HAL Id: hal-02415747
https://hal.archives-ouvertes.fr/hal-02415747
Submitted on 17 Dec 2019
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
In-situ measurements in microscale gas flows – conventional sensors or something else?
Juergen Brandner
To cite this version:
Juergen Brandner. In-situ measurements in microscale gas flows – conventional sensors or something else?. Micromachines, MDPI, 2019, 10 (5), pp.292. �10.3390/mi10050292�. �hal-02415747�
Micromachines 2018, 9, x; doi: FOR PEER REVIEW www.mdpi.com/journal/micromachines
Review
1
In‐situ measurements in microscale gas flows –
2
conventional sensors or something else?
3
Juergen J. Brandner 1,*
4
1 Karlsruhe Institute of Technology (KIT), Institute for Microstructure Technology (IMT), Staff Position
5
Microstructures and Process Sensors (MPS), Hermann‐von‐Helmholtz‐Platz 1, 76344
6
Eggenstein‐Leopoldshafen, Germany; juergen.brandner@kit.edu
7
* Correspondence: juergen.brandner@kit.edu; Tel.: +49‐721‐608‐23963
8
Received: date; Accepted: date; Published: date
9
Abstract:
10
Within the last decades miniaturization was a driving force in almost all areas of technology,
11
leading to tremendous intensification of systems and processes. Information technology provides
12
now data density several orders of magnitude higher than a few years ago, and the smartphone
13
technology includes, amongst simple communication, things like internet, video and music
14
streaming, but also implementation of global positioning system, environment sensors or
15
measurement systems for individual health. So‐called wearables are everywhere, from
16
physio‐parameter sensing wrist smart watch up to measurement of heart rate by underwear. This
17
trend holds also for gas flow applications, where complex flow arrangements and measurement
18
systems formerly designed for macro scale have been transferred into miniaturized versions. Thus,
19
those systems took advantage of the increased surface to volume ratio as well as of the improved
20
heat and mass transfer behavior of miniaturized equipment. In accordance, disadvantages like gas
21
flow mal‐distribution on parallelized mini‐ or micro tubes or channels as well as increased pressure
22
losses due to the minimized hydraulic diameters and an increased roughness‐to‐dimension ratio
23
have to be taken into account. Furthermore, major problems are arising for measurement and
24
control to be implemented for in‐situ and/or in‐operando measurements. Currently, correlated
25
measurements are widely discussed to obtain a more comprehensive view to a process by using a
26
broad variety of measurement techniques complementing each other. Techniques for correlated
27
measurements may include commonly used techniques like thermocouples or pressure sensors as
28
well as more complex systems like gas chromatography, mass spectrometry, infrared or ultraviolet
29
spectroscopy and many others more. Some of them can be miniaturized, some of them cannot yet.
30
Those should, nevertheless, measure at the same location and the same time, preferably in‐situ and
31
in‐operando. Therefore, combinations of measurement instruments might be necessary, which will
32
provide complementary techniques to access local process information. A recently more
33
intensively discussed additional possibility is the application of nuclear magnetic resonance (NMR)
34
systems, which might be useful in combination with other, more conventional measurement
35
techniques. NMR is currently undergoing a tremendous change from large‐scale to benchtop
36
measurement systems, and it will most likely be further miniaturized. NMR allows a multitude of
37
different measurements, which are normally covered by several instruments. Additionally, NMR
38
can be combined very well with other measurement equipment to perform correlative in‐situ and
39
in‐operando measurements. Such combinations of several instruments would allow to retrieve an
40
“information cloud” of a process. This paper will present a view to some common measurement
41
techniques and the difficulties to apply them on one hand in miniaturized scale, on the other hand
42
in correlative mode. Basic suggestions to reach the above mentioned objective by combination of
43
different methods including NMR will be given.
44
Keywords: miniaturization, gas flows in micro scale, measurement and control, integrated micro
45
sensors, advanced measurement technologies
46
47
1. Introduction
48
The interest to gas flows in miniaturized systems has grown tremendously in the last couple of
49
years. Driven by, i.e., automotive, semiconductor, chemical and pharmaceutical industry, modelling,
50
precise measurement and control of the flow of gaseous compounds, mixtures and reactive systems
51
through mini‐ and micro‐structured devices gained importance for various applications like heat
52
transfer [1‐16], gas‐liquid or gas‐solid contacting [17‐24] as well as chemical reactions [25‐30].
53
Amongst correct design and manufacturing of the miniaturized devices, all the named topics need
54
precise measurement and control of the processes taking place inside microstructures. The final
55
objective of these efforts is to provide an almost comprehensive description of the process to be able
56
to model and simulate as well as to predict by software. The following description will focus on
57
measurement and close out control, to make it not too exhaustive.
58
While measurement of gaseous flows in macroscale is not trivial but manageable, it turns out to
59
be much more problematic in micro scale. There are several reasons for this. The fact that gas is a
60
compressible medium per se makes it more complex to measure certain behavior and parameters of
61
a flow inside confined or miniaturized systems [31‐35]. Additionally, the scale down of
62
conventionally‐sized macro scale tubes or channels into the micro scale makes it more complex to
63
measure. This is, on one hand, due to sensors which are simply too large to be inserted into the micro
64
devices. Figure 1 (a) & (b) show examples for this. For temperature measurement in macro‐scale
65
tubes, a thermocouple is simply located into the flow, measuring the gas flow temperature. This
66
changed by scaling down the tube diameter. While Figure 1 (a) shows a mini heat pipe cut open,
67
Figure 1 (b) provides a view to a so‐called “micro‐thermocouple”. It is quite obvious that the
68
thermocouple will block the inner diameter to a major extend and, therefore, change the fluidic
69
behavior totally. Thus, no precise measurement would be possible. More problems occur due to the
70
low density of gases, the change in viscosity, the small specific heat capacity and the need for
71
increased leak‐tightness of microstructures while handling gases (and here, depending to the gas,
72
the acceptable leak rate QL can vary in orders of magnitude!). However, in many cases Micro
73
Electromechanical Systems (MEMS) have been applied as measurement tools to be implemented
74
into microstructure devices [33]. This is often a practical solution, in other cases it is not, as will be
75
shown in the later discussion.
76
(a)
(b)
Figure 1. Microstructure fluidic device. (a) Micro heat pipe, opened to see the inside structure. Inner
77
diameter is 0.8mm. From: KIT‐IMVT; (b) Miniaturized Type K thermocouple. Although this sensor
78
has an outer diameter of 0.4mm only, a major part of the structure in (a) would have been blocked.
79
From: www.ninomiya‐ew.co.jp.
80
Nevertheless, from 1964 until today, roughly about 45,000 papers have been published on
81
miniaturized gas sensors and sensor systems [36]. A large variety of relevant technologies is
82
available which will not be described in this paper in details, more can be found in comprehensive
83
textbooks or, i.e., in [37].
84
The following will present an overview on parameters of gas flows in microstructures to be
85
measured as well as measurement methods for these. By no means the presented overview is
86
complete. The envisaged field of measurement is very much in move, thus, new technologies and
87
improved methods are rapidly developing, especially by improving sensitivity and selectivity of the
88
measurements. One of these developments is the use of nanomaterials for sensing opportunities.
89
Here, numerous different technologies based on nanostructured materials or nanostructures in
90
specific materials have been described and are used now for sensing in gases. Examples are given in
91
[38‐47]. The sensitivity and selectivity of nanostructure / nanomaterials sensors has been improved
92
significantly in the last years [48‐50].
93
Another trend is a combination of sensing elements in micro or nano scale for measurement of
94
biological parameters or environment (biosensing, [51‐56]). This field is relatively new, and lots of
95
developments are to expect in the next future.
96
As mentioned before, an additional point is the correlation of a multitude of measurement
97
methods to achieve a more dense “information cloud” and, therefore, reach a better understanding
98
of the effects and behavior in gas flows. This leads to the necessity of in‐situ and in‐operando
99
combination of several measurement methods with similar (ideally: identical) timely and spatial
100
resolution. All measurements have to be taken simultaneously and at the same location to provide
101
the highest possible information density of the process taking place at this point. As an example for
102
such a process a heterogeneously catalyzed gas phase reaction could be taken. Here, gas flow, gas
103
composition, catalyst‐gas‐surface interaction, temperature, pressure and product concentration
104
should be measured at the same time and the same location, to name just a few of the parameters of
105
such a process.
106
The data obtained from each of such monitoring systems will then be correlated to those of all
107
others, in time as well as in space, to get a description of an n‐dimensional parameter space as
108
comprehensive as possible. Within the overview presented here several possible technologies to
109
correlate measurements are very briefly and rudimentarily described. Possibilities, advantages and
110
disadvantages of each method are presented, and a possible use in combination with other methods
111
is evaluated. This overview is not intended to give a vast description of the measurement methods,
112
their advantages and disadvantages, their working principles or underlying functionalities – this is
113
left to much more specified papers or textbooks. This collection, without going deep into details of
114
the methods themselves, should provide a quick overview for possible correlative measurement
115
technologies only. For this objective, the following section deals with the parameters which most
116
regularly are measured, as well as with some most common methods for those and with scaling, the
117
problem of miniaturization itself [57‐59].
118
2. Parameters, measurement methods and scaling
119
Numerous parameters of gas flows in micro scale can be measured. However, many of them are
120
only useful in specific applications. Amongst those are parameters like chemical reactivity, pH,
121
solubility in liquids, speed of sound etc. This publication will focus onto more common parameters
122
like temperature (or temperature difference), heat flux, pressure (or pressure difference), mass flow,
123
flow velocity, mixture composition and concentration (concentration gradients, respectively) of
124
species inside micro scale gas flows.
125
The measurement methods can in general be split in electrical measurements using
126
conventional wire systems [60, 61], electrical measurements using micro‐electromechanical systems
127
(MEMS) [62], spectroscopic measurements or optical measurement methods [63]. There are
128
numerous papers on each of those different measurement methods, thus, it shall not be the objective
129
of this publication to provide more detailed information on them.
130
Wire‐system based methods use conventional sensors which are, in many cases, macroscopic.
131
In general, these are thermocouples or thermistors (see below), pressure sensors, mass flow meters
132
and similar, which are well known measurement devices in style as well as in behavior. These
133
sensors are mounted aside of the process tubing, e.g. thermocouples before and after a possible
134
reaction vessel as well as pressure sensors or mass flow meters. An in‐situ or in‐operando monitoring
135
is hardly possible.
136
MEMS‐based measurements show the possibility to be implemented for in‐situ and in‐operando
137
measurements. Pressure sensors, temperature sensors, flow sensors and others more are available of
138
the shelf. They are, in general, Si‐based, small, provide a short response time, good reliability and
139
long lifetime. Aside, most of those systems are cheap in production. However, they need to be
140
coupled with a visualization module or a data logger to make the measurements available for the
141
operator. This is regularly done by wiring. There are some cases of wireless MEMS sensors [64].
142
However, due to the continuous developments in semiconductor technology, wireless solutions are
143
becoming more and more popular, and their use is greatly enhanced by, i.e., radio frequency
144
identification (RFID) and near field communication (NFC) technology [65‐69].
145
What was mentioned above for wire‐based electrical sensor in principle holds also for optical
146
sensors, where either an optical fiber is implemented into the measurement location, or an active
147
optical component (light emitting diode LED, laser or other light source) is integrated there – in
148
many cases also an optical detector like a photo diode or similar. Examples are presented in [70‐76].
149
Most of the measurement methods of the sensors named above have been derived from
150
macroscopic standards, adapted and improved for mini scale and then applied and even further
151
revised for micro scale; and almost all of them (but the wireless MEMS devices) share the same
152
problem of scaling. This holds for all type of sensors, electrical or optical.
153
Scaling a measurement method simply down from macro to micro scale will normally not
154
generate the desired results. Either is the fluid influenced too strongly by the measurement method
155
(as it was shown in the example given in Fig. 1), or the precision of measurement suffers from the
156
miniaturization. It is, in many cases, not easy to decide where in the fluid flow to place the sensors to
157
obtain correct measurements. A good example is the measurement of temperatures. If a
158
MEMS‐based measurement system is placed into the sidewall of a microfluidic system, the wall
159
temperature is measured, or maybe the temperature of the gas flow near the wall. The same result
160
will be obtained with an optical element placed inside the sidewall of a microfluidic system. The
161
temperature in the core of the flow is not determined in any way. Moreover, the MEMS system may
162
alter the wall‐gas interaction (because the material might be different), and therefore not even
163
provide a representative measurement signal for the sidewall temperature of the gas flow [77]. This
164
might not be so much the case for the optical sensor. Fig. 2 shows an example of a silicon MEMS
165
sensor system used for temperature measurement inside a microchannel gas flow under ambient or
166
slightly rarefied conditions [78].
167
168
Figure 2. Silicon microstructure chip for temperature measurement in a gas flow. The chip is inserted
169
into a sealed housing and forms the fourth side of a rectangular microchannel. Numerous
170
thermopiles have been implemented into the chip, to measure in combination with precision
171
resistors the temperature of a gas flow [78]. www.gasmems.eu
172
If the process is well known, the core temperature of the flow might be calculated correctly. Not
173
knowing the process precisely, it might be necessary to measure the core flow temperature – which
174
is possible with a sensor implemented into the flow. Placing a very thin wired sensor or a thin
175
optical fiber inside the flow core may lead to a more precise measurement of the temperature there,
176
but also could lead to a deflection of the measurement sensor due to insufficient stiffness, or to
177
generate local turbulences and eddies and, therefore, disturb the flow. The same holds for a wireless
178
sensor, which has to be mounted and fixed somehow in the flow. In any case, scaling down fluidic
179
systems for gas flows into the micro scale needs a very careful consideration of the location and the
180
precision of the sensors applied. This holds for all parameters to be measured.
181
2.1. Temperature measurement
182
One of the most common tasks in gas flows is the measurement of temperatures, as mentioned
183
above. At the same time, it is a relatively complex measurement, which is, in many cases,
184
underestimated [79‐85]. In the following, some possible methods will be described and evaluated as
185
a general example for other parameters to be measured in micro scale.
186
2.1.1. Measurements using conventional intrusive sensors
187
Conventional sensors in this case means either thermistors or thermocouples. Optical sensors
188
shall not be considered here, because they are in this case more delicate to handle and in general
189
more expensive [86‐88]. Both thermocouples and thermistors are commercially available in various
190
shapes, sizes and forms, which makes them flexible and handy tools. Moreover, the efforts to obtain
191
some results are limited. In both cases, wired standard solutions as well as wired MEMS are
192
available.
193
Thermistors are temperature‐dependent electrical resistors. They can be separated in devices
194
with positive temperature coefficient retrieving a higher resistance with increasing temperature, and
195
with negative temperature coefficient, which lowers the resistance by increasing temperature. In any
196
case, the dependency between the change of resistance and the temperature is very linear (or can be
197
linearized in an easy way) [89]. The most common example is the PT100, a platinum resistor with a
198
nominal value of 100Ω at standard conditions (variations with 10Ω or 1000Ω are also common). The
199
PT‐sensor is used as part of a Wheatstone bridge circuit, which makes a very precise measurement
200
possible [90].
201
Temperature sensors based on the resistor principle are wide‐spread in any application,
202
because they can be manufactured as discrete devices as well as integrated circuits in semiconductor
203
technology. Thus, extremely small sensors can be generated using standard semiconductor
204
manufacturing processes. This holds, in some cases, also for the second common temperature
205
measurement principle, which is thermocouples.
206
A thermocouple is an electrical device consisting of two dissimilar electrical conductors
207
forming electrical junctions at differing temperatures. A thermocouple produces a
208
temperature‐dependent voltage as a result of the thermoelectric effect, and this voltage can be
209
interpreted to measure a temperature [78, 91]. Some types of thermocouples can also be produced in
210
micro‐ or even nano scale by use of semiconductor technology, others cannot [37]. However, the two
211
main disadvantages of thermocouples are their limited signal strength and the need of a reference
212
temperature. While the signal strength can be enhanced by creating thermopiles (numerous
213
thermocouples circuited in a row, see Fig. 2), the reference temperature need is not to avoid. This fact
214
makes it not easy to apply thermocouples in miniaturized equipment, which regularly shows a more
215
or less isothermal behavior, short distances in the mm or μm range and very short temperature
216
spreading times. Thus, the difference between reference temperature and measured temperature
217
might be diminished rapidly.
218
Both measurement principles mentioned above are regularly done with wired sensors. Wireless
219
temperature sensors, especially in micro scale, are not so widely common yet, but will be more
220
common in future [92‐94]. However, all the sensors presented in this subsection need a wire
221
connection to the outer environment, thus, sealings and interconnections are necessary. This might
222
be, depending on the supervised process, a source of leakage and uncertainty. Therefore, wireless
223
measurement would be a better option, as it was mentioned before. Additionally, it was mentioned
224
that introducing a sensor into the gas flow would disturb the flow mode or even change the process
225
parameters significantly. Thus, non‐intrusive methods have to be used.
226
2.1.2. Measurements using non‐intrusive methods
227
As the example of the use of conventionally sized resistors or thermocouples (presented in Fig.
228
1 (b)) shows, this is, in many cases, not an option, mainly for size reasons as it was described before.
229
Thus, it has to be asked which other techniques and methods could be used for such a simple
230
measurement task as retrieving the temperature?
231
One option is the use of non‐intrusive optical systems (i.e. infrared (IR) radiation, ultra violet
232
(UV) fluorescence, others) or ultrasound [95‐97]. Here, surface temperatures are measured,
233
providing a non‐intrusive possibility to acquire data on the thermal state of a body. However, these
234
methods have several disadvantages. They are not applicable for all cases, since certain materials are
235
not transparent for IR radiation (i.e. common glasses, metals etc.) or reflecting or damping‐out
236
ultrasound very strongly (i.e. dense liquids, metals etc.). In such cases, a wrong signal, a strongly
237
reduced signal or no signal at all can be obtained, leading to a very low signal‐to‐noise ratio (SNR).
238
With this, a measurement might be useless. A further disadvantage is that the measured signal
239
represents the surface temperature of a device obtained by conduction, convection or radiation from
240
the inner side, but not the core temperature. This means, that the inner parts of a body could differ
241
significantly in temperature from the outer surface. This cannot be seen and measured by the chosen
242
measurement technology. Same holds for other measurement technologies using optical
243
fluorescence to measure temperatures [98].
244
Another option is Raman spectrometry [99]. With this technique, temperature distributions
245
inside of miniaturized structures are well to measure, if the surrounding is transparent for the
246
Raman light. All limits named before for IR and ultrasound measurements hold here also, because it
247
still is an optical method. Efforts are high, and the Raman systems are expensive and complex to
248
handle. If all these drawbacks are given for measurement of an all‐day parameter like temperature,
249
does this also hold for other measurement needs like pressure, flow, density or more complex sets of
250
parameters like local concentration? And what might be feasible in terms of in‐situ, in‐operando and
251
correlated measurements?
252
2.2. Measurements of other parameters
253
As pointed out in the sections before, measurements of several parameters leads to application
254
of different techniques at a single process. The process pressure is measured with pressure
255
transducers, in most cases located outside the process vessel. The process flow through tubes or
256
vessels is obtained by anemometers or similar systems. Electrochemical sensors are used to measure
257
pH or conductivity. Viscosity is obtained with a rheometer (in general, outside of the process run),
258
and so on. Additional analytical methods like gas chromatography (GC) [100, 101], infrared (IR)
259
spectrometry [102], ultraviolet (UV) spectrometry [103], the above mentioned Raman spectrometry
260
[104], mass spectrometry (MS) [105, 106] or Nuclear Magnetic Resonance spectroscopy (NMR)
261
[107‐111] are used to determine the components of the mixture, depending on the applicability of the
262
respective method to the gas. Visualization of flow processes can be obtained by, i.e., high speed
263
videography [112‐114]. An example for a microstructure used for high speed videography is given
264
in Figure 3.
265
(a)
(b)
Figure 3. Bifurcative microchannel distribution system for liquids and gases. (a) Overview to a major
266
part of the system; (b) Magnification of one of the splits from (a). A step between the single incoming
267
channel and the two outgoing channels is clearly visible, which is due to manufacturing. From [112].
268
All of these systems work, in most cases, independently, are time consuming and costly. In‐situ
269
measurements or in‐operando measurements inside of micro scale systems are possible, but rarely
270
done, in most cases the analysis is performed offline with a specific sample taken from the process
271
flow. A correlation of results of these analytical methods to gain more complete process information,
272
as it was proposed in the introduction section, with measurement results of the process parameters
273
obtained as described before is at least difficult [115‐117], in many cases it is almost impossible.
274
Thus, it might be a good idea to have a combination of several complementary measurement
275
and analytical systems which can provide all the information wanted within a single measurement
276
campaign at once.
277
2.3. Combined and correlated measurement system
278
The objective mentioned above, namely to measure in‐situ, in‐operando, simultaneously at the
279
same location, can be obtained by combining several measurement and analytical systems.
280
However, aside of the tremendous efforts to be undertaken, this will result most likely in a spatial
281
correlation only. Due to different measurement time constants of the different sensors and systems, a
282
timely correlation is very unlikely to reach. All the methods named above show drawbacks for
283
in‐situ, in‐operando, combined and correlated measurements. While Raman spectroscopy will need
284
an optical access, IR spectroscopy as well as UV spectroscopy will need additionally specimen active
285
in the named wavelength region. In‐situ and in‐operando GC or GC‐MS is possible (see i.e. [118]), but
286
in general done in macro scale. To scale those techniques down into micro scale might be feasible but
287
is generally not available yet. NMR measurement systems [111] are, in general, big machines with
288
super‐cooled magnets and a huge bunch of external equipment, which is one of the major drawbacks
289
of this technology. Figure 4 shows an example of a 400 MHz (9.4T) NMR device made by the Bruker
290
company.
291
However, with an NMR spectrometer like this, a sample can be characterized for various
292
parameters in a single measurement campaign, correlating time and location with each of the
293
separate measurements taken. The question arising now is whether such a system is able to
294
characterize the desired parameters with a spatial and time resolution high enough to be useful in
295
micro scale.
296
For the spatial resolution, a value of about 10μm is given in literature (see i.e. [119]). The time
297
resolution is named to be around 20ms (see i.e. [120]). Assuming this to be valid for a multitude of
298
measurement objectives performed by NMR, spatial as well as time resolution given in literature are
299
in principle good enough to run NMR systems as well‐chosen measurement and analytical tools for
300
numerous applications, reaching from chemistry to materials research, pharmacy and food
301
engineering to biomedical applications or physics. However, it has to be considered that many
302
measurements will not be done after the 20ms given above, but take much longer and averaging a
303
multitude of single measurement shots.
304
Thus, beside the drawbacks of size, costs and running efforts for such a device in terms of liquid
305
helium and nitrogen for cooling down the magnets etc., it is not clear yet whether all desired
306
parameters can be measured in reasonable time, or if even all parameters important for a process to
307
be characterized can be measured. Aside of this, not all materials can be used with NMR. The
308
technology is very selective, but limited to non‐magnetic materials for devices. If process parameters
309
can only be reached using a vessel made of magnetic material NMR measurement methods fail due
310
to the magnetic properties of the device. Another drawback is that not all desired parameters can be
311
measured directly, some can be retrieved indirectly only.
312
While the measurement of temperature with NMR is in some cases precise and simple [121], the
313
process pressure can be retrieved by pressure‐sensitive materials only [122]. Viscosity, density, mass
314
flow and phase, phase changes or particle content can be measured directly (i.e. [123]), as well as the
315
electrochemical potential of compounds (see [124]), concentrations of single mixture compounds
316
[125] or the pH of a mixture.
317 318
319
Figure 4. 400MHz (9.4 Tesla) NMR system of the Bruker company. www.bruker.com
320
Additionally, this technology allows not only to measure different parameters non‐intrusively
321
in very small confinements like micro channels but also in living tissues [126, 127]. This is an add‐on
322
not provided by a multitude of the other techniques presented before. Another add‐on is the
323
possibility to visualize processes by magnetic resonance imaging (MRI, see: [128]). An example of
324
this is presented in Figure 5, showing a cut through a human head and imaging the brain.
325
Thus, NMR seems to provide the possibility to cover lots of measurement tasks of other
326
methods in a single machine, being non‐invasive, non‐intrusive and non‐detrimental to the
327
measured object. By all the possibilities provided herewith a discussion is needed on at least two
328
topics: is measurement with an NMR correlative? And, moreover, is NMR measurement the best
329
possible solution for every measurement task?
330
331
Figure 5. Magnetic resonance tomography (MRT) cut picture of a human head as an example for the
332
MRI possibilities. From: Private.
333
3. Discussion
334
As it was pointed out in section 2, lots of different methods for measurement of process
335
parameters are available and explored. Miniaturization of process systems results in major problems
336
for scaling them down, as it was mentioned and described before. However, with larger efforts,
337
smart design and good planning of experiments, scaling down of measurement systems and
338
obtaining precise results of the process parameters is possible to a certain extent, using various
339
methods for process parameter characterization. Thus, miniaturization and scaling is a problem for
340
intrusive measurement methods, while precise measurement of the parameter distribution is a
341
problem for non‐intrusive methods. This has clearly been described in section 2. In almost all cases,
342
measurements are performed sequentially, having each of the measurement and analysis systems
343
running after the one before. This type of performance, aside of the difficulties given by scaling,
344
allows to achieve good and reliable measurement results, but not to correlate them to each other and
345
to obtain a more fundamental understanding of the interactions between process parameters and
346
underlying principles. This is shown schematically in Figure 6, in which the current state‐of‐the‐art
347
measurement methodology is presented.
348
Thus, correlation of the measurement results is the major point to deal with. One of the future
349
perspectives of measurement is to obtain correlated information in situ and in operando by as many
350
different techniques as possible. Correlated measurement means timely and spatially, and with
351
similar resolutions for all applied techniques. This is necessary to gain an exhaustive view to the
352
examined process, to understand the underlying principles and actions, and therefore make them
353
accessible for modelling and simulation as well as in silico prediction and optimization. The latter
354
will lead to major reductions in resource consumption. Figure 7 shows a scheme of a possible future
355
measurement environment, in which a multitude of instruments are involved in parallel, means at
356
the same time and location, measuring in situ and in operando and, therefore, deliver results which
357
can directly be correlated. With this method, a much deeper understanding of links between actions
358
and effects can be achieved in a more appropriate way. Thus, correlative measurement, generation
359
of linked information and interpretation, management and feedback of connected data will be one of
360
the main tasks in future. The measurement structure schematically shown in Figure 7 can support
361
this task.
362
Some of the measurement methods presented above are useful for stand‐alone measurements
363
only, some can be performed simultaneously. However, the methods and techniques described in
364
sections 2.1 and 2.2 can hardly be done in a correlated way. It is possible to measure temperature,
365
pressure and chemical composition of a flow at the same position using, i.e., thermocouples,
366
pressure sensors and Raman spectrometry. To obtain a precise time correlation by using a trigger is
367
difficult, mainly due to different time constants of the measurement methods. Additional
368
visualization of the process adds some more problems, because for high speed videography as
369
described in section 2, very high light intensities are necessary, which might cope with other optical
370
measurement methods; therefore a combined high speed videography – Raman measurement is not
371
an option.
372
All this holds for all of the conventional methods described above. Without precise triggering,
373
correlative measurement is most likely impossible; and even with a time trigger, the location of
374
measurement as well as the spatial resolution most likely differs largely.
375
NMR makes it partly easier in this particular point. As described before, in principle a
376
multitude of parameters can be retrieved with NMR, all of them at the same location. Thus, spatial
377
resolution is consistent. However it is doubtful that a timely consistence can be reached because
378
some measurements simply take much longer than others, as pointed out before. Thus, truly
379
correlative measurements in terms of time and space cannot be generated for the most cases, this
380
problem stays. However, the data measured by NMR are to a certain extent per se correlated,
381
because several parameters are measured with the same method and then averaged on a time
382
period, which is a huge advantage compared to conventional measurement equipment. Process
383
pressure has to be measured via pressure‐sensitive materials (see description above and [122]),
384
which makes this a special case to be dealt with. Additionally, with MRI a handy visualization tool is
385
available, which provides even more in‐sight view to processes and flows. Thus, many of the
386
problems attached to conventional measurement systems are less applying for NMR. However, the
387
time correlation problem is kept, and due to the limitation of applicability to non‐magnetic device
388
materials NMR is not a panacea. It is a good toolbox for lots of possible measurement solutions
389
correlating the obtained data directly internally. A future vision would be to retrieve all process
390
parameters at the same time, with the same precision, the same spatial and time resolution, in‐situ
391
and in‐operando, as it was requested. Thus, NMR might be a nice additional tool for certain
392
measurement applications, but it cannot considered to be a stand‐alone tool for generation of
393
correlated data on processes yet. There’s need for combinations of methods, to obtain the desired
394
“information cloud”. Only with this, a reasonable combination of different measurement and
395
analytical tools, more comprehensive descriptions of processes might be feasible.
396 397
398
Figure 6. Current state‐of‐the‐art sequential measurement and analysis method. The timely and
399
locally separated measurement makes it difficult or even impossible to correlate measurement
400
results to each other and retrieve by this deeper insight into the underlying mechanisms.
401
Recently ideas for further correlated measurements came up, like MR‐optical combinations, or
402
integration of atomic force microscopy (AFM) and NMR into a single device. Moreover, proposals
403
on combining NMR technology with additive manufacturing, i.e. 3D printing methods, came up.
404
Here, the characterization as well as the measurement would be integrated in a single process,
405
allowing to generate miniaturized systems in a way which has not been thought of ten years ago.
406
407
Figure 7. Future improved parallel measurement and analysis method. The detectors are triggered
408
automatically in a way that they all measure at the same location, and deliver the measured data
409
with the same spatial and time resolution. Data are then transferred to an analysis level, where they
410
are correlated.
411
4. Conclusions
412
Macroscale measurement of gas flow parameters with conventional technology like
413
thermistors, pressure sensors etc. is, by no means, anymore a problem, but scaling those systems
414
down into the mini or micro scale still is. Due to either insufficient size of classical sensors or
415
insufficient possibilities to position micro sensors in the gas flow, the measurement influences the
416
process itself largely. Non‐intrusive methods are possible, but show other limitations, like optical
417
accessibility, the need for high power light exposure or non‐magnetic equipment materials.
418
One of the most interesting measurement problems is to achieve timely and spatially correlated
419
results. By conventional methods involving thermistors, GC, MS, Raman, IR or UV spectroscopy,
420
pressure transducers etc., this is almost impossible. Either the measurement location largely differs,
421
or the time resolution is simply not suitable. Aside, the combination of data from different
422
measurement sources is not trivial. The use of nuclear magnetic resonance systems (NMR), which in
423
principle allow measurement of a multitude of parameters at least at the same location, and in most
424
cases with the same spatial resolution, eases this problem slightly. Here, results can be obtained from
425
the same measurement location, show consistent data format, and can therefore be easily correlated.
426
Even visualization is relatively easily achieved by combining regular NMR with magnetic resonance
427
imaging MRI, which adds some more valuable data to the parameter cloud obtained by the
428
measurement. The scale range for NMR / MRI systems reaches from macro to nano scale. Spatial and
473
time resolution is not depending on the scale but on the magnetic system, which is also an advantage
474
compared to other measurement methods.
475
Recently, NMR and MRI systems themselves have been target of miniaturization. Meanwhile,
476
benchtop low field NMR systems up to 80MHz resonance frequency (about 1.88T) are available
477
which can easily ran in a conventional lab environment, used for stationary or flow‐through data
478
achievement. These systems are easy to handle and deliver a bunch of analytical possibilities.
479
Nevertheless, miniaturization went on, and meanwhile developments are reaching into the direction
480
of mini or micro NMR systems [126, 128‐130]. Even wireless systems are under research now, thus, it
481
is to expect that application of miniaturized and integrated NMR in correlative measurements will
482
be greatly enhanced in future.
483
It stays a vision to have measurements of different parameters by NMR with the same time
484
resolution, as well as the possibility to measure all the desired parameters of a process. Thus,
485
combinations of a multitude of analysis instruments will be necessary, ideally within a single
486
system, controlled by a combined measurement electronic system, which triggers internally all
487
methods.
488 489
Acknowledgments
490
The author wish to acknowledge Prof. Dr. Jan Korvink and Dr. Neil MacKinnon of KIT‐IMT for lots
491
of fruitful discussions, which led to some of the thoughts briefly described in this paper.
492 493
Funding
494
This research received funding by the European Commission H2020 Marie Skłodowska‐Curie
495
Actions under the Grant Agreement No. 643095 “MIGRATE”.
496 497
498
© 2018 by the authors. Submitted for possible open access publication under the terms
499
and conditions of the Creative Commons Attribution (CC BY) license
500
(http://creativecommons.org/licenses/by/4.0/).
501
502
[1] S. G. Kandlikar et al., Journal of Heat Transfer 2013, 135 (9), 091001-091001. DOI:
503
10.1115/1.4024354.
504
[2] G. L. Morini, International Journal of Thermal Sciences 2004, 43 (7), 631-651. DOI:
505
https://doi.org/10.1016/j.ijthermalsci.2004.01.003.
506
[3] H. A. Mohammed, G. Bhaskaran, N. H. Shuaib, R. Saidur, Renewable and Sustainable 507
Energy Reviews 2011, 15 (3), 1502-1512. DOI: https://doi.org/10.1016/j.rser.2010.11.031.
508
[4] Y. Han, Y. Liu, M. Li, J. Huang, Energy Procedia 2012, 14, 148-153. DOI:
509
https://doi.org/10.1016/j.egypro.2011.12.910.
510
[5] T. Dixit, I. Ghosh, Renewable and Sustainable Energy Reviews 2015, 41, 1298-1311.
511
DOI: https://doi.org/10.1016/j.rser.2014.09.024.
512
[6] P. Rosa, T. G. Karayiannis, M. W. Collins, Applied Thermal Engineering 2009, 29 (17), 513
3447-3468. DOI: https://doi.org/10.1016/j.applthermaleng.2009.05.015.
514
[7] A. Mohammed Adham, N. Mohd-Ghazali, R. Ahmad, Renewable and Sustainable 515
Energy Reviews 2013, 21, 614-622. DOI: https://doi.org/10.1016/j.rser.2013.01.022.
516
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio Feldfunktio
[8] S. Colin, Journal of Heat Transfer 2011, 134 (2), 020908-020908-13. DOI:
559
10.1115/1.4005063.
560
[9] S. Yin et al., Journal of Thermal Spray Technology 2016, 25 (5), 874-896. DOI:
561
10.1007/s11666-016-0406-8.
562
[10] M. Asadi, G. Xie, B. Sunden, International Journal of Heat and Mass Transfer 2014, 79, 563
34-53. DOI: https://doi.org/10.1016/j.ijheatmasstransfer.2014.07.090.
564
[11] B. H. Salman, H. A. Mohammed, K. M. Munisamy, A. S. Kherbeet, Renewable and 565
Sustainable Energy Reviews 2013, 28, 848-880. DOI:
566
https://doi.org/10.1016/j.rser.2013.08.012.
567
[12] Y. Zhai, G. Xia, Z. Li, H. Wang, Experimental Thermal and Fluid Science 2017, 83, 568
207-214. DOI: https://doi.org/10.1016/j.expthermflusci.2017.01.005.
569
[13] A. M. Sahar et al., Applied Thermal Engineering 2016, 93, 1324-1336. DOI:
570
https://doi.org/10.1016/j.applthermaleng.2015.08.087.
571
[14] B. Kim, International Journal of Heat and Fluid Flow 2016, 62, 224-232. DOI:
572
https://doi.org/10.1016/j.ijheatfluidflow.2016.10.007.
573
[15] O. Mokrani, B. Bourouga, C. Castelain, H. Peerhossaini, International Journal of Heat 574
and Mass Transfer 2009, 52 (5), 1337-1352. DOI:
575
https://doi.org/10.1016/j.ijheatmasstransfer.2008.08.022.
576
[16] A. M. Sahar et al., Applied Thermal Engineering 2017, 115, 793-814. DOI:
577
https://doi.org/10.1016/j.applthermaleng.2017.01.018.
578
[17] V. Hessel, P. Angeli, A. Gavriilidis, H. Löwe, Industrial & Engineering Chemistry 579
Research 2005, 44 (25), 9750-9769. DOI: 10.1021/ie0503139.
580
[18] G. N. Doku, W. Verboom, D. N. Reinhoudt, A. van den Berg, Tetrahedron 2005, 61 581
(11), 2733-2742. DOI: https://doi.org/10.1016/j.tet.2005.01.028.
582
[19] Y. Su, Y. Zhao, G. Chen, Q. Yuan, Chemical Engineering Science 2010, 65 (13), 583
3947-3956. DOI: https://doi.org/10.1016/j.ces.2010.03.034.
584
[20] A. Woitalka, S. Kuhn, K. F. Jensen, Chemical Engineering Science 2014, 116, 1-8. DOI:
585
https://doi.org/10.1016/j.ces.2014.04.036.
586
[21] M. N. Kashid, A. Renken, L. Kiwi-Minsker, Chemical Engineering Science 2011, 66 587
(17), 3876-3897. DOI: https://doi.org/10.1016/j.ces.2011.05.015.
588
[22] F. Reichmann, A. Tollkötter, S. Körner, N. Kockmann, Chemical Engineering Science 589
2017, 169, 151-163. DOI: https://doi.org/10.1016/j.ces.2016.10.028.
590
[23] C. Yao, Z. Dong, Y. Zhao, G. Chen, Chemical Engineering Science 2015, 123, 137-145.
591
DOI: https://doi.org/10.1016/j.ces.2014.11.005.
592
[24] S. Haase, D. Y. Murzin, T. Salmi, Chemical Engineering Research and Design 2016, 593
113, 304-329. DOI: https://doi.org/10.1016/j.cherd.2016.06.017.
594
[25] J. Yue, Catalysis Today 2018, 308, 3-19. DOI:
595
https://doi.org/10.1016/j.cattod.2017.09.041.
596
[26] A. Tanimu, S. Jaenicke, K. Alhooshani, Chemical Engineering Journal 2017, 327, 597
792-821. DOI: https://doi.org/10.1016/j.cej.2017.06.161.
598
[27] X. Yao et al., Renewable and Sustainable Energy Reviews 2015, 47, 519-539. DOI:
599
https://doi.org/10.1016/j.rser.2015.03.078.
600
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
[28] G. Kolb, Chemical Engineering and Processing: Process Intensification 2013, 65, 1-44.
643
DOI: https://doi.org/10.1016/j.cep.2012.10.015.
644
[29] H. Pennemann, G. Kolb, Catalysis Today 2016, 278, 3-21. DOI:
645
https://doi.org/10.1016/j.cattod.2016.04.032.
646
[30] I. Rossetti, M. Compagnoni, Chemical Engineering Journal 2016, 296, 56-70. DOI:
647
https://doi.org/10.1016/j.cej.2016.02.119.
648
[31] J. Zhao et al., Scientific Reports 2016, 6, 32393. DOI: 10.1038/srep32393.
649
[32] J. C. Harley, Y. Huang, H. H. Bau, J. N. Zemel, Journal of Fluid Mechanics 2006, 284, 650
257-274. DOI: 10.1017/S0022112095000358.
651
[33] C.-M. Ho, Y.-C. Tai, Annual Review of Fluid Mechanics 1998, 30 (1), 579-612. DOI:
652
10.1146/annurev.fluid.30.1.579.
653
[34] G. Hetsroni, A. Mosyak, E. Pogrebnyak, L. P. Yarin, International Journal of Heat and 654
Mass Transfer 2005, 48 (10), 1982-1998. DOI:
655
https://doi.org/10.1016/j.ijheatmasstransfer.2004.12.019.
656
[35] R. Goldstein, Fluid Mechanics Measurements, Second Edition. Taylor & Francis, 2017.
657
[36] W. o. Science, 2019 ed., Clarivate Analytics,2019.
658
[37] Z. Yunusa, M. N. Hamidon, A. Kaiser, Z. Awang, Gas Sensors: A Review. 2014.
659
[38] E. Llobet, Sensors and Actuators B: Chemical 2013, 179, 32-45. DOI:
660
https://doi.org/10.1016/j.snb.2012.11.014.
661
[39] S. S. Varghese et al., Sensors and Actuators B: Chemical 2015, 218, 160-183. DOI:
662
https://doi.org/10.1016/j.snb.2015.04.062.
663
[40] F.-L. Meng, Z. Guo, X.-J. Huang, TrAC Trends in Analytical Chemistry 2015, 68, 37-47.
664
DOI: https://doi.org/10.1016/j.trac.2015.02.008.
665
[41] S. Pandey, Journal of Science: Advanced Materials and Devices 2016, 1 (4), 431-453.
666
DOI: https://doi.org/10.1016/j.jsamd.2016.10.005.
667
[42] P. Kuberský et al., Procedia Engineering 2015, 120, 614-617. DOI:
668
https://doi.org/10.1016/j.proeng.2015.08.746.
669
[43] H. Elhaes, A. Fakhry, M. Ibrahim, Materials Today: Proceedings 2016, 3 (6), 670
2483-2492. DOI: https://doi.org/10.1016/j.matpr.2016.04.166.
671
[44] S. Basu, P. Bhattacharyya, Sensors and Actuators B: Chemical 2012, 173, 1-21. DOI:
672
https://doi.org/10.1016/j.snb.2012.07.092.
673
[45] K. Toda, R. Furue, S. Hayami, Analytica Chimica Acta 2015, 878, 43-53. DOI:
674
https://doi.org/10.1016/j.aca.2015.02.002.
675
[46] K. Xu et al., Instrumentation Science & Technology 2018, 46 (2), 115-145. DOI:
676
10.1080/10739149.2017.1340896.
677
[47] R. Bogue, Sensor Review 2014, 34 (1), 1-8. DOI: doi:10.1108/SR-03-2013-637.
678
[48] J. Zhang, X. Liu, G. Neri, N. Pinna, Advanced Materials 2016, 28 (5), 795-831. DOI:
679
doi:10.1002/adma.201503825.
680
[49] N. Joshi et al., A review on chemiresistive room temperature gas sensors based on metal 681
oxide nanostructures, graphene and 2D transition metal dichalcogenides. 2018.
682
[50] X. Liu, T. Ma, N. Pinna, J. Zhang, Advanced Functional Materials 2017, 27 (37), 683
1702168. DOI: doi:10.1002/adfm.201702168.
684
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio
Feldfunktio