An intelligent artificial throat with sound-sensing ability based on laser induced graphene

Abstract

Traditional sound sources and sound detectors are usually independent and discrete in the human hearing range. To minimize the device size and integrate it with wearable electronics, there is an urgent requirement of realizing the functional integration of generating and detecting sound in a single device. Here we show an intelligent laser-induced graphene artificial throat, which can not only generate sound but also detect sound in a single device. More importantly, the intelligent artificial throat will significantly assist for the disabled, because the simple throat vibrations such as hum, cough and scream with different intensity or frequency from a mute person can be detected and converted into controllable sounds. Furthermore, the laser-induced graphene artificial throat has the advantage of one-step fabrication, high efficiency, excellent flexibility and low cost, and it will open practical applications in voice control, wearable electronics and many other areas.

ARTICLE

Received 26 Aug 2016 | Accepted 13 Jan 2017 | Published 24 Feb 2017

Lu-Qi Tao1,*, He Tian1,2,*, Ying Liu1, Zhen-Yi Ju1, Yu Pang1, Yuan-Quan Chen1, Dan-Yang Wang1, Xiang-Guang Tian1, Jun-Chao Yan1, Ning-Qin Deng1, Yi Yang1 & Tian-Ling Ren1

Traditional sound sources and sound detectors are usually independent and discrete in the human hearing range. To minimize the device size and integrate it with wearable electronics, there is an urgent requirement of realizing the functional integration of generating and detecting sound in a single device. Here we show an intelligent laser-induced graphene articial throat, which can not only generate sound but also detect sound in a single device. More importantly, the intelligent articial throat will signicantly assist for the disabled, because the simple throat vibrations such as hum, cough and scream with different intensity or frequency from a mute person can be detected and converted into controllable sounds. Furthermore, the laser-induced graphene articial throat has the advantage of one-step fabrication, high efciency, excellent exibility and low cost, and it will open practical applications in voice control, wearable electronics and many other areas.

DOI: 10.1038/ncomms14579 OPEN

An intelligent articial throat with sound-sensing ability based on laser induced graphene

1 Institute of Microelectronics and Tsinghua National Laboratory for Information Science and Technology (TNList), Tsinghua University, Beijing 100084, China.

2 Ming Hsieh Department of Electrical Engineering, University of Southern California, Los Angeles, California 90089, USA. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to Y.Y. (email: mailto:[email protected]

Web End [email protected] ) or to T.-L.R.(email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:14579 | DOI: 10.1038/ncomms14579 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14579

Owing to diseases and unexpected accidents, a lot of people in this world are not able to speak out with languages. Some technologies have been developed to help them

express themselves in other ways. For example, eye tracking technology is developed to convert the eye movements into language expressions. They can use their eyes to focus on their wanted characters1,2. However, these technologies are not widely applied due to their complexities and extreme costs. In fact, many mute people have the ability to produce some kind of sounds such as a cough, hum or scream, which do not have clear meaning to other people. It is of great signicance to develop an easy-to-use articial throat that can convert the sound with unclear meaning into controllable and precise sound in languages. It means that the articial throat should have the capabilities of both detecting and generating sound. However, acoustic transducers that can achieve this usually work with narrow bandwidth in the ultrasound range3 or under water4,5. Traditional sound sources and detectors are usually isolated from each other and are discrete within the human hearing range. Besides, they have the drawbacks of resonant peak and poor exibility, which are not suitable for wearable applications.

The research of nanomaterials and nanotechnologies are contributing to the development of novel acoustic devices based on new mechanisms. For example, thermoacoustic sound sources based on graphene612, carbon nanotube13,14, silver nanowires15 and indium tin oxide16 have attracted a lot of attentions due to their high transparency, low heat capacity, broad frequency and excellent exibility. However, the complex fabrication process, low yield and high cost will restrict the commercial and practical applications of thermoacoustic sources. Except novel sound sources, some nanomaterial-based novel sound detectors1722 have been developed based on the piezoresistive effect. These novel sound detectors have responses towards the movement of throat. However, the responses are indistinctive and there is still a long way to make them for practical applications. Most importantly, the different working mechanisms of thermoacoustic sound sources and piezoresistive sound detectors make it hard to realize the functional integration of the sound source and detector in a single device. On one hand, the subsistent sound detectors based on piezoresistive materials18,2123 cannot work as sound sources. Most of these sensors were encapsulated in the elastic polymers such as Ecoex20 or polydimethylsiloxane22, which made it impossible to emit Joule heating into the air so that they cannot work as sound sources. The rest sensors18,21,23 have very high original resistances (up to MO); therefore, little Joule

Heating could be generated. Previous reported sound detectors cannot work as sound sources; thus, these kinds of sound detectors can be replaced by conventional microphones. On the other hand, the thermal sound sources that were developed previously11 cannot work as sound detectors, because the gauge factor of laser scribed graphene is only 0.11 (ref. 24), which means it has a poor sensitivity and is hard to work as a sound detector; thus, these kinds of sound sources can be replaced by conventional speakers. However, none of the previously demonstrated nanomaterial-based acoustic device can have both sound-emitting and -detecting ability in hearing range, and this kind of multifunctional device would not be replaced by any conventional devices. Nevertheless, the functional integration of generating and detecting hearable sound in a single device has not been explored until now.

It is known that laser scribing technology has been demonstrated as an efcient method to fabricate graphene-based devices2530. Some graphene devices based on laser scribing technology has been developed11,27, which shows advantages of large scale and customized pattern. However, the preparation of graphene oxide and the spin coating process still cause a waste of

time. Lin et al.31 developed a method of converting polyimide (PI) into porous graphene lms via one-step process. This method is further applied to develop supercapacitor energy storage devices32 and exible strain sensors33,34. The high thermal conductivity and low heat capacity of laser induced graphene (LIG) is ideal for thermoacoustic sound sources. Besides, the porous structure will have a sensitive response towards weak vibrations, which is suitable for sound detection. Therefore, the simple and one-step fabrication will be suitable for the realization of this novel articial throat.

Herein, we develop a one-step, wearable and low-cost LIG articial throat by direct laser writing of PI and it shows good performance in generating and detecting sounds. This LIG articial throat has completely different working mechanism compared with conventional acoustic transducers, which usually utilize piezoelectric effect and inverse piezoelectric effect3537. When working as a sound source, LIG articial throat can generate wide-band sound with frequency from 100 Hz to 40 kHz. A thinner LIG will produce higher sound pressure level (SPL). When working as a sound detector, LIG articial throat shows unique responses towards different kinds of sounds and throat vibration modes. LIG can recognize cough, hum and scream with different tones and volumes. Besides, it also has the capability of recognizing words and sentences. Beneting from its capability of generating and detecting sounds, the intelligent LIG articial throat will signicantly assist for the disabled. The throat vibration with different volume and frequency can be converted into controllable and predesigned sounds. Furthermore, the LIG articial throat has the advantage of one-step process, high efciency, excellent exibility and low cost. The LIG can be acquired by using a portable and low-power laser platform, which will reduce the risk under operation. The LIG articial throat has promising applications in the elds including voice control, wearable electronics and many others.

ResultsFabrication and characterization of LIG articial throat. Laser direct writing technology promotes the fast growth of porous graphene, and a low-cost and portable laser platform is chosen in this work (Supplementary Fig. 1). Figure 1a shows the one-step process of the LIG. The PI lm is located under the 450 nm laser and converted into LIG by direct irradiation of laser. The XY directional motors control the movement of the laser so that a predesigned pattern can be irradiated at precise locations (see the Methods section for details). For instance, Tsinghua Universitys logo and a 6 cm 4 cm rectangle are imported into the computer

control software, and the same patterns are generated on the PI lm by converting PI into graphene (Supplementary Fig. 2). Here, a simple rectangle LIG is produced to work as an articial throat. As illustrated in Fig. 1b, the articial throat has the integrated functions of emitting and detecting sounds. When an AC voltage is applied on the device, the periodic joule heat will cause the expansion of air, resulting in sound waves. When a low bias voltage is applied on the device, the vibration of throat cords will cause the change of the devices resistance, resulting in a uctuation of the current. Therefore, the devices can work as sound source and detector at the same time. The working principle of the LIG articial throat is shown in Fig. 1c. A cough, hum or scream will cause the vibration of throat cords, which can be detected by the LIG articial throat, and then LIG articial throat will generate controllable sounds accordingly. Therefore, the LIG articial throat can realize the conversion from meaningless sounds to controllable and predesigned sounds.

The photograph of LIG generated at different laser power P ranging from 20 to 350 mW is shown in Fig. 1d. The most

2 NATURE COMMUNICATIONS | 8:14579 | DOI: 10.1038/ncomms14579 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14579 ARTICLE

a c

b

Electrode

Acoustictransducer LIG

Input

Output

450 nm laser

Polyimide

LIG

Scribing platform

d e g

f

High power

h j

i

Low power

Figure 1 | Schematic illustration of the fabrication process and the morphology of LIG. (a) One-step fabrication process of LIG. PI is directly converted into LIG by the irradiation of the 450 nm laser. Scale bar, 2.5 cm. (b) LIG has the ability of emitting and detecting sound in one device. (c) The articial throat can detect the movement of throat and generate controllable sound, respectively. (d) Six LIG samples produced by 450 nm laser with different power ranging from 20 to 350 mW. Scale bar, 5 mm. (e) The morphology of LIG sample produced at 290 mW under scanning electron microscopy. Scale bar, 150 mm. (f) The morphology of LIG sample produced at 290 mW under high magnication. Scale bar, 5 mm. (g) Cross-sectional view of LIG sample produced at 290 mW. Scale bar, 12.5 mm. (h) The morphology of LIG sample produced at 125 mW under scanning electron microscopy. Scale bar, 150 mm.

(i) The morphology of LIG sample produced at 125 mW under high magnication. Scale bar, 5 mm. (j) Cross-sectional view of LIG sample produced at 125 mW. Scale bar, 12.5 mm.

bottom LIG is irradiated at P 20 mW and no obvious LIG is

observed (as shown in Supplementary Fig. 3). The second one and the fourth one from the bottom, which are generated at P 125 mW and P 290 mW, respectively, are chosen to show

their scanning electron microscope images because of their typical morphologies. As shown in Fig. 1ej, it can be noticed that ridgy lines are formed orderly along the scanning trace of laser from up to down. The line width is around 100 mm, which is similar to the focus spot size of the laser. Morphological differences can be clearly observed with the increasing of laser power. More pores are produced at P 290 mW. From the high-

magnication image, the differences between the structures become more distinctive. A thin layer of carbon sheet with porous polygon appears at P 125 mW and more porous and

irregular structure is produced at P 290 mW, because local high

temperature during laser heating conduct the pyrolysis of PI38,39, breaking the CO, C O and NC bonds, and causing the

venting of some carbonaceous and nitric gases. From the side view, we can notice that the thickness increases from B10 to B40 mm at P 125 mW and P 290 mW, respectively. The

structure generated at 350 mW (Supplementary Fig. 4) becomes more porous and thicker.

The Raman spectrums of PI lm and samples generated at P 125 mW and P 350 mW are performed for further

investigation (Supplementary Fig. 5). The spectrums obtained at P 125 mW and P 350 mW show the similar characteristics

with a D, G and 2D peak at 1,350, 1,580 and 2,700 cm 1 respectively. The Raman spectrum is clearly different from amorphous carbon, proving the existence of randomly graphene stacks. With the increasing of the laser power, the intensity of D peak increases, indicating that high-power laser makes the structure more defective and disordered. Moreover, the high-resolution transmission electron microscope image presents the lattice fringes with an interspace of B3.4 , corresponding to the interplanar spacing of (002) plane in graphic materials (Supplementary Fig. 6).

LIG articial throat working as a sound source. Four samples of LIG articial throats, which are produced at different laser powers, are used to test the performance of emitting sound. The area of the LIG is around 1 2 cm2. The laser powers are 125,

200, 290 and 350 mW. The average thickness of LIG is 8, 22,

38 and 60 mm. The LIG articial throat is clamped under a

NATURE COMMUNICATIONS | 8:14579 | DOI: 10.1038/ncomms14579 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14579

a

b

c

0.012 20 kHz 10 kHz 20 kHz 10 kHz

70

60

30

20

Sound pressure (Pa)

0.009

SPL (dB)

50

0.006

40

0.003

0.000 0.0 0.1 0.2 0.3 0.4

10 k 20 k 30 k 40 k

Power (W) Frequency (Hz)

0 20

d

e

f

0.030

70

60

30

10 kHz

60

0.024

SPL (dB)

Sound pressure (Pa)

50

50

0.018

SPL (dB)

40

40

0.012

30

100 Hz 10 kHz 20 kHz 40 kHz

0.006

20

10k 20k 30k 40k

0

20 40 60 80

40

60

80

100

120

140

160

180

Frequency (Hz)

dLIG (m)

Time (min)

Figure 2 | The performance of emitting sound. (a) The LIG is clamped under a commercial microphone to test the performance of emitting sound. Scale bar, 1 cm. (b) The plot of the SP versus the input power at 10 and 20 kHz. The square is the experimental result and the line is the theoretical result. (c) The output SPL versus the frequency of LIG generated by the laser with different power. The four curves are normalized with the input power of 1 W. (d) The SPL versus the frequency showing that the model agrees well with experimental results. (e) The plot of the SP versus the thickness of LIG at 10 and 20 kHz. The square is the experimental result and the line is the theoretical result. (f) The stability of output SPL over time.

commercial microphone as shown in Fig. 2a. The distance between LIG and microphone is 2.5 cm (see the Methods section for details). As shown in Fig. 2b, the output SP of LIG (generated under 125 mW) increases with the input power and the LIG has a higher efciency at 20 kHz. The tting line shows that the SP has a linear relation with the input power. The relationships between SPL and the frequency of different samples are demonstrated in Fig. 2c. The experimental results are normalized with the same power (1 W). It is noticed that the SPL becomes lower with the increasing of laser power and the thickness of LIG. A similar phenomenon was found in the work by Tian et al.6.

The theoretical model of thermal acoustic is rst built by Arnold and Crandall40. In recent years, a new model of the thermal acoustic effect is proposed based on the energy conservation41, which makes it easy to analyse and calculate. The core of thermal acoustic is the propagation of the heat. SP is decided by the thermal energy diffused into air:42

prms

mair f 2

2

p CpT0r

_

Qair 1

where mair is the molecular weight of air, f is the frequency of the acoustic, Qair is the thermal energy diffused into air, Cp is the heat capacity at constant pressure, T0 is the room temperature and r is the measuring distance from the source.

The thermal energy is calculated as:41

_

Qair

2 2 2

p x 2x2 2

p x3 4 1 2

p x x22

Pe1 w

2

where x p2

p

rsCsdsf 0:5

p

rPICv;PIkPI

p , w

rPICv;PIkPI

rairCv;airkair

q , and

r; Cv and k are the density, heat capacity at constant pressure and conductivity of each material, respectively. ds is the thickness of the LIG and Pe is the input power. From equation (2), we can get the SPL as a function of frequency and the thickness of LIG.

Figure 2d is the comparison of theoretical curve and the experimental results. The experimental analysis matches well with the theory model. Figure 2e shows the experimental data and theoretical curve as a function of the thickness of the LIG under the frequency of 10 and 20 kHz. The four samples with the thickness of 8, 22, 38 and 60 mm would produce the sounds with the SP of 0.01775, 0.01276, 0.01019 and 0.00706 Pa at the frequency of 10 kHz and 0.0235, 0.0183, 0.01346 and 0.01 Pa at the frequency of 20 kHz, respectively. The theoretical curve matches well with the experimental results. As the decrease of the thickness, the leakage of thermal energy will be reduced and more thermal energy will be propagated into the air, implying that a thinner LIG will generate a higher SP. The experimental data t well with the theoretical model. The working temperature is tested by a thermal infrared camera (Fluke Ti 200). The working temperature are 23.7 C with 0.4 W input power and below 29 C even with 1.225 W input power, which are acceptable when attached to the skin (Supplementary Fig. 7). The SPL is measured for a long time to investigate the stability of LIG. SPL at 100 Hz, 10 kHz, 20 kHz and 40 kHz are shown in Fig. 2f. There are no signs of degradation or changes in the device performance in 3 h.

LIG articial throat working as a sound detector. Except for emitting sound, the LIG articial throat also has excellent responses when detecting sound. PI with different thickness will have different performance of recognition. A 25 mm-thick PI is

rairCv;airkair

4 NATURE COMMUNICATIONS | 8:14579 | DOI: 10.1038/ncomms14579 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14579 ARTICLE

a

b

c

Resistance (a.u.)

Resistance (a.u.) Resistance (a.u.)

Resistance (a.u.) Resistance (a.u.)

0 2 4

6 8 10 12

Time (s)

0 0 5 10 15 20 25 30 35

2 4 6 8 10 12 14

Time (s)

0 2 4 6 8 10 12 14

Time (s)

Time (s)

d

e

f

Resistance (a.u.)

0 0 2 4 6 8 10

1 2 3 4 5

Time (s)

Time (s)

Figure 3 | Responses towards different audios from a loudspeaker. The LIG is placed 3 cm away from the loudspeaker. The orange insets above indicate the sound wave proles of the original audios. Relative resistance changes show almost synchronous response to proles of the original audios when the loudspeaker plays the audio of (a) recrackers, (b) a cow, (c) a piano, (d) a helicopter, (e) a bird and (f) a drum.

chosen to generate LIG because of the obvious resistance change compared to the PI with the thickness of 75 and 180 mm (Supplementary Fig. 8). The sensitivity is high enough to detect SP produced by a loudspeaker. The 25 mm PILIG is xed by two clamps free-standingly and the loudspeaker is placed 3 cm away from the articial throat. The audio tests with six kinds of audios including recracker, cow, piano, helicopter, bird and drum are performed. The results are shown in Fig. 3. Although the sampling frequency of our articial throat is 100 Hz, which is far lower than the frequency of sound, we can still notice that the responses of the transducer are well synchronous to these original audio signals (orange parts). Especially, the characteristic peaks are retained and reected with high delity. Besides, the volume of loudspeaker has a great effect on the amplitude of the signal. With the volume increases, the vibration will be strengthened, causing a more obvious change of resistance (Supplementary Fig. 8). The lm is clamped and bent under the force (Supplementary Fig. 9). The curvature radius is 1.27 cm and the stress is 0.07 N. The bending degree is higher than the vibration degree, which is caused when speaking. The testing results after 0 cycle, 1,000 cycles and 10,000 cycles are shown in Supplementary Fig. 9. We can see the device shows excellent durability under high strain.

After identifying some kinds of audio clearly, LIG throat is used to detect the vibration of throat cords. As shown in Fig. 4a, the tester makes two successive coughs, hums and screams, and then the tester does swallowing and nod actions in two times. The repeatability of the detection is excellent according to the two-time successive testing. Besides, the swallowing and the nod can cause the muscle movement, which can also result in the change of the resistance. Fortunately, the waveforms of these kinds of muscle movements also have recognizable characteristics. Different movement has its unique characteristic waveform as shown in Fig. 4a; thus, we can get the useful waveforms by relying on the pattern recognition and machine learning. The interference by some other activities can be recognized and eliminated by training many times in advance. Then, the tester makes the hums with four different tones as shown in Fig. 4b; we

can see that different tones also have different responses, increasing the variety of the language of mute persons. Especially, the hum tone 2 is same with the hum in Fig. 4a. Furthermore, as shown in Fig. 4c, the resistance increases as the sound intensity increases, resulting from the increase of the mechanical vibration of throat cords.

When the device is attached on the throat, it can detect both SP and throat vibration. An experiment is performed to compare whether the mechanical vibration of throat cords or the SP contributes most to the relative resistance change of the LIG articial throat. The LIG is attached on the throat of the tester and the tester makes some hums with different volumes, and these sounds are recorded and played by a loudspeaker. The LIG then is placed 3 cm away from the loudspeaker, to acquire the signal induced by pure SP. According to Supplementary Fig. 10, the relative resistance change of the device placed 3 cm away from the loudspeaker is only 0.5% when the loudspeaker plays a sound of 90 dB, which means the pure SPL of 90 dB will cause a 0.5% relative resistance change. However, the relative resistance change of the device attached on the throat can be 8.2% when a person makes a hum of 90 dB. The relative resistance change of the device attached on the throat is 16 times larger than that caused by pure SP, which means the vibration of throat cords contributes more than the SP.

Besides, the LIG articial throat also has the capability of voice recognition. As shown in Supplementary Fig. 11, some words including Graphene, Material and Industry are pronounced by different testers including an elder, a boy, a man and a woman, respectively. The wave curves of different words in the time domain show apparently different characteristics, which is helpful to distinguish different words. Besides, the wave curves of a same word pronounced by different persons show similar but different characteristics, which can be a key factor for identity authentication by voice recognition. Furthermore, a long sentence, Graphene is a carbon-based material with huge potential for industry, is spoken repeatedly by a woman for ve times. The articial throat shows excellent repeatability and reliability to

NATURE COMMUNICATIONS | 8:14579 | DOI: 10.1038/ncomms14579 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14579

a

Resistance (a.u.) Resistance (a.u.)

10 15 20 25 30

Time (s)

0 5

b c

R/R(%)

20

16

12

8

4

0

0 0

4 8 12 16 10 20 30 40

Time (s)

Time (s)

Figure 4 | Responses towards different kinds of throat vibrations. (a) The LIGs resistance changes towards the throat vibrations of the tester who makes two successive coughs, hums, screams, swallowing and nods. (b) The LIGs resistance change caused by four different kinds of hum tones and the hum tone 2 is same with the hum in a. (c) The relative resistance change of LIG increases with the increase of the sound intensities of the hum.

work as an acoustic detector. Especially, from the magnied image of the sentence, we can notice that the three words, graphene, material and industry, are almost identical to the individual pronunciation by the woman. Compared with some other nanomaterial-based acoustic detectors1722, our device demonstrates superior predominance in voice recognition because of its excellent repeatability and reliability.

Working as sound source and sound detector simultaneously. The LIG articial throat developed in this work has great potential to bring a revolution in the eld of acoustic. As we know, most mute persons are born deaf and they cannot speak words. However, their throat cords can vibrate and they are capable of making noises in their own ways, which are meaningless to the normal. The intelligent LIG articial throat is developed to transform the meaningless noise into controllable and understandable sound signals. The testers need to be trained rst. When they produced a specic cough, hum or scream, we told them the corresponding meaning by gesture language. They can be accustomed to the sound intensity by multi-time repeated training. The training process is similar with the process of importing the ngerprint into iPhone. Then, these waveforms can be recognized by pattern recognition and machine learning. Besides, we set some threshold ranges in advance. The software can recognize the sound intensity when it lies on the threshold range. The accuracy of recognition can be guaranteed by the above two ways. After the training process, the different volume, frequency and last time of hum, scream or cough can be converted into specic meanings. Here we simply demonstrated the conversion from three kinds of different hums to three controllable sounds. The working principle of LIG articial throat is shown in Fig. 5a (see the Methods section for details). It works as a sound detector at rst, then the tester makes three kinds of

hums to imitate a mute person and the resistance of the transducer will be changed by the throat cords vibration correspondingly. A microcontroller is used to detect the resistance change of the transducer and realize the hum judgement. After that, the transducer begins to work as a sound source and corresponding sound signal is generated. The LIG articial throat is attached on the throat of a tester as Fig. 5b shows.

As shown in Fig. 5c, a high-volume hum, a low-volume hum and an elongated hum are pronounced twice by the tester. Correspondingly, a high-volume 10 kHz, a low-volume 10 kHz and a low-volume 5 kHz sound signal are produced. The output sound frequency is twice the input electric signal; thus, a DC bias is added to generate the fundamental tone11. The high-volume 10 kHz, the low-volume 10 kHz and the low-volume 5 kHz sound are generated by 10 V DC 10 V 10 kHz AC, 5 V DC 5 V

10 kHz AC and 5 V DC 5 V 5 kHz AC, respectively (Supple

mentary Movie 1). From the magnied images (Fig. 5df), the amplitude and the frequency of the output sound signals are shown as we expected. Finally, an intelligent LIG articial throat is demonstrated to convert meaningless noises into useful sound signals with controllable frequencies and volumes. It will be possible for mute persons to express themselves with the assistance of LIG articial throat.

DiscussionIn summary, a one-step fabricated wearable articial throat based on LIG has been developed. The low-power laser with the wavelength of 450 nm can induce the conversion from PI to LIG. The LIG realize the functional integration of emitting and detecting sound on a single device because of its superior thermoacoustic and piezoresistive properties. As a sound source, the SPL of the LIG articial throat has been demonstrated from 100 Hz to 40 kHz. The thickness of LIG will have an obvious

6 NATURE COMMUNICATIONS | 8:14579 | DOI: 10.1038/ncomms14579 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14579 ARTICLE

a b c

2 High

1

Working as sound detector

Artificial throat

Intensity (a.u.)

0

1

5

High-volume hum

Elongated hum

2 0 10 15 20 25 30

Time (s)

Low-volume hum

d e f

Working as sound source

Working as sound source

Working as sound source

Intensity (a.u.)

0.2

0.1

0

0.1

0.2

2.545 2.546 13.455 13.456 22.455 22.456 Time (s) Time (s) Time (s)

Intensity (a.u.)

0.2

0.1

0

0.1

0.2

Intensity (a.u.)

0.2

0.1

0

0.1

0.2

10 kHz high volume

10 kHz low volume

5 kHz low volume

Figure 5 | An intelligent LIG articial throat. (a) The working procedure of the articial throat. (b) The tester wearing the LIG articial throat. Scale bar, 1 cm. (c) high-volume, low-volume and elongated tone hum are detected by LIG throat and converted into high-volume 10 kHz, low-volume 10 kHz and low-volume 5 kHz sound, respectively. (d) The magnied wave of high-volume 10 kHz sound. (e) The magnied wave of low-volume 10 kHz sound. (f) The magnied wave of low-volume 5 kHz sound.

inuence on SPL according to our theory and a thinner LIG will generate sound with higher SPL. The LIG articial throat has a relatively broad frequency spectrum because of resonance-free oscillations of the sound sources. Besides, as a sound detector, the LIG articial throat can capture the mechanical vibration of throat cords with a ne repetition. It can clearly differentiate the characteristics of cough, hum and scream with different tones and volumes according to their unique waveforms. Besides, it also has the capability of voice recognition because of its outstanding mechanical properties. The intelligent LIG articial throat will signicantly assist for disabled person. The LIG articial throat can generate volume and frequency controllable sound by detecting different kinds of imitative hum of the tester and realize the conversion from meaningless hum to controllable sound, which has signicantly practical potentials. Furthermore, the one-step fabricated wearable LIG articial throat will open widely practical applications in voice control, wearable electronics and many other areas due to its high sensitivity, excellent repetition, good exibility and simple fabrication process.

Methods

Fabrication of articial throat. A custom-designed platform equipped with a laser diode (OSRAM) with wavelength of 450 nm was used to convert PI into LIG. Figure 1a shows the schematic diagram of the mask-free process. The movement of laser was controlled by two stepper motors in XY direction. The beam size of laser and the minimum displacement of stepper motor are 100 mm. The scanning speed of laser is 8.5 mm s 1 and the maximum power is 500 mW. Commercial PI lms (Kapon) with thickness of 25, 75, 125 and 180 mm were used as the substrate without any further treatment. After the laser irradiation, the LIG articial throats are wired out by copper wire by using silver paste.

Characterization. The surface morphology of the LIG is observed by a Quanta FEG 450 scanning electron microscope (FEI Inc.). Raman spectroscopy is performed using a laser with a wavelength of 532 nm (HORIBA Inc.). Transmission electron microscopic images are taken by JEM2100F (JEOL Inc.).

Testing of articial throat. The sound emitting testing platform consisted of a standard microphone and a dynamic signal analyser. A 1/4 inch standard microphone (Earthworks M50) was chosen to measure the SPL of the LIG articial throat. The microphone has a very at frequency response within 40 kHz and a high sensitivity of 31 mV Pa 1. The dynamic signal analyser (Agilent 35670A) was

used to generate swept-frequency signals from 100 Hz to 40 kHz, to drive the LIG articial throat and to record the value of the SPL of it. The sound-detecting testing platform consisted of a commercial loudspeaker and a digital multimeter. LIG articial throat was placed 3 cm away from the loudspeaker. The digital multimeter (ROGOL DM3068) was used to the relationship between the resistance and the time.

Generating and detecting sound in a single device. The LIG articial throat was attached on the throat of a tester. The LIG articial throat worked in two modes: detecting mode and emitting mode. During the detecting mode, a microcontroller was used to detect the amplitude and last time of the meaningless hums. When the amplitude or last time exceeded the thresholds, the microcontroller would stop detecting and the digital function generator would be applied on the LIG articial throat for 3 s. After 3 s, the microcontroller closed the digital function generator by a relay and LIG articial throat works in detecting mode again. Different digital function generators would be applied on the LIG articial throat respectively according to the amplitude and last time of the meaningless noise.

Data availability. The data that support the ndings of this study are available from the corresponding author upon request.

References

1. Jain, M., Puri, S. & Unishree, S. Eyeball motion controlled wheelchair using IR sensors. World Acad. Sci. Eng. Technol. Int. J. Comput. Electr. Autom. Control Inf. Eng. 9, 906909 (2015).

2. Thakur, B. & Kulshrestha, K. in Proceedings of IEEE International Conference on Computational Intelligence and Computing Research (ICCIC) 16 (Coimbatore, India, 1820 December 2014).

3. Brown, L. F. & Carlson, D. L. Ultrasound transducer models for piezoelectric polymer lms. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 36, 313318 (1989).

4. Marciniak, R. D. Unidirectional underwater-sound pressure-gradient transducer. IEEE Trans. Sonics Ultrason. 18, 8994 (1971).

5. Meeks, S. W. & Timme, R. W. Rare earth iron magnetostrictive underwater sound transducer. J. Acoust. Soc. Am. 62, 11581164 (1977).

6. Tian, H. et al. Graphene-on-paper sound source devices. ACS Nano 5, 48784885 (2011).

7. Suk, J. W., Kirk, K., Hao, Y., Hall, N. A. & Ruoff, R. S. Thermoacoustic sound generation from monolayer graphene for transparent and exible sound sources. Adv. Mater. 24, 63426347 (2012).

8. Tian, H. et al. Single-layer graphene sound-emitting devices: experiments and modeling. Nanoscale 4, 22722277 (2012).

9. Tian, H. et al. Static behavior of a graphene-based sound-emitting device. Nanoscale 4, 33453349 (2012).

NATURE COMMUNICATIONS | 8:14579 | DOI: 10.1038/ncomms14579 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14579

10. Tian, H. et al. A exible, transparent and ultrathin single-layer graphene earphone. RSC Adv. 5, 1736617371 (2015).

11. Tian, H. et al. Graphene earphones: entertainment for both humans and animals. ACS Nano 8, 58835890 (2014).

12. Tian, H., Yang, Y., Xie, D., Ge, J. & Ren, T.-L. A reduced graphene oxide sound-emitting device: a new use for Joule heating. RSC Adv. 3, 1767217676 (2013).

13. Xiao, L. et al. Flexible, stretchable, transparent carbon nanotube thin lm loudspeakers. Nano Lett. 8, 45394545 (2008).

14. Mason, B. J., Chang, S., Chen, J., Cronin, S. B. & Bushmaker, A. W. Thermoacoustic transduction in individual suspended carbon nanotubes. ACS Nano 9, 53725376 (2015).

15. Tian, H. et al. Flexible, ultrathin, and transparent sound-emitting devices using silver nanowires lm. Appl. Phys. Lett. 99, 253507 (2011).

16. Tian, H. et al. Transparent, exible, ultrathin sound source devices using indium tin oxide lms. Appl. Phys. Lett. 99, 043503 (2011).

17. Wang, Y. et al. Wearable and highly sensitive graphene strain sensors for human motion monitoring. Adv. Funct. Mater. 24, 46664670 (2014).18. Tang, Y. et al. Highly stretchable and ultrasensitive strain sensor based on reduced graphene oxide microtubes-elastomer composite. ACS Appl. Mater. Interfaces 7, 2743227439 (2015).

19. Wang, Y. et al. Ultra-sensitive graphene strain sensor for sound signal acquisition and recognition. Nano Res. 8, 16271636 (2015).

20. Wang, C. et al. Carbonized silk fabric for ultrastretchable, highly sensitive, and wearable strain sensors. Adv. Mater. 28, 66406648 (2016).

21. Gong, S. et al. Highly stretchy black gold e-skin nanopatches as highly sensitive wearable biomedical sensors. Adv. Electron. Mater. 1, 1400063 (2015).

22. Park, J. J., Hyun, W. J., Mun, S. C., Park, Y. T. & Park, O. O. Highly stretchable and wearable graphene strain sensors with controllable sensitivity for human motion monitoring. ACS Appl. Mater. Interfaces 7, 63176324 (2015).

23. Cheng, Y., Wang, R., Sun, J. & Gao, L. A stretchable and highly sensitive graphene-based ber for sensing tensile strain, bending, and torsion. Adv. Mater. 27, 73657371 (2015).

24. Tian, H. et al. Scalable fabrication of high-performance and exible graphene strain sensors. Nanoscale 6, 699705 (2014).

25. El-Kady, M. F., Strong, V., Dubin, S. & Kaner, R. B. Laser scribing of high-performance and exible graphene-based electrochemical capacitors. Science 335, 13261330 (2012).

26. El-Kady, M. F. & Kaner, R. B. Direct laser writing of graphene electronics. ACS Nano 8, 87258729 (2014).

27. Tian, H. et al. Cost-effective, transfer-free, exible resistive random access memory using laser-scribed reduced graphene oxide patterning technology. Nano Lett. 14, 32143219 (2014).

28. Tian, H. et al. Wafer-scale integration of graphene-based electronic, optoelectronic and electroacoustic devices. Sci. Rep. 4, 3598 (2014).

29. Gao, W. et al. Direct laser writing of micro-supercapacitors on hydrated graphite oxide lms. Nat. Nanotechnol. 6, 496500 (2011).

30. El-Kady, M. F. & Kaner, R. B. Scalable fabrication of high-power graphene micro-supercapacitors for exible and on-chip energy storage. Nat. Commun. 4, 1475 (2013).

31. Lin, J. et al. Laser-induced porous graphene lms from commercial polymers. Nat. Commun. 5, 5714 (2014).

32. In, J. B. et al. Facile fabrication of exible all solid-state micro-supercapacitor by direct laser writing of porous carbon in polyimide. Carbon 83, 144151 (2015).

33. Luo, S., Hoang, P. T. & Liu, T. Direct laser writing for creating porous graphitic structures and their use for exible and highly sensitive sensor and sensor arrays. Carbon 96, 522531 (2016).

34. Rahimi, R., Ochoa, M., Yu, W. & Ziaie, B. Highly stretchable and sensitive unidirectional strain sensor via laser carbonization. ACS Appl. Mater. Interfaces 7, 44634470 (2015).

35. Desilets, C. S., Fraser, J. D. & Kino, G. S. The design of efcient broadband piezoelectric transducers. IEEE Trans. Sonics Ultrason. 25, 115125 (1978).

36. Kossoff, G. The effects of backing and matching on the performance of piezoelectric ceramic transducers. IEEE Trans. Sonics Ultrason. 13, 2030 (1966).

37. Manthey, W., Kroemer, N. & Magori, V. Ultrasonic transducers and transducer arrays for applications in air. Meas. Sci. Technol. 3, 249261 (1992).

38. Wynn, G. H. & Fountain, A. W. Development and characterization of electrochemical devices using ultraviolet laser induced carbonization of polyimide lms. J. Electrochem. Soc. 144, 37693772 (1997).

39. Brannon, J. H. & Lankard, J. R. Pulsed CO2 laser etching of polyimide. Appl. Phys. Lett. 48, 12261228 (1986).

40. Arnold, H. D. & Crandall, I. B. The thermophone as a precision source of sound. Phys. Rev. 10, 2238 (1917).

41. Xie, Q. Y. et al. A point acoustic device based on aluminum nanowires. Nanoscale 8, 55165525 (2016).

42. Asadzadeh, S. S., Moosavi, A., Huynh, C. & Saleki, O. Thermo acoustic study of carbon nanotubes in near and far eld: theory, simulation, and experiment.J. Appl. Phys. 117, 095101 (2015).

Acknowledgements

This work was supported by National Natural Science Foundation (61574083 and 61434001), National Basic Research Program (2015CB352101), National Key Research and Development Program (2016YFA0200400), National Key Project of Science and Technology (2011ZX02403-002) and Special Fund for Agroscientic Research in the Public Interest (201303107) of China. We are thankful for the support of the Independent Research Program (2014Z01006) of Tsinghua University, and Advanced Sensor and Integrated System Lab of Tsinghua University Graduate School at Shenzhen under project number ZDSYS20140509172959969.

Author contributions

L.-Q.T. and H.T. discovered the procedures, designed most of the data and wrote the manuscript with equal contribution. Y.L. performed scanning electron microscopy and Raman characterizations. Z.-Y.J. conducted the modelling and theory. Y.P. performed transmission electron microscope characterizations. Y.-Q.C. and D.-Y.W. conducted some gures. X.-G.T., J.-C.Y. and N.-Q.D. provided valuable discussion. Finally, Y.Y. andT.-L.R. provided guidance to the research.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Tao, L.-Q. et al. An intelligent articial throat with sound-sensing ability based on laser induced graphene. Nat. Commun. 8, 14579 doi: 10.1038/ncomms14579 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2017

8 NATURE COMMUNICATIONS | 8:14579 | DOI: 10.1038/ncomms14579 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications