The measurement of gas and liquid flowrates is an essential requirement in many industrial and commercial applications. In 1975 Hayward estimated there are more than one hundred different types of sensors with a mode of operation based on almost any physical domain [1]. Although several large scale flowmeter types are commercially available, the continuous development of three dimensional techniques of micro-fabrication, with consequent cost reduction and quality improvement, has rapidly extended the market of micromachined flow sensors. The growing interest in research activities related to MEMS is demonstrated by the vast scientific literature and the several hundred companies specifically dedicated to micromachined systems. This growth is mainly due to some crucial advantages of micromachined flow sensors compared to large-scale ones, such as better dynamic characteristics, low power consumption, reduced mass, small size, and cost-effectiveness thanks to batch-fabrication, among other attributes. Although often associated to silicon, because of its frequent use, its economical characteristics and its desirable mechanical properties, micromachining has been performed on several materials including metals, polymers and glasses. The concept of micro-electro-mechanical systems (MEMSs) is still considered to be an emerging technology, but it was born in the early sixties [2], and almost fourteen years (1974) passed until a milestone study appeared, describing the first integrated silicon-based sensor for gas flow measurement [3]. A high growth of research works in the field of MEMS and micromachined flow sensors took place in the eighties and about a decade was needed to develop the integration of many microfluidic devices into a single chip (e.g., micro pumps, valves and flow sensors). Given that microflow devices have very small volume, the result of this challenge was a new class of micromachined flow sensors with an integrated flow micro channel [4]; which Petersen introduced for the first time into this design [5].

The small size and certain valuable characteristics (e.g., high sensitivity, accuracy and precision, low power consumption) coupled with the chance of providing a better outcome for the patients and lower health care cost, make the potential of micro-fabricated devices in medical applications enormous. In fact, some medical fields, ranging over a wide variety of applications in surgical, diagnostic devices and therapeutic areas, are involved in the continuous expansion of micromachined devices [6]. The vast majority of MEMS implemented in biomedical applications are sensors for monitoring many physical parameters such as pressure, acceleration, and fluid flow among others. They are commonly used in orthopedic research field in the study of muscles and patient’s posture, in the monitoring of blood flow, in the measurement of pressure, such as intravascular blood pressure [7], in microsurgery [8], bladder and intraocular applications [9] and in measurement of cerebro-spinal fluid pressure [10]. They are also utilized to monitor the outcome of abdominal aortic aneurysms surgery [11], in the long term monitoring of prosthetic devices, in respiratory monitoring to measure gas flows in spirometric devices [12], to diagnose salivary gland [13] and stomach diseases [14], and in microfabricated drug delivery devices [15]. In this review we do not exhaustively present each application of micromachined sensors in the biomedical field, rather we focus our attention on some applications of micro flow sensors in the monitoring of physiological parameters or used as a feedback in delivery microsystems [16].

This paper is divided into subsections where a concise description of the measurement principle of micromachined flow sensors is presented along with the main biomedical applications. A detailed analysis of sensor’s performances and principles of measurement are also reported.

We will follow the most common classification for micromachined flow sensors, based on the working principles, that distinguishes two groups: the former contains flow sensor based on heat exchange, named “thermal flow sensors”; all others flow meters, based on different working principles of thermal exchange, are grouped into the so-called “non-thermal flow sensors” group. This classification is a consequence of the huge number of thermal-based flow sensors reported in the literature [4]. In the following two sections, both groups of micro flow sensors and their use in specific medical applications are reported. The performances of the sensing methods are also presented.

The principles of measurement of thermal flow sensors include transport principles of three of the major branches of physics, based on mechanical-thermal-electrical transport phenomena: the interaction between the measurand (mechanical parameter) and the core of the sensing element, heated by Joule effect (electrical phenomenon), causing a thermal exchange (thermal phenomenon). Therefore, the mechanical measurand modulates, through thermal exchange, one parameter of the sensing element; the modulation produces an output voltage signal, as schematically reported in Figure 1.

Thermal flow sensors are usually divided in two subgroups:

A further principle is based on the flow dependence of time of flight taken by a heat pulse over a known distance, although it is rarely mentioned in published papers [4].

The first integrated silicon anemometer was presented by van Putten et al. [3]. The same author found a technical solution to drift, by using an integrated double bridge anemometer [18]. The higher performances obtained by Tai et al. [19] and the fabrication of the first flow sensor with integrated micro channel [20], encouraged the interest of researchers in the development of micromachined flow sensors. This is confirmed by numerous and interesting works in the eighties and nineties, which presented several flow sensors with various principles of measurement and performances [17,21,22,23,24,25]. During the last decade the interest of the scientific community is still growing and some works with valuable peculiarities (e.g., multi-range sensors to facilitate the measurement of both low and high gas velocity, very low power consumption sensors, miniaturized sensors) have been reported [26,27,28,29]. For example Glaninger et al. fabricated a thermal flow sensor using a thin film germanium thermistor with a wide range of measurement (±0.01 m∙s⁻¹ to ±200 m∙s⁻¹) and a response time of less than 20 ms [30]. Other interesting works, which have introduced technical solutions to improve some sensor characteristics (e.g., accuracy, sensitivity, and range of measurement), have been published in the last few years [31,32,33,34].

In summary, this family of sensors shows heterogeneous performances:

The main advantages of thermal flow sensors are high sensitivity and the wide measurement range; moreover the introduction of solutions to thermally isolate the sensing element is essential to improve some characteristics (e.g., accuracy). The drawbacks of many thermal flow sensors are substantially related to the non-linearity of the calibration curve: although it could represent a valuable characteristic because of the abovementioned high sensitivity at low flow rates, on the other hand, non-linearity causes an appreciable sensitivity decrease with fluid flow. In some cases, a further issue is related to possible presence of dust or humidity in gas flow that strongly affects sensor accuracy.

In the following section a detailed description of micromachined thermal flow sensors utilized in some biomedical applications is reported. The measurement principles of the sensors, and their performances are analyzed.

All the transducers whose measurement principle is not based on the modulation of thermal exchange by fluid flow are considered non-thermal flow sensors. They are also divided in subgroups. In an exhaustive and recent review, Wang et al. [2] classified non-thermal flow sensors depending on the principle of measurement used: resonating, differential pressure-based, lift-force-based, and cantilever type flow sensors. A brief description of these working principles and a more detailed description of laser Doppler anemometry (LDA) will be provided. This laser measurement technique is of particular interest in medicine (e.g., for blood flow monitoring) and in the last decade the fabrication of integrated laser Doppler flowmeters has been one of the main challenges in micromachined devices.

√ Lift-force, drag-force or cantilever flow sensors

These principles of measurement are based on the force acting on a body located in a fluid flow. Actually, the fluid flow exerts on the body two perpendicular force components: the drag force, F_drag, i.e., the force acting in the flow direction; and the lift force, F_lift, i.e., the force perpendicular to the flow stream. Both are proportional to the square of the mean flow velocity ():

    (3)

where C_D is the drag-force coefficient and C_L is the lift-force coefficient, ρ is the fluid density, l and w are the body length and width. C_L strongly depends on the angle of attack and is experimentally estimated. The study of Svedin et al. is a milestone in the design of micromachined flow sensors based on lift-force [59,60]: the deflection of two thin plates of single-crystalline silicon induced by the lift force is utilized to modulate the electrical resistances of an array of strain gauges connected in a Wheatstone bridge configuration. In a successive work, Svedin et al. combined the lift-force with a thermal principle of measurement to obtain the detection of flow direction and to improve the sensor sensitivity in a wider range (up to 70 L∙min⁻¹) [61,62]. The lift force is insensitive to flow direction therefore, a dual chip configuration to detect the direction was introduced. Moreover, both thermal and lift-force sensors have a non-linear calibration curve with a considerable difference: thermal flow sensor has high sensitivity at low flow (accordingly to King’s law), whereas the lift-force principle makes sensor’s output increase with flow, therefore both principles were combined to increase the sensitivity in a wide flow rate range.

√ Drag-force or cantilever type flow sensors

As far as it concerns drag-force-based and cantilever flow sensors, an interesting work was presented by Gass et al. where the drag force exerted by fluid flow onto a cantilever beam allowed to measure flow ranging from 5 µL∙min⁻¹ to 500 µL∙min⁻¹ [63]. Similar approaches were used by Su et al. [64] and Kao et al. [65]. A cantilever flow sensor, allowing velocity measurement up to 45 m∙s⁻¹, with a response time of about 0.53 s and a power consumption of 0.02 mW, was realized by Wang et al. [66]. They upgraded their valuable work by integrating temperature compensation and the ability to discriminate flow direction [67], as shown in Figure 3. The principle of measurement, based on the deflection of a cantilever operated by fluid flow, allows a voltage output obtained by a Wheatstone bridge with resistance of piezoresistors depending on the cantilever deflection, thus on the flow value. Recently, the development of multi-sensor integrated technology is a trend in MEMS followed by Du et al., among others, who integrated a previously designed drag force sensor [68] with a thermal one to realize a micro-wind sensor. Also in this case, the sensor utilizes the thermal part for low wind velocity (<2 m∙s⁻¹), the mechanical one for high velocity (>2 m∙s⁻¹) [69].

√ Differential pressure-based flow sensors

In the early twentieth century, this measurement principle has been introduced in large-scale gas flowmeters with different approaches (e.g., orifice or Venturi flowmeters and Fleisch pneumotachographs), where flow estimation is carried out by the measurement of pressure drop across an ad hoc designed sensing element (an orifice or a linear resistance). When a fluid flow passes through a duct, or over a surface, it produces a pressure drop depending on the mean velocity of the fluid. Several micromachined sensors relying on this principle are fabricated: as an example, Berbering et al. developed a differential-based sensors realizing flow-velocity detection up to 23 m∙s⁻¹ by measurement of the pressure difference between the stagnant fluid pressure in front of the sensor and static pressure of fluid around the sensor [70]. The pressure gradient causes the deflection of a membrane that constitutes one of the two plates of a capacitor, and results in a decrease of capacitance between the electrodes converted in a voltage signal by an electrical circuit.

√ Resonating flow sensors

In 1989, the influence of temperature on the resonance frequency of a vibrating membrane was used to develop a micromachined flow sensor with a 2 µm thick membrane [71]. Successively, the same authors realized a resonating microbridge to improve sensor performance [72]. Microflow sensors based on resonance frequency shift of the output signal are relatively robust against noise levels and signal drift. Different approaches were used to realize these sensors, for example, Seo et al. [73] have recently presented a self-resonant flow sensor were the vibration of a microcantilever is modulated by surface stress acting on the beam due to fluid drag force. A cantilever, placed in the fluid stream, undergoes mechanical oscillations related to the vortex shedding frequency. This phenomenon, known as “flow-induced or vortex induced vibration”, can be expressed as:

    (4)

being f_s the vortex shedding frequency, V the flow velocity, St the Strouhal number, and D the characteristic length of the cantilever. In MEMS application the fluid flow can be turbulent or the cantilever beam can be smaller than the boundary layer of the fluid flow, and the cantilever beam oscillates by the “turbulence-induced vibrations”. In this case the microstructure vibration is modulated by the resonant frequency of the cantilever itself and shows a resonance frequency depending on the axial force induced by the fluid drag force. The vibrations are sensed by a piezoelectric material placed on a silicon cantilever beam.

√ Laser Doppler flowmeter

This working principle will be described in more detail because of its large use in the medical field and its potential in applications related to blood flow monitoring. The design of these sensors is based on the Doppler shift described by Christian Doppler in 1842. The phenomenon is due to the interaction between an electromagnetic or acoustic wave and a moving object: the wave is reflected back showing a frequency different from the incident one. The frequency shift, Δf, depends on the velocity of the moving object, therefore, if the object moves within the fluid flow, it represents a flow indirect measurement:

    (5)

Equation (5) gives a direct relation between the frequency shift and the velocity (U) of the moving object; λ is the wavelength of the incident wave, α the angle between the axis of the incoming wave and the observer, and β the angle between the moving object velocity direction and the bisector of the angle between the axis of the incoming wave and the segment connecting the object and the observer, as depicted in Figure 4(a).

The “dual-beam approach” is often implemented to improve sensor’s performances: the radiation is split into two beams and successively crossed into an intersection region through a mirror (Figure 4(b)). This region defines a sampling volume, placed into the fluid flow, where the particle velocity is measured. In the intersection region, an interference pattern, characterized by interference fringes, is produced. When a particle passes through the fringes, it causes periodic variations of the intensity of the scattered light [74]. The particle speed can be expressed as a function of the frequency fluctuations (f_r) of the scattered intensity and of the distance between two contiguous fringes (γ) as follows:

    (6)

being θ the angle between the two beams.

Some works were carried out to realize compact laser Doppler-based instruments [75]. The aim was to create devices that integrate all the optical components of the laser Doppler flowmeter. This approach allows avoiding the main drawback affecting the most common techniques used to conduct the laser light in the site of measurement and from the site to the photodetector (e.g., by a fiber optic): i.e., the need of alignment between individual optical components [76].

In summary, some non-thermal flow sensors present the main drawback of the response’s dependence on the fluid density.

Micromachined devices represent the next phase in industrial measurement and monitoring. At present, these devices are mainly implemented in industrial applications such as automotive ones. Although their introduction in medical field is just at the beginning, they demonstrate enormous potential for surmounting common issues encountered in medicine and are considered as an attractive area of research. Interest is encouraged by the valuable characteristics of micromachined devices with a number of possibilities to be discovered in the near future: miniaturization is an essential requirement in minimally-invasive medical applications. Moreover, good accuracy and fast response are also crucial factors for monitoring physiological parameters in surgical, diagnostic and therapeutic areas. At present micromachined sensors are mainly used in medicine for pressure monitoring, but this technology could revolutionize some medical applications. The aim of this review is to give an idea of the potentialities of micromachined flow sensors in medical applications without presuming to exhaustively present all the numerous publications on this topic.

In particular their introduction could improve the flow rate monitoring of blood, in large vessels or microcirculation, and respiratory flows. They present some crucial characteristics to be used in these fields, such as short response time, good sensitivity and accuracy; moreover, their small size could introduce some advantages in minimally invasive measurements. Regardless of their huge potentialities, the greatest challenges for commercial success of this technology on the medical market are cost reduction and robust design.