semi合集-English.pdf - 第838页
Percentage = 10 0 10 (dB/10) . For example, app roximately 94.8% ( –0.232 dB of loss) of the power at 13.56 MH z will be transmitted through a cable assembly made from RG-217 cable with a physical lengt h of 15.24 mete…
output encoder value that corresponds to the position of
a variable tuning element in a Matching Network. For
example, the voltage from a rotary potentiometer on the
rotating shaft of a variable capacitor (the “Tuning
Element”) would be referred to as the capacitor’s
“Position”. In this example, the position/voltage
corresponds to a certain shaft location or position.
6 Ordering Information
6.1 Semiconductor manufacturers may use this
standard when procuring processing equipment to
specify RF power delivery system performance and
documentation. The equipment suppliers may also use
this document to specify the RF system components
and subassemblies.
6.2 Orders for equipment in accordance with this
standard shall include:
6.2.1 This specification number and date of issue.
6.2.2 Any certification showing passage of
qualification tests required to be provided (optional).
6.2.3 Any test results required to be included in reports
to be provided (optional).
7 Requirements
7.1 Semiconductor processing equipment shall be
designed and built with a reliable RF power delivery
system. The requirements defined in this specification
address the integrated RF system as well as the
following RF power delivery system components: the
RF cable assemblies, the matching network, the
generator, and the RF applicator/interface, as well as
specific components of these assemblies or systems.
7.2 The mean (productive) time between failures
(MTBF
p
), as outlined in SEMI E10, of the generator,
matching network, and the RF power delivery system
as a whole shall be provided by the supplier, as well as
the test method used to determine MTBF
p
.
7.3 RF Cable Assemblies (Transmission Lines)
7.3.1 Many RF power delivery systems use RF cable
assemblies to transfer power from the generator to the
input of the matching network and, in some cases, from
the output of the matching network to the process
chamber (e.g., the chuck assembly). All cable
assemblies shall adhere to industry standards for cables
and connectors (e.g., MIL-C-17G, MIL-PRF-39012D,
IEEE-STD-383) in terms of rated power/voltage/
current, allowed bending radius, expected variation in
characteristic impedance, and attenuation at the
expected operating frequency(s) for the specific cable
type (e.g., RG-217, RG-218) per the cable
manufacturer’s specifications. SEMI E114 or an
equivalent test method(s) shall be used to determine the
electrical length, power dissipation (loss), and
characteristic impedance of the RF cable assemblies.
7.3.2 The length of the cable assemblies used to
connect portions of the RF power delivery system shall
be specified in terms of electrical length at the nominal
operating frequency, in addition to the nominal physical
length. The electrical length shall be given in terms of
degrees of phase shift. Variations in the dielectric
properties (e.g., phase velocity, characteristic
impedance) of the cable assemblies can cause the
electrical length to differ between assemblies.
7.3.3 The electrical length of cables used between the
matching network and the chamber (if any) shall be
specified to within 0.25° of the standard electrical
length value at the operating frequency for fixed
frequency systems or at the midpoint frequency for
variable frequency systems, for frequencies less than or
equal to 13.56 MHz. For frequencies above 13.56
MHz, the length should be specified to a value (in
degrees) equal to the frequency in MHz multiplied by
0.018 (e.g., the specification for 60 MHz would be
60 0.018 = 1.1°). For example, at 13.56 MHz, the
physical length of typical cable (RG-217)
corresponding to 0.25° of phase shift (0.5° of reflection
coefficient phase shift) would be approximately 1 cm.
This specification would result in impedance variations
seen by the matching network of less than
approximately 1 percent between processing chambers.
7.3.4 The length of cables used between the matching
network and the generator shall be specified to within
1.0° of the standard electrical length value at the
operating frequency for fixed frequency systems or at
the midpoint frequency for variable frequency systems,
for frequencies less than or equal to 13.56 MHz. For
frequencies above 13.56 MHz, the length shall be
specified to a value (in degrees) equal to the frequency
in MHz multiplied by 0.074 (e.g., the specification for
60 MHz would be 60 0.074 = 4.44°). For example, at
13.56 MHz, the physical length of typical cable (RG-
217) corresponding to 1.0° of phase shift (2.0° of
reflection coefficient phase shift) would be
approximately 4 cm. This specification would
minimize the impedance transformation at the harmonic
frequencies (10° of phase shift at the 10
th
harmonic
frequency).
7.3.5 The expected power dissipation in the cable
assembly(s) supplied with the system shall be provided
as a function of the operating frequency(s) of the
system. The amount of power transmitted through the
cable assembly shall be given in terms of dB (decibels)
and percentage. The relationship between dB and
percentage is given as:
SEMI E113-1104 © SEMI 2001, 2004 3
Percentage = 100 10
(dB/10)
.
For example, approximately 94.8% (–0.232 dB of loss)
of the power at 13.56 MHz will be transmitted through
a cable assembly made from RG-217 cable with a
physical length of 15.24 meters (electrical length of
375.74°). In other words, 5.2% of the power is
dissipated in the cable assembly at 13.56 MHz.
7.4 Matching Networks
7.4.1 Matching networks transform the impedance of
the plasma load to match the impedance of the
generator, which is nominally 50 ohms. The
transformation is achieved by the use of various
reactive components, such as capacitors and inductors,
that “tune” the matching network by transforming the
impedance at the output of the network to the input of
the matching network. For a given frequency, there is a
fixed relationship between the impedance at the input of
the matching network and the impedance at the output
of the matching network (the load impedance). In other
words, the impedance to which the matching network is
matched (the load impedance) can be determined if the
transform properties of the network are known for a
given frequency. Typically, there are two methods of
impedance transformation. The first method uses a
fixed frequency (e.g., 13.56 MHz) and variable tuning
elements, while the second method uses a variable
frequency and fixed tuning elements. Specifications for
both methods are given below. SEMI E115 or an
equivalent test method(s) shall be used to determine the
load impedance and efficiency of matching networks
that are designed to operate at fixed frequency with a
50-ohm input impedance.
7.4.2 For those systems that use fixed frequency
operation, matching networks typically have two
variable tuning elements, which are usually variable
capacitors and/or inductors. These variable tuning
elements typically have an output voltage or encoder
value that corresponds to a certain value of the element
(i.e., an output voltage will correspond to a specific
value of capacitance or inductance in the matching
network). The value of the output voltage or encoder
value is referred to as the tuning element position. To
obtain information on the operation and performance of
the matching network, the tuning element positions
must be provided as outputs.
7.4.3 For those systems that use variable frequency
operation, matching networks typically have fixed
tuning elements (no variation) or tap points between
fixed tuning elements, and the frequency is varied to
obtain the best matched condition. In some cases, the
tap points are also varied for matching. To obtain
information on the operation and performance of these
types of networks, the operating frequency and the
input impedance (or reflection coefficient magnitude
and phase angle) shall be provided. For those systems
that have variable tap points, the tap point position shall
also be provided as an output.
7.4.4 For matching networks run at a fixed frequency
with variable tuning elements, the load impedance shall
be provided as a function of the tuning element
positions. In addition, the efficiency of the matching
network shall be provided as a function of the tuning
element positions. The information shall be provided in
tabular form. The increment of the tuning element
positions shall be in steps equal to or less than 10% of
the full range. For example, a voltage increment of 1
volt or less shall be used in the case where the full
range of the tuning element position is 10 volts. A plot
providing an example of the real part of the load
transformed to 50 ohms by the matching network as a
function of tuning element position indicator is shown
in Figure 1 and a plot of the reactive part of the load
impedance is shown in Figure 2. A plot of the power
efficiency is shown in Figure 3. An example of data to
be provided in tabular form is shown in Table 1. The
required uncertainty of the magnitude of the impedance
is 1.5%, the required uncertainty in the phase angle of
the impedance is 0.35°, and the required uncertainty
of the power efficiency is 2.0%. For example, a load
impedance of 2.0 – j20 ohms would have an uncertainty
in the real part of 0.15 ohms and an uncertainty in the
reactive part of 0.32 ohms.
7.4.5 For matching networks run at a variable
frequency with fixed tuning elements and/or tap points,
the efficiency and load impedance as a function of its
output parameters (frequency, input impedance, and tap
point value (if any)) shall be provided. The information
shall be provided in tabular form. The increment of the
output parameters shall be in steps equal to or less than
10% of the full range. For example, if the operating
frequency parameter is given as an output voltage, a
voltage increment of 1 volt or less shall be used in the
case where the full range of the output parameter is 10
volts. The required uncertainty of the magnitude of the
impedance is
1.5%, the required uncertainty in the
phase angle of the impedance is 0.35°, and the
required uncertainty of the power efficiency is 2.0%.
7.4.6 The tuning element positions/values for each
matching network (of the same type/model) shall be
adjusted to provide consistent efficiency and matching
network input impedance when the matching network is
connected to a fixed load impedance. The load
impedance used shall be one that is typically
encountered by the matching network during processing
and/or shall be an impedance that is near the mid-range
of the tuning space (e.g., 2 – j20 ohms). A simple way
to ensure network-to-network consistency is to adjust
SEMI E113-1104 © SEMI 2001, 2004 4
the tuning element positions/values to have consistent
values between matching networks when connected to a
common/fixed load impedance. The variation in tuning
element positions between matching networks shall be
less than 1% of full scale for a common fixed load
impedance and shall be less than 2% of full scale over
80% of the operating range. For example, if the fixed
load impedance is chosen such that the tuning element
positions on matching network A are both 5 volts (for a
10 volt range), the tuning element positions for
matching network B shall be within 0.1 volts of 5.0
volts.
7.4.7 The maximum power that the system can safely
sustain during nominal, steady state processing
conditions shall be provided as a function of the
positions of the tuning elements (i.e., the plasma load
conditions). The operating specifications of the
components used in the matching network shall be
provided (voltage and current capability), and peak
values of voltage and current shall not exceed 80% of
the rated voltage and current at the operating frequency
for these components over the operating range of the
matching network. This specification does not pertain
to transient (e.g., plasma strike) conditions.
7.4.8 The maximum power the system can safely
sustain without plasma (pre-ignition conditions) shall
be provided to the end user, as well as the maximum
time limit allowed at the peak power.
7.4.9 To reduce variation between matching networks,
the frequency response of the matching network at the
harmonic frequencies shall be repeatable between
networks up to the 5th harmonic frequency (inclusive).
Over the expected operating range of the network, the
reflection coefficient measured at the output of the
matching network, when the input of the network is
terminated with both an open and a short circuit, shall
vary less than 5% in magnitude and phase at the
harmonic frequencies. Comparisons between networks
shall be made at three operating points, which include a
nominal operating point (e.g., at the midpoint of the
tuning range), a point within 10% of the maximum of
the tuning range, and a point within 10% of the
minimum operating range. Unless specifically
designed to, power dissipation in the network of greater
than 10% (reflection coefficient of 0.95) at the
harmonic frequencies shall be avoided.
7.4.10 The matching network shall be able to reach a
stable tuning solution for any given available preset
condition (tuning element position/frequency) in less
than 3 seconds when the network is operated into a test
load that is in the middle of the network’s tuning range.
Measurements of the time to tune shall be taken with
the preset conditions (tuning elements positions/
frequency) varied in increments equal to 20% or less of
full scale. The input power level during this bench test
shall be in the range of 20% to 100% of the nominal
power rating of the matching network.
7.4.11 There shall be only one tuning solution over the
entire tuning range of the matching network. In other
words, multiple tuning points for the same loading
conditions shall be avoided.
7.5 Generators
7.5.1 To maintain consistent process conditions, the
generators providing RF power to establish and sustain
process plasmas and provide required wafer bias
conditions must deliver power to load that can vary in
time and value. In general, testing the generator output
when driving a 50-ohm load is not sufficient to
determine proper operation. In the semiconductor
process environment, the generator may be exposed to
power from the match/plasma system at the harmonics
of the generator output frequency and needs to provide
stable output power when operating into typical
processing conditions.
7.5.2 The generator shall maintain consistent output
impedance at each of the harmonic frequencies. The
value of the reflection coefficient measured at the
output of the generator shall be consistent at each
frequency to within 5% in both magnitude and phase
up to at least the 10
th
harmonic frequency (i.e., up to
10 fundamental frequency) over the lifetime of the
generator. This specification should not be interpreted
as meaning that the reflection coefficient shall be the
same value for all the harmonic frequencies, but the
reflection coefficient shall maintain its particular value
at each particular harmonic frequency to within 5%
over time.
7.5.3 Generators shall deliver power to a nominal 50
ohm fixed load within 2.0% of true power from 10%
to 100% of the maximum rated output power.
7.5.4 Generators shall provide consistent power
delivery with a variation of less than 1.0% of the
requested power level (over time and during steady
state wafer processing conditions) when operated into a
matched load.
7.5.5 Generators shall deliver power with a power
variation of less than 1.5% when operated into a load
impedance with a reflection coefficient of at least 0.33
at any phase angle (VSWR of at least 2.0) from 10% to
100% of the maximum rated output power. For
example, a 25-ohm resistive load with various lengths
of transmission line between the load and the generator
can be used to produce the required phase shift
variation. An example plot of forward power output vs
reflection coefficient phase angle is shown in Figure 4.
SEMI E113-1104 © SEMI 2001, 2004 5