`Projected-Capacitive
`Touch Technology
`
`Geoff Walker
`Senior Touch Technologist
`Intel Corporation
`
`June 1, 2014
`
`Must use exact
`capitalization!
`
`File Download: www.walkermobile.com/Touch_Technologies_Tutorial_Latest_Version.pdf
`
`DISPLAY WEEK ‘14
`
`v1.2
`
`1
`
`TPK 2008
`Wintek v. TPK Touch Solutions
`IPR2013-00567
`
`
`
`Agenda
`
` Introduction
` Basic Principles
` Controllers
` Sensors
` ITO-Replacement Materials
` Modules
` Embedded
` Large-Format
` Stylus
` Software
` Conclusions
` Appendix A: Historical Embedded Touch
`
`DISPLAY WEEK ‘14
`
`2
`
`
`
`Introduction
` P-Cap History
` P-Cap Penetration
` P-Cap by Application
` Touch User-Experience
`
`File Download: www.walkermobile.com/Touch_Technologies_Tutorial_Latest_Version.pdf
`
`DISPLAY WEEK ‘14
`
`3
`
`Must use exact
`capitalization!
`
`
`
`P-Cap History
`
`Company
`UK Royal Radar
`Establishment
`(E.A. Johnson)
`CERN (Bent Stumpe)
`
`Dynapro Thin Films
`(acquired by 3M Touch
`Systems in 2000)
`Zytronic (first license from
`Ronald Binstead, an
`inventor in the UK)
`
`Visual Planet (second
`license from Ronald
`Binstead)
`Apple
`
`Significance
`First published application of transparent
`touchscreen (mutual-capacitance p-cap on
`CRT air-traffic control terminals)
`Second published application of mutual-
`capacitance p-cap (in the control room of
`the CERN proton synchrotron)
`First commercialization of mutual-
`capacitive p-cap (renamed as Near-Field
`Imaging by 3M)
`First commercialization of large-format
`self-capacitive p-cap;
`first commercialization of large-format
`mutual-capacitive p-cap
`Second commercialization of large-format
`self-capacitive p-cap
`
`Year
`1965
`
`1977
`
`1995
`
`1998
`
`2012
`
`2003
`
`First use of mutual-capacitive p-cap in a
`consumer electronics product (the iPhone)
`
`2007
`
`
`
`DISPLAY WEEK ‘14
`
`4
`
`
`
`P-Cap Penetration
`
`% of Units Shipped
`
`Embedded
`= P-Cap
`
`Source: DisplaySearch Touch-Panel Market Analysis Reports 2008-2014
`
`DISPLAY WEEK ‘14
`
`5
`
`
`
`P-Cap Forecast by Application…1
`(Consumer)
`
`Million Units
`120
`
`100
`
`80
`
`60
`
`40
`
`20
`
`0
`
`PDA
`Desktop Monitor
`Video Camera
`All‐in‐one PC
`Portable Game
`Still Camera
`EPD eReader
`Media Player
`Smart Watch
`Navigation Device
`Notebook PC
`
`2018
`2017
`2016
`2015
`2014
`2013
`2012
`2018: Phones = 1.8 Billion Units; Tablets = 447 Million Units
`Source: DisplaySearch Touch-Panel Market Analysis Report 1Q-2014
`
`DISPLAY WEEK ‘14
`
`6
`
`
`
`P-Cap Forecast by Application…2
`(Commercial)
`
`Million Units
`8.0
`
`7.0
`
`6.0
`
`5.0
`
`4.0
`
`3.0
`
`2.0
`
`1.0
`
`0.0
`
`2012
`2013
`2014
`2015
`2016
`2017
`2018: Automobile Monitor = 42 Million Units
`Source: DisplaySearch Touch-Panel Market Analysis Report 1Q-2014
`
`2018
`
`DISPLAY WEEK ‘14
`
`7
`
`Education/Training
`Point of Interest
`Ticketing/Check‐in
`Casino Game
`Medical Equipment
`ATM Machine
`Office Equipment
`Retail and POS/ECR
`Factory Equipment
`
`
`
`P-Cap Defines the Standard
`for Touch User-Experience
` Smartphones and tablets have set the standard
`for touch in SEVERAL BILLION consumers’ minds
` Multiple simultaneous touches
`(robust multi-touch)
` Extremely light touch (zero force)
` Flush surface (“zero-bezel”
`or “edge-to-edge”)
`
` Excellent optical performance
` Very smooth & fast scrolling
` Reliable and durable
` An integral part of the
`device user experience
`
`DISPLAY WEEK ‘14
`
`8
`
`Source: AP / NBC News
`
`
`
`Basic Principles
` Self Capacitive
` Mutual Capacitive
` Mutual Capacitive Electrode Patterns
`
`DISPLAY WEEK ‘14
`
`9
`
`
`
`Self-Capacitance
`
` Capacitance of a single electrode to ground
` Human body capacitance increases the capacitance
`of the electrode to ground
` In a self-capacitance sensor, each electrode is measured
`individually
`
`Source: The author
`
`DISPLAY WEEK ‘14
`
`10
`
`
`
`The Problem with Self-Capacitance
`
` Touches that are diagonally separated produce
`two maximums on each axis (real points & ghost points)
` Ghost points = False touches positionally related to real touches
`
`Self Capacitance
`
`Mutual Capacitance
`
`Source: Atmel
`
`DISPLAY WEEK ‘14
`
`11
`
`
`
`Self-Capacitance and
`Pinch/Zoom Gestures
` Use the direction of movement of the points rather
`than the ambiguous locations
`
`Y3
`
`Y2
`
`Y1
`
`X1
`
`X2
`
`X3
`
`X4
`
`Source: The author
`
`DISPLAY WEEK ‘14
`
`12
`
`
`
`Self-Capacitance Electrode Variations
`
`20 measurements
`
`Source: 3M
`
`20 measurements
`
` Multiple separate pads
`in a single layer
` Each pad is scanned
`individually
`
` Rows and columns of electrodes
`in two layers
` Row & column electrodes are
`scanned in sequence
`
`DISPLAY WEEK ‘14
`
`13
`
`
`
`Self-Capacitance
`Advantages & Disadvantages
`
`Self-Capacitive Advantages Self-Capacitive Disadvantages
`Simpler, lower-cost sensor
`Limited to 1 or 2 touches with ghosting
`Can be a single layer
`Lower immunity to LCD noise
`Long-distance field projection
`Lower touch accuracy
`Can be used with active guard Harder to maximize SNR
`Fast measurement
`
`
` Where it’s used
` Lower-end smartphones and feature-phones with touch
`● Becoming much less common due to single-layer p-cap
` In combination with mutual capacitance to increase capability
`
`
`
`
`
`DISPLAY WEEK ‘14
`
`14
`
`
`
`Self-Capacitance for Hover
`
` Self-capacitance is used to produce “hover”
`behavior in some smartphones (in addition to
`mutual-capacitance for contact-touch location)
` Also used for automatically detecting glove vs. fingernail vs. skin,
`and for dealing with water on the screen
`
`Source: Panasonic
`
`Source: Cypress
`
`DISPLAY WEEK ‘14
`
`15
`
`
`
`Multi-Touch Self-Capacitance
`Using Active Guard Concept…1
` Guarding is a well-known technique for reducing the
`effects of electrical current leakage
`
`Source: Fogale
`
`DISPLAY WEEK ‘14
`
`16
`
`
`
`Multi-Touch Self-Capacitance
`Using Active Guard Concept…2
` Another contender: zRRo
`
`3D single-touch
`for smartphones
`
`3D multi-touch
`for smartphones
`and tablets
`
`DISPLAY WEEK ‘14
`
`Source: zRRo
`
`17
`
`
`
`Mutual Capacitance
`
` Capacitance between two electrodes
` Human body capacitance “steals charge” which decreases
`the capacitance between the electrodes
` In a mutual-capacitance sensor, each electrode intersection
`is measured individually
`
`Source: The author
`
`DISPLAY WEEK ‘14
`
`18
`
`
`
`Mutual Capacitance
`Electrode Patterns…1
` Rows and columns of
`electrodes in two layers
`
` In the real world…
` “Bar and stripe”, also called
`“Manhattan” or “Flooded-X”
`(LCD noise self-shielding)
`
`11 x 9 = 99 measurements
`
`Source: 3M
`
`DISPLAY WEEK ‘14
`
`19
`
`4 x 10 = 40 measurements
`
`Source: Cypress
`
`
`
`Mutual Capacitance
`Electrode Patterns…2
` Interlocking diamond pattern
`with ITO in “one layer” with bridges
`4.5 mm typical
`
`Source: 3M
`
`Source: The author
`
`DISPLAY WEEK ‘14
`
`20
`
`
`
`More On Mutual Capacitance…1
`
` BTW, there isn’t just one mutual capacitance…
`
`Source: Cypress
`
`DISPLAY WEEK ‘14
`
`21
`
`
`
`More On Mutual Capacitance…2
`
` And there are more capacitors than just the Cm’s…
`
`Source: Cypress
`Source: ELAN, modified by the author
`
`DISPLAY WEEK ‘14
`
`22
`
`
`
`More On Mutual Capacitance…3
`
`Mutual-Capacitive Advantages Mutual-Capacitive Disadvantages
`2 or more unambiguous touches More complex, higher-cost controller
`Higher immunity to LCD noise
`2 layers (or 1 with bridges) for >3 pts
`Higher touch accuracy
`
`More flexibility in pattern design
`
`Easier to maximize SNR
`
`
` Where it’s used
` Mid & high-end smartphones, tablets,
`Ultrabooks, AiOs, commercial products
`● Standalone self-capacitive is becoming increasingly rare
`in consumer electronics (except for buttons)
` With “true single-layer” sensors in low-end smartphones
`
`
`
`
`
`DISPLAY WEEK ‘14
`
`23
`
`
`
`Mutual Capacitance
`Electrode Patterns…3
` Bars & stripes require bridges too…
`
`DISPLAY WEEK ‘14
`
`Source: Synaptics
`
`24
`
`
`
`Mutual Capacitance
`Electrode Patterns…4
` And so does this unusual diamond pattern…
` 102, 106, 108, 210
`● Drive (X) electrodes
` 114 & 202
`● Sense (Y) electrodes
` 110
`● Bridges
` 120 & 230
`● Dummy (floating) ITO
` 200 & 206
`● Optional dummy ITO
` 212
`● Blank (no ITO)
`
`Source: STMicro
`
`DISPLAY WEEK ‘14
`
`25
`
`
`
`Mutual Capacitance
`Electrode Patterns…5
` Claimed advantages of this particular
`pattern over traditional interlocking diamond
` Reduction in sense electrode area reduces LCD noise pickup
` “Finger projections” (0.1 – 0.2 mm) increase the perimeter of
`interaction between drive and sense electrodes, which
`increases sensitivity
` Linearity is improved due to more uniform coupling across channels
` Floating separators aid in increasing the fringing fields, which
`increases sensitivity
`
`DISPLAY WEEK ‘14
`
`26
`
`
`
`Mutual Capacitance
`Electrode Patterns…6
` Holy Grail: True single-layer mutual capacitance sensor
` “Caterpillar” pattern
` Everybody’s single-
`layer patterns are
`proprietary
` Requires fine
`patterning, low sheet
`resistance & low
`visibility
` Benefits: Narrow
`borders, thin stack-
`ups, lower cost, can
`reliably handle 2-3
`touches
`
`Source: Synaptics
`
`DISPLAY WEEK ‘14
`
`27
`
`
`
`Mutual Capacitance
`Electrode Patterns…7
` ELAN’s caterpillar pattern
`
`DISPLAY WEEK ‘14
`
`Source: ELAN
`
`28
`
`
`
`Mutual Capacitance
`Electrode Patterns…8
` An alternative true single-layer pattern from ELAN
` This is a very small portion
`of a much larger sensor
`
`DISPLAY WEEK ‘14
`
`29
`
`Source: ELAN
`
`
`
`Controllers
` Architecture
` Touch Image Processing
` Key Characteristics
` Signal-to-Noise Ratio
` Noise Management
` Innovation Areas
` Suppliers
`
`DISPLAY WEEK ‘14
`
`30
`
`
`
`Mutual Capacitance
`Touch System Architecture
`
`Source: The author
`
` Making X*Y measurements is OK, but it’s better
`to measure the columns simultaneously
` Controllers can be ganged (operate in a
`master-slave relationship) for larger screens
`
`DISPLAY WEEK ‘14
`
`31
`
`
`
`Touch Image Processing
`
`Raw data including noise
`
`Filtered data
`
`Gradient data
`
`Touch region coordinates
`and gradient data
`
`Touch regions
`
`“10 fingers,
`2 palms
`and
`3 others”
`
`Source: Apple Patent Application #2006/0097991
`
`DISPLAY WEEK ‘14
`
`32
`
`
`
`Key Controller Characteristics…1
`
` Node count (x channels + y channels)
` Given typical electrode spacing of 4.5 to 5 mm, this determines
`how large a touchscreen the controller can support (w/o ganging)
` Scan rate
` Frames per second (fps) – faster reduces latency for a better UX
` Windows logo requires 100 fps; Android is unspecified
` Signal-to-noise ratio (SNR)
` More info on upcoming slides
` Operating voltage & current
` OEMs continue to request lower-power touchscreen systems
` Win8 “Connected Standby” is a significant influence
` Internal core (micro/DSP)
` Varies from small 8-bit micro to ARM-7 or higher
`
`DISPLAY WEEK ‘14
`
`33
`
`
`
`Key Controller Characteristics…2
`
` Number of simultaneous touches
` Windows Logo requires 5 (except AiO = 2); Android is unspecified
` Market trend is 10 for tablets and notebooks
` Support for unintended touches
` “Palm rejection”, “grip suppression”, etc.
` Rarely specified, but critically important
` For a 22” screen, even 50 touches isn’t too many in this regard
` Amount of “tuning” required
` Never specified – more info on upcoming slide
`
`DISPLAY WEEK ‘14
`
`34
`
`
`
`Signal-to-Noise Ratio (SNR)…1
`
` SNR = Industry-standard performance metric
`for p-cap touchscreen systems
` However, no standard methodologies exist for measuring,
`calculating, and reporting SNR
` The two components (signal & noise) depend heavily on
`the device under test
` Noise from displays (LCDs & OLEDs) and from
`USB chargers is spiky – it doesn’t have a normal
`(Gaussian) distribution – and spikes create jitter
` Yet marketers typically specify SNR in the absence of noise,
`using the RMS noise (standard deviation) of analog-to-digital
`convertors (ADCs)
` With Gaussian noise, you can multiply the RMS noise by 6 to
`calculate the peak-to-peak noise with 99.7% confidence
`
`DISPLAY WEEK ‘14
`
`35
`
`
`
`Signal-to-Noise Ratio (SNR)…2
`
` Typical system (raw ADC data, no digital filters applied)
`
`Noise (CNS)
`
`Source: Cypress
`(modified by the author)
`
`DISPLAY WEEK ‘14
`
`36
`
`
`
`Signal-to-Noise Ratio (SNR)…3
`
` SNR of system in previous slide
` CFinger = Mean (Finger) - Mean (NoFinger)
` CFinger = 1850 - 813 = 1037
`
` CNS (Standard Deviation) = 20.6 counts
` CNS (Peak-to-Peak) = Max (NoFinger) - Min (NoFinger) +1
` CNS = 900 - 746 +1 = 155 counts
`
` SNR (Peak-to-Peak) = 1037/155 = 6.7
` SNR (Standard Deviation) = 1037/20.6 = 49.9
` Highest SNR currently reported by marketer = 70 dB (3,162*)
`
`* Signal amplitude ratio in dB = 20log10 (A1 / A0)
`
`DISPLAY WEEK ‘14
`
`37
`
`
`
`Noise Management…1
`
` Charger noise is common-mode
` A smartphone on a desk (not handheld) isn’t grounded, so the
`entire phone moves relative to earth ground as it follows the noise
` A touching finger provides an alternative path to ground, which
`is equivalent to injecting the noise at the finger location
` The noise signal can be 10X to 100X that of the signal
`generated by the touching finger
`
`Can be
`as high
`as 60 V
`p-p for
`non-EN
`chargers
`
`Source: Cypress
`
`DISPLAY WEEK ‘14
`
`38
`
`
`
`Noise Management…2
`
` Examples of charger noise spectra
` Effect of noise is false or no touches, or excessive jitter
`
`DISPLAY WEEK ‘14
`
`39
`
`Source: Cypress
`
`
`
`Noise Management…3
`
` Variation in common-mode noise spectra in 2
`different chargers at 3 different loads
`
`DISPLAY WEEK ‘14
`
`40
`
`Source: Cypress
`
`
`
`Noise Management…4
`
` Techniques to combat charger noise
` Multiple linear and non-linear filters
` Adaptive selection of the best operating frequency (hopping)
` Increased drive-electrode voltage
`● Going from 2.7 V to 10 V increases SNR by 4X
` Many proprietary methods
`
` Display noise
` LCD noise is similar across the display; the high correlation of noise
`signals across all sensor signals allows relatively easy removal
` Very high noise in embedded touch can require synchronization
`of the touch controller with the LCD driver (TCON)
`
`DISPLAY WEEK ‘14
`
`41
`
`
`
`Controller Innovation Areas
`
` More information in upcoming slides
` Finger-hover
` Glove-touch
` Pressure sensing
` Other touch-objects
` Faster response (reduced latency)
` Adaptive behavior
` Water resistance
` Software integration
` Automated tuning
`
` More information later in this course
` Passive and active stylus support
`
`DISPLAY WEEK ‘14
`
`42
`
`
`
`Finger-Hover…1
`
` There are two ways of emulating “mouseover” on
`a p-cap touchscreen
` Hover over something to see it change, then touch to select
` Press lightly on something to see it change, then press harder
`to select
` The industry is moving towards hover because nobody
`has been able to implement pressure-sensing in a way
`that works well and that OEMs are willing to implement
` Startup: NextInput
`● Force-sensing using an array of organic transistors where pressure
`changes the gate current
` Startup: zRRo
`● Multi-finger hover detection
`
`DISPLAY WEEK ‘14
`
`43
`
`
`
`Finger-Hover…2
`
` What can you do with hover?
` Enlarge small links when you hover over them
` Make a passive stylus seem to hover like an active stylus
` Magnify an onscreen-keyboard key as you approach
`rather than after you’ve touched it, or even use a “Swipe”
`keyboard without touching it
` Preview interactive objects such as an array of thumbnails
` Use as an alternative to standard proximity detection
` Use multi-finger gestures for more complex operations
` And more…
`
`DISPLAY WEEK ‘14
`
`44
`
`
`
`Glove-Touch
`
` Can be accomplished by
`adding self-capacitive to
`existing mutual-capacitive
` Mutual-capacitive provides
`touch location
` Self-capacitive provides
`proximity sensing
` Glove-touch causes the finger
`to remain a constant distance
`above the screen; proximity
`sensing can detect that without
`the user manually switching
`modes
`
`DISPLAY WEEK ‘14
`
`45
`
`Source: ELAN
`
`
`
`Pressure Sensing
`
` Pressure-sensing is an alternative selection method
` True absolute pressure-sensing in p-cap doesn’t exist today
` Some (including Microsoft) believe that “touch lightly to view
`choices then press to select” is more intuitive than hover
`● It has never been implemented successfully in a mobile device
` Blackberry Storm (2 models!) failed due to terrible implementation
` Nissha/Peratech (QTC) collaboration never made it into mass-production
` Multiple startups are working on smartphone pressure-sensing
`● NextInput
` Uses an array of pressure-sensitive organic transistors under the LCD
`● FloatingTouch
` Mounts the LCD on pressure-sensing capacitors made using a 3M material
`
`DISPLAY WEEK ‘14
`
`46
`
`
`
`Other Touch Objects
`
` You will soon be able to touch with a fine-tipped (2 mm)
`passive stylus, long fingernails, a ballpoint pen, a #2
`pencil, and maybe other objects
` This is being accomplished through higher signal-to-noise
`(SNR) ratios
`● Much of this improvement may come from enhancing the controller
`analog front-end in addition to focusing on the digital algorithms
` This enhancement to the UX will be the end of “finger-only” p-cap
`
`DISPLAY WEEK ‘14
`
`47
`
`
`
`Faster Response
`
` Make touch more natural by reducing latency
` The shorter the time is between a touch and the response,
`the better the user feels about the touch system
`● If an object lags behind your finger when you drag it, or ink lags
`behind a stylus when you’re drawing, it doesn’t feel real
` Latency today is typically 75-100 ms;
`studies have shown that humans
`need less than 10 ms for comfort
`● Synaptics has addressed the problem
`by creating a direct path between the
`touch controller and the TCON to
`allow limited instant screen updates
`● Tactual Labs (startup) has a method
`of reducing latency to just a few
`milliseconds
`
`Android
`lag!
`
`Source: Gigaom.com
`
`DISPLAY WEEK ‘14
`
`48
`
`
`
`Adaptive Behavior: Noise Immunity
`
` Adaptive noise-management by N-Trig
`
`Finger-Touch
`Detection
`
`One
`Finger
`Only?
`
`Yes
`
`Normal Operation: Multi-Touch
`with Frequency-Hopping
`
`Medium
`
`No
`
`Noise
`Level?
`
`Medium-
`High
`
`Reduce
`Touch
`Report
`Rate
`
`Extreme
`
`Single-
`Touch
`Operation
`
`DISPLAY WEEK ‘14
`
`49
`
`
`
`Water Resistance…1
`
` The basic concept is combining self-capacitive and
`mutual-capacitive sensing (again)
`
`Water drops on the screen
`
`Source: ELAN
`
`Water is not detected
`in self-capacitive mode
`
`Water is detected in
`mutual-capacitive mode
`
`DISPLAY WEEK ‘14
`
`50
`
`
`
`Water Resistance…2
`
` A large amount of water with single-touch
`
`DISPLAY WEEK ‘14
`
`51
`
`Source: ELAN
`
`
`
`Water Resistance…3
`
` A large amount of water with two touches
`
`DISPLAY WEEK ‘14
`
`52
`
`Source: ELAN
`Source: ELAN
`
`
`
`Software Integration
`
` Make more resources available to the touch controller
` Run touch algorithms on the GPU instead of the controller micro
`● Algorithm-writers can take advantage of much larger resources on
`the host device (MIPS and memory)
` This can support higher frame-rate, reduced latency, reduced
`power consumption, easier support of different sensor designs, etc.
`● Algorithmic code is easier and faster to change when it’s in a “driver”
`than when it’s in firmware in an ASIC
` Most touch-controller suppliers never change the firmware in the
`touch controller once it ships in a device; N-Trig is the sole exception
`● Cost-reduction by elimination of one micro
` Even more cost reduction for large screens by elimination of slave chips
` Something similar to this has already been done in NVIDIA’s
`“Direct Touch”, but it hasn’t been widely used in actual devices
`
`DISPLAY WEEK ‘14
`
`53
`
`
`
`Automated Tuning
`
` For true “touch everywhere”, p-cap has to become
`like resistive: Just slap it on and you’re done
` We’re far from that point today
` Atmel says that the typical first integration of a p-cap touch-panel
`into a new product takes one full day of tweaking up to 200
`individual parameters
` That badly needs to be automated so that small commercial
`product-makers have easier access to p-cap
`
`DISPLAY WEEK ‘14
`
`54
`
`
`
`P-Cap Controller Suppliers
`
` In order by estimated 2013 revenue
`Company
`Country
`Broadcom (Apple) USA
`Atmel
`USA
`Synaptics
`USA
`TI
`USA
`FocalTech
`China & Taiwan
`Melfas
`Korea
`Cypress
`USA
`Goodix
`China
`ELAN
`Taiwan
`Mstar
`Taiwan
`EETI
`Taiwan
`Zinitix
`Korea
`SiS
`Taiwan
`Ilitek
`Taiwan
`Imagis
`Korea
`Sentelic
`Taiwan
`Weida
`Taiwan
`Sitronix
`Taiwan
`
`
`DISPLAY WEEK ‘14
`
`55
`
`Top 7 (30%)
`account for
`about 85% of
`total revenue
`
`And a few others…
` AMT
` Avago
` Pixcir
` Silicon Labs
` STMicro
` Weltrend
`
`
`
`Sensors
` Substrates
` Structures
` Sheet vs. Piece Method
` More on OGS
` Glass Strengthening
` Surface Treatments
` ITO Index Matching
` Suppliers
`
`DISPLAY WEEK ‘14
`
`56
`
`
`
`Sensor Substrates…1
`
` ITO film substrates are usually PET1 or COP2
` Thickness has dropped from 100 µm to 50 µm
` Lowest practical ITO sheet resistivity is currently ~100 Ω/□
`
` ITO glass substrates
` Standard thickness for GG is 0.33 mm and 0.4 mm
` Some makers have developed a thinning process (like for LCDs)
`that reduces glass thickness to 0.2 mm
` Corning and AGC have developed 0.1 mm glass but it hasn’t
`been used in volume sensor production yet
` Lowest practical ITO sheet resistivity on glass is ~50 Ω/□
`
`1 = Polyethylene Terephthalate
`2 = Cyclic Olefin Polymer
`
`DISPLAY WEEK ‘14
`
`57
`
`
`
`Sensor Substrates…2
`
` PET film versus glass
`
`PET
`
`Glass Transition Temperature 70°C
`Aging Effects
`Yellowing, curling,
`surface deformation
`85%
`10-30 µm
`Thinner
`Lighter
`Good
`Excellent
`None
`
`Transparency
`Resolution Capability
`Stackup
`Weight
`Moisture Resistance
`Lamination Yield
`Mechanical Strengthening
`
`Cost
`
`
`
`$$ (was < glass)
`
`Glass
`570°C
`No known effect
`
`=>90%
`1 µm
`Thicker
`Heavier
`Excellent
`Good
`Chemical, heat,
`ion-exchange
`$
`
`DISPLAY WEEK ‘14
`
`58
`
`
`
`Sensor Structures…1
`
` Sensor structure abbreviations (for reference)
`Symbol Meaning
`(G)
`Cover-glass (or plastic or sapphire)
`G
`Cover-glass, or sensor-glass with ITO on one side, or
`plain glass for film lamination
`Cover-glass + one sensor-glass (without ITO location)
`GG
`GGG Cover-glass + two sheets of sensor-glass (rare)
`G#
`# = Number of ITO layers on one side of sensor-glass
` (G2 = “One Glass Solution” = OGS = SOC = SOL, etc.)
`F = Sensor-film with ITO on one side, laminated to glass
`FF = Two sensor-films, laminated to glass
`1 = Two ITO layers on one side of sensor-film,
`laminated to glass (also called GF-Single)
`2 = One ITO layer on each side of sensor-film,
`laminated to glass (also called GFxy with metal mesh)
`ITO on one side of substrate (single-sided);
`usually includes metal bridges for Y to cross X
`ITO on both sides of substrate (double-sided)
`F1 = Single-sided sensor-film on top of CF glass;
`T = Transmit (drive) electrodes on TFT glass
`(LG Display’s hybrid in-cell/on-cell)
`
`G1F
`GFF
`GF#
`
`SITO
`
`DITO
`F1T
`
`
`DISPLAY WEEK ‘14
`
`59
`
`
`
`Sensor Structures…2
`
` Glass-only structures
`
`Structure Names
`Comments
`
`Example Products
`
`GGG
`Single ITO layer on
`each piece of glass;
`Obsolete
`None
`
`GG or G-SITO
`Single ITO layer
`with bridges
`
`Kindle Fire,
`B&N Nook;
`Nokia Lumia 800
`
`GG , G-DITO or G1G
`ITO layer on each
`side of 1 glass; or ITO
`on one side of 2 glass
`iPhone-1; iPad-1
`(GG); Lenovo AiOs
`(G1G)
`
`OGS or SOC
`Single ITO layer
`with bridges
`
`Google Nexus 4/7;
`Xiaomi 2;
`Nokia Lumia 920
`
` SITO = Single-sided ITO layer; usually means there’s a bridge
` DITO = Double-sided ITO layer (Apple patent)
` OGS = One Glass Solution (sensor on cover-glass)
` SSG = Simple Sensor Glass (OGS without cover-glass shaping & finishing)
`
`DISPLAY WEEK ‘14
`
`60
`
`
`
`Sensor Structures…3
`
` Glass-and-film structures
`
`G1F
`Structure Names
`Comments Single ITO layer on
`glass; single ITO
`layer on film
`Many Samsung
`products in 2013;
`Microsoft
`Surface RT
`
`Example Products
`
` Why would a touch-module maker use a sensor structure
`that requires having both glass- and film-handling equipment?
`» One reason is that there was a shortage of ITO film in 2013
`
`DISPLAY WEEK ‘14
`
`61
`
`
`
`Sensor Structures…4
`
` Film-only structures
`
`Structure Names
`Comments
`
`Example Products
`
`GFF
`Bare glass and two
`single-sided ITO films;
`performance is better
`than GF1
`Samsung Galaxy Tabs
`and Notes; Google
`Nexus 10
`
`GF2 or DITO-Film
`Bare glass and one
`double-sided
`ITO film
`
`Apple iPads; next
`iPhone if Apple can’t get
`good yield on in-cell
`
`GF1
`Bare glass with true
`single-layer complex
`pattern on film
`(e.g., “caterpillar”)
`Many low-end
`smartphones, especially
`in China
`
`GF Triangle
`Bare glass with true
`single-layer triangle
`pattern on film
`(e.g., “backgammon”)
`Low-end products with
`“gesture touch”, not
`multi-touch
`
` Single-layer caterpillar pattern is used to support “real” multi-touch with 2-3
`touches, typically in a smartphone (that’s not enough touches for a tablet)
` Single-layer backgammon pattern is used to support “gesture touch” on
`low-end devices, i.e., the ability to detect pairs of moving fingers but not
`always resolve two stationary touches
`
`DISPLAY WEEK ‘14
`
`62
`
`
`
`Sensor Structures…5
`
` Why do touch-module makers choose one structure
`over another?
` Transmissivity
` Thickness & weight
` Border width due to routing
` Cost & availability of ITO film or deposition
` Lamination experience & yields
` Existing equipment and/or method experience
`
`DISPLAY WEEK ‘14
`
`63
`
`
`
`Sensor Structure by Application
`
` Tablets & Notebooks All-in-Ones
` Smartphones
`Structure
`Share
`Structure Share
`Structure
`Share
`GFF
`44%
`GG SITO
`81%
`GFF
`42%
`GF2/DITO Film
`19%
`GFF
`13%
`OGS/G2
`16%
`OGS/G2
`18%
`SSG
`6%
`GF1/Single-Layer
`12%
`GG DITO
`11%
`
`GG SITO
`11%
`GG SITO
`3%
`
`GF Triangle
`5%
`G1F
`2%
`
`GG DITO
`5%
`GF1/Single-Layer
`1%
`
`G1F
`4%
`SSG
`1%
`
`PF
`3%
`
`
`PFF
`2%
`Data based on DisplaySearch’s “Q1-2014 Quarterly Touch-Panel
`Market Analysis Report”, with adjustments by the author
`
`
`
`
`
`
`
`
`
`
`DISPLAY WEEK ‘14
`
`64
`
`
`
`Sheet vs. Piece Method…1
`(Wintek Sheet Example - OGS)
`
`DISPLAY WEEK ‘14
`
`Source: Wintek
`
`65
`
`
`
`Sheet vs. Piece Method…2
`(Wintek Piece Example - Discrete)
`
`DISPLAY WEEK ‘14
`
`Source: Wintek
`
`66
`
`
`
`More On OGS
`
` One-Glass Solution (OGS)
` Also called “touch on lens” (TOL), “sensor on cover” (SOC),
`“direct patterned window” (DPW) and many other names
` Advantages
`● Eliminates a fourth sheet of glass (G-DITO), making the end-product
`thinner and lighter
`● Competitive weapon against embedded touch from LCD suppliers
` Disadvantages
`● Requires close cooperation with cover-glass makers, or increased
`vertical integration (preferable)
`● Yields are lower (more complex operations)
`● Bendable cover glass can affect touch performance
`● Harder to shield touchscreen from LCD noise
` Note: There is no generic name (yet) for touch sensors built on the
`cover-glass without direct ITO deposition (“OGS-type”)
`
`DISPLAY WEEK ‘14
`
`67
`
`
`
`Glass Strengthening
`
` Heat strengthened
` Less-rigorous version of fully tempered; does not “dice” when
`broken; 2X as strong as standard glass
` Fully tempered
` Uses heat; requires glass > 3 mm, so not used for consumer
`touchscreens; glass “dices” when broken (think auto windows);
`4X to 6X as strong as standard glass
` Chemical strengthened (CS)
` Uses ion-exchange in a salt bath; best for glass < 3mm; glass does
`NOT “dice” when broken; 6X to 8X as strong as standard glass
` High ion-exchange aluminosilicate glass
` 6X to 8X as strong as standard glass (same as CS glass)
` Corning Gorilla®, Asahi Dragontrail™, Schott Xensation™
`
`DISPLAY WEEK ‘14
`
`68
`
`
`
`Sensor Surface Treatments…1
`
` Historically most common treatment is anti-glare (AG)
` Changes specular reflection into diffuse reflection
` Used mostly for commercial & enterprise, not consumer (“glossy”)
` Three methods, roughly equal cost
`● Chemical etching
`● Application of sol-gel containing silica particles
`● Mechanical abrasion
` Level of anti-glare can be very little to a lot
`
` Anti-fingerprint (AF) treatment is rapidly growing
` Many different forms (spray-on, rub-on, sputter, etc.); also
`called “anti-smudge” (AS)
` Demand is increasing
` Cost is dropping (currently ~$8.50/m2)
`DISPLAY WEEK ‘14
`
`69
`
`
`
`Sensor Surface Treatments…2
`
` Anti-reflection (AR) treatment is still a problem
` Reduces specular reflection to range of 2% to 0.4%
` Durability is typically < 1 year
` It’s expensive (currently ~$34.50/m2)
` Yet it’s really important for outdoor viewing, particularly of
`consumers’ glossy screens (ideal is AF+AR = ~$43/m2)
` Other coatings are available but less common
` Anti-corruption (allows permanent Sharpie ink to be wiped off)
` Anti-microbial/anti-bacterial (AM/AB, for healthcare applications)
` Hard coating (can be made up to 9H for glass-like anti-scratch)
` Anti-stiction (reduces finger-sticking friction)
` Anti-crack coating (increases durability at lower cost than Gorilla
`glass; uses atomic layer deposition [ALD])
`
`DISPLAY WEEK ‘14
`
`70
`
`
`
`ITO Refractive-Index Matching
`
` Reduce the reflectivity of ITO by compensating for the
`difference in index of refraction of ITO vs. glass/PET
` Limited to 2 layers on PET; more can be used on glass
` Alternating layers of material with low and high refractive index
` Layer thicknesses (typically between ¼ and ½ of the wavelength
`of light) are chosen to produce destructive interference in reflected
`light, and constructive interference in transmitted light
`
`ITO (RI = ~2.0)
`TiO2 (RI = 2.48)
`SiO2 (RI = 1.45)
`Glass (RI = 1.52)
`or PET (RI = 1.65)
`
`DISPLAY WEEK ‘14
`
`71
`
`Source: The author
`
`
`
`Sensor Suppliers
`
` Many touch-module makers manufacture their
`own sensors
` The remainder are made by the following companies,
`in order by estimated 2013 revenue
`Company
`Country
`Nissha Printing
`Japan
`HannsTouch
`Taiwan
`Dongwoo Fine Chemical Korea
`Cando
`Taiwan
`Innolux
`Taiwan
`CSG
`China
`Token
`China
`CPT
`Taiwan
`DNP
`Japan
`Young Fast
`Taiwan
`AimCore
`Taiwan
`
`And at least one more…
` Laibao (China)
`
`
`DISPLAY WEEK ‘14
`
`72
`
`
`
`ITO-Replacement Materials
` ITO
` Metal Mesh
` Silver Nanowires
` Carbon Nanotubes
` Conductive Polymers
` Graphene
` Summary
`
`DISPLAY WEEK ‘14
`
`73
`
`
`
`ITO Replacements…1
`
` Why replace ITO?
` Costly to pattern & needs high temperature processing
` Highly reflective (IR = 2.6) & tinted yellow; brittle & inflexible
` NOT because we’re going to run out of it!
` Replacement material objectives
` Solution processing (no vacuum, no converted LCD fab)
` Better performance than ITO (transmissivity & resistivity)
` Lower material & process cost than ITO
` Five replacement candidates
` Metal mesh
` Silver nanowires
` Carbon nanotubes
` Conductive polymers
` Graphene
`
`DISPLAY WEEK ‘14
`
`74
`
`
`
`ITO Replacements…2
`
` ITO-replacement materials are having a definite
`market impact – 11% in 2014!
` See the latest IHS market report on non-ITO films
`
` Ag halide is simply
`another method of
`making a silver mesh,
`so the mesh total is
`85% vs. 15% for
`nanowire
`
` The value is performance and cost
`● Both unit cost and CAPEX
`
`DISPLAY WEEK ‘14
`
`75
`
`
`
`Metal Mesh…1
`
` Metal mesh is shipping in touchscreens, and it’s
`looking very promising!
` Brief history of first-movers
` MNTech in Korea was the first to ship metal-mesh at the
`end of 2012 – but their factory burned down
` Atmel (partnered with CIT in the UK) was the second to ship metal-
`mesh (XSense™) for a smartphone and a 7” tablet in 1H-2013
` FujiFilm started production of their silver-halide-based
`metal-mesh product in 2Q-2013
`
`DISPLAY WEEK ‘14
`
`76
`
`
`
`Metal Mesh…2
`
`4-5 mm sense electrode (bottom surface)
`
`Top
`layer
`(red)
`
`Bottom
`layer
`(white)
`
`Source:
`Photo by Unipixel,
`annotation by
`the author
`
`DISPLAY WEEK ‘14
`
`77
`
`4-6 µm wide
`conductors
`with spacing
`of 100-400 µm
`
`4-5 mm
`drive
`electrode
`(top surface)
`
`Intentional
`gaps in lines
`
`
`
`Metal Mesh…3
`
` Metal mesh has significant advantages
` Patterning via roll-to-roll printing allows both operating and
`capex cost to be very low – it’s going to beat both litho and laser!
`● Electrodes and border connections are printed simultaneously,
`which allows borders as narrow as 3 mm (typically 9 mm with ITO)
` Sheet resistivity is much lower than ITO (under 10 ohms/square)
`● Reduces p-cap charge time, which allows larger touchscreens
` Transparency is better than ITO
` Mesh pattern creates electrical redundancy, which improves yields
` Highly flexible – bend radius typically 4 mm
`
`DISPLAY WEEK ‘14
`
`78
`
`
`
`Metal Mesh…4
`
` O-film is the “800-pound gorilla” of metal mesh!
` Largest touch-module maker in China, #3 globally
` Like “the TPK of film”; innovative and aggressive
` New roll-to-roll printing method
` “Hybrid printing” or “micro-imprinting”
`
`Source: O-film
`
`Cross-section of
`embedded metal line
`
`Impressions
`
`Metal
`roller
`(mold)
`
`Silver nano-
`particle ink
`
`UV cure
`
`UV resin
`
`Source: The author
`
`Source: O-film
`
`PET
`
`DISPLAY WEEK ‘14
`
`79
`
`
`
`Metal Mesh…5
`
` O-film technical details
` Additive process with little waste
` < 2 µm line width
` < 10 Ω/□
` Randomized mesh design (one method of eliminating moirés)
` Top surface of embedded metal line is blackened & sealed
` Embedded metal reduces haze and eliminates peel-off
` Producing > 1.5M touch sensors per month (size not stated)
` O-film’s success makes visible a developing aspect of
`the ITO-replacement business
` A vertically-integrated sensor & module-maker is in a much better
`position to