preventive
maintenance. (See also Appendix F, Item 1.2).

2. Bath temperature (record steady state readings with and without carrier gas flow)
2.1. The Stage 1 bubbler preheats the carrier gas and moisture content in order to minimize cooling in
the Stage 2 bubbler. The Stage 1 bath temperature is therefore maintained as close to 100OC as
possible, running at maximum power and therefore no finer temperature control is necessary. A
typical controlled level in the range 94 98 OC is suitable. A digital thermometer is adequate for
setup and spot check.
2.2. The Stage 2 bubbler temperature, on the other hand, needs to be consistent at nominally 95OC to
maximize vapour pressure without the sporadic effect on gas flowrate due to boiling. As the
vapour pressure of water rises rapidly with temperature near boiling point, the control of Stage 2
bath and bubbler temperatures is more critical. The Stage 2 bath temperature is set to 98 + 1 OC
(read by a digital thermometer), which typically results in a bubbler temperature in the range 95
96 OC, depending on room temperature and consistent water level, and controlled to + 0.5 OC
without carrier gas flow. Typical temperature drop at the maximum recommended carrier gas
flowrate is 2 degrees C.

3. Bubblers temperature
3.1. The Stage 1 bubbler temperature is monitored only by an immersed Type J thermocouple
connected to a computer-interfaced digital gauge (Omega Engineering HH506RA) to monitor
temperature near boiling point.
3.2. The Stage 2 bubbler temperature, with or without gas flow, is critical since it determines the
relative humidity of the carrier gas immediately before injection into the furnace tube. A Type J
thermocouple probe is immersed in the bubbler and connected to a gauge with computer-
interfaced monitoring (see calibration instruction below) and datalogging to facilitate process
history recording. There is provision to plot temperature vs. time while running.
3.3. Thermocouple gauge calibration against an ice bath should be done annually as follows:
3.3.1. Fetch crushed ice (ice chest and crusher in 4th floor core in Stirling Hall) in a thermos. Do
not use mere ice in water or ice cubes.
3.3.2. Temporarily remove the Type J thermocouple probe from the Stage 1 or 2 bubbler by
loosening the screw-cap adapter. Immerse in crushed ice bath and wait at least 1 min.
3.3.3. Set the calibration (i.e. temperature compensation) of the thermocouple gauge so as to
obtain a steady reading of 0OC (ice point). Wait longer if the reading has not yet settled. (*
Procedure specific to a given gauge see manual)
3.3.4. Retrieve the thermocouple probe from the ice bath and record steady room temperature for
reference.
3.3.5. Re-install the thermocouple probe on Stage 1 or 2 bubbler, tightening the screw-cap adapter.

18
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

APPENDIX F. DISCUSSION OF EXPERIMENTAL TECHNIQUES

1. Apparatus dependence
1.1. As the behaviour of a moisture-loaded gas in a furnace near 1000oC is quite complex, actual
oxidation rates are very apparatus specific such that comparison of oxidation kinetics data with
published values [Refs.3, 4, 6, 7] may not show agreement. In order to manage laboratory time
appropriate to the given apparatus, preliminary short runs at each of the 5 specified furnace
temperatures for a single effective oxidation time of 30 min. are recommended as early as
possible to obtain some idea of oxidation rates. This would allow an early test of Arrhenius
relation (15), leaving the test of linearity to the remainder of available lab. time at selected
temperatures but with more accurate time budgets.
1.2. Conventional furnaces generally show a rather flat quiescent temperature profile in the middle
section, i.e. after disturbances such as boat loading/unloading have settled. To quantify the extent
and quality of the flat zone, one can distribute multiple boats and chips along the middle section
and run a dry-oxidation test, which is solely dependent on chip temperature and not oxidant gas
transport. Since dry oxidation is very slow, this only deals with steady state conditions but not
transients.

2. Crucibles
2.1. Alundum crucibles suffer thermal shock upon entering or exiting the furnace especially without a
buffer period (e.g. 1 min.) sitting at the mouth of the furnace. However, parasitic oxidation may
occur during the buffer period even under N2 purge as atmospheric oxygen can diffuse in. While
applying the endcap during the buffer period would limit atmospheric oxygen access, the crucible
and chip temperatures would also be raised, somewhat countering the purpose of minimizing
parasitic oxidation. One might consider a short experiment to assess the significance of such
parasitic oxidation.
2.2. A hot Alundum crucible also suffers thermal shock when unloaded from the furnace tube directly
to the cold porcelain end station. A workable technique is to hold chips at room temperature in a
smaller porcelain crucible, which in turn sits in the Alundum crucible. The porcelain crucible
may serve as a better buffer with lower thermal mass and better thermal conductivity. If two
porcelain crucibles are used alternately, it would greatly facilitate batch sample loading and
unloading without complete withdrawal of the Alundum crucible from the mouth of the furnace
tube.
2.3. Multiple sets of crucibles can be loaded and selectively unloaded from the furnace as long as they
are all comfortably in the flat zone (see 1.2 above).

3. Redundancy
3.1. The threat of chip sample contamination (by particulates) or accidental loss in handling can be
partly met by having multiple samples per run.
3.2. As gas dynamics effects are not obviously discernable, having multiple samples at several sites
(say 3) along the length of the crucible and running them for the same duration would be
informative. See also optional test bb. Mass flow and temperature effects under Section 4.
Experiments.

4. Dynamics of planning and execution

4.1. Given the dynamics of the experiment, it is recommended that chip samples be measured by one
partner as soon as available while the other partner runs the process. The roles can be switched to
broaden experience.
4.2. Data is promptly plotted to gain increasing insight and to identify problems in a timely manner so
plans can be adjusted.

5. Estimate of effective wet oxidation time - Since there are switching transients between carrier and
purge gases when initiating or terminating a wet oxidation cycle, the stop watch and the TC2 trace give
only crude estimates of the actual wet exposure time. By holding a glass slide at the spout of the beaker
endcap to intercept the effluent gas, one can estimate the time delay between appearance (or
disappearance) of condensation and wet O2 ON (or N2 purge OFF). Make systematic corrections.

19
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

APPENDIX G. DETAILS OF WATER SOURCE PNEUMATICS

Flowmeter
(Coarse)
Dry Channel
Flowmeter

Vent 1 Manual
Open
Vent 2 N2
O2 40 psi Purg
N2 65 psi
Dry Pneumatic
e
Open
ON

Thermometers Thermometers

Wet
Wet Channel
Flowmeters

N2

Manual
Closed
STAGE 2 STAGE
1 O2 40 psi

Pneumatic details (Dry N2 Purge Mode)

1. In (dry) N2 purge mode:

1.1. The 4-way valve supplies N2 to the dry channel flowmeters as well as pneumatic pressure to keep
Vent 2 open so moisture from the Stage 2 flask would escape there.
1.2. The manual valve at the dry channel flowmeter can be preadjusted and left manually open.
1.3. The manual valve at the wet channel flowmeter must be manually closed.
1.4. Vent 1 continues to relieve Stage 1 hot water vapour pressure.

2. In (wet) oxidation mode:

2.1. The 4-way valve cuts off N2 supply to the dry N2 purge and allows the pneumatic pressure to
collapse through the valve at Vent 2, which is then normally closed. The moisture carrier gas (wet
O2) then flows only to the furnace tube.
2.2. The manual valve at the wet channel flowmeter must be manually open and adjusted slowly to
the desired flowrate in order to avoid injecting liquid water into the furnace a very important
aspect for safety and cleanliness.
2.3. The steady state oxygen flowrate, affects the degree of cooling in both Stage 1 and 2 flasks, the
water temperature of the latter being most important to sustain. Typical temperature drop in both
flasks due to bubbling is only two degrees C.
2.4. While maximizing the steady state oxygen flowrate operating range, Stage 1 can be
overpressured (against the flow impedance of the Stage 2 bubbler) from both inlet gas pressure
and the vapour pressure of near-boiling water such that Vent 1 is required at least for safety. Vent
1 has a needle valve that can be adjusted for optimum pressure maintenance and relief. Stage 1
being a heat exchanger itself, the adjustment of flowrate and Vent 1, while having an effect on
Stage 1 water temperature, should not impact Stage 2 temperature excessively. The possibility of
adjusting Vent 1 also means the oxygen flowmeter reading (Tube 602, glass or stainless steel
bead) does not uniquely determine carrier gas throughput in the Stage 2.

20
Physics 450/453 Silicon Oxidation Studies Using Ellipsometry Rev. 0504/2010 M. Watt

2.5. However, the apparatus has proved satisfactory with the Vent 1 needle valve wide open hence a
default setting. Under this condition, a maximum carrier gas flow rate (as measured by the
150mm 602 flow tube) of 110 mm by the S.S. float is recommended empirically.

Flowmeter
(Coarse)
Dry Channel
Flowmeter

Vent 1 Manual
open
Vent 2 N2
O2 40 psi Purge
N2 65 psi
Dry Normally
closed OFF

Thermometers Thermometers

Wet
Wet Channel
Flowmeters

N2

Manual
Open
STAGE 2 STAGE
1 O2 40 psi

Pneumatic details (Wet Oxidation Mode)

3. There are thus systematic switching transients in Stage 2 water temperature between idling and
oxidation cycles, which impose lower limits on the duration of both (about 10 min.). The associated
systematic errors in oxidation time and Stage 2 bath temperature are best estimated from datalogged
traces of the latter since an approximate 2 degrees temperature excursion is synchronized with the
oxidation-purge cycling.

21