Study of Lithographically Defined Data Track and Servo Patterns

Abstract

We report on fabrication of discrete tracks on perpendicular magnetic recording (PMR) media with an e-beam lithographical process. We studied the recording performance of the e-beam media on a spinstand in parallel with conventional PMR media. Discrete track media show significant reduction in adjacent track erasure (ATE). We studied and quantitatively measured the source of the ATE improvement, and developed a triple track geometrical model to calculate achievable track density for both discrete track recording (DTR) and continuous media. From the model, we identify two factors of DTR that contribute to reaching a higher TPI. Using the same fabrication technique, we also studied servo burst design and its playback waveform quality. At 250 ktpi, we compare DTR servo bursts with servo bursts written with a conventional method. DTR servo bursts show better edge definition, which can translate to better position error signal sensitivity and support higher TPI in the future.Discrete tracks are fabricated on conventional PMR media with an e-beam litho graphical process. The recording performance is studied on a spinstand in parallel with conventional PMR media. Discrete track media shows significant reduction in adjacent track erasure (ATE). The source of the ATE improvement is studied and quantitatively measured. A triple track geometrical model is developed to calculate achievable track density for both DTR and continuous media. From the model, we identify two factors of DTR, which contribute to reaching a higher TPI. Using the same fabrication technique, we also study servo burst design and its playback waveform quality. At 250 ktpi, we compare DTR servo bursts and servo bursts written with a conventional method. DTR servo bursts show better edge definition, which can translate to better PES signal sensitivity and support higher TPI in the future.

Full text

Page 1

4106 IEEE TRANSACTIONS ON MAGNETICS, VOL. 43, NO. 12, DECEMBER 2007
Study of Lithographically Defined Data Track and Servo Patterns
Xiaodong Che?, Ki-Seok Moon?, Yawshing Tang?, Na-Young Kim?, Soo-Youl Kim?, Hyung Jai Lee?,
Matthew Moneck?, Jian-Gang Zhu?, and Nobuyuki Takahashi?
Samsung Information Systems America, San Jose, CA 95134 USA
Electrical and Computer Engineering Department, Carnegie Mellon University, Pittsburgh, PA 15213-3890 USA
Fuji Electric Device Technology Co., Ltd., Nagano 390-0821, Japan
We report on fabrication of discrete tracks on perpendicular magnetic recording (PMR) media with an e-beam lithographical process.
Westudiedtherecordingperformanceofthee-beammediaonaspinstandinparallelwithconventionalPMRmedia.Discretetrackmedia
show significant reduction in adjacent track erasure (ATE). We studied and quantitatively measured the source of the ATE improve-
ment, and developed a triple track geometrical model to calculate achievable track density for both discrete track recording (DTR) and
continuous media. From the model, we identify two factors of DTR that contribute to reaching a higher TPI. Using the same fabrication
technique, we also studied servo burst design and its playback waveform quality. At 250 ktpi, we compare DTR servo bursts with servo
bursts written with a conventional method. DTR servo bursts show better edge definition, which can translate to better position error
signal sensitivity and support higher TPI in the future.
Index Terms—Electron beam lithography, patterned media, perpendicular magnetic recording, servo.
I. INTRODUCTION
P
granular films [1]. With the recent technology transition from
longitudinal magnetic recording (LMR) to PMR, data storage
areal density is continuously increasing without crossing the
super-paramagnetic limit [2], [3]. New PMR media concepts,
such as exchange coupled composite (ECC) media, have been
introduced and demonstrated to further improve thermal sta-
bility without increasing magnetic switching field or sacrificing
writing capability. Based on magnetic modeling studies, these
new media concepts can store thermally stable information at
areal densities up to 1 Tb/in [4].
Advanced PMR mediastructure usually includes a highly ex-
changecoupledsoftcappinglayer[5].Thiscappinglayercauses
magnetic particles to form clusters and forces local magnetic
grains to switch coherently. The PMR nucleation field is in-
creased and dc media noise is reduced. However, such mag-
neticlayerstructuremayresultinarelativelylargeerasingband.
Erasing bandwidth is fundamentally determined by media mag-
netic structure and head field gradient crossing the track. They
are not scaled down with writer pole width. As TPI increases,
the erasing band becomes an increasingly large portion of the
recording track.
Discretetrackrecording(DTR)mediahasgroovesseparating
data tracks magnetically and physically [6]–[13]. This concept
potentially offers the following two advantages:
1) on-track signal-to-noise ratio (SNR) gain by eliminating
track edge noise and reoptimizing head writer/reader
geometry;
ERPENDICULAR magnetic recording (PMR) technology
significantly improves the thermal stability of magnetic
Digital Object Identifier 10.1109/TMAG.2007.908279
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
2) reduction of adjacent track erasure by physically sepa-
rating adjacent data tracks.
The first claim has been studied extensively by simulations
[8], [14]. The level of SNR gain strongly depends on PMR
media design [15]. Practical measurement is difficult since
the optimal writer and reader design is different for DTR
and continuous PMR media [16]. The second claim has been
confirmed by recording measurements [8]. Our previous work
found that data tracks on a DTR media are magnetically more
stable against adjacent track erasure (ATE) [17]. However, that
study mainly focused on on-track signal amplitude change; it
did not address how much TPI can increase due to such ATE
performance improvement.
In this work, our focus is on the assessment of TPI gain. We
first quantify adjacent track erasing width difference between
DTR and continuous PMR media. This is an extension of our
previous work [17]. Then, we apply the measurement result to
a geometrical model for TPI calculation. A simple equation can
be derived relating the maximal achievable TPI for DTR and
continuous media.
DTRalsoraiseschallengesforharddiskdrive(HDD)integra-
tion. It is reasonable to assume that servo and Gray code infor-
mation should be fabricated in sync with the data track patterns
[8]. Such an approach eliminates the servo write/copy process
during HDD manufacture and is cost effective as TPI increases.
Here, several commonly used servo patterns are studied. The
servo patterns are fabricated with the same e-beam lithograph-
ical method as the data tracks. The playback signals from these
servo patterns are analyzed on a Guzik spinstand. At 250 ktpi
(100-nm servo track pitch), conventional servo patterns gener-
ated by shingle writing are also evaluated as a comparison.
This paper is organized in the following way.
1) E-beam lithographical DTR process: The key advantage of
this process is its flexibility and good feature definition for
data tracks and servo bursts.
2) Measurement of erasing width for DTR and continuous
PMR media: This measurement quantifies the advantage
of DTR in adjacent track writing.
0018-9464/$25.00 © 2007 IEEE

Page 2

CHE et al.: STUDY OF LITHOGRAPHICALLY DEFINED DATA TRACK AND SERVO PATTERNS4107
Fig. 1. Schematic process flow for the fabrication of discrete tracks on a per-
pendicular magnetic disk. (1) Sputter deposit media and overcoat. (2) Sputter
deposit Al. (3) Spin coat PMMA. (4) E-beam lithography. (5) lon mill Al and
media. (6) Strip Al/PMMA.
3) Geometrical model determining TPI for DTR and conven-
tional media: An equation is derived relating achievable
TPIs for DTR and continuous media.
4) Evaluation of e-beam fabricated servo bursts.
II. E-BEAM LITHOGRAPHIC FABRICATION
Integrationofdiscretetracksandservoburstpatternstogether
on one disk requires a flexible process capable of accommo-
dating a variety of patterns while maintaining the integrity of
the magnetic media. Our previous work has led us to develop a
novelfabricationprocesscapableofproducingflyablepatterned
magnetic disks by combining electron beam lithography and Ar
ion milling with a protective Al sacrificial layer [18]. By uti-
lizing this process, discrete tracks with various pitches, as well
asseveraldifferentservoburstpatterns,werefabricatedonPMR
media.
The fabrication process flow, shown in Fig. 1, begins with a
sputter-deposited 2.5-in (65-mm) magnetic disk, containing a
soft underlayer (SUL), Ru interlayer, CoCrPt media layer, and
diamond-like carbon overcoat (DLC). A 15-nm layer of Al is
sputtered onto the media surface both as a protective layer and
as a sacrificial layer used at the end of the fabrication process.
Once all of the films are deposited, the disk is spin-coated with
PMMA 950-A2 from Microchem Corporation and the various
patternsarewrittenwitha30-kVelectronbeamatdosesranging
from 350 to 700 C/cm . The latent images are developed with
a 1:3 MIBK:IPA solution, leaving behind well-defined litho-
graphic tracks and servo burst patterns. Pattern transfer into the
underlying Al, DLC, and media layers is achieved via Ar ion
milling with a 40-mA, 500-V beam at an angle of 22.5 . Pat-
terningiscompletedbyliftingofftheAlsacrificiallayerandany
remainingPMMAwithAZ400KphotoresistdeveloperfromAZ
Electronic Materials, thus leaving behind patterned magnetic
Fig. 2. SEM images of patterned (a) 50-nm and (b) 100-nm discrete tracks.
Fig. 3. SEM images of patterned AB servo bursts fabricated from (a) nonmag-
netic holes and (b) isolated magnetic islands, as well as (c) patterned chevron
and (d) short length AB servo patterns.
features with their DLC overcoat intact. A final cleaning is per-
formed with PVA foam and a DI water rinse.
The fabrication process results in patterned magnetic disks
containing50-nmdiscretetracksona250-nmpitchand 100-nm
tracks on a 300-nm pitch, shown in Fig. 2. Disks were also fab-
ricated with four different servo patterns shown in Fig. 3. The
servo patterns consist of 100-nm AB bursts assembled from
nonmagnetic holes and magnetic islands, magnetic chevrons on
a 300-nm pitch, and short length AB bursts. The scanning elec-
tron microscope (SEM) images of Figs. 2 and 3 demonstrate the
patterns haverelativelysmooth edgeswith edge roughnesstypi-
callylessthan10nm. Atomicforcemicroscopeheightscaleim-
agesandaugeranalysisprovethemediahasacleansurfacewith
no residual Al and that all of the magnetic material in the trench
regions has been etched away during the ion milling process.
The result is a flyable magnetic disk with a clean undamaged
surface and no signal degradation [18].
III. DTR AND CONTINUOUS PMR MEDIA ERASING
WIDTH MEASUREMENT
The maximal achievable TPI for a given head/media combi-
nation is mainly determined by adjacent track erasure (ATE).

Page 3

4108IEEE TRANSACTIONS ON MAGNETICS, VOL. 43, NO. 12, DECEMBER 2007
Fig. 4. Erasing width, measured by ? , is a combination of data track width
? and two erasing bandwidths ?
and ?
widths can be different, depending greatly on head skew angle.
. For PMR, two erasing band-
Fig. 5. 2-D playback signal map. (a) shows the signal map after the reference
track is recorded. The DTR track is on the left and the continuous media track
signal immediately follows on the right. (b) shows the signal map after the adja-
cent trackiswrittenbelowthe referencetrack.Thewritingfrequencyistoohigh
to be seen in the color map. However, its erasing effect on the reference track is
clearly shown. The erasing edges are identified for DTR and continuous PMR
media in two horizontal lines. ?
on the left is the offset of these two edges.
ATE is usually measured by two criteria: adjacent track single
writing and adjacent track multiple writing. In this study, we
limit our discussion to single adjacent track writing. As shown
inFig.4,whenaheadwritesonetrack,itcreatesadatatrackand
two erasing bands where the head field is high enough to erase
old information but too weak to record new information. Total
erasing width
is the combination of data track width and
two erasing bandwidths.
ultimately determines how close
adjacent tracks can be positioned.
Erasing width is traditionally measured by a triple track pro-
file method [19]. For DTR media, where data track location and
widtharepredetermined,theconventionaltripletrackmethodis
not applicable. To resolve this problem, we choose to measure
erasing band outer edge location. The following is the method.
A. Erasing Edge Measurement
To identify erasing edges (two limit lines defining the width
in Fig. 4), we write a frequency
the following continuous media. The writing position is aligned
to the DTR track center. We call this track the reference track.
A 2-D signal map of the reference track is shown in Fig. 5(a).
Then, we move the head to a given off-track position and write
with an erasing frequency
. As shown in Fig. 5(b), after
recorded below the reference track, the bottom portion of the
reference track is erased. Illustrated by horizontal lines, two
on a DTR track and
is
TABLE I
MEASUREMENT PARAMETERS
erasingedgesareformedinDTRandcontinuousmediaregions,
respectively. The erasing edge locations for DTR and contin-
uous PMR media are different. The offset labeled as
measured.
is the erasing width difference between DTR and
continuous PMR media.
Erasing edge position is calculated from the reference track
profile: the track edge is defined as the 50% amplitude level
of the track profile [19]. As long as the reference track is not
severely erased, its width is larger than the reader width, thus,
such measurement of erasing edge position is accurate [20].
Test conditions and head/media parameters are summarized
in Table I.
can be
B. Erasing Edge Versus Adjacent Track Location
Fig. 6(a) shows DTR and continuous PMR media erasing
edge location versus adjacent track position
zontal axis is the adjacent writing position relating to the refer-
ence track center. The vertical axis is the erasing edge location
relating to one edge of the DTR track: “0” refers to the erasing
edge aligned with the DTR physical edge.
For continuous media, as adjacent track writing gets closer to
the reference track (
decreases), the erasing edge moves
accordingly. The edge position movement is exactly equal to
the reduction of
. This is a result of the following two
factors:continuousmediaishomogeneous,andtheerasingedge
is created at a fixed point under the writer where the minimal
erasing field is reached.
TheerasingprocessforDTRmediaisverydifferentandmore
complex. As shown in Fig. 6(a), when
erasing edge is actually the discrete track physical edge: adja-
cent writing does not erase any of the reference track. When
nm, theerasingedgestartstomoveawayfromthe
physical track edge: the DTR reference track is partially erased.
It is interesting to notice that once erasure starts, the erasing
edge moves more than the reduction of
cates that a somewhat higher field is needed to initiate erasure
and once it starts, the erasing edge extends rapidly across the
whole discrete track.
To compare DTR and continuous media, we divide adjacent
writing offset
into the following three regions as shown
in Fig. 6.
. The hori-
nm, the
. This result indi-

Page 4

CHE et al.: STUDY OF LITHOGRAPHICALLY DEFINED DATA TRACK AND SERVO PATTERNS 4109
Fig. 6. Erasing edge versus adjacent track position for DTR (open square, red
in color) and continuous PMR (solid circle, blue in color). (a) is the case for
reference frequency ? ? ?? Mflux/s, (b) is the case for reference frequency
? ? ??? Mflux/s. DTR performance advantage in ATE is quantitatively mea-
sured as ?
. For both cases, ?
? ?? nm.
1)nm: The adjacent writing erases the refer-
ence track on continuous media only. The erasing edge for
continuous media is outside the DTR data track region.
Therefore, even if the minimal magnetic erasing fields
were the same for DTR and continuous media, the head
field would be too small to erase the DTR reference track.
nmnm: The continuous media ref-
erence track is further erased. The erasing edge is pushed
to pass the discrete track physical edge as shown clearly
by the 2-D playback signal map in Fig. 5(b). In this re-
gion, the DTR reference track remains intact. This means
a higher adjacent writing field is required to erase a DTR
track. Given the micromagnetic film properties for DTR
and continuous PMR regions are identical, this phenom-
enonisnotexpected.WesuspecttheDTRphysicalfeatures
alter the magnetic erasing mechanism. Further study is re-
quired to fully understand the magnetic erasing process at
the edge of the data track for DTR media.
In Fig. 6(a), ATE advantage for DTR media can be defined
by a single parameter
recording media, introducing a DTR physical track (edge)
2)
. For a givenwriter and PMR
would allow adjacent tracks to be
erencetrack.Herewewouldliketoemphasizethatforcon-
tinuous media, the closest adjacent writing position is de-
finedasthepointwheretheerasingedgeisalignedwiththe
DTR physical edge rather than where the reference track
edge erasure starts. Therefore, the value of
not depend on writer or reader widths.
In Fig. 6(b),
is measured for the reference track
recording frequency
is approximately equal to 30 nm. This indicates
that
is a physical parameter that is independent of
the recording data pattern.
nm: The adjacent writing starts to erase the
DTR media. The erasing edge moves faster than the adja-
cent writing location. For a proper discrete track recording
design, this region has to be avoided. Analysis of this re-
gion is beyond the scope of this paper.
closer to the ref-
does
Mflux/s. For both cases,
3)
IV. GEOMETRICAL MODEL FOR RECORDING TRACK DENSITY
In the following geometrical model, we set three geomet-
ricalconditionsthatrelatemagneticwriterwidth,erasingwidth,
reader width, and their corresponding TPI value. The three con-
ditions are the following.
1) Servo linearity: Magnetic reader width has to be wide
enough to give a proper linear position error signal (PES)
region for each servo burst.
2) Maximal on-track SNR: Playback signal does not contain
erasing band noise (write wide, read narrow). For DTR
media,suchaconditionismodifiedasthewholeDTRtrack
should be written and read (both writer and reader widths
are larger than DTR track width).
3) Minimal adjacent track interference: Adjacent track
writing does leave the remaining reference track narrower
than the reader width. For DTR media, one more criterion
is required since its reader can be as large as the writer: no
signal is picked up from the side tracks.
To simplify calculation and focus on recording mechanisms,
both writing and erasing regions are assumed to be a “box”
shape across the track: within the box region, the media is per-
fectly recorded or erased; outside the box region, the media
remains intact. Magnetic reader width is MRW. We assume
both write and read widths are independent of the recording
frequency.
Many complex issues, such as ATE due to multiple writing
and track mis-registration (TMR) for both writing and reading,
are not included in this model. These factors will affect DTR
media recording as well as continuous media recording. For a
relative comparison between DTR and continuous media, we
choose not to include them in the model.
A. Continuous Media TPI Model
Fig. 7 shows a triple track layout for the continuous media
case. Based on the three conditions, we can generate the fol-
lowing equations:
(1)
(2)
(3)

Page 5

4110IEEE TRANSACTIONS ON MAGNETICS, VOL. 43, NO. 12, DECEMBER 2007
Fig.7. Tripletracklayoutforcontinuousmedia.ComparedwithFig.4,erasing
bandwidths on both sides of the track are equal; adjacent tracks are written after
the center track to show the maximal track squeezing condition.
Fig. 8. Triple track layout for DTR media. Track width is physically defined.
It is equal to the sum of data track width and groove width:?? ? ? ?? .
For a given TPI (or track pitch TP), acceptable values
for MWW and MRW can be obtained based on these three
equations.
B. DTR TPI Model
Fig. 8 shows a triple track layout for DTR media. Data track
width is defined physically rather than by the writer dimensions
. Assuming that writing is always at the center
of the physical data track, to meet the triple track TPI criteria
above, we have the following four equations:
(4)
(5)
(6a)
(6b)
Equations (4) and (5) are for servo and on track SNR require-
ments, respectively. Equations 6(a) and (b) are for the third re-
quirement: minimal adjacent track interference. Equation 6(a)
is to avoid adjacent track erasure. Equation 6(b) is to avoid ad-
jacent track reading interference.
In Fig. 8, the erasing band is not shown explicitly for DTR
recording. The erasing band observed on continuous PMR
media is not created during writing due to the existence of
groove regions. Here,
in 6(a) is a conceptual variable.
Based on erasing edge measurement in the previous section,
DTR media could allow adjacent track writing to be
closer to the reference track. Mathematically, it means that
is the difference between the “conceptual” erasing
width for DTR and continuous media:
(7)
We need to emphasize that (7) is correct only in one extreme
case: it is the tipping point discussed in previous section, where
the adjacent writing offset is at its minimal value before the ref-
erence track is erased. At this tipping point, the highest possible
TPI for DTR media is reached.
C. Continuous Media and DTR Comparison
TPI gain for DTR can be obtained from (3), 6(a), and 6(b).
Subtracting (3) from the sum of 6(a) and 6(b), both MWW and
MRW are eliminated. Utilizing (7), we obtain a simple relation:
(8)
Replacingtrackpitch(TPandTP’) in(8)withitscorresponding
TPI value, (8) is changed to the following form:
(9)
where
quantitatively state the DTR media advantage in track density
performance.
is the normalized groove width. It is a physical character-
istic parameter. Its value is from “0” (continuous media) to “1”
(DTR with infinitesimal data track width).
NIL process capability. Currently, it has a value from 0.3 to
0.5. Large
corresponds to a narrow data track, hence reducing
on-track playback signal and eventually lowering achievable
linear density. Such linear density roll-off and its tradeoff with
track density gain should be investigated for DTR areal density
improvement over continuous media. This determines the op-
timal target for
.
is the normalized magnetic characteristic parameter. It
is determined by media film properties and magnetic and
physical data track edge properties. It is a new quantified factor
found for DTR media. In this study,
nmnm. We believe
determine whether DTR has significant technical advantage
over continuous media. If PMR media optimization results in
increasing , DTR media will become an attractive solution to
match with PMR technology in the future.
and. These two factors
is determined by
is approximately 0.25
will ultimately
V. LITHOGRAPHICALLY FABRICATED SERVO PATTERN FOR
DISCRETE TRACK MEDIA
In HDDs, the media surface is currently formatted into two
major regions: data and servo blocks. The servo region contains
head positioning information. For conventional media, servo