2 D Wfr Meas Avs Oct 2008

Time-resolved2-D wafer surface measurementsfor process optimizationand control

Cal Gabriel and Jeff ShieldsSpansion, Inc.

Greg RocheKLA-Tencor

2 © 2008 Spansion Inc.

Introduction

• Current plasma etching process optimization and mon itoring

– 1-D data provided by tool (reflected power, V dc, optical emission intensity, etc.)

– 2-D data provided by post-etch measurement (critica l dimensions or film thickness changes)

• Better to monitor plasma condition across wafer sur face in near-real-time: KLA-Tencor PlasmaVolt sensor wafer to me asure surface voltage (“V rf”)

– Vrf is influenced by plasma density, sheath voltage, an d frequency

– 2-D array of sensors across the surface of a 300 mm wafer

– Time-resolved—collects data throughout multi-step pl asma process

• We compare V rf with V dc and show effects of process parameter and hardware changes

• Wafer-level V rf measurements are valuable for plasma monitoring, chamber matching, and process optimization


Plasma etch processes and tools

• Processes

– Fluorocarbon-based plasmas

– For etching inorganic dielectric film stacks to for m dual damascene trenches/vias for 45 nm Flash technologie s

• Multi-frequency, capacitively coupled industrial pl asma etching systems

– Tool A: 2, 27, and 60 MHz power supplies connected to lower electrode (grounded upper electrode)

– Tool B: 13.56 MHz connected to lower electrode; 60 MHz connected to upper electrode (with option of increa sing DC bias on upper electrode)


Polyimide layer

Measurement capacitor

wafer

“V RF”

IRF

Plasma

PlasmaVolt wafers

• 300 mm wafers

• Autonomous wireless wafer-level data collection

• Precision instrumented electrical measurements of t he plasma

• Measurement capacitor is connected to RF peak detec tion circuitry

– No ground reference

– Measurement is calibrated in a 13.56 MHz electric f ield


Wafer (V dc)

IRF

“V RF”

Capacitor is in series with sheath,

hence VRF ∝ J0 RF ~ Vs·ω2·ns (McVittie, Titus)

VRF ~ Vs·ns·ω2

where:

Vs is DC component of sheath voltage

ns is plasma density at edge of sheath

ω is RF driving frequency

In practice, VRF responds to source power, bias power, pressure, flow, and other parameters

Surface voltage measurement, “V RF”


Sample PlasmaVolt data

Vary pressure Vary total flow

14 measurementsites (maps areinterpolations)


0

Vdc (~ Vpp/2)

-

+

Vol

tage

Vpp(peak-to-peak voltage)

-- -- -+ + + + + + + + + + + + + + +

-- --- -- -- -

Vdc, Vpp, and ion energy

• Commercial etchers generally display a single volta ge measured at the wafer electrode: V dc or V pp

• RF applied to the wafer electrode causes a self-bia s to develop (to maintain charge balance)

– Voltage waveform shifts negative, becoming positive just long enough to collect electrons that counterbalance the positive ion flux

– If there are no collisions in the sheath, the maxim um ion energy will be close to the DC bias, V dc (or about V pp/2)


2000

3000

4000

5000

6000

7000

8000

9000

10000

0 200 400 600 800 1000 1200 1400

| Vdc | (V)

Vrf

(V

)

Vary 2 MHzpower

Vary 60 or 27 MHz power

Vary pressure

• No—Vrf is affected by plasma density, not just ion energy

– Vrf and Vdc may be directly proportional (vary low frequency power)

– Vrf and Vdc may be inversely proportional relationship (vary pressure)

– Sometimes the relationship is nonlinear (e.g. when mixing powers, which affects both ion energy and density)

• No—Vrf is spatially mapped, not a single measurement

• No—Vrf measurement is independent of etch tool hardware de sign

Is Vrf the same as the wafer electrode voltage reported by the tool?

Vrf


Effect of RF frequency and power

• Increasing power of any frequency raises V rf, but not always V dc

• Non-linear relationship between V rf and Vdc suggests plasma density plays a role in V rf

2500

3000

3500

4000

4500

5000

5500

6000

6500

600 800 1000 1200 1400

| Vdc | (V)

Vrf

(V

) 27 MHz

60 MHz

PlasmaVoltresponding to plasma density

PlasmaVoltresponding to ion energy andplasma density

2500

3000

3500

4000

4500

5000

5500

6000

6500

0 200 400 600 800 1000

RF power (W)

Vrf

(V

)

2 MHz

27 MHz

60 MHz

600

700

800

900

1000

1100

1200

1300

1400

0 500 1000 1500

RF power (W)

| Vdc

| (V

)

27 MHz

60 MHz

60 MHz power affects plasma density, not bias voltage

27 MHz power also affects bias voltage


Effect of chamber pressure

• Decreasing pressure causes higher V dc as expected but lower (and more nonuniform) Vrf

• Vrf vs. Vdc is inversely proportional—V rf is responding to plasma density, not just ion energy

400

450

500

550

600

650

700

750

800

0 50 100 150 200 250 300

Pressure (mTorr)

| Vdc

| (V

)

4000

4500

5000

5500

6000

6500

7000

7500

8000

0 50 100 150 200 250 300

Pressure (mTorr)

Vrf

(V

)

-1s

+1s

4000

4500

5000

5500

6000

6500

7000

7500

8000

400 500 600 700 800

| Vdc | (V)

Vrf

(V

)

0%

2%

4%

6%

8%

10%

12%

0 50 100 150 200 250

Chamber pressure (mTorr)

Vrf

non

unifo

rmity

(%

1s)

2 D Wfr Meas Avs Oct 2008

Technology

Transcript of 2 D Wfr Meas Avs Oct 2008