Journal of Manufacturing Processes: Xiangcheng Kong, Jia Deng, Jingyan Dong, Paul H. Cohen T

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Journal of Manufacturing Processes 50 (2020) 47–56

Contents lists available at ScienceDirect

Journal of Manufacturing Processes


journal homepage: www.elsevier.com/locate/manpro

Study of tip wear for AFM-based vibration-assisted nanomachining process T


Xiangcheng Kong, Jia Deng, Jingyan Dong*, Paul H. Cohen
Edward P. Fitts Department of Industrial and Systems Engineering, North Carolina State University, Raleigh, North Carolina, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Nanofabrication technologies have many applications in science and engineering. Among different nanofabri-
Tip-based nanomachining cation technologies, the tip-based vibration-assisted nanomachining using an Atomic Force Microscope (AFM)
Atomic force microscope (AFM) provides a low-cost, easy-to-setup approach for the production of nano-scale structures. As the resolution and
Tip wear monitoring quality of the machined features are greatly affected by the radius and sharpness of the tip, it is critical to
Tip wear modeling
investigate the behavior of tip wear during the nanomachining process and to estimate the tip life. In this work,
the evolvement of the tip wear was characterized and modeled to predict tip wear and tip life for the nano-
machining process. Besides the direct inspection of the tip radius using a Scanning Electron Microscope (SEM),
the pull-off force between the AFM tip and the sample surface was found to correlate well with the tip radius,
which enabled the measurement of tip wear directly without unloading the tip from the AFM. To study the tip
wear at different conditions, the tip radius was measured from the pull-off force under a wide range of machining
conditions. The change rates of the tip radius were significantly affected by the machining parameters, such as
setpoint force and feed rate. Moreover, during the nanomachining process, three regions were identified for the
tip wear evolvement as initial tip wear region, transition region, and tip failure region. Regression models were
developed to describe the tip wear at different stages, and to estimate the tip life (i.e. when the tip needs to be
changed), which provide usefully information for future process planning and process optimization.

1. Introduction is a low-cost approach for the production of nano-scale structures. The


simplest application of tip-based nanomachining is direct scratching
Nanofabrication, especially the fabrication of master patterns and using a sharp AFM tip, in which the sample surface is modified me-
masks, is critical for emerging nanotechnology applications including chanically by cutting or plastic deformation. Direct scratching is easy to
fundamental physics, chemistry, electronics, materials, and biology. apply, but generally has low manufacturing throughput. Since the size
There are many nanofabrication methods that were able to fabricate of the produced feature is primarily affected by the tip radius, multiple
nanopatterns and nanostructures, including X-ray lithography [1–3], scratching steps have to be used to machine large features.
EUV lithography [4,5], e-beam lithography [6,7], nanoimprint litho- Furthermore, in the direct scratching process, a large normal force is a
graphy [8,9], Dip-pen nanolithography [10] etc. Despite impressive prerequisite to ident the tip into the sample for mechanical machining
resolution from these methods, many of them rely on the masks and [11–13], which causes severe tip wear. To overcome these dis-
molds fabricated by e-beam lithography to enable their processes. The advantages, researchers have developed and demonstrated vibration-
e-beam lithography systems are very expensive to acquire and to assisted nanomachining approaches for low-cost high-rate nanofabri-
maintain. Compared with the expensive e-beam lithography system cation [14–16]. Three-dimensional nanostructures were successfully
with very high hourly rate, tip-based nanofabrication with Atomic fabricated using this nanomachining process [17,18]. With the assis-
Force Microscope (AFM) uses lower-cost equipment with easy-to-setup tance of vibration, the machining force as well as tip wear were sig-
system and tool to produce nanoscale features. The hourly rate for using nificantly reduced. To understand the mechanism of the tip-based vi-
an AFM is much lower than that for using an e-beam system, although bration-assisted nanomachining process, the tip-sample interaction
the short tip life could increase the cost of AFM based nanofabrication. forces (i.e. the machining forces) involved in this process and the re-
Other than the cost, the AFM based nanofabrication provides compar- sultant feature dimensions under given machining parameters have
able resolution to that from the e-beam lithography. been analyzed and modeled in previous research [19,20], which pro-
Tip-based nanomachining using an Atomic Force Microscope (AFM) vided useful guidelines for selecting the machining parameters and


Corresponding author at: Department of Industrial and Systems Engineering, North Carolina State University, 414-C Daniels Hall, Campus box 7906, Raleigh,
North Carolina, 27695-7906, USA.
E-mail address: [email protected] (J. Dong).

https://doi.org/10.1016/j.jmapro.2019.12.013
Received 25 June 2019; Received in revised form 6 December 2019; Accepted 9 December 2019
1526-6125/ © 2019 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
X. Kong, et al. Journal of Manufacturing Processes 50 (2020) 47–56

Fig. 1. (a) Schematics of the vibration assisted nanomachining system. (b) Tip vibration in xy-plane control the feature width in one machining pass. (c) The top view
of machining process shows the tip-sample engagement during one rotation cycle.

planning the process properly to improve productivity. In these models, these studies, the tip wear were mostly observed by a scanning electron
the tip radius was assumed constant during the machining process, microscope (SEM) or transmission electron microscopy (TEM) by
which is reasonable for brand new tips that have only undergone light measuring the tip radius and the evolvement of tip profile [37[36]].
machining tasks. However, when a tip was used in nanomachining for Although SEM and TEM provide very accurate measurement of the tip
high volume production, the tip can be severely worn resulting in a wear and tip radius, the machining operation has to be interrupted
much larger tip radius than that of a new tip. Such tip wear can sig- frequently to take off the tip from the AFM for the off-line SEM or TEM
nificantly affect the machining performance and feature resolution. It is inspection and measurement. To overcome such inconvenience and to
critical to understand the tip wear and estimate the tip life for the na- monitor the tip radius continuously, the pull-off forces between the
nomachining process. AFM tip and the sample surface was measured and correlated with the
For the tool wear at the conventional scale, many monitoring and AFM tip radius [37] to study the tip wear under simple sliding between
modeling methods have been developed to predict its behavior, in- the tip and the sample.
cluding analytical modeling, finite element analysis, and empirical In this paper, the tip wear behavior for the vibration assisted na-
modeling techniques. These modeling methods have been applied to nomachining process was systematically studied with process models
different machining processes like turning, milling, and other cutting being developed to understand the correlation between process condi-
processes [21–25]. Analytical models have been developed to describe tions and corresponding tip life. In addition to the SEM inspection, the
the evolution of tool wear during different machining processes pull-off force between the AFM tip and the sample surface was char-
[26–28]. The cutting conditions and average tool flank wear during acterized and ultilized to measure the tip wear directly without un-
milling process were correlated in the model, in which the cutting loading the tip from the AFM and interupting the machining process. To
power was employed to calculate the normal cutting power for tool study the tip wear at different conditions, the tip radius was measured
wear monitoring. Due to the difficulties to build analytical models, from the pull-off force under a wide range of machining conditions.
numerical methods have also been adapted to predict the tool wear During the nanomachining process, three regions (i.e. initial tip wear
during machining processes. A thermo-mechanical discrete element region, transition region, and tip failure region) were observed as the
model has been developed to study the cutting process [29], which tip was gradully worn off. Regression models were developed to de-
considered both mechanical and thermal phenomena and their re- scribe the tip wear at different stages, and to estimate the tip life (i.e.
ciprocal influence for the cutting process. Finite element method (FEM) when the tip needs to be changed), which provide usefully information
has also been used to investigate and optimize the tool wear in drilling for future proces planning and process optimization.
process for the difficult-to-cut nickel-based superalloy [30]. The effect
of the process variables including cutting speed, feedrate and tool 2. Experimental setup for vibration-assisted nanomachining
diameter on tool wear were analyzed and characterized. A tool wear process
model for twist drill was then built to study the mechanism of drill
wear. The result from the FEA model was demonstrated to match well The nanomachining experiments were performed on a commercial
with the experimental data. AFM, Park XE-70 (Park Systems Corp®) with a customized nano-vi-
For AFM imaging and tip-based fabrication processes, the tip radius bration system (Fig. 1(a)) [14]. The lab customized nano-vibration
is a critical factor that determines the image quality or dimensional system provided the tip-sample vibration to implement the vibration-
accuracy of the machined features. However, the tip wear is an in- assisted nanomachining process. To monitor the machining process, the
evitable result of both the imaging and tip-based nanomachining pro- signals from four-quadrant photodetector in the AFM were used to
cesses. The consequences of tip wear include the degradation of the measure the normal deflection and torsion of the cantilever as A–B and
image resolution, false measurement, reduced fabrication resolution, C–D signals. The normal deflection and torsion of the cantilever, re-
and reduced fabrication capability. Due to the significant differences in presenting the normal force and lateral forces respectively during the
the tools and machining mechanisms between the conventional scale nanomachining process, were acquired by LabView through a data
machining and micro/nanometer scale tip-based machining, many of acquisition device (NI USB-6295). The command signals (i.e. synchro-
the tool wear models for the conventional machining process cannot be nized sinusoid signals with 90° phase difference) were also generated by
directly applied for micro/nano scale machining processes. A few the data acquisition device to drive the xy-piezo actuators for xy in-
atomic-level modeling methods [[31–35]] have been developed to plane circular vibration.
study the micro/nano-scale machining process, such as molecular dy- We used a tapping mode cantilever with a nominal stiffness of 48 N/
namics method (MD), Monte Carlo (MC) simulation method. Some in- m and a resonant frequency of 190 kHz as the tool for the nano-
itial parametric and exprimental studies were performed to observe and machining process. A PMMA film was used as the sample to perform
understand the tip wear in nanomachining applications [31,32]. In tip-based vibration-assisted nanomachining. The PMMA film was spin-

48
X. Kong, et al. Journal of Manufacturing Processes 50 (2020) 47–56

coated on a silicon substrate with about 40 nm in thickness after baking which high-quality structures were reliably produced. When the tip was
to dry. The sample was mounted on the nano-vibration system to be used for certain machining operations and gradually worn, a transition
vibrated in the xy plane at a frequency of 2 KHz. To simplify the stage was observed. At this stage, machining results became unreliable
modeling process, vertical z-vibration was not applied in this study. and the trench depth reduced rapidly to below 5 nm (Fig.3(b)) due to
With the xy-vibration, the width of the machined trench can be easily the larger tip radius. Finally, at the tip failure stage, the tip was severely
controlled in one machining pass, as shown in Fig. 1(b). Moreover, with worn and tip radius became very large. The worn tip cannot remove
the assistance of the xy-vibration, the material removal was distributed materials effectively (Fig. 3(c)).
to each rotational cycle, only a small slice of material was removed in
each cycle (Fig. 1(c)), which effectively reduced the resulting ma- 3.2. In-situ tip wear detection
chining force and increased the overall machining speed [14,15].
Although the SEM images can be used to detect tip wear and mea-
3. Detection of tip wear during nanomachining process sure tip-radius effectively, this process is time-consuming and cannot be
performed in-situ. To take SEM images of the AFM tip, the tip has to be
3.1. Tip wear measurement and impact on features produced removed from the AFM, resulting in the interruption of the machining
process and reduction in the productivity. To solve this problem, a low-
During the nanomachining process, tool wear is generated by the cost, easy-to-implement method is needed to detect the tip radius and
direct contact and sliding motion between the cutting tool (AFM tip) tip wear without interrupting the nanomachining process.
and the workpiece. Along with the development of tip wear, the ma- A pull-off (or adhesive) force is seen when the AFM tip is retracted
chining force, as well as the cutting temperature, will increase, which from the sample, which provides a promising method to estimate tip
will degrade the dimensional accuracy and surface quality of the pro- radius and to detect tip wear [38]. The pull-off force can be con-
duced features. Moreover, the tip has to be replaced when the ma- veniently measured from the force-distance response curve when ap-
chining performance become unsatisfactory. The change of the tools in proaching the tip to the sample first and then retracting the tip from the
the nanomachining process increases the machining cost and reduces sample. A typical force-distance curve is shown in Fig. 4. The vertical
the productivity. As a result, tool wear is a critical factor that greatly axis represents the normal force applied on the cantilever due to the
affects the quality and productivity of the nanomachining process. For interaction between the tip and the sample, while the horizontal axis is
the tip-based vibration-assisted nanomachining, the effect of tool wear the displacement of the cantilever. When the tip is far away from the
is even more significant as the dimensions of the machined patterns are sample surface, the cantilever has no deflection and zero tip-sample
affected by the dimensions of the AFM tip. The radius of the AFM tip is interaction force. As the tip gradually approaches the sample, the
usually around tens of nm, while the depth of machined patterns is at cantilever begins to deflect toward the sample, due to an attractive
the same range. As a result, when the radius of an AFM tip increased by force between them. After the tip physically touch and push the sample,
a small amount, the dimensions of machined patterns are greatly in- the cantilever is deflected away from the surface due to the repulsive
fluenced. Unlike conventional scale machining, in which many models interaction force between the tip and the sample. Finally, when the
have been studied to analyze the tool wear during the machining pro- cantilever is retracted from the sample, the tip-sample contact breaks at
cess, little research has been performed for the tip wear in the nan- a value defined by the pull-off force, which is the adhesion force be-
ometer scale machining process. tween the tip and the sample. This adhesion force is related to the tip-
To study the tip wear systematically during the nanomachining sample contact and interaction. For a blunt tip with a large tip radius,
process, parallel experiments were conducted. Six new tips were ap- there is a larger contact between the tip and the sample, causing a
plied to machine 0, 2, 4, 8, 12 and 20 identical patterns. Each designed larger force to retract the tip. As a result, the pull-off force is a good
pattern includes 32 identical trenches with a length of 5 μm and a width indicator of the tip radius.
of 60 nm that were machined using the same machining parameters In this study, the force-distance curve can be easily measured im-
(with the setpoint force, feed rate, and XY vibration amplitude to be mediately following each machining operation. After machining on the
1000 nN, 2 μm/s and 40 mV respectively) to study the evolution of tip PMMA film, the AFM tip was moved to an area of silicon to measure the
wear and the resulting machining performance. Each tip was observed force-distance curve using an “indentation mode.” Since the humidity
in the SEM after the machining of the requisite number of patterns. level could significantly affect the measured pull-off force, the mea-
Fig. 2(a) provides the SEM images of six tips at different wear stages. surement experiments were conducted when the ambient humidity was
With the ThreePointCircularROI, a plugin of Image J software, a circle between 16% and 21%. During the measurement, both the approach
was created based on three selected points on the cutting edge of the and retraction speed were set as 0.01 μm/s, and the applied normal
tip, whose radius was used to estimate the corresponding tip radius. force was set as 200 nN. After the tip pressed the sample at 200 nN, the
With this method, the radius of these tips were measured, and the cantilever was slowly retracted. The force-distance curves were ob-
change in tip radius during the machining process was recorded, as tained for the both approach and retraction stages, which were used to
shown in Fig. 2(b). Clearly, after a tip was used to produce more pat- calculate the pull-off force.
terns, the tip experienced more wear, resulting a larger tip radius. In Fig. 5(a), the measured pull-off force and the tip radius measured
When using a brand-new tip with sharp tip radius, the trenches by the SEM were compared in the same plot. As can be observed, the
produced had large and uniform depths ranging from 20 nm to 25 nm changing trend of the pull-off forces matches very well with that of the
(Fig.3(a)). During machining, the tip gradually wore yielding a larger tip radius, and the good correlation between these two variables can be
tip radius. The blunt tip with larger radius also reduced the trench clearly observed. Therefore, the pull-off force was used as an alternative
depth. When the tip was severely worn and tip radius was too large, its for measuring the tip radius, which can be used to monitor and study
machining capability to produce nanoscale features was significantly the tip wear without interrupting the machining process. The pull-off
degraded, because with a large tip radius, the given setpoint force forces measured at different moments provided detailed information
cannot produce enough stress for the tip to indent into the sample. For about the evolvement of the tip wear during the nanomachining pro-
the recorded depth in Fig. 3(d), the depth of each trench was measured cess. Fig. 5(b) showed the development of tip wear and tip radius (re-
at five different locations. The variations from these measurements flected by the pull-off force) for a typical nanomachining condition with
were pretty small (less than 1−2 nm), which is close to the measure- the setpoint force and feed rate at 1000 nN and 2 μm/s respectively. As
ment resolution of the AFM system. From our observation of experi- the tip worn from the machining operation, the tip radius became
mental results, the evolvement of the tip wear can be categorized into larger, and a larger pull-off force was measured during the tip retraction
three stages. When a new tip was used, the tip was sharp enough from process.

49
X. Kong, et al. Journal of Manufacturing Processes 50 (2020) 47–56

Fig. 2. (a) SEM images for tip 1–6 after machining different number of groove patterns. (b) Measured tip radius at different total machining distance (one pattern is
160 μm in machining distance).

4. Tip wear behaviors under different machining conditions tip radius, the depth of the machined features decreased. After certain
numbers of the patterns were machined for each machining condition,
The tip wear behaviors were different under different process con- the depth of the machined features decreased rapidly, indicating severe
ditions, such as setpoint force and feedrate in nanomachining. It is tip wear and large tip radius that became ineffective to remove material
critical to understand the effects of these machining parameters on the at the given machining condition. Moreover, from the evolution of the
tip wear and develop the tip wear models to estimate the tip life and the pull-off force in Figs. 6–8, two regions (at the beginning of the ma-
machined distance before the tip has to be changed. To understand the chining experiments vs. at the end of machining experiments) can be
impact of the machining parameters (i.e. setpoint force and feed rate) clearly observed, which have quite different wear rates (i.e. change rate
on the tip wear, a full factorial experiment with these two factors was of the tip radius or pull-off force). These two regions corresponded to
designed and conducted. For each factor, three different values were the two different stages of the nanomachining, as the initial tip wear
assigned corresponding to the low, medium, and high level of this region and the tip failure region as we discussed before. For the first
factor, as shown in Table 1. stage (i.e. initial tip wear region), the tip was still sharp enough to
To implement this factorial design experiment, nine new tips were remove material, while at the late stage (i.e. the tip failure region), the
applied to machine trenches with different machining parameters, as tip became very blunt and cannot effectively produce trenches on the
shown in Table 1. For each machining condition, the tip was used to sample. The tip wear behavior with and without effective machining
machine multiple patterns until it was fully worn out, and cannot was quite different. During effective machining when the tip can be
produce noticeable features on the sample. The machined patterns were used to remove materials and produce trenches with large depth, the
imaged with their feature depth measured using the same AFM at the change rate of the pull-off force (i.e. tip radius) was smaller compared
tapping mode. After finish machining each pattern with 32 trenches, with the change rate when the tip cannot effectively remove material
the pull-off forces were measured using the “indentation mode” to re- and produce trenches on the sample. For the later condition, the tip
flect the tip wear at that moment. simply slid on the sample surface, resulting in a faster wear rate. In
The values of measured pull-off force and average feature depth between these two regions, a transition region was defined to describe
during the nanomachining process at three different setpoint forces the change between these two stages, in which the depth of the ma-
(500 nN, 1000 nN and 1500 nN respectively) and three feedrate (1 μm/ chined features decreased rapidly, and the machining process became
s, 2 μm/3, and 3 μm/s) were plotted from Figs. 6–8. Clearly, from these unreliable.
figures, similar trends can be observed for the change in the pull-off The process conditions have strong effects on the tip wear behavior,
force (i.e. tip radius). For each machining condition, the pull-off forces as shown in Figs. 6–8. For these different machining conditions, the
increased when more patterns were machined using that tip, indicating change rates of the pull-off forces can be estimated when the tips were
the increased tip radius. In the meantime, along with the increase of the capable to machine the patterns on the sample. Clearly, the change

50
X. Kong, et al. Journal of Manufacturing Processes 50 (2020) 47–56

Fig. 3. (a) The pattern machined with a brand-new tip during initial tip wear region. (b) The pattern machined with a partially worn tip during transition region. (c)
The pattern machined with a severely worn tip during tip failure regions. (d) The measured feature depth during machining process from different tips, represented
by different colors.

Fig. 4. Force-distance curve obtained from an AFM when approaching the tip to
and retracting the tip from the sample. The pull-off force, as the maximum
attraction force during retraction, is proportional to the tip radius.

rates of the pull-off forces (that is the rate of the tip wear) during na-
nomachining increased with a larger setpoint force and a larger feed
rate. This behavior is easy to understand, as a larger machining load
resulted in more severe wear of the tip used in the nanomachining.

5. Modeling of the tip wear and tip life Fig. 5. (a) Plot of the measured tip radius using SEM and the pull-off force. The
pull-off force is strong correlated with the tip radius. (b) A typical evolvement
To successfully apply and plan the tip-based nanomachining pro- of the measured pull-off forces during the nanomachining process under 1000
nN setpoint force and 2 μm/s feed rate. In the plots, one pattern is 160 μm in
cess, it is critical to understand tip wear at different machining condi-
machining distance.
tions and estimate the life of an AFM tip, so as to decide when to change
the tip to ensure the machining performance and plan the

51
X. Kong, et al. Journal of Manufacturing Processes 50 (2020) 47–56

Table 1 the tip is relatively sharp and can effectively remove materials to ma-
Factorial design experiments for tip wear study. chine trenches on the sample with large depth. Both the pull-off forces
Test number Setpoint force/nN Feed rate/μm/s and the change rates of the pull-off forces are small and the quality of
machined patterns in this region is good. After the tip finished ma-
1 500 1 chining a number of patterns, the transition region is observed. In the
2 500 2
transition region, the pull-off force (i.e. an indication of the tip radius)
3 500 3
4 1000 1
increases rapidly, indicating the quick wear of the tip. In the meantime,
5 1000 2 the depth of the machined features decreases rapidly. In the final tip
6 1000 3 failure region, due to the large tip radius, with the given machining
7 1500 1 condition, features/trenches cannot be effectively machined on the
8 1500 2
sample. The tip is simply sliding on the surface of the sample without
9 1500 3
removing any material. The change rate of the pull-off force is much
larger than that in the initial tip wear region, which indicates a fast
nanomachining process efficiently with the minimized cost. As can be wear rate of the tip.
seen in Figs. 6–8, the change rates of the pull-off force and the ma- Fig. 9 provides a typical example of different tip wear regions
chined feature depth were different under different machining condi- during the machining process. For the machining of the first nine pat-
tions. To model the effect of different process conditions on the tip wear terns, the machined feature depth decreases slowly, and the pull-off
rate, we need to study and quantify the change rate of the pull-off force force also increases slightly. Between the machining of the 10th pattern
and compare the results among different machining conditions. and the 13th pattern, the depth of machined features rapidly decreases.
In Fig. 9, to ease the observation of the evolution of the tip wear and Finally, after the 13th pattern, no noticeable trenches can be machined
the resulting feature depth at each wear stage, the pull-off force and the with feature depth close to zero while the pull-off force continuously
corresponding feature depth (x10) for one machining condition (i.e. increases with a steeper slope than before.
1000 nN setpoint force and 2 μm/s feedrate) were plotted together. Since each pattern machined in the experiments included 32 lines
Based on the change rate of the pull-off force and the machined feature with the length of each line of 5 μm, it is more formal to describe
depths, the nanomachining process can be roughly divided into three change of pull-off force with respect to the total machined linear dis-
regions: Initial tip wear region, Transition region, and Tip failure re- tance (160 μm for each pattern machined), regardless of whether the
gion. The initial tip wear region starts when a new tip begins to be used machined trench is visible or not. The pull-off forces during the initial
for nanomachining and produces the first few patterns. In this region, tip wear region and the tip failure region for all the machining condi-
tions (Figs. 6–8) were analyzed. Fig. 10 is an example for one

Fig. 6. Measured pull-off forces and the average depth of the machined trenches as a function of the number of machined patterns when the setpoint force is 500 nN
at three feedrates (1, 2, 3 μm/s). In the plots, one pattern is 160 μm in machining distance.

52
X. Kong, et al. Journal of Manufacturing Processes 50 (2020) 47–56

Fig. 7. Measured pull-off forces and the average depth of the machined trenches as a function of the number of machined patterns when the setpoint force is 1000 nN
at three feedrates (1, 2, 3 μm/s). In the plots, one pattern is 160 μm in machining distance.

machining condition as given in Fig. 9. The regression models fit the happens at the bottom surface, which rapidly increases the tip radius, as
data very well with goodness of fit R-squared at about 0.95. The X axis indicated by the large change rate of the pull-off force.
is the machined linear distance (μm), while the Y axis is the pull-off To study the effect of the machining conditions (i.e. setpoint force
force (nN) measured after each machined pattern. For both the initial and feedrate) on the tip wear, process models were developed to de-
tip wear region and tip failure region, linear regression lines were fitted scribe the progression of the tip wear and estimate the transition be-
to correlate the pull-off forces with the machined linear distance under tween the initial region and the tip failure region. The transition from
each machining condition, which were defined as the local fitting lines. the initial region to the tip failure region is an indicator of tip life. From
The slopes of these local fitting lines represent the change rate of the Figs. 6–8, approximately linear trends are observed for the tip wear
pull-off forces with respect to the machined distance. For the initial tip with respect to the machined distance for both the initial region and the
wear region, the intercepts of the line with the vertical axis represent tip failure region. Linear regression models were developed to correlate
the extrapolated pull-off force for a brand-new tip before machining. the pull-off forces with the machined linear distance during both the
Table 2 lists the change rates of the pull-off forces for the initial tip initial tip wear region and tool failure region. In the linear regression
wear region and tip failure region under all nine different machining model, both the slope and intercept were correlated with machining
conditions. Clearly, the change rates of the pull-off forces (that reflects parameters that were setpoint force and feedrate, which was shown as
the wear rate of the tip) increased with a larger setpoint force and a Eq. 1.
larger feed rate. This behavior is easy to understand, as a larger ma-
chining load resulted in more severe wear of the tip used in the nano- Fp = slope × Dlinear +intercept
machining. Moreover, for each machining condition, the tip wear rate is F k2

much higher (more than three times) at the tip failure region than that slope = k1 ⎛ set ⎞ feed k3 + k 4
⎝ 1000 ⎠
from the initial tip wear region. The difference in the wear rate for these F k6
two regions can be explained intuitively. At the initial tip wear region, intercept= k5 ⎛ set ⎞ feed k7 + k8
⎝ 1000 ⎠ (1)
the tip is sharp enough to indent into the sample’s surface for me-
chanical machining. Both the bottom of the tip and the side of the tip Fp is the pull-off forces measured during the initial tip wear region (nN),
interact with the sample. Tip wear happens at the bottom surface and Dlinear (μm) is the machined linear distance, Feed is the feed rate (μm/s),
the side surface, although the wear rate at the bottom tends to be larger. Fset is the applied setpoint force (nN). In the model, the setpoint force is
The tip wear on the side surface sharpens the tip and effectively reduces scaled (by a factor of 1000) to make the value of the force and feedrate
the increase rate of the tip radius. At the tip failure region, the tip is too of comparable magnitude. slope and intercept represent the slope and
blunt to be indented into the sample to have effective machining. The intercept of the linear regression model where k1, k2, k3, k 4 and k5 are
tip simply slides on the sample, and the resulting tip wear mainly regression coefficients. With the pull-off forces and corresponding

53
X. Kong, et al. Journal of Manufacturing Processes 50 (2020) 47–56

Fig. 8. Measured pull-off forces and the average depth of the machined trenches as a function of the number of machined patterns when the setpoint force is 1500 nN
at three feedrates (1, 2, 3 μm/s). In the plots, one pattern is 160 μm in machining distance.

Fig. 9. The measured pull-off forces and the feature depth (x10) as a function of
the number of machined patterns. The blue dots represent the machined feature
depth, while the boxes represent the pull-off forces. Based on the rate of change
of pull-off forces or feature depth, the tip wear process can be identified as three
regions: Initial tip wear region, transition region and tip failure region. In the
plot, one pattern is 160 μm in machining distance. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web
version of this article.)

Fig. 10. The plot of both global fitting line and local fitting line for (a) initial tip
machining parameters applied, this regression model was calibrated wear region and (b) tip failure region when the setpoint force and feed rate
and all the coefficients were determined, as shown in Eq. 2. were set as 1000 nN and 2 μm/s. The blue dots represent the measured pull-off
forces, the red lines represent the local fitting lines and the green lines represent
Finitial = slope *Dlinear +intercept the global fitting lines. (For interpretation of the references to colour in this
2.76
F figure legend, the reader is referred to the web version of this article.)
slope=0.019*⎛ set ⎞ *feed 0.23 + 0.0083
⎝ 1000 ⎠
intercept = 26.56 (2) the tips used in the experiments before machining.
Similarly, a regression model was developed and calibrated for the
As can be observed, the slope (or the wear rate) increases with the tip failure region, which is shown as Eq. 3.
larger setpoint force and feed rate, and the intercept is calculated as
26.56 nN, which is close to the average of measured pull-off forces for

54
X. Kong, et al. Journal of Manufacturing Processes 50 (2020) 47–56

Table 2 indication of tip life for that specific machining condition. From the
The wear rate measured by the change rate of the pull-off forces (nN/ μm) for regression model, two lines were used to describe the evolvement of tip
the initial tip wear region and tip failure region under different machining wear along with the machined distance. The intersection point of the
conditions. two lines for both the initial tip wear region and tip failure region
Test # Setpoint force Feed rate Wear rate at initial Wear rate at failure provided an estimation of the transition point, from which the change
(nN) (μm/s) region (nN/μm) region (nN/μm) rate of the pull-off force increased rapidly. The specific machined linear
distance at the intersection point can also be calculated from Eqs. 2 and
1 500 1 0.0097 0.0292
2 500 2 0.0083 0.0426
3 when the calculated Ffailure equal to Finitial at each machining condi-
3 500 3 0.0185 0.0667 tion. Fig. 11 provided an example showing the lines from regression
4 1000 1 0.0209 0.0623 model (with goodness of fit R-squared at 0.93) and the transition point
5 1000 2 0.0289 0.1123 indicating the tip life when the setpoint force and feed rate were 1000
6 1000 3 0.0382 0.1379
nN and 2 μm/s respectively. Table 3 listed the estimated transition
7 1500 1 0.0691 0.1528
8 1500 2 0.0755 0.1971 point in both the machined distance and the number of the machined
9 1500 3 0.0819 0.2519 patterns. The machined feature depth around the transition point was
also listed in Table 3. Compared with the evolvement of the tip wear
from Figs. 6–8, these transition points from the regression model pro-
vided a rough estimation about of tip life about when to change the tip
to achieve acceptable machining performance.

6. Summary

In this work, the tip wear during the tip-based vibration-assisted


nanomachining process was systematically studied, and process models
were developed to understand the correlation between process condi-
tions and corresponding tip life. In addition to the SEM inspection, the
Fig. 11. Lines from regression model for the initial regions and tip failure re-
pull-off force between the AFM tip and the sample surface was applied
gion. The intersection of the two lines provide an estimation of the transition
point where the tip need to be changed. to measure the tip wear directly without unloading the tip from the
AFM and without interupting the machining process. To study the effect
of the machining parameters, such as setpoint force and feed rate, on
Table 3 the tip wear process, a full factorial design experiment was performed
The estimated transition point from the regression model and the feature depth
to study the effects of these machining parameters on the tip wear, and
at the transition point.
to develop the tip wear models to estimate the tip life and the machined
Test # Setpoint Feed Machined Number of Machined distance for the tip change. During the nanomachining process, three
force (nN) rate distance (μm) patterns feature depth
regions (i.e. initial tip wear region, transition region, and tip failure
(μm/s) machined (nm)
region) were observed as the tip was gradully worn off. The wear rate of
1 500 1 1328 8.3 5 the tip was higher with a larger setpoint force and a larger feed rate, as
2 500 2 1456 9.1 2 a larger machining load resulted in more severe wear of the tip used in
3 500 3 1504 9.4 1 the nanomachining. Moreover, the tip wear rate was much high at the
4 1000 1 2032 12.7 6
5 1000 2 1952 12.2 2
tip failure region than that from the initial tip wear region. Regression
6 1000 3 1872 11.7 1 models were developed to describe the tip wear behavior for the initial
7 1500 1 2624 16.4 4 tip wear region and the tip failure region, which provided a metho-
8 1500 2 2320 14.5 4 dology to estimate the tip life (i.e. when the tip needs to be changed).
9 1500 3 2208 13.8 5
The model of the tip wear and tip life can potentially provide usefully
guidelines for future proces planning and process optimization.
Ffailure = slope *Dlinear +intercept
1.78
Declaration of Competing Interest
F
slope=0.066*⎛ set ⎞ *feed 0.55 − 0.0089
⎝ 1000 ⎠ The authors declare that they have no known competing financial
1.72
F interests or personal relationships that could have appeared to influ-
intercept=-101.57*⎛ set ⎞ *feed 0.62 + 33.66
⎝ 1000 ⎠ (3) ence the work reported in this paper.

where Ffailure is the pull-off force measured during the tip failure region Acknowledgement
(nN). Similar to the regression model for the initial tip wear region, the
slope (or wear rate) increases with the larger setpoint force or feed rate. This work was supported in part by the National Science Foundation
In Fig. 10, the pull-off forces were plotted versus the machined linear under Grant Award NSF CMMI-1233176.
distance when the setpoint force and feed set as 1000 nN and 2 μm/s
separately. As can be observed, for both the initial tip wear region and References
tip failure region, the lines from the regression models fits well to the
changing trend of the measured pull-off force. [1] Smith HI, Schattenburg ML. X-ray lithography from 500 to 30 nm: X-ray nano-
Using the regression model, the life of the tip can be roughly esti- lithography. IBM J Res Dev 1993;37(3):319–29.
mated as well. For the tip-based nanomachining process, the patterns [2] Silverman JP. X-ray lithography: status, challenges, and outlook for 0.13 μm. J Vac
Sci Technol B Microelectron Nanometer Struct Process Meas Phenom
machined during the initial tip wear region are usually acceptable, 1997;15(6):2117–24.
while during the tip failure region, the tip rapidly wears out, and no [3] Griffiths SK. Fundamental limitations of LIGA x-ray lithography: sidewall offset,
trenches can be reliably machined. To provide the guideline for the tip slope and minimum feature size. J Micromechanics Microengineering
2004;14(7):999.
change, we need to identify a machined distance where pattern quality [4] Wagner C, Harned N. EUV lithography: Lithography gets extreme. Nat Photonics
rapidly degrades, which is termed as the “transition point” and is an 2010;4(1):24.

55
X. Kong, et al. Journal of Manufacturing Processes 50 (2020) 47–56

[5] Tallents G, Wagenaars E, Pert G. Optical lithography: Lithography at EUV wave- [22] Iliescu D, Gehin D, Gutierrez ME, Girot F. Modeling and tool wear in drilling of
lengths. Nat Photonics 2010;4(12):809. CFRP. Int J Mach Tools Manuf 2010;50(2):204–13.
[6] Tseng AA, Chen K, Chen C, Ma KJ. Electron beam lithography in nanoscale fabri- [23] Gu J, Barber G, Tung S, Ren-Jyh G. Tool life and wear mechanism of uncoated and
cation: recent development. IEEE Trans Electron Packag Manuf 2003;26(2):141–9. coated milling inserts. Wear 1999;225:273–84.
[7] Chen Y. Nanofabrication by electron beam lithography and its applications: a re- [24] Rawat S, Attia H. Wear mechanisms and tool life management of WC–Co drills
view. Microelectron Eng 2015;135:57–72. during dry high speed drilling of woven carbon fibre composites. Wear 2009;267(5-
[8] Chou SY, Krauss PR, Renstrom PJ. Nanoimprint lithography. J Vac Sci Technol B 8):1022–30.
Microelectron Nanometer Struct Process Meas Phenom 1996;14(6):4129–33. [25] Tansel IN, Arkan TT, Bao WY, Mahendrakar N, Shisler B, Smith D, et al. Tool wear
[9] Guo L. Nanoimprint lithography: methods and material requirements. Adv Mater estimation in micro-machining.: part I: tool usage–cutting force relationship. Int J
2007;19(4):495–513. Mach Tools Manuf 2000;40(4):599–608.
[10] Pine RD, Zhu J, Xu F, Hong S, Mirkin CA. Dip-pen nanolithography. Science [26] Remadna M, Rigal JF. Evolution during time of tool wear and cutting forces in the
1999;283(5402):661–3. case of hard turning with CBN inserts. J Mater Process Technol
[11] Yongda Y, Sun T, Liang Y, Dong S. Investigation on AFM-based micro/nano-CNC 2006;178(1–3):67–75.
machining system. Int J Mach Tools Manuf 2007;47(11):1651–9. [27] Yen Y-C, Söhner J, Lilly B, Altan T. Estimation of tool wear in orthogonal cutting
[12] Göbel H, Von Blanckenhagen P. Atomic force microscope as a tool for metal surface using the finite element analysis. J Mater Process Technol 2004;146(1):82–91.
modifications. J Vac Sci Technol B Microelectron Nanometer Struct Process Meas [28] Karagüzel U, Uysal E, Budak E, Bakkal M. Analytical modeling of turn-milling
Phenom 1995;13(3):1247–51. process geometry, kinematics and mechanics. Int J Mach Tools Manuf
[13] Yan Y, Hu Z, Zhao X, Sun T, Dong S, Xiaodong L. Top‐down nanomechanical ma- 2015;91:24–33.
chining of three‐dimensional nanostructures by atomic force microscopy. Small [29] Rojek J. Discrete element thermomechanical modelling of rock cutting with va-
2010;6(6):724–8. luation of tool wear. Comput Part Mech 2014;1(1):71–84.
[14] Zhang L, Dong J. High-rate tunable ultrasonic force regulated nanomachining li- [30] Kolahdoozan M, Azimifar F, Yazdi SR. Finite Element Investigation and
thography with an atomic force microscope. Nanotechnology 2012;23(8):085303. Optimization of Tool Wear in Drilling Process of Difficult-to-Cut Nickel-Based
[15] Li Z, Dong J, Cohen PH. Material-insensitive feature depth control and machining Superalloy using Response Surface Methodology. Int J Adv Des Manuf Technol
force reduction by ultrasonic vibration in AFM-Based nano-machining. IEEE Trans 2014;7(2).
Nanotechnol 2013;12(5):743–50. [31] Gozen BA, Burak Ozdoganlar O. Wear of ultrananocrystalline diamond AFM tips
[16] Gozen BA, Burak Ozdoganlar O. A rotating-tip-based mechanical nano-manu- during mechanical nanomanufacturing by nanomilling. Wear 2014;317(1-2):39–55.
facturing process: nanomilling. Nanoscale Res Lett 2010;5(9):1403. [32] Chung K-H. Wear characteristics of atomic force microscopy tips: A review. Int J
[17] Deng J, Zhang L, Dong J, Cohen PH. AFM-based 3D nanofabrication using ultra- Precis Eng Manuf 2014;15(10):2219–30.
sonic vibration assisted nanomachining. J Manuf Process 2016;24:195–202. [33] Islam S, Ibrahim R. Mechanism of abrasive wear in nanomachining. Tribol Lett
[18] Deng J, Dong J, Cohen PH. Development and characterization of ultrasonic vibra- 2011;42(3):275.
tion assisted nanomachining process for three-dimensional nanofabrication. IEEE [34] Gnecco E, Bennewitz R, Socoliuc A, Meyer E. Friction and wear on the atomic scale.
Trans Nanotechnol 2018;17(3):559–66. Wear 2003;254(9):859–62.
[19] Kong X, Dong J, Cohen PH. Modeling of the dynamic machining force of vibration- [35] Cheng K, Luo X, Ward R, Holt R. Modeling and simulation of the tool wear in
assisted nanomachining process. J Manuf Process 2017;28:101–10. nanometric cutting. Wear 2003;255(7-12):1427–32.
[20] Kong X, Cohen PH, Dong J. Predictive modeling of feature dimension for tip-based [36] Skårman B, Wallenberg LR, Jacobsen SN, Helmersson U, Thelander C. Evaluation of
nano machining process. J Manuf Process 2016;24:338–45. intermittent contact mode AFM probes by HREM and using atomically sharp CeO2
[21] Özel T, Karpat Y. Predictive modeling of surface roughness and tool wear in hard ridges as tip characterizer. Langmuir 2000;16(15):6267–77.
turning using regression and neural networks. Int J Mach Tools Manuf [37] Gotsmann B, Lantz MA. Atomistic wear in a single asperity sliding contact. Phys Rev
2005;45(4–5):467–79. Lett 2008;101(12):125501.

56

You might also like