micromachines

Article
In Situ Image Acquisition and Measurement of
Microdroplets Based on Delay Triggering
Xuefeng Chang 1,2,3 , Kang Zheng 2,3 , Dan Xie 1,2,3, *, Xiayun Shu 1,2,3, * , Keyu Xu 1,2,3 ,
Wenhuan Chen 1,2,3 , Bo Li 1,2,3 and Changjian Wu 1,2,3
1 School of Mechanical and Automotive Engineering, Xiamen University of Technology,
Xiamen 361024, China; xfchang@xmut.edu.cn (X.C.); xukeyu@stu.xmut.edu.cn (K.X.);
chenwenhuan@stu.xmut.edu.cn (W.C.); libo@stu.xmut.edu.cn (B.L.); wuchangjian@stu.xmut.edu.cn (C.W.)
2 Key Laboratory of Precision Actuation and Transmission, Fujian Province University, Xiamen 361024, China;
zhengkang163@163.com
3 Program for Innovative Research Team in Science and Technology in Fujian Province University,
Fujian Province University, Xiamen 361024, China
* Correspondence: xiedan@xmut.edu.cn (D.X.); shuxiayun@xmut.edu.cn (X.S.);
Tel.: +86-592-6291-386 (D.X. & X.S.)




Received: 19 January 2019; Accepted: 20 February 2019; Published: 22 February 2019


Abstract: An in situ image acquisition apparatus based on delay triggering for visualizing
microdroplets formation is described. The imaging system includes a charge-coupled device camera,
a motion control card, a driving circuit, a time delay triggering circuit, and a light source. By adjusting
the varying trigger delay time which is synchronized with respect to the signal for jetting, the steady
sequential images of the droplet flying in free space can be captured real-time by the system.
Several image processing steps are taken to measure the diameters and coordinates of the droplets.
Also, the jetting speeds can be calculated according to the delay time interval. For glycerin/water
(60:40, mass ratio), under the given conditions of the self-made pneumatically diaphragm-driven
drop-on-demand inkjet apparatus, the average of diameter and volume are measured as 266.8 µm and
9944 pL, respectively, and the maximum average velocity of the microdroplets is 0.689 m/s. Finally,
the imaging system is applied to measure the volume of 200 microsolder balls generated from the
inkjet apparatus. The average diameter is 87.96 µm, and the relative standard deviation is 0.83%.
The results show good reproducibility. Unlike previous stroboscopic techniques, the present in
situ imaging system which is absence of instantaneous high intensity light employs two control
signals to stimulate the microdroplet generator and the charge-coupled device (CCD) camera. Hence,
the system can avoid the desynchronization problem of signals which control the strobe light-emitting
diode (LED) light source and the camera in previous equipment. This technology is a reliable and
cost-effective approach for capturing and measuring microdroplets.

Keywords: microdroplet jetting; in situ; image processing; delay triggering

1. Introduction
Drop-on-demand microjetting is a noncontact printing method for depositing microdroplets on
various substrates. This technology has been widely used in a variety of applications including solders,
ceramics, polymers, biopolymers arrays, and so on [1,2]. In order to be a reliable production technology,
the jetting performance should be captured and controlled accurately. Then, the approach to obtain
the sequential images of the microdroplet formation process and measure the droplet characteristics
including size and speed are two essential crucial factors. However, the whole stages during drop
formation last less than 3 ms. In this process, even a 104 FPS (frames per second) camera can capture

Micromachines 2019, 10, 148; doi:10.3390/mi10020148 www.mdpi.com/journal/micromachines

Micromachines 2019, 10, 148 2 of 16

only one or two images. Therefore, the common high-speed camera simply cannot capture the details of
the microdroplet formation processes. The ultrahigh-speed camera can directly implement sequential
photograph for jetting process. But the price is extremely expensive [3,4]. Accordingly, a reliable and
low-cost image system for visualizing the microdroplet formation dynamic motion is necessary.
In recent years, the emergence of stroboscopic technique provided an effective approach to
record microjetting. Along with this technology was put forward, many image acquisition systems of
microdroplets based on delay triggering have been proposed, and their components and features are
illustrated in Table 1. Dong et al. [5] described a visualizing apparatus based on flash photography
including a pulsed laser, a charge-coupled camera, and signal generators. Sharp images with sufficient
temporal resolution had been obtained. Hutchings et al. [6] described a technique which had been
developed for analyzing the shape and development of ink jets and drops. Kwon et al. [7] proposed
an image measurement system to observe and analyze the image of the droplet. The components
were the multifunction I/O unit, a strobe light-emitting diode (LED) light and a charge-coupled
device (CCD). The two counters in multifunction I/O unit could generate two digital pulse trains.
The first digital pulse was to stimulate the waveform for the jetting. The second was used as
a trigger signal for controlling a strobe LED light which was synchronized with the signal used
to produce the ink jet. By adjusting the delay time, the microdrop formation process could be
obtained. The in situ measurement system built by Fan et al. [8] was composed of a CCD camera, a zoom
lens, and a stroboscopic light. The strobe light was simultaneously triggered by the same frequency
as the function generator which actuated the piezoelectric ceramic film. By selecting a proper shutter
speed, a steady image of the droplet could clearly be viewed and captured by the camera. Jo et al. [9]
also employed the stroboscopic method to get the pictures of drop deformation by changing the
delay time between the pulse inception and the flash of LED light. The precise real-time measurement
technology proposed by Thurow et al. [10] was implemented in most appliances with the aid of
a stroboscope camera system. The hardware was made up of a camera with an enlarging lens and
a LED stroboscope. In the international standards of IEC 62899-302-1:2017 Printed electronics Part
302-1 and IEC 62899-302-2:2018 Printed electronics Part 302-2, the imaging based measurement of jetted
drop speeds and drop volume was described [11,12]. By means of the above techniques, the speed
and diameter of the microdroplet could be calculated based on the CCD camera image and the
delay time [13–15]. Martin et al. published analysis of inkjet droplet image processing techniques
and errors [16,17]. Laser holographic methods in 3D are described by Martin et al. [18] Hsiao et al.
has published imaging work on solder balls [19]. Jung described a high-speed imaging which was
used to analyze the impact and spreading the drops onto smooth glass surfaces [20]. Jiang et al. used
a low-cost microcontroller board (typical cost is less than a hundred US dollars) as the pulse generator
in a flash photographic system [21]. Martin et al. also described the measurement of the volumes of
small (10–100 µm) liquid drops is important in a number of fields including inkjet printing, liquid
dispensing, and spraying [22].
Among these systems, the print heads are all piezoelectric transducers and the image systems
include the stroboscopic illumination source and CCD camera. Then, three control signals are applied
to operate the ejection of the microdroplet, fire the strobe, and trigger the CCD. Therefore, the camera
exposure moment must strictly coincide with the firing time of the LED light because photographing
needs instantaneous high intensity light. Otherwise, the microdroplet image will be blurred due to
insufficient illumination. However, the desynchronization problem about the signals will inevitably
affect measurement quality and accuracy. Consequently, reducing the number of control signals and
employing continuous intense light source of stable luminous flux are ideal techniques to improve the
reliability and precision of this type of low-cost image system.
This paper proposed a novel in situ image acquisition and measurement of microdroplets based
on delay triggering. Using a high-speed CCD camera, a motion control card, a driving circuit, a time
delay triggering circuit, and a light source, an imaging system was built to obtain the sequential images
of microdroplets generated through the pneumatically diaphragm-driven actuator. By adjusting the
Micromachines 2019, 10, 148 3 of 16

delay time, stable microdroplet fabrication process could be captured. The image processing methods
have been adapted to measure the diameters, coordinates, and speeds of the microdroplets in real-time.

Table 1. Image acquisition system of microdroplets based on delay triggering and their features.

Number of
Ref. Printhead Type Jetting Material Control Object by Trigger Signals
Signals
Glycerin and water Trident printhead
Trident printhead Pulsed Cu
[5] mixture in the 3 by piezo CCD camera
by piezo transducer vapor laser
weight ratio 48/52 transducer
Xaar 126-200 print
head with a linear Semi-transparent
[6] 3 Print head Flash source Camera
array of 50 µm UV-curable ink
diameter nozzles
Nanoparticles
Piezo inkjet Piezo inkjet Strobe LED
[7] dissolved in DI 3 CCD camera
printhead printhead light
(deionized) water
Piezo inkjet Piezo inkjet Stroboscopic
[8] Liquid crystal 3 CCD camera
printhead printhead light
Piezoelectric sleeve Piezoelectric
Glycerol and water Strobe LED
[9] transducer 3 sleeve transducer CCD camera
mixture light
printhead printhead
Water, DMSO, NMP,
Piezo inkjet Acetonitrile, Piezo inkjet Strobe LED
[10] 3 CCD camera
printhead Dichloromethane, printhead light
Methanol
Pneumatically Glycerin and water Pneumatically
Proposed
diaphragm-driven mixture (60/40, 2 diaphragm-driven CCD camera
article
actuator mass ratio) actuator

2. Experimental Set-Up

2.1. Pneumatically Diaphragm-Driven Actuator

Figure 1 shows photos of drop-on-demand microjetting system and pneumatic diaphragm
microdroplet actuator. The actuator consists of a reservoir, a pneumatic chamber connecting with
a solenoid valve, a driving chamber separated by a diaphragm made of stainless steel, and a microglass
nozzle. When a voltage pulse is applied to the solenoid valve, the generated pneumatic pressure pulse
will act on the diaphragm. Then, the diaphragm will bend and its deformation will push a certain
amount of liquid
Micromachines out9, of
2018, x the nozzle and form the microdroplet. 4 of 17

Figure
Figure 1. (a)1.Drop-on-demand
(a) Drop-on-demand microjettingsystem,
microjetting system,(b)
(b)side
side view,
view, (c) top
top view,
view,and
and(d)
(d)bottom
bottomview
view of
of pneumatic
pneumatic diaphragm
diaphragm microdroplet
microdroplet actuator.
actuator.

2.2. In Situ Imaging System

Figure 2 shows the in situ imaging system setup used for observing the microdroplet
performance. The system is composed of a high-speed CCD camera, a 4× zoom lens, a light source
with the accessory of the fiber light guide, a motion control card, a driving circuit, and a time delay
 Figure 1. (a) Drop-on-demand microjetting system, (b) side view, (c) top view, and (d) bottom view
of pneumatic
Micromachines diaphragm
2019, 10, 148 microdroplet actuator. 4 of 16

2.2. In Situ Imaging System

2.2. In Situ Imaging System
Figure 2 shows the in situ imaging system setup used for observing the microdroplet
Figure
performance. 2 shows the in situ
The system imaging system
is composed setup usedCCD
of a high-speed for observing
camera, athe 4×microdroplet
zoom lens, aperformance.
light source
The system is composed of a high-speed CCD camera, a 4 × zoom
with the accessory of the fiber light guide, a motion control card, a driving circuit, lens, a light and
source with
a time the
delay
accessory
circuit. TheofVolpi
the fiber lightseries
Intralux guide, a motion
light control card,
source provides a driving
a reliable, circuit, high-intensity
dependable, and a time delay cold
circuit. The Volpi Intralux series light source provides a reliable, dependable,
halogen light. The Intralux includes infrared reflecting filters to ensure the coldest light and the high-intensity cold
halogen light. The
unique crescent Intralux
shaped includes infrared
diaphragm, reflectingconstant
which provides filters tocolor
ensure the coldest light-emitting
temperature, light and the
unique crescent
angle, and shapedintensity
continuous diaphragm, which provides
adjustment. constant
The selected coloristemperature,
model IX 4000-1, and light-emitting angle,
the specifications
and continuous intensity adjustment. The selected model is IX 4000-1, and
are shown in Table 2. The external trigger signal is simultaneously stimulated by the same the specifications are shown
in Table 2. The
frequency as external trigger which
the generator signal isactuates
simultaneously stimulated
the solenoid by the
valve. same frequency
Although the measuredas the
generator
microdropletswhich areactuates the solenoid
fast moving objects,valve. Although
a steady image the measured
of the microdroplets
microdroplet are fast
can clearly moving
be captured
objects,
by the system. When the time of exposure is too long, the image edge will be blurred. The time
a steady image of the microdroplet can clearly be captured by the system. When the time of
of
exposure
CCD exposureis too long, the image
has been set toedge
35 μs,willand
be blurred. The time
the brightness andof the
CCD exposure
contrast arehas been
both set to 35 µs,
adjustable on
and the brightness and the contrast are both adjustable on demand.
demand.

Figure 2.
Figure 2. In situ image acquisition and measurement
measurement system.
system.

Table 2. The specifications of Volpi Intralux Model 4000-1.

Illumination Uniformly
Rated Max
Intensity at Illuminated Color Light
Lamp Housing Lamp-EPV
Light Guide Light Guide Temperature Adjustment
Life Temp
Receptacle Diameter
Crescent Halogen
3150 ◦ K,
320,000 ftc 10 mm 1000 hrs shaped 100 ◦ K reflector,
constant
diaphragm 14.5 V/90 W

For the CCD camera, a Photonfocus MV-D750E with resolution of 1024 × 760 pixels is used.
The camera includes free running and external trigger mode. When the acquisition of an image needs
to be synchronized to an external event, a trigger mode can be used. As seen in Figure 3, camera control
signals (CC-signals) can be defined by the camera manufacturer to provide certain signals to the
camera. There are four CC-signals available and are all unidirectional with data flowing from the
frame grabber to the camera. For example, the external trigger is provided by a CC-signal.
 For the CCD camera, a Photonfocus MV-D750E with resolution of 1024 × 760 pixels is used. The
be synchronized to an external event, a trigger mode can be used. As seen in Figure 3, camera control
camera includes free running and external trigger mode. When the acquisition of an image needs to
signals (CC-signals) can be defined by the camera manufacturer to provide certain signals to the
be synchronized to an external event, a trigger mode can be used. As seen in Figure 3, camera control
camera. There are four CC-signals available and are all unidirectional with data flowing from the
signals (CC-signals) can be defined by the camera manufacturer to provide certain signals to the
frame grabber to the camera. For example, the external trigger is provided by a CC-signal.
camera. There are four CC-signals available and are all unidirectional with data flowing from the
Micromachines 2019, 10, 148 5 of 16
frame grabber to the camera. For example, the external trigger is provided by a CC-signal.

Figure 3. Trigger delay visualization from the trigger source to the camera.
Figure 3.
Figure Trigger delay
3. Trigger delay visualization
visualization from
from the
the trigger
trigger source
source to
to the
the camera.
camera.
In this work, the motion control card and the time delay circuit are used to generate digital
pulse trains as Figure 4 shown. Pulse Iand
is generated fromcircuit
motion arecard. Pulse II is used aspulse
pulse
In this
In this work,
work,thethemotion
motioncontrol
control card
card thethe
and time delay
time delay circuit used to generate
are used digital
to generate digital
signalasfor
trains solenoid
Figure 4 shown.valvePulse
in order to generate themotion
waveform for theIIjetting. Pulse III is used foras
pulse trains as Figure 4 shown.I Pulseis generated from
I is generated from card.
motion Pulse
card. isPulse
usedII asispulse
usedsignal
as pulse
externalvalve
solenoid trigger
in signaltofor
order controlling
generate the CCD. Thefor
waveform delay
the time can
jetting. theIIImodulated
Pulse is used as manually
external via an
trigger
signal for solenoid valve in order to generate the waveform for the jetting. Pulse III is used as
adjustable
signal resistance
fortrigger
controlling in the
CCD. time delay circuit.
theDue to the time delay via in Pulse III, the droplet at inthe
external signal for The delay
controlling time
CCD.can The modulated
delay timemanually
can the modulatedan adjustable resistance
manually via an
delayed
the time circuit.
time delay can appear
Due to bethesteady.
time When in setting a certain minimal interval time step, we can to get
adjustable resistance in the time delaydelay
circuit. Pulse
Due to III, thetime
the droplet at the
delay delayed
in Pulse III,time can appear
the droplet at the
bethe combination
steady. When pictures
setting a of theminimal
certain ejectinginterval
process.timeBecause
step, thecan
we ejecting
get theprocess pictures
combination are captured
pictures of the
delayed time can appear to be steady. When setting a certain minimal interval time step, we can get
at different
ejecting times,
process. reproducibility
Because isprocess
the prerequisite for captured
the technology. By the imaging system, the
the combination pictures the ejecting
of the ejecting pictures
process. are
Because the ejectingat different times,
process pictures reproducibility
are captured
isejected
at the
different
trajectoryfor
prerequisite canthe
betechnology.
captured and
times, reproducibility is the
the
the droplet
Byprerequisite
imagingfor diameters
system, and
the velocities
ejected
the technology.
can becan
trajectory
By the imaging
measured
be captured
system,
from
the
these
and the pictures.
droplet The actual
diameters and time delayscan
velocities between
be the drop
measured from ejection
these pulses and
pictures. The the external
actual time trigger
delays
ejected trajectory can be captured and the droplet diameters and velocities can be measured from
pulses can
between be captured
the drop ejection with anand oscilloscope, as trigger
Figure pulses
5 shown. can Figure 6 is the operation interface
these pictures. The actualpulses
time delays thebetween
external the drop ejection be captured
pulses and with an oscilloscope,
the external trigger
asofFigure
microjetting
5 shown. system
Figure developed
6 is the by Visual
operation C++ 6.0.
interface of The camerasystem
microjetting image developed
can be importedby into
Visual C++the
pulses can be captured with an oscilloscope, as Figure 5 shown. Figure 6 is the operation interface
software using
6.0.microjetting
The camera systemthe frame
image can grabber.
be imported into the software using the frame
of developed by Visual C++ 6.0. The camera imagegrabber.
can be imported into the
software using the frame grabber.

Micromachines 2018, 9, x 4. (a) Pulse signal flow chart, (b) pulse signal timing sequence diagram.
Figure 6 of 17
Figure 4. (a) Pulse signal flow chart, (b) pulse signal timing sequence diagram.

Figure 4. (a) Pulse signal flow chart, (b) pulse signal timing sequence diagram.

Figure 5. Pulse signal for solenoid valve and external trigger

trigger signal
signal for
for charge-coupled
charge-coupleddevice
device(CCD).
(CCD).

MicroJetting v1.0

Output pulse signal Cursor location

 Figure
Micromachines 5. Pulse
2019, signal for solenoid valve and external trigger signal for charge-coupled device
10, 148 6 of 16
(CCD).

MicroJetting v1.0

Output pulse signal Cursor location

Options
Return to Ref-point Motor start-up
Template setting Mode
Motor setting Jetting
Jetting
PARM detection
Detecting
Set initial point Automatch
Set jetting steps
Set jetting position

Fine motion Coarse motion

Start
Motor zero Parameter Note
Array position X：0 X origin Y origin Motor speed Basic value
Array position Y：0
Jetting position X：0 Jet origin Jet position Template Position
Jetting position Y：0
Jetting time： ms
X:276 Y:89 Num

Figure6.6.Operation
Figure Operation interface of microjetting
interface of microjettingsystem.
system.

3. Image Acquisition
3. Image Acquisition
Time-lapse
Time-lapse images
images of microdroplet
of microdroplet generation cancan
generation be easily acquired
be easily by the
acquired byinthe
situ
inimaging system.
situ imaging
When setting the minimal interval time step as 100 µs, though adjusting
system. When setting the minimal interval time step as 100 μs, though adjusting the variablethe variable resistance in the
time delay circuit, we can get the combination pictures of the ejecting process by
resistance in the time delay circuit, we can get the combination pictures of the ejecting process by adjusting the interval
from 0 to 5000
adjusting the or more.from
µsinterval In our experiment,
0 to 5000 μs or 50 samples
more. In our ofexperiment,
the microdroplet are captured
50 samples and evaluated
of the microdroplet
at each moment.and
are captured And 5 sets of at
evaluated sequence diagramAnd
each moment. are obtained
5 sets of to illustrate
sequence the flight
diagram are and formation
obtained to
process of microdroplets, as the Supplementary Materials shown. Figure 7 describes
illustrate the flight and formation process of microdroplets, as the supplementary materials shown. the time-lapse
Figure
images of7microdroplet
describes the generation
time-lapse images
with one of microdroplet
primary droplet. generation with one
The solution is primary droplet. (60:40,
glycerin/water The
solution
mass ratio).isThe
glycerin/water
viscosity and (60:40, masstension
surface ratio). are
The10viscosity
MPa·s and and surface
69 mN/m, tension are 10 MPa·
respectively. s andthe
When
69 mN/m,
driving respectively.
pressure was 0.39When Mpa, the
the driving
time of pressure
high level was
for0.39
inputMpa, thewas
pulse time1.7 of ms
highand
level
thefor input
operating
pulse was 1.7 ms and the operating frequency was 2 Hz; the ejecting process from
frequency was 2 Hz; the ejecting process from 900 µs to 3800 µs is shown in Figure 7. It can be clearly 900 μs to 3800 μs
is the
seen shownstepsinofFigure 7. Itelongation,
ejection, can be clearly seen breakup,
necking, the stepsoscillation,
of ejection,and
elongation,
formation; necking,
only one breakup,
primary
droplet came into being under these conditions. Since the ejected trajectory can be captured, the Since
oscillation, and formation; only one primary droplet came into being under these conditions. droplet
the ejected
diameters and trajectory
velocities can
can be
be captured,
measured the
fromdroplet
these diameters
pictures and velocities
through the can be
method of measured
image from
processing.
Micromachines 2018, 9, x 7 of 17
these pictures through the method of image processing.

Figure 7. Time-lapse images of microdroplet generation (one primary droplet).

Figure 7. Time-lapse images of microdroplet generation (one primary droplet).

4. Image Processing
The digital image signals captured directly by CCD generally contain a variety of noise and
distortion. The typical defects include gray concentration caused by insufficient illumination and
noisy signal during the transmission process from CCD to the computer. Although the
Micromachines 2019, 10, 148 7 of 16

4. Image Processing
The digital image signals captured directly by CCD generally contain a variety of noise and
distortion. The typical defects include gray concentration caused by insufficient illumination and noisy
signal during the transmission process from CCD to the computer. Although the improvement of
image capture hardware can effectively reduce the defects, image processing technology on the initial
image for correction, enhancement, noise reduction, and so on, it is an essential way to get a distinct
image and obtain the accurate data. The image processing algorithms and extraction methods of
microdroplet are described as follows.

4.1. Pretreatment of Droplet Images

Because the collected original images have salt-and-pepper noise, and the target (droplet) and
background contrast are not high, the acquired image needs to be enhanced. The purpose of image
enhancement is as follows. Firstly, the visual effect of the image and the clarity of the image can be
improved. Secondly, the image can be converted into a more suitable form for human or machine
analysis and processing. According to the purpose and effect of enhancement, there are many
methods for image enhancement. Image enhancement can be divided into grayscale correction, image
smoothing, image sharpening, image enhancement, and color enhancement according to processing
space. The median filtering method is used to preprocess the collected images. The reasons why the
median filtering method is an appropriate method are as follows.
(1) Median filtering is a typical nonlinear filtering technique, which can overcome the image detail
blur caused by linear filters (such as the mean filter) under certain conditions.
(2) Median filter is a typical low-pass filter, which is mainly used to suppress impulse noise. It can
completely filter out the interference noise of sharp wave, and at the same time has the characteristics
of better protection of the edge of the target image.
(3) The median filter is nonlinear and has a strong filtering effect on salt-and-pepper noise or
pulsed interference.
The three characteristics of the median filter method are exactly in line with the requirements
of “protecting the edge of the target image and eliminating salt-and-pepper noise”. The image
preprocessing achieved good results. Subsequently, in order to further the contrast between the
droplet and the background, a grayscale linear transformation was performed on the median-filtered
picture. Finally,
Micromachines the
2018, 9, x small area where the droplets are located is obtained by means of threshold 8 of 17
segmentation, feature extraction, and morphology.
The purpose
purposeofofthe theabove
abovemethod
methodis to
is reduce
to reducethe scope of image
the scope processing
of image to a range
processing to a of radius
range of
(R + ∆r) (R refers to the droplet radius, ∆r > 0 is the droplet radius extension value), so
radius (R + Δr) (R refers to the droplet radius, Δr > 0 is the droplet radius extension value), so that that only the
next
only step is to deal
the next step with
is to the
dealsmall
witharea
the where the droplets
small area are droplets
where the located. The specific steps
are located. are shown
The specific as
steps
Figure 8. Figure
are shown 9 shows
as Figure the steps
8. Figure of droplet
9 shows image
the steps preprocessing.
of droplet image preprocessing.

Original
Original Median
Median Feature
Feature
Binarization
Binarization
image
image filtering
filtering selection
selection

Image
Image Region
Region Region
Region fill
fill
difference
difference dilation
dilation

Figure 8. Image preprocessing flow chart.

chart.
 Image
Image Region
Region Region
Region fill
fill
difference
difference dilation
dilation
Micromachines 2019, 10, 148 8 of 16
Figure 8. Image preprocessing flow chart.

Figure 9. Droplet image preprocessing. (a) original image, (b) median filter, (c) binarization, (d) connection,
Figure 9. Droplet image preprocessing. (a) original image, (b) median filter, (c) binarization, (d)
(e) feature selection, (f) region fill, (g) region dilation, and (h) image difference.
connection, (e) feature selection, (f) region fill, (g) region dilation, and (h) image difference.
4.2. Segmentation of Droplet Images
4.2. Segmentation of Droplet Images
Image segmentation is the technique and process of segmenting an image into a number of specific
Image
regions withsegmentation
unique properties is the and
technique and the
extracting process
targetofofsegmenting
interest. It is anaimage
crucialintostepafrom
number imageof
specific regions
processing withanalysis.
to image unique properties and extracting
After the previous the targetaofsmall
preprocessing, interest.
range It of
is aregions
crucialcontaining
step from
image processing to image analysis. After the previous preprocessing, a
only droplets can be obtained, which reduces the interference factor of the subsequent segmentation small range of regionsof
containing only droplets can be obtained, which reduces the
the droplet image and simplifies the difficulty of droplet contour extraction. interference factor of the subsequent
segmentation
The existingof the droplet
image image and simplifies
segmentation methods arethe difficulty of droplet
mainly divided intocontour extraction.
the following categories:
The existing image segmentation methods are mainly divided
threshold-based segmentation methods, region-based segmentation methods, edge-based into the following categories:
threshold-based
segmentation methods, segmentation methods,
and segmentation region-based
methods segmentation
based on specific theories.methods, edge-based
segmentation
Based on methods,
the advantagesand segmentation
of the Cannymethods
algorithm based on specific
and the theories.
characteristics of the droplet image, the
Canny operator in the direction operator is used to detect the edge. The directionofoperator
Based on the advantages of the Canny algorithm and the characteristics the droplet image,
is based on
the Canny operator in the direction operator is used to detect the edge. The direction
the differentiation in a particular direction to detect the edge. In the following three standard senses, operator is
based on the differentiation in a particular direction to detect
the Canny operator is optimal for step edges that are affected by white noise. the edge. In the following three
standard senses,standards:
(1) Testing the Cannyno operator
importantis optimal forlost,
edges are stepandedges thatshould
there are affected by white
be no false edges.noise.
1)
(2)Testing standards:
Positioning nothe
criteria: important
deviation edges are lost,
between the and there
actual edgeshould
and thebe no false edges.
detected edge position is
2) Positioning
the minimum. criteria: the deviation between the actual edge and the detected edge position is
the minimum.
(3) Single response criteria: reduce multiple responses to a single edge response.
3) Single response criteria: reduce multiple responses to a single edge response.
The edge detection is performed by using the canny operator of Figure 9 to obtain the sub-pixel
Theof edge
outline the detection
droplet, is performed
as shown in Figureby 10. using the canny operator of Figure 9 to obtain the
Micromachines 2018, 9, x 9 of 17
sub-pixel outline of the droplet, as shown in Figure 10.

Figure 10. Droplet contour extraction (sub-pixel accuracy).

4.3. Circle Fitting

4.3. Circle Fitting of
of the
the Droplet
Droplet Profile
Profile
Since
Since the
theimage
imagewaswasacquired
acquiredunder
understeady-state conditions,
steady-state the droplets
conditions, gradually
the droplets became
gradually stable
became
during the flight
stable during theuntil
flighta standard sphere was
until a standard formed
sphere (assuming
was formed the droplets
(assuming were flying
the droplets werefarflying
enough).
far
Therefore,
enough). Therefore, the outline of the droplets in the steady state is approximately circular. of
the outline of the droplets in the steady state is approximately circular. If the contour the
If the
contour of the droplet extracted in Section 4.2 has a roundness of 0.98 or more, it can be considered
that the droplet under this condition is in a stable state, and that its morphology can be
approximated as a standard sphere. However, due to the presence of the reflective surface of the
droplet and the presence of salt-and-pepper noise, the extracted edges are not smooth. In order to
facilitate the calculation of the liquid volume and the accuracy of the edge extraction, the shape of
 Figure 10. Droplet contour extraction (sub-pixel accuracy).

4.3. Circle Fitting of the Droplet Profile

Since the image was acquired under steady-state conditions, the droplets gradually became
stable during
Micromachines the
2019, 10,flight
148 until a standard sphere was formed (assuming the droplets were flying 9 offar
16
enough). Therefore, the outline of the droplets in the steady state is approximately circular. If the
contour of the droplet extracted in Section 4.2 has a roundness of 0.98 or more, it can be considered
droplet
that theextracted
droplet inunder
Section 4.2 condition
this has a roundness
is in ofa 0.98 or more,
stable state, itand
can that
be considered that the droplet
its morphology can be
under this condition is in a stable state, and that its morphology can be approximated
approximated as a standard sphere. However, due to the presence of the reflective surface of the as a standard
sphere.
droplet and However, due to of
the presence thesalt-and-pepper
presence of thenoise,
reflective surface of
the extracted the droplet
edges are not and the presence
smooth. In order ofto
salt-and-pepper noise, the extracted edges are not smooth. In order to facilitate
facilitate the calculation of the liquid volume and the accuracy of the edge extraction, the shape ofthe calculation of the
liquid volume edge
the sub-pixel and the accuracy
obtained of theisedge
above extraction,
fitted. the shape
A circle-fit image ofofthethesub-pixel
droplet edge obtained
profile above
as shown in
is fitted.11Awas
Figure circle-fit image of the droplet profile as shown in Figure 11 was obtained.
obtained.

Figure 11. Circle fitting of the droplet profile (sub-pixel accuracy).

accuracy).

5. Measurement of
5. Measurement of Micro-Drops
Micro-Drops

5.1. The Volume of the Drops

5.1. The Volume of the Drops
As shown
Micromachines 2018,in
9, Figure
xFigure11, the fitted curve is an accurate circle; the microdroplet can be described 10 of as
17
As shown in 11, the fitted curve is an accurate circle; the microdroplet can be described
a sphere. For a spherical model, its volume derives directly from the radius of the two-dimensional
as a sphere. For a spherical model, its volume derives directly from the radius of the
contour profile. In
According to this
thestudy,
resultstheofdroplet volume
the circle is calculated
fitting by the
in Section 4.3,radius,
we can theeasily
estimated
get volume
the pixelis
two-dimensional contour profile. In this study, the droplet volume is calculated by the radius, the
coordinates and radius of the droplet. In this study, one pixel corresponds to ~2.01 μm. The
estimated volume is 4 3
obtained center coordinates and radius are shown V 4= inπR Figure 12. (1)
33
Therefore, the diameter and volume of the3droplet under this condition can be calculated as (1)
𝑉 = π𝑅
According to the results of the circle fitting in Section 4.3, we can easily get the pixel coordinates
D = 66.379 × 2.01 × 2 ≈ 266.8 µ m
and radius of the droplet. In this study, one pixel corresponds to ~2.01 µm. The obtained center
coordinates and radius are shown V = (4/3)in×3.14
Figure 12.
× (66.379×2.01) 3 × 10−3 ≈ 9944 pL

Figure 12. Center coordinates and radius of the droplet (unit: pixel).

The canny algorithm is used for droplet contour extraction, which achieves the sub-pixel
accuracy, then the radius deviation value is approximately ±1 pixel (1 pixel = 2.01 μm). Compared
with barycentric coordinates and center coordinates extracted from the fitting circle, the coordinate
deviation value is approximately ±3 pixels (1 pixel = 2.01 μm). Then, the error range of radius is ±1.5%
and the error range of center coordinate is ±2.26%. These shows that the errors were nearly all less
than ±3%.
Micromachines 2019, 10, 148 10 of 16

Therefore, the diameter and volume of the droplet under this condition can be calculated as

D = 66.379 × 2.01 × 2 ≈ 266.8 µm

V = (4/3) × 3.14 × (66.379 × 2.01)3 × 10−3 ≈ 9944 pL

The canny algorithm is used for droplet contour extraction, which achieves the sub-pixel accuracy,
then the radius deviation value is approximately ±1 pixel (1 pixel = 2.01 µm). Compared with
barycentric coordinates and center coordinates extracted from the fitting circle, the coordinate deviation
value is approximately ±3 pixels (1 pixel = 2.01 µm). Then, the error range of radius is ±1.5% and the
error range of center coordinate is ±2.26%. These shows that the errors were nearly all less than ±3%.

5.2. The Speed of the Drops

The microdroplet speed is calculated by varying trigger delay time of CCD which is synchronized
in regard to the stimulating signal for jetting. The droplet speed can be calculated from the two CCD
camera images with the trigger delay time of CCD. The analyzed results need to be converted to
distance in micrometers in order to calculate the speed of a droplet. Therefore, the velocity of the
droplet can be calculated according to the following formula.

py2 − py1
ν= × µm/pixel (2)
t2 − t1

where, py is pixel location of the droplet’s center in y direction and t1 and t2 are the respective trigger
delay times according to the two CCD images.
In order to calculate the velocity change of the process in which droplets are ejected from the
nozzle until a stable sphere is formed in the air, the steps to calculate the velocity of the liquid at
different times are as follows.
(a) Collecting an image at the nozzle where no droplets are generated and obtaining the area
where the nozzle
Micromachines x located by image processing, the red area as shown in Figure 13 at t = 900 µs.
2018, 9,is 11 of 17

Figure 13. The area where the nozzle is located when there is no injection.

(b) Collecting images of different time delays and obtaining

obtaining the area of the nozzle and liquid
through image processing, the extracted area
area is
is shown
shown in
in the
the red
red area
area in
in Figure
Figure 14.
14.
 Figure 13. The area where the nozzle is located when there is no injection.

(b) Collecting
Micromachines images
of different time delays and obtaining the area of the nozzle and 11
2019, 10, 148 liquid
of 16
through image processing, the extracted area is shown in the red area in Figure 14.

Figure 14. The area where the nozzle and droplet are located at different times.
Figure 14. The area where the nozzle and droplet are located at different times.
It should be noted that the images acquired in steps (a) and (b) need to be in the same lighting
It should be noted that the images acquired in steps (a) and (b) need to be in the same lighting
environment and the image processing parameters should be consistent.
environment and the image processing parameters should be consistent.
(c) Subtracting the region obtained in the step (b) and the region obtained in (a), respectively.
(c) Subtracting the region obtained in the step (b) and the region obtained in (a), respectively.
Obtaining a region of the liquid outside the nozzle at different times, the results obtained after the
Obtaining a region of the liquid outside the nozzle at different times, the results obtained after the
regional subtraction
Micromachines 2018, 9, x are shown in Figure 15. 12 of 17
regional subtraction are shown in Figure 15.

Figure 15. The shape of the droplets outside the nozzle at different times.
Figure 15. The shape of the droplets outside the nozzle at different times.
(d) Calculating the barycentric coordinates of each region by the operator in HALCON, the results
(d) Calculating the barycentric coordinates of each region by the operator in HALCON, the
are shown in Table 3.
results are shown in Table 3.

Table 3. Average abscissa, ordinate, and velocity of droplets at different times and their
corresponding standard deviations.

Horizontal Average Vertical

Average Average Velocity
Micromachines 2019, 10, 148 12 of 16

Table 3. Average abscissa, ordinate, and velocity of droplets at different times and their corresponding
standard deviations.

Average Horizontal Average Vertical

Average Velocity
Horizontal Coordinate Vertical Coordinate
Time (µs) Velocity Standard
Coordinate Standard Coordinate Standard
(m/s) Deviation
(pixel) Deviation (pixel) Deviation
900 141.6 0.100 21.3 0.134 0 0
1000 138.0 0.487 55.3 0.415 0.683 0.008
1100 137.7 0.239 64.1 0.383 0.177 0.012
1200 137.7 0.274 69.0 0.503 0.097 0.012
1300 138.1 0.288 77.2 0.546 0.166 0.017
1400 139.3 0.319 83.3 0.711 0.122 0.016
1500 143.5 0.235 89.8 0.460 0.131 0.012
1600 143.3 0.182 94.6 0.546 0.097 0.018
1700 143.1 0.148 99.8 0.502 0.103 0.014
1800 143.7 0.055 102.7 0.555 0.060 0.014
1900 143.6 0.055 106.6 0.327 0.077 0.006
2000 143.5 0.045 114.3 1.122 0.154 0.026
2100 143.4 0.055 120.3 0.349 0.122 0.022
2200 143.3 0.055 128.4 0.349 0.162 0
2300 143.3 0.055 139.1 2.472 0.217 0.054
2400 143.0 0.045 143.9 0.329 0.096 0.046
2500 143.0 0.084 152.7 0.409 0.176 0.010
2600 143.0 0.130 158.2 1.242 0.111 0.027
2700 142.9 0.045 169.0 0.365 0.218 0.020
2800 142.9 0 177.6 0.785 0.172 0.019
2900 142.9 0.045 185.7 1.447 0.163 0.045
3000 142.8 0.055 194.9 1.158 0.183 0.045
3100 142.8 0.045 200.8 0.404 0.120 0.031
3200 142.7 0.084 207.6 1.568 0.135 0.036
3300 142.8 0.055 219.0 1.816 0.233 0.057
3400 142.8 0.055 224.1 0.377 0.101 0.036
3500 142.8 0.055 232.3 0.230 0.166 0.007
3500 142.5 0 304.1 0.518 0.166 0.007
3600 141.9 0.822 300.7 1.055 −0.068 0.025
3700 142.1 0.418 326.2 0.321 0.513 0.021
3800 142.1 0.313 360.4 0.522 0.689 0.010

When t = 3500 µs, there are two barycentric coordinates and, meanwhile, only one droplet is
formed. When calculating the velocity of the droplet at this moment, the generated droplet and the
liquid column below the nozzle are regarded as a whole; the reference point is calculated using the
overall center of barycentric coordinate as the velocity.
(e) By substituting the barycentric coordinates obtained in step (d) into formula (2), the velocity of
the liquid at different time delays can be obtained, as shown in Table 3.
According to Table 3, a line chart of average velocity, average abscissa, and ordinate varying
with time can be plotted. In addition, in order to reflect the droplet barycentric coordinates (pixel
coordinates, including horizontal and longitudinal coordinates) and the fluctuation size of velocity
at each moment more clearly, the error bars of velocity, horizontal, and longitudinal coordinates
are obtained according to Table 3 and are shown in Figure 16. The technique has higher accuracy
compared with the method of calculating velocity with the microdroplet terminal coordinate as the
reference point.
 Micromachines 2019, 10, 148 13 of 16
Micromachines 2018, 9, x 14 of 17

Figure 16. The speed and ordinate of the microdroplet at different times. (a) speed and horizontal
Figure 16. VS
ordinate Thetime,
speed
(b)and ordinate
speed of the ordinate
and vertical microdroplet at different times. (a) speed and horizontal
VS time.
ordinate VS time, (b) speed and vertical ordinate VS time.
Figure 16 describes the variation relationship between droplet velocity and coordinates over time.
TheFigure
velocity 16 of
describes
the droplet the just
variation
ejected relationship
from the nozzle between droplet velocity
is relatively and coordinates
large, reaching ~0.68 m/s, over
then
time. The decrease
rapidly velocity of the droplet
under just ejected
the influence from theresistance
of the viscous nozzle isand relatively
surfacelarge, reaching
tension ~0.68 m/s,
of the solution and
then rapidly
reaches decreasestable
a relatively under the (0.18
speed influence
m/s).ofAfter the that,
viscous resistance
the liquid columnandissurface
necked tension of theto
and fractured
solution
form a and reaches
droplet, andathe relatively
velocitystable
of thespeed
process (0.18 m/s). Afterreduced
is gradually that, thetoliquid column
zero and is necked by
accompanied and the
fractured to form a droplet, and the velocity of the process is gradually
rebound phenomenon of the droplet. This is the reason why the velocity has a negative value. Finally, reduced to zero and
accompanied
the droplets by arethe rebound
in flight. phenomenon
According to Figureof the
16,droplet. This is the
the conclusions canreason
be drawnwhyasthe velocity has a
follows.
negative(1)value. Finally, trend
The variation the droplets
of velocity are isinbasically
flight. According
consistent to Figure
with 16, the conclusions
the formation process of can be
droplets.
drawn as follows.
That is, in the initial stage, the droplets are rapidly reduced by a larger velocity, and then enter the
1) Theand
necking variation
breaking trend
stageof at
velocity is basically
a relatively consistent
stable velocity. with the
Finally, theformation processare
broken droplets ofin
droplets.
flight.
That is,(2)
inThethe variation
initial stage,trendtheofdroplets
droplet are rapidly reduced
velocity–time curve by a larger
in five groups velocity, and then enter
is approximately the
the same,
necking
which and breaking
indicates thatstage at a relatively
the jetting conditionstable
is stablevelocity.
underFinally, the broken
the setting droplets
parameters (seeare in flight.
Supplementary
2) The variation trend of droplet velocity–time curve in five groups is approximately the same,
Material).
which (3)
indicates
Duringthat thethe jetting condition
formation is stable
of the droplet, theunder
ratio the setting
of the parameters
maximum offset(see supplementary
of the droplet in the
material).
horizontal direction (taking the central axis of the nozzle as the reference) to the total height of the
drop3) is
During
~1.6%,the formation
which indicates of that
the droplet,
the nozzle theis ratio
absenceof the
from maximum offset of surrounding
the disturbance the droplet inairflow.
the
horizontal
Therefore,direction (taking
the influence the central
is almost axis of the nozzle as the reference) to the total height of the
negligible.
drop isIn~1.6%,
addition,whichFigureindicates
16 alsothat the error
shows nozzlebars is absence from
of velocity and thethedisturbance
horizontal ofand surrounding
longitudinal
airflow. Therefore,
coordinates of thethe influence
droplets is almost
at each moment. negligible.
In addition, Figure 16 also shows error bars of velocity and the horizontal and longitudinal
coordinates of the droplets at each moment.
tendency of “small middle and big ends”. The reason is that the droplets are greatly influenced by
the accumulation of liquid from the end surface of the nozzles when the droplets are just ejected
from the nozzles. Since then, as the droplets continue to stretch, the influence of the accumulated
fluid is getting smaller
Micromachines 2019, 10,and
148 smaller, and the value is kept within a smaller range. At approximately 14 of 16
t = 3500 μs, one droplet is generated. At this time, the droplet is out of the binding of the liquid
column, andHorizontal
is more susceptible
coordinate errorto the disturbance
bar: in the process of of
theformation
airflow,and andflight
its value is slightly
of droplets, there increased.
is
However, the droplets
a tendency of “smallin middle
flight gradually
and big ends”. achieve a balance
The reason is that under the are
the droplets coaction
greatly of kinetic energy,
influenced
by thepotential
gravitational accumulation of liquid
energy, from the
surface freeend surfaceand
energy of the nozzles
heat when thefrom
generated droplets
air are just ejected
resistance. Thus, the
from the nozzles. Since then, as the droplets continue to stretch, the influence
error bar value is getting smaller and smaller. Until t = 3800 μs, the relative standard deviation of of the accumulated
fluid is getting smaller and smaller, and the value is kept within a smaller range. At approximately
the abscissa is 0.22% (standard deviation/average value), indicating that the position oscillation of
t = 3500 µs, one droplet is generated. At this time, the droplet is out of the binding of the liquid column,
the droplet
and in the horizontal
is more susceptible to direction is very
the disturbance of small at this
the airflow, andmoment.
its value is slightly increased. However,
Longitudinal coordinate error bar: during the
the droplets in flight gradually achieve a balance under the coaction entire injection
of kineticprocess, the longitudinal
energy, gravitational
potential
coordinate errorenergy,
bar ofsurface free energy
the droplet and heat generated
is maintained from airrange,
in a small resistance.
which Thus, the error bar
indicates thatvalue
the droplet
is getting smaller and smaller. Until t = 3800 µs, the relative standard deviation
position in vertical direction is almost constant at the same delay time. This is another illustration of the abscissa is
0.22% (standard deviation/average value), indicating that the position oscillation of the droplet in the
for demonstrating jetting stability.
horizontal direction is very small at this moment.
VelocityLongitudinal
error bar:coordinate
there is aerrorgreatbar:fluctuation approximately
during the entire injection process,t =the3500 μs. At this
longitudinal moment, the
coordinate
velocity error
of the bardroplet reveals
of the droplet a significant
is maintained change
in a small (reduces
range, to zero that
which indicates or even a negative
the droplet positionvelocity)
in vertical direction is almost constant at the same delay
because the rebound phenomenon occurs after the droplet breaks. After that, the dropletstime. This is another illustration for in flight
graduallydemonstrating
achieve a jetting
balance stability.
in the mutual conversion of various kinds of energy. The shape of
Velocity error bar: there is a great fluctuation approximately t = 3500 µs. At this moment, the
droplets is increasingly closer to the standard sphere; meanwhile, the velocity error bar will
velocity of the droplet reveals a significant change (reduces to zero or even a negative velocity)
decreasebecause
rapidly. When tphenomenon
the rebound = 3800 μs,occurs the after
relative standard
the droplet breaks.deviation
After that,of thethe droplet
droplets velocity is
in flight
reduced gradually
to 1.4%, which
achieve indicates
a balance in thatthethe stateconversion
mutual of the droplet is very
of various stable
kinds at thisThe
of energy. moment.
shape of
droplets is increasingly
In summary, just at t = closer
3800 to μs,thethe
standard
shapesphere;
of themeanwhile,
droplet and the velocity
the jettingerrorbehavior
bar will decrease
have reached
rapidly. When t
to a stable state. Then, the working position of the substrate can be set near the height 1.4%,
= 3800 µs, the relative standard deviation of the droplet velocity is reduced to point at this
which indicates that the state of the droplet is very stable at this moment.
moment, so as to fabricate a good quality pattern on the substrate.
In summary, just at t = 3800 µs, the shape of the droplet and the jetting behavior have reached
to a stable state. Then, the working position of the substrate can be set near the height point at this
5.3. Microsolder
moment,Ball
so asVolume Measurement
to fabricate a good quality pattern on the substrate.

The5.3.
pneumatically diaphragm-driven
Microsolder Ball Volume Measurement actuator is also applied to generate microsolder balls.
The solder brand is 63/37, with 63% tin and 37% lead. The melting point is 183 °C. At 200 °C and the
The pneumatically diaphragm-driven actuator is also applied to generate microsolder balls.
viscosityThe
and surface
solder tension
brand arewith
is 63/37, 2.8863%
mPa·s and37%
tin and 0.435
lead.N/m.
The Figure
melting 17a
pointshows
is 183 ◦microsolder
C. At 200 ◦ C balls in
silicone and
oil the
andviscosity
Figureand 17b shows
surface microsolder
tension balls
are 2.88 mPa ·s andarray (20 × Figure
0.435 N/m. 20) on17astainless steel plate. The
shows microsolder
balls in silicone oil and Figure 17b shows microsolder balls array (20 ×
detailed volume data of 200 microsolder balls in Figure 17a are plotted in Figure 18. plate.
20) on stainless steel The average
The detailed volume data of 200 microsolder balls in Figure 17a are plotted in Figure 18. The average
diameter is 87.96 μm, the standard deviation is 0.7265 μm, and the relative standard deviation is
diameter is 87.96 µm, the standard deviation is 0.7265 µm, and the relative standard deviation is 0.83%.
0.83%. The average volume is 356.36 pL, the mean standard deviation is 8.8218 pL, and the relative
The average volume is 356.36 pL, the mean standard deviation is 8.8218 pL, and the relative standard
standarddeviation
deviation is 2.5%.
is 2.5%.

(a) (b)
Figure 17. (a) Microsolder balls in silicone oil, (b) microsolder balls array on stainless steel plate.

Figure 17. (a) Microsolder balls in silicone oil, (b) microsolder balls array on stainless steel plate.
 Micromachines 2019, 10, 148 15 of 16
Micromachines 2018, 9, x 16 of 17

Figure 18. Measured diameters and volumes of 200 microsolder balls.

Figure 18. Measured diameters and volumes of 200 microsolder balls.
6. Conclusions
6. Conclusion
This paper proposed a novel in situ image acquisition and measurement of microdroplets based
on delay triggering. Using a high-speed CCD camera, a motion control card, a driving circuit, a time
Thisdelay
paper proposed a novel in situ image acquisition and measurement of microdroplets
triggering circuit, and a light source, an imaging system is built to obtain the sequential images
based onofdelay triggering.
microdroplets Usingthrough
generated a high-speed CCD camera,
the pneumatically a motion control
diaphragm-driven actuator. card, a driving
By adjusting the circuit,
a time delay triggering
delay time, the stable circuit, andfabrication
microdroplet a light process
source,can an imagingThesystem
be captured. is builtmethods
image processing to obtain the
have been adapted to measure the coordinates, diameters and speeds
sequential images of microdroplets generated through the pneumatically diaphragm-driven of the microdroplets real-time.
actuator.For
Byglycerin/water
adjusting the (60:40,
delaymass ratio),the
time, under the given
stable condition, the
microdroplet average of diameter
fabrication processand canvolume
be captured.
are 266.8 µm and 9944 pL, respectively. The maximum average velocity of the microdroplets is
The image processing methods have been adapted to measure the coordinates, diameters and
0.689 m/s. Finally, the pneumatically diaphragm-driven actuator is applied to generate microsolder
speeds of theThemicrodroplets
balls. results show the real-time.
system processes For good
glycerin/water
reproducibility.(60:40, massdiameter
The average ratio), isunder the given
87.96 µm,
condition,
the the average
standard of isdiameter
deviation 0.7265 µm,and and thevolume
relative are 266.8
standard μm and
deviation 9944ThepL,
is 0.83%. respectively.
in situ imaging The
system employs two control signals to stimulate the microdroplet
maximum average velocity of the microdroplets is 0.689 m/s. Finally, the pneumatically generator and the CCD camera and
is absence of actuator
diaphragm-driven instantaneous high intensity
is applied light. Hence,
to generate the system can
microsolder avoid
balls. Thetheresults
desynchronization
show the system
problem of signals which control the strobe LED light source and the camera in previous techniques.
processes good reproducibility. The average diameter is 87.96 μm, the standard deviation is 0.7265
The system with low cost and simple configuration is especially applicable to capture and analyze the
μm, andflight
the relative
process ofstandard deviation
microdroplets involving is fairly
0.83%. The
large in(~200
sizes situ µm)
imaging
moving system employs
at low speed two control
(~1 m/s).
signals This
to stimulate
technology isthe microdroplet
a reliable generator
and cost-effective and and
for acquiring themeasuring
CCD camera and is absence of
microdroplets.
instantaneous high intensity light. Hence, the system can avoid the desynchronization problem of
Supplementary Materials: The following are available online at http://www.mdpi.com/2072-666X/10/2/148/s1.
signals which control the strobe LED light source and the camera in previous techniques. The
Author Contributions: D.X., X.S., and X.C. conceived and designed the experiments; K.Z., K.X., and
system with low costthe
B.L. performed and simple configuration
experiments; K.Z. conducted image is especially
processing applicable
optimization; to capture
W.C. analyzedand analyze the
the data;
C.W. contributed reagents; and X.C. wrote the paper. All authors discussed the results and commented on
flight process of microdroplets involving fairly large sizes (~200 μm) moving at low speed (~1 m/s).
the manuscript.
This technology is awork
Funding: This reliable and cost-effective
was supported by the Natural for acquiring
National Scienceand measuring
Foundation microdroplets.
of China (NSFC) (Grant No.
51875491, No. 51475400), the Natural Science Foundation of Fujian Province (Grant No. 2017J01492), and the
Supplementary Materials:
Xiamen Municipal Theand
Science following arePlan
Technology available online
Project (Grant No.at3502Z20179030,
www.mdpi.com/xxx/s1.
No. 3502Z20183049). This work
was also supported by the Program for Innovative Research Team in Science and Technology in Fujian Province
Author Contributions: D.X., X.S.,
University. The experimental and
work X.C. conceived
is mainly carried out at and designed
the Institute the experiments;
of Micro-Systems in School K.Z., K.X., and B.L.
of Mechanical
Science and Engineering in Huazhong University of Science and Technology. The financial
performed the experiments; K.Z. conducted image processing optimization; W.C. analyzed the data; C.W. and technique supports
are gratefully acknowledged.
contributed reagents; and X.C. wrote the paper. All authors discussed the results and commented on the
Conflicts of Interest: The authors declare no conflict of interest.
manuscript.

Funding:References
This work was supported by the Natural National Science Foundation of China (NSFC) (Grant No.
51875491,1.No.Dong,
51475400),
H.; Carr,the Natural
W.W.; Morris,Science Foundation
J.F. An experimental of of
study Fujian Province (Grant
drop-on-demand No. 2017J01492),
drop formation. Phys. Fluids and the
Xiamen Municipal Science
2006, 18, and Technology Plan Project (Grant No. 3502Z20179030, No. 3502Z20183049). This
51–59. [CrossRef]
work was also supported by Chevalier,
2. Khalil, A.; Puel, F.; the Program for Innovative
Y.; Galvan, Research
J.M.; Rivoire, A.; Klein, J.P.Team
Study in Science
of droplet sizeand Technology
distribution duringin Fujian
an emulsification
Province University. process usingwork
The experimental in situis
video probecarried
mainly coupledoutwithatantheautomatic image
Institute analysis. Chem. Eng.
of Micro-Systems inJ.School of
2010, 165, 946–957. [CrossRef]
Mechanical Science and Engineering in Huazhong University of Science and Technology. The financial and
technique supports are gratefully acknowledged.

Conflicts of Interest: The authors declare no conflicts of interest.

References
Micromachines 2019, 10, 148 16 of 16

3. Kwon, K.S.; Kim, W. A waveform design method for high-speed inkjet printing based on self-sensing
measurement. Sens. Actuators A 2007, 140, 75–83. [CrossRef]
4. Shin, Y.J.; Lee, J.B. Machine vision for digital microfluidics. Rev. Sci. Instrum. 2010, 81, 014302. [CrossRef]
[PubMed]
5. Dong, H.; Carr, W.W.; Morris, J.F. Visualization of drop-on-demand inkjet: Drop formation and deposition.
Rev. Sci. Instrum. 2006, 77, 085101. [CrossRef]
6. Hutchings, I.; Martin, G.; Hoath, S. High speed imaging and analysis of jet and drop formation. J. Imaging
Sci. Technol. 2007, 51, 438–444. [CrossRef]
7. Kwon, K.S. Speed measurement of ink droplet by using edge detection techniques. Measurement 2009, 42,
44–50. [CrossRef]
8. Fan, K.C.; Chen, J.Y.; Wang, C.H.; Pan, W.C. Precision in situ volume measurement of micro droplets. J. Opt.
A Pure Appl. Opt. 2008, 11, 015503. [CrossRef]
9. Jo, B.W.; Lee, A.; Ahn, K.H.; Lee, S.J. Evaluation of jet performance indrop-on-demand (DOD) inkjet printing.
Korean J. Chem. Eng. 2009, 26, 339–348. [CrossRef]
10. Thurow, K.; Krü, T.; Stoll, N. An optical approach for the determination of droplet volumes in nanodispensing.
J. Anal. Methods Chem. 2009, 2009, 1–9. [CrossRef] [PubMed]
11. IEC 62899-302-1:2017 Printed Electronics—Part 302-1: Equipment—Inkjet—Imaging Based Measurement of
Jetting Speed. Available online: https://webstore.iec.ch/preview/info_iec62899-302-1%7Bed1.0%7Den.pdf
(accessed on 21 February 2019).
12. IEC 62899-302-2:2018 Printed Electronics—Part 302-2: Equipment—Inkjet—Imaging-Based Measurement of
Droplet Volume. Available online: https://webstore.iec.ch/preview/info_iec62899-302-2%7Bed1.0%7Den.
pdf (accessed on 21 February 2019).
13. Jia, C.; Chen, G.; Li, X.; Liu, J.; Zhao, L. Droplet Surface Tension Calculation Method Based on Contour
Fitting. J. Comput. Inf. Syst. 2013, 9, 2075–2082.
14. Fan, K.C.; Wang, N.; Wang, Z.W.; Zhang, H. Development of a roundness measuring system for microspheres.
Meas. Sci. Technol. 2014, 25, 064009. [CrossRef]
15. King, B.A.; Winward, T.W.; Bjorneberg, D.L. Comparison of drop size and velocity measurements by a laser
precipitation meter and low-speed photography for an agricultural sprinkler. Appl. Eng. Agric. 2014, 30,
413–421.
16. Ravasio, C.S.; Hoath, S.D.; Martin, G.D.; Boltryk, P.; Dorrestijn, M. Meniscus Motion Inside A DoD Inkjet
Print-Head Nozzle. In Proceedings of the 32nd International Conference on Digital Printing Technologies
(NIP): Printing for Fabrication 2016, Manchester, UK, 12–16 September 2016.
17. Martin, G.D.; Castrejónpita, J.R.; Hutchings, I.M. Holographic Measurement of Drop-on-Demand Drops in
Flight. In Proceedings of the 27th International Conference on Digital Printing Technologies, Minneapolis,
MN, USA, 2–6 October 2011.
18. Castrejón-García, R.; Castrejón-Pita, J.R.; Martin, G.D.; Hutchings, I.M. The shadowgraph imaging technique
and its modern application to fluid jets and drops. Rev. Mex. Fis. 2011, 57, 266–275.
19. Hsiao, W.K.; Chun, J.H.; Saka, N. The Effects of Wetting and Surface Roughness on Liquid Metal Droplet
Bouncing. J. Manuf. Sci. Eng. 2009, 131, 267–272. [CrossRef]
20. Jung, S. Fluid Characterisation and Drop Impact in Inkjet Printing for Organic Semiconductor Devices.
Ph.D. Thesis, University of Cambridge, Cambridge, UK, April 2011.
21. Jiang, H.; Merritt, B.A.; Tan, H. A Low-Cost Flash Photographic System for Visualization of Droplets in
Drop-on-Demand Inkjet. J. Imaging Sci. Technol. 2018, 62, 60502-1–60502-9. [CrossRef]
22. Martin, G.D.; Price, W.C.; Hutchings, I.M. Measurement of Inkjet Drop Volume Estimation—The Role of
Image Processing. In Proceedings of the 32nd International Conference on Digital Printing Technologies
(NIP): Printing for Fabrication 2016, Manchester, UK, 12–16 September 2016.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).