Upload
others
View
15
Download
2
Embed Size (px)
Citation preview
R. B. Darling / EE-527 / Winter 2013
EE-527: MicroFabrication
Positive Photoresists
R. B. Darling / EE-527 / Winter 2013
Advantages of Positive Photoresists
• They are more commonly used in the IC industry. • They are superior to negative photoresists because:
– They do not swell during development. – They are capable of finer resolution. – They are reasonably resistant to plasma processing operations such
as dry etching.
R. B. Darling / EE-527 / Winter 2013
Phenolic Resins - 1• Phenolic resins are condensation polymers of aromatic
alcohols and formaldehyde. • Bakelite was the first thermosetting plastic. • Phenolic resins are readily cross-linked by thermal
activation into rigid forms. • Most phenolic resins are readily dissolved by aqueous
alkaline solutions: – NaOH; contains Na+ so it is NOT CMOS compatible. – KOH; contain K+ so it is NOT CMOS compatible. – NH4OH; it is CMOS compatible, but it is hard to keep at a constant
concentration due to evaporation of NH3. – TMAH (CH3)4NOH; it is CMOS compatible and the most
common base for use in positive photoresist developers.
R. B. Darling / EE-527 / Winter 2013
Phenolic Resins - 2OH
HC
H
O
phenol
formaldehyde
OH OH OHH2C
H2C
H2C
bakelite
HO
Hwater
+
+
R. B. Darling / EE-527 / Winter 2013
Phenolic Resins - 3OH
HC
H
O
formaldehyde
OH OH OHH2C
H2C
H2C
HO
Hwater
+
+
CH3
CH3 CH3 CH3
para-cresol
novolac
R. B. Darling / EE-527 / Winter 2013
Important Properties of the Base Phenolic Resin
• average molecular weight– typically in the range of 1000 to 3000 g/mole– (8 to 25 repeating units in the polymer chain)
• dispersity of the molecular weights• isomeric composition of the cresols
– ortho-cresol– meta-cresol– para-cresol
• relative position of the methylene linkages (--CH2--)
R. B. Darling / EE-527 / Winter 2013
Cresol IsomersOH
OHH2C
CH3
CH3
para-cresol
OHOH
CH3
CH3
ortho-cresol meta-cresol
OH
CH3
OH
CH3
H2C
H2C
ortho-cresol resin meta-cresol resin para-cresol resin
R. B. Darling / EE-527 / Winter 2013
Cresol Isomer Properties
• Single isomers and smaller molecular weights are desirable. • The manufacture of positive photoresist relies heavily upon
obtaining only a single isomer of the resin, usually para-cresol.
• Each monomer is [C8H8O] (120.151 g/mole)
Isomer Methylene Link Molecular Weight Dissolution Rate Plastic Flow Temp.
ortho-cresol 3 2100 g/mole 2.7 A/sec 85 C
meta-cresol 1 15000 g/mole 0.7 A/sec 73 C
para-cresol 1 1600 g/mole 3.0 A/sec 119 C
R. B. Darling / EE-527 / Winter 2013
Photoreaction in a Positive Photoresist
O
N2
COOH
+
+
H2O
N2
h
diazonaphthaquinone (DQ)
indene carboxylic acid (ICA)
R. B. Darling / EE-527 / Winter 2013
Dissolution of Phenolic Resins - 1
• Because of the OH groups, phenolic resins are hydrophilic and are readily dissolved by aqueous alkaline solutions.
• Diazonaphthaquinone (DQ) is a hydrophobic and non-ionizable compound.
• When phenolic resins are impregnated with DQ, they become hydrophobic and their dissolution is greatly inhibited.
• After exposure, DQ is converted into indene carboxylic acid (ICA) which is hydrophilic and very ionizable. – This allows the developer to wet and penetrate the novolac resin.
• Phenolic resins which contain ICA instead of DQ are readily dissolved by aqueous alkaline developers.
R. B. Darling / EE-527 / Winter 2013
Dissolution of Phenolic Resins - 2
0 10 20 30 40
0.1
1.0
10
100
1000
exposed novolac + DQ
unexposed novolac + DQ
DQ concentration, weight percent
Dis
solu
tion
rate
, nm
/sec
in 0
.15
M N
aOH
R. B. Darling / EE-527 / Winter 2013
DQ Primary PhotoreactionO
N2
O
N2+
+ H2O
CO COOH
diazonaphthaquinone (DQ) a carbene
a ketene indene carboxylic acid (ICA)
h
ring contraction
hydration
photolysis
See Otto Suess, 1944 and 1947 papers in Annalen.
R. B. Darling / EE-527 / Winter 2013
DQ Side Reactions - 1
CO
a ketene
+ HO CH3
H2C
novolac resin
C
O
O CH3
H2C
cresol ester
Lack of humidity in a clean room will dehydrate the photoresist and promote this reaction.
This is one reason why positive photoresist is so sensitive to humidity! Ketene scavengers, e.g. amines, are sometimes added to reduce this.
The ketene can form cross-links with the novolac if insufficient water is present-- just the opposite of what is desired...
R. B. Darling / EE-527 / Winter 2013
DQ Side Reactions - 2
+ OHH3C
CH2
novolac resin
OHH3C
CH2
O
N2
diazonaphthaquinone (DQ)
O
N N
red azodye
This reaction always occurs to some extent, it causes the red color of photoresist, and it is benign as long as the DQ content is not overly depleted.
This reaction is more prevalent in the unirradiated areas where the DQ has not been consumed by the photolysis reaction.
This reaction is one of the basic shelf-life limits to positive photoresist.
R. B. Darling / EE-527 / Winter 2013
DQ Side Reactions - 3
+
H2
COOH
indene carboxylic acid (ICA)
CO2
indene
+ 2 CO2 +indene dimer
Decarboxylation of ICA occurs when the exposed photoresist is heated, and this usually occurs as a normal part of the process during postbake.
The decarboxylated indene or its dimer are both hydrophobic and non-ionizable, so these once again act as a dissolution inhibitor, although now they are no longer photosensitive.
This reaction can be used as the basis for an image reversal process, especially when it is catalyzed by a compound such as imidazole.
N
Nimidazole (C3H3N2)
R. B. Darling / EE-527 / Winter 2013
DQ BallastingO
N2
SO2
O
O
N2
SO2
O
OR
diazoquinone sulfonyl groups (DQO-)
C
O
polyhydroxybenzophenone (dibenzoketone)
C
O
DQO
DQO DQOtri-diazonaphthaquinone sulfonyl benzophenone
Most positive photoresists use a derivative of the basic DQ photosensitive dissolution inhibitor. Diazoquinone sulfonyl groups (DQO) are added to a “ballast” compound such as dibenzoketone to produce a higher molecular weight compound such as tri-diazonaphthaquinone sulfonyl benzophenone.
This is the most common photosensitive dissolution inhibitor which is used in common photoresist families such as AZ-1300, AZ-1500, AZ-4000, Microposit 1300, Microposit 1400, and Microposit 1800.
R. B. Darling / EE-527 / Winter 2013
Positive Photoresist Image Reversal Process1. Standard masked exposure and conversion of DQ into ICA:
2. Baking and imidazole catalyst decarboxylates the ICA into indene:
3. Flood expose to convert the remaining DQ into ICA:
4. Development will dissolve away the regions which still contain ICA:
CO2 CO2
This is very useful when a negative image and an undercut resist profile are desired.
R. B. Darling / EE-527 / Winter 2013
Physical Requirements on the Photoactive Component
• Need an overlap of the absorption spectrum with the emission spectrum of the exposure source, e.g. a Hg lamp.
• Need bleachability at the exposure wavelength so that the photoreaction is able to reach the resist-substrate interface.
• Need compatibility with the base resin (novolac) so that the two form a single, miscible phase.
• Need thermal stability so that the photoactive dissolution inhibitor does not break down at prebake temperatures.
• Photoactive dissolution inhibitors are often modified to alter their spectral absorption, thermal stability, and miscibility characteristics.
R. B. Darling / EE-527 / Winter 2013
Spectral Absorption of Novolac, DQ, and ICA
200 300 400 500 600
1.0
10
100
1000
10,000
Wavelength, , nm
313
365
405
437
Hg arc lamp linesnovolac
DQ (unexposed photoinhibitor)
ICA (exposed photoinhibitor)
R. B. Darling / EE-527 / Winter 2013
Primary Components of a Positive Photoresist• Non-photosensitive base phenolic resin (~48% by weight)
– usually novolac
• Photosensitive dissolution inhibitor (~4% by weight)– usually a DQ-derived compound
• Coating solvent (~48% by weight)– usually a mixture of:
• ~80% 2-ethoxyethyl acetate, this is the hazardous component• ~10% n-butyl acetate, and • ~10% xylene.
• Above weight ratios are typical for a 1-2 m thick resist viscosity. – Thicker resists will have less coating solvent, thinner resists more.
R. B. Darling / EE-527 / Winter 2013
Positive Photoresist Coating Solvents
• 2-ethoxyethyl acetate, (Cellosolve® acetate)– CH3COOCH2CH2OCH2CH3, flammable and problematic (see the MSDS!)– Is the majority (~80-90%) of most common positive photoresist solvent– OSHA PEL = 100 ppm TWA– ACGIH TVL = 5 ppm TWA, a well-known carcinogen and blood poison!
• n-butyl acetate, – CH3COOCH2CH2CH2CH3, flammable but not very toxic
• xylene, – CH3C6H4CH3, flammable but not very toxic
• alternative solvents to 2-ethoxyethyl acetate: – 2-ethoxyethanol, CH3CH2OCH2CH2OH, a glycol ether– ethyl acetate, CH3COOCH2CH3, a fruity flavored ester, low toxicity
R. B. Darling / EE-527 / Winter 2013
Secondary Components of a Positive Photoresist
• Antioxidants• Radical scavengers• Amines to absorb O2 and ketenes• Wetting agents• Dyes to alter the spectral absorption characteristics• Adhesion promoters• Coating aids
R. B. Darling / EE-527 / Winter 2013
What To Do With Waste Positive Photoresist• The hazardous component in positive photoresist is the
2-ethoxyethyl acetate solvent. – 99.9% of this is removed during spinning. USE A FUME HOOD!– Nearly all of the remainder is removed during prebake.
• Dried photoresist contains only novolac resin and some combination of DQ and ICA, all of which are comparatively innocuous.
• Disposal depends upon whether the resist is wet or dry: – Excess wet resist is poured into marked bottles and labeled as hazardous
waste. – Pipettes containing wet resist are collected and marked as hazardous
waste. – Resist slung off into the spinner bowl is allowed to dry under the fume
hood; afterwards, the resist soiled foil can go into the trash. – Wipers containing wet resist and/or acetone are allowed to dry under the
fume hood; afterwards, the resist soiled wipers can go into the trash.
R. B. Darling / EE-527 / Winter 2013
Sensitometric Curve for a Positive Photoresist
1 10 100 1000 10,000
0.0
0.5
1.0
Exposure Dose, D, mJ/cm2
Developed Resist Thicknessnormalized to 1.0
working pointfor the resist
R. B. Darling / EE-527 / Winter 2013
Latent Image Formation - The Dill Equations
• M(z,t) = inhibitor fraction remaining, ranges from 0 to 1. • I(z,t) = radiation intensity.
BtzMAtzIz
tzI
CtzMtzIt
tzM
),(),(),(
),(),(),( C expresses the photoreaction speed.
[AM(z,t) + B] plays the role of theoptical absorption coefficient.
Initial Conditions:
M(z,0) = 1
I(z,0) = I0e-(A+B)z
Boundary Conditions:
I(0,t) = I0
M(0,t) = e-I0Ct
The {A,B,C} Dill parameters characterize a given positive photoresist.
This image formation model was developed by Fred Dill at IBM Corp.
R. B. Darling / EE-527 / Winter 2013
Bleaching of a Positive Photoresist – 1– The solution to the coupled Dill equations predicts a sharp
boundary between exposed and unexposed regions of the resist. The boundary is the front of a bleaching edge which propagates downward to the substrate as the resist is exposed. This makes the wall angle more dependent upon the {A,B,C} Dill parameters than upon the exposure wavelength, and gives positive photoresists very high resolution.
z
0
d
0 1 M(z)
z
0
d
0 I0 I(z)
Inhibitor Fraction Radiation Intensity
substrate
bleachedregion
positive photoresist
The exposing light bores an optical hole through the photoresist, creating steeper sidewalls than what normal diffraction limits would allow.
R. B. Darling / EE-527 / Winter 2013
Bleaching of a Positive Photoresist – 2• The photo-bleaching of positive photoresists may appear
as a subtle point, but it has had huge consequences in the microelectronics industry.
• Physical optics calculations of the printing resolution show that a limit of about 0.65 µm should be reached with i-line exposure. Yet the semiconductor industry has used i-line exposure systems all the way down to 90 nm CMOS. How is this possible?
• The answer is that the photoresist bleaches! • This nonlinear exposure characteristic has “rescued”
photolithography across many technology nodes. • Good physical chemistry can break the laws of physics…
R. B. Darling / EE-527 / Winter 2013
Bleaching of a Positive Photoresist – 3
• And now for the subtle points: • The latent image in an exposed positive photoresist (prior
to development) is slightly visible. – This is once again a consequence of the resist bleaching.
• The photobleaching process is another way of creating an optical ROM.
R. B. Darling / EE-527 / Winter 2013
The Dill Parameters
• Typical parameters for an AZ-1500 series photoresist: – A = 1.00 m−1
– B = 0.20 m−1
– C = 0.025 cm2/mJ
• Both real and imaginary parts of the refractive index are changed by bleaching: – unbleached: n = 1.7123 k = 0.0358– bleached: n = 1.6994 k = 0.0058
• Optical absorption coefficient: – unbleached: M = 1, = A + B– bleached: M = 0, = B
• Bleaching produces ~5-6 fold increase in transparency.
k4
@ = 365 nm
R. B. Darling / EE-527 / Winter 2013
Cauchy Coefficients
• The Cauchy coefficients provide a good empirical fit to the refractive index of a photoresist resin versus wavelength:
43
22
1 NNNn
N1 N2 (m2) N3 (m4)
unbleached 1.5996 0.013498 0.000188
bleached 1.5966 0.003758 0.002450
Typical values for an AZ-1500 series positive photoresist:
R. B. Darling / EE-527 / Winter 2013
Novolac Dissolution – 1
• A minimum concentration of [OH-] is required to produce a net forward rate:
H3C OH
CH2
+ OH-
H2O+H3C O-
CH2
The dissolution rate is R = kCn, where C is the base concentration.
For NaOH solutions, R = (1.3 x 105) [Na+]1 [OH-]3.7 Angstroms/second.
R. B. Darling / EE-527 / Winter 2013
Novolac Dissolution – 2
Dissolution Rate, Angstroms/secondSolutionUnexposed Exposed
0.15 M NaOH 20 14000.15 M KOH 10 860
0.15 M NaOH +0.1 M Na2SiO3
270 3400
0.15 M NaOH +0.1 M Na3PO4
350 2800
0.15 M NaOH +0.1 M Na2CO3
270 2400
Typical data for different developer solutions:
R. B. Darling / EE-527 / Winter 2013
Novolac Dissolution - 3• NaOH and KOH based developers are not compatible with
CMOS processing, due to Na+ and K+ contamination of gate oxides.
• NH3OH based developers are difficult to maintain because of their evaporation of NH3.
• CMOS-compatible positive photoresist developers usually used tetramethyl ammonium hydroxide (TMAH) as the base, (CH3)4NOH. – TMAH is an anisotropic etchant for silicon, but it has a very low
etch rate on SiO2 and Si3N4 and Aluminum. – Most common positive photoresist developers contain ~2.4%
TMAH by weight to give a 0.261 Normal solution. – Development times are ~60-120 sec. @ 20-25C for 1-2 m of PR. – TMAH has recently been found to be a neurotoxin!
R. B. Darling / EE-527 / Winter 2013
Developing Methods• Automated
– Spray– Spray & Puddle
• Manual– Puddle– Immersion
• Spray & puddle development are performed on a spinner chuck. – Spray performed @ ~100-200 rpm– Puddle performed with chuck stopped– Rinsing with DI water performed @ ~100-200 rpm– Drying performed @ ~2000-3000 rpm
• Spray is only useful for automated systems – too messy to do by hand. • Immersion requires much more developer, but provides accurate stop
and start timing control.
R. B. Darling / EE-527 / Winter 2013
Positive Photoresist Exposure Latitude
-0.4
0.0
+0.4
Exposure Dose, D, mJ/cm2
-0.2
+0.2
Critical Dimension Shift, m
200180160 220150 170 190 210 230
Exposure Latitude
BLOAT
SHRINK
lines and islands
spaces and windows
Kodak Micro Positive 820
30 min. @ 95 C prebake in convection oven
30 sec. @ 71 C develop with agitation, 1:2 mix
working point: 185 mJ/cm2
R. B. Darling / EE-527 / Winter 2013
Positive Photoresist Prebake Latitude
-0.4
0.0
+0.4
Prebake Temperature, degrees C
-0.2
+0.2
Critical Dimension Shift, m
Prebake Latitude
BLOAT
SHRINK
lines and islands
spaces and windows
Kodak Micro Positive 820
185 mJ/cm2 exposure
30 sec. @ 71 C develop with agitation, 1:2 mix
working point: 95 C
90 95858075 100 105 110 115
R. B. Darling / EE-527 / Winter 2013
Photoresist Tuning – 1• While standard recipes exist and are a good starting point,
optimal image formation and development requires small adjustments to prebake time, exposure time, and development time.
• The optimal photoresist recipe is pattern dependent! • 50:50 line and space patterns are often used to tune resist
since they provide a quick visual indication of the process. • Usually the process conditions are kept constant (prebake
temperature, exposure intensity, developer concentration), and the three key times are adjusted: – 1. Development time, tD
– 2. Exposure time, tE
– 3. Prebake time, tP
R. B. Darling / EE-527 / Winter 2013
Photoresist Tuning – 2A typical line and space pattern set for tuning the photoresist:
R. B. Darling / EE-527 / Winter 2013
Photoresist Tuning – 3• If the line/space pattern comes out over or under, which
knob do you turn? • Essential considerations:
– The prebake needs to be long enough to drive off all of the coating solvent but not too long where it starts to thermally cross-link.
– The exposure needs to bleach the resist all the way to the bottom, but not further.
– The development needs time to clear the bottom of the features, and the edges and corners, but not so long as to etch laterally.
• While detailed theory can provide some guidance, in the lab the empirical approach often wins.
• Design of Experiments: (DOE) – Run a permutation of trials from a nominal operating point to
explore which direction produces the best results.
R. B. Darling / EE-527 / Winter 2013
Single Component Positive Photoresists
• Use a photosensitive resin. • Radiation produces chain scission, rendering region
soluble to a developer.
RO2S R'
R SO2+ + R'
R+ SO2+ R'++
e-beam radiation
e-beam radiation
H2C
HC
CH2
CH3
O2S
npolybutene-1-sulfone
Good for research; generally not yet mainstream for microfabrication.
R. B. Darling / EE-527 / Winter 2013
Stripping Positive Photoresist
• High processing temperatures and irradiation can induce excessive thermal cross-linking of a photoresist, making it very difficult to remove.
• Subsequent processing relies heavily upon the complete removal of all prior photoresist and other organic residues. – Organic residues must never enter furnace tubes!
• Chemically-based stripping is used to achieve most of this. • Oxygen plasma ashing is used to achieve final cleaning.
– UV + ozone (O3) is also becoming popular for this.
R. B. Darling / EE-527 / Winter 2013
Positive Photoresist Strippers - 1
• Acetone, CH3COCH3– Generally used only in research labs, but very safe and effective
when the photoresist is still fairly soft. Requires liberal flushing!
• Piranha Etch, 4:1 H2SO4 : H2O2 @ 90C– Extremely aggressive etch that will remove all organics, but it has
a rather short pot life and must be handled very carefully.
• Cyantek Nanostrip or Nanostrip 2X– Usually a better alternative to piranha etch: safer, long storage life,
premixed, can be used either at room temperature or heated up to ~90C, if needed.
• Sulfuric acid H2SO4 90%• Peroxymonosulfuric acid H2SO5 5%• Hydrogen peroxide H2O2 <1%• Water H2O 5%
R. B. Darling / EE-527 / Winter 2013
Positive Photoresist Strippers - 2
• Alkaline Strippers– AZ Kwikstrip, 300T, 400T positive photoresist strippers– Based on TMAH, (CH3)4NOH – They behave like very strong developers. – They are comparatively safe to use.
• Phenol-based Strippers– Indus-Ri-Chem J-100 was the industry workhorse for several decades.
• It is now marketed by EKC. – It is composed of phenol, C6H5OH, (a solid crystal, also known as
carbolic acid) dissolved into another solvent such as TCE or TCA– It is quite aggressive as a solvent, but has fairly high toxicity.
• A fume hood is mandatory for its use!– It has the advantage of long pot life (several weeks) in the laboratory.
R. B. Darling / EE-527 / Winter 2013
Plasma Ashing• When simple chemical treatments fail, plasma ashing is
usually the technique used to strip stubborn photoresist and its residues. – It is frequently used as a final clean up step, regardless.
• Plasma ashing is (dry) plasma etching using oxygen. – Barrel (full cassette) or single wafer chambers are most common. – Operating pressures are typically 1-15 mTorr. – Relatively low RF power is required, ~50-200 Watts. – Cycle times are also fairly short, ~5-15 minutes.
• Ashing is a bit of a misnomer. – No “ash” is left behind. – The O= oxygen radicals are extremely reactive with organic
hydrocarbons, usually producing volatile H2O and CO2 which gets removed in the vapor phase by the vacuum pump.