Article pubs.acs.org/JPCC
Laser Ablation of Sub-10 nm Silver Nanoparticles Alexander Zinovev,*,† Jerome F. Moore,‡ Sergey V. Baryshev,§ J. Albert Schultz,∥ Ernest Lewis,∥,⊥ Bruce Brinson,⊥ Michael McCully,∥ and Michael Pellin† †
Materials Science Division, Argonne National Laboratory, 9700 Cass Avenue, Lemont, Illinois 60439, United States Robot Nose Corporation, 15700 103rd Street Suite 220, Lemont, Illinois 60439-9610, United States § Euclid TechLabs, 365 Remington Boulevard, Bolingbrook, Illinois 60440, United States ∥ Ionwerks Inc., 1200 Binz, Suite 1230, Houston, Texas 77004, United States ⊥ Chemistry Department, Rice University, 6100 Main Street, Houston, Texas 77005, United States ‡
ABSTRACT: Laser ablation of silver nanoparticles (NPs) was studied with laser post-ionization (LPI) time-of-flight mass spectrometry (TOF MS). Silver NPs containing ∼15 000 Ag atoms (4 nm radius) were deposited by soft landing (energy 3 eV/atom) onto indium tin oxide (ITO)/glass substrates. Laser ablation was performed using frequency-doubled Ti:sapphire nanosecond pulsed laser irradiation at three different wavelengths (371, 401, and 421 nm), whereas for post-ionization, pulses from an F2 laser were used. Laser fluences and time delay dependencies of Ag and In signals were obtained. Using these data, the temperature of the desorption source as well as its time duration were calculated. It was found that the peak temperature of NPs was above their melting point and they cooled down slowly, with temperature decay time of several hundreds of nanoseconds. This anomalous behavior was explained based on a model where the semiconducting ITO substrate is initially transparent to the desorption laser radiation but starts to adsorb it due to the temperature increase arising from heat exchange with NPs. Poor heat conduction in the ITO film creates conditions for long-lived hot spots on the surface and initiates further optical damage of the substrate. No difference in the ablation process due to plasmon resonance was detected, likely due to thermal expansion and melting of NPs during laser irradiation, which then broadens the plasmon absorption band enough to cover all wavelengths used. These results clearly demonstrate that the process of NP interaction with laser radiation is governed not only by initial optical and thermophysical parameters of NPs and the surrounding media, but also by their alteration due to temperature increases during the irradiation process.
1. INTRODUCTION The history of the study of metal clusters is decades long1 and remains a major topic of interest. As the building blocks of numerous modern nanomaterials, metal clusters and nanoparticles (NPs) also have unique physical and chemical properties that sometimes make them dramatically more valuable than bulk metals.2 This has motivated development of methods for studies of these species by a variety of techniques.1 Mass spectrometry (MS) has the advantage of being sensitive both elementally and isotopically. On the other hand, the method is inherently destructive, and for that reason is not as widely used for such studies. As a result, there are very few studies related to laser desorption (LD) or ablation of NPs from surfaces studied using MS methods.3−5 Nevertheless, the growing interest in NP ablation by laser irradiation is connected to the development of new methods in MS of bio- and organic materials such as tissues. The search for alternatives to matrix assisted laser desorption/ionization (MALDI)6 resulted in the development of a similar technique, surface assisted laser desorption/ionization (SALDI),7,8 and © 2017 American Chemical Society
more recently matrix implanted laser desorption ionization (MILDI).9,10 Both of these more recent techniques are based on substrate surface modification, particularly by adding nanoparticles to the analyte or having them on the surface already before the analyte is added. Because metal NPs have much stronger light adsorption than surrounding media, subsequent irradiation of the target by pulsed laser can heat the NPs to high temperatures and then transfer adsorbed energy to surrounding molecules, causing their desorption and ionization. This is a simplified model of SALDI and MILDI processes and there are many important yet poorly understood mechanistic details that have limited the adoption of these methods (in contrast with MALDI) for routine, regular procedures for the study of biological materials. The main thing that makes NPs so useful for SALDI and MILDI is their unique optical and thermal properties in Received: February 2, 2017 Revised: March 23, 2017 Published: April 13, 2017 9552
DOI: 10.1021/acs.jpcc.7b01061 J. Phys. Chem. C 2017, 121, 9552−9559
Article
The Journal of Physical Chemistry C
Figure 1. Schematics of experimental setup. 1, target; 2, reflectron front optics; 3, Schwarzschild microscope objective; 4, reflectron; 5, MCP detector; 6, TV camera; 7, beamsplitters; 8, steering mirrors; 9, Ti:sapphire lasers; 10, YLF pump laser. (b,c) Timing diagram of instrument pulsing. DLP, desorbing laser pulse; ILP, ionizing laser pulse; EP, TOF extraction pulse.
these analytical methods, understanding how NPs undergo heating and desorption is relevant to other applications of nanotechnology, such as the use of NPs under laser irradiation for medical treatment,21 processing of NPs in order to form exotic nanostructures, and in surface engineering.22 The primary goal of this work is to study the dynamics and parameters of NP heating and cooling during pulsed laser irradiation in the situation when the substrate is nominally transparent and does not absorb radiation. Therefore, an indium−tin oxide (ITO) film on glass system was chosen as templates for Ag NPs, since silver has become established for SALDI/MILDI applications due to its low reactivity and its propensity as a cluster or NP to strongly absorb in the blueviolet region of the spectrum.
comparison with bulk materials of the same composition. Plasmon resonances in the optical spectra of metal NPs increase their optical absorption allowing for a reduction of the desorbing laser fluence and consequent reduction in the optical and thermal fragmentation of subsequently desorbed organic molecules. To validate this concept, the processes of NP heating by laser radiation and consequent energy transfer to surrounding media should be studied in detail. The problem of heating and heat transfer of different kinds of NPs was studied extensively over the last 15 years.11−17 However, most of these studies were performed using some model assumptions that sometimes stray far from the experimental conditions specific for SALDI or MILDI. The most common approach uses spherical NPs embedded into some uniform media that is subjected to a laser pulse.11−13,15−18 Compared with SALDI and MILDI techniques,10,19 it is easy to understand that these approaches are far from real experimental conditions. In SALDI, NPs are deposited on the substrate surface with the use of some common techniques (for example, metal magnetron sputtering19,20) and analyte solution is then deposited onto it.19 In this case, NPs are not spherical and are in contact with two different media, the substrate surface and the analyte. MILDI is similar, but includes the deposition of biological samples such as tissues onto the substrate surface and subsequent softlanding of charged NPs atop of them.9,10 The shape and NPsubstrate contact area are also far from ideal conditions. In the context of these approaches, many important physical− chemical parameters of the combined NP−substrate system remain unknown, which makes theoretical modeling difficult to apply to specific experiments. Presented in this paper are experimental studies of NP laser ablation in a specific environment, conducted to provide an understanding of NP−substrate interaction. An understanding of these processes will facilitate the further development of SALDI and MILDI methods. In addition to applicability of
2. EXPERIMENTAL SETUP The experiments were conducted with the use of a custom TOF mass spectrometer SARISA23 that was slightly modified for this experiment (Figure 1). The illumination of the front side of the sample by desorption laser was arranged through an in-vacuum Schwarzschild microscope objective, and laser ablated material was post-ionized by a pulse from an F2 (GAM100-EX) laser. The instrument was operated at repetition rate 200 Hz, while the F2 laser was used at 100 Hz only. The ionized part of the laser desorbed plume was captured by the electric field at the input of time-of-flight MS (TOFMS) ion optics, and mass spectra were detected for each desorption laser shot in interleaved fashion. Therefore, the acquired mass spectra consisted of two groups: purely laser desorbed, and laser desorbed then post-ionized spectra. In this fashion, the ionization efficiency in the LD process, that is the relative ratio of ionized and neutral desorbed particles, could be estimated for each measurement cycle, minimizing sampling error. 9553
DOI: 10.1021/acs.jpcc.7b01061 J. Phys. Chem. C 2017, 121, 9552−9559
Article
The Journal of Physical Chemistry C
Figure 2. (a) SEM image of Ag NPs deposited on ITO substrate; (b) absorption spectrum of Ag NPs. Arrows are marking desorption laser wavelengths. (c) TEM image of Ag NPs obtained at the same conditions as the NP size distribution (d).
The produced photoions were mass separated by TOF MS, detected by fast ion detector (Gen2, Photonis) and the resulting signals were acquired by two digitizers (FastFlight-2, Ortec) that generated two data streams: one for post-ionized signal and the other for laser desorption only. One measurement cycle consists of 512 laser pulses (256 post-ionized and 256 pure LD signals) that were summed by the digitizers and transferred to the main computer for further processing.
For laser desorption, three different tunable home-built Ti:sapphire lasers, pumped by a single commercial diodepumped YLF laser (DM60, Photonics Inc.), were coaxially aligned to the time-of-flight axis of the instrument (Figure 1). The output radiation frequency of each of these lasers was doubled by LBO (lithium triborate) crystals and the resulting desorption wavelengths were 371, 401, and 421 nm, respectively, with laser pulse durations τL = 15 ns at full width at half-maximum (fwhm). The energy of each laser were tuned so that at the instrument input window its maximum value was ∼8 μJ/pulse. The Schwarzschild microscope objective focused this radiation onto the target surface in a spot of elliptical shape with the axes sizes ∼(10 × 8) μm2 at 1/ e. Because of the slightly different divergence of these three lasers, the exact shape of each profile was measured separately and was used to calculate the fluence density of irradiation. The delay between the desorbing laser pulse and the start of the extraction pulse of the TOF instrument could be continuously varied within the interval 0−30 μs. To improve signal statistics and minimize sample damage during each measurement, the laser spot was rastered on the sample surface over an area ∼70 × 70 μm2 with the use of 2D-galvoscanner (OFH-15, Nutfield Tech.). For the power study, the energy of desorption laser radiation was varied with the use of the set of calibrated neutral density (ND) optical filters with increment of 0.1 optical density (OD) and 5% RMS error. The post-ionizing laser pulse had ∼2 mJ/pulse energy and was focused above the target with MgF2 lens with focal distance 650 mm. The beam waist has a rectangular shape with the size 3 × 10 mm2, with the longer axis parallel to the sample surface. The delay between this pulse and TOF extraction pulse was optimized to 100 ns by maximizing the signal.
3. SAMPLE PREPARATION AND CHARACTERIZATION Our NPs samples were prepared with the use of a soft landing technique comparable to the one used in MILDI processes. The substrate was a commercially prepared glass slide covered by 60-nm-thickness indium tin oxide (ITO) (Delta Technologies). ITO is highly transparent in the near-UV and visible area of spectrum and also has high damage threshold for laser irradiation. The conductive ITO substrate also allowed us to obtain images of NPs using an SEM, perform NPs soft landing, and also ensured field uniformity for ion extraction into the TOFMS. The sample was prepared by soft landing predefined cluster ions with mass of 1.6 MDa (∼1.5 × 104 Ag atoms) on the ITO covered target by NPlanter built and operated by Ionwerks and located at Rice University. A detailed description of the NPlanter is available elsewhere.24 In brief, a magnetron sputter source with aggregation zone and quadrupole filter (Mantis Deposition, UK) was extensively customized and outfitted with rastering and acceleration/deceleration lenses, current monitoring, and a sample introduction and manipulation chamber. For the present work, Ag NPs were formed from a 2-in.-diameter 99.99% Ag target in the source, and Ag NPs of 8 nm diameter 9554
DOI: 10.1021/acs.jpcc.7b01061 J. Phys. Chem. C 2017, 121, 9552−9559
Article
The Journal of Physical Chemistry C
Figure 3. Typical power dependencies of Ag and In signals obtained at 401 nm laser irradiation. Dashed lines in (a) denote the areas of In sublimation (1), ITO melting (2), and evaporation of In from the liquid ITO phase (3).
equation, as a thermally driven kinetic process is a reasonable assumption in the nanosecond time domain, and therefore, the flux of ablated silver atoms is described as
were preferentially delivered at 500 eV to the target, with a raster area of 21 × 21 mm2. Again, silver is a popular choice for analysis because of its low reactivity and high absorption cross section. An SEM image of the sample is shown in Figure 2a, which confirms the uniformity of coverage and absence of NPs aggregation over an extended area of 4 × 4 μm2, i.e., over the area that is on the order of the size of the desorption laser beam spot. The resolution of Figure 2a is not enough to measure the size of all NPs, but presumably because they were mass separated before soft landing to the mass 1.6 MDa, it is easy to estimate that the diameter should be around 8 nm. A TEM micrograph (JEOL 2010F) is also shown that precisely indicates the size and distribution of the NPs. Larger field energy dispersive spectrometry (not shown) also verified that a uniform coverage of NPs was obtained over hundreds of μm2 area. Samples were characterized by their optical transmission spectra with the use of a spectrophotometer (Shimadzu UV2600) and results are presented in Figure 2b. It has a relatively narrow resonance that agrees with the narrow NP size distribution.
IAg = C·exp( −UAg /kBT )
and the temperature T is given by T=μ
3Q abs 4cρR
F
(1) 14,19
(2)
Here, c and ρ are specific heat and mass density of Ag, F is laser fluence, C is the constant, kB is Boltzmann constant and Qabs is the light absorption efficiency, UAg is the silver atom enthalpy of vaporization equal to 2.64 eV/atom,26 and μ is an adjustable parameter that was used to take into account heat exchange between NPs and substrate. However, eq 2 has to be used with caution, because the heat exchange parameter μ obviously depends on temperature, which consequently makes eq 2 nonlinear. Nevertheless, Figure 3a demonstrates power-law growth of the signal in a narrow range of laser fluences (0.1−0.2 J/cm2), where the Arrhenius equation seems to be valid. Figure 3b demonstrates the same dependencies replotted in (1/T) coordinates where T is the Ag NP temperature that was found by fitting the exponentially rising part of the Ag signal with the use of eqs 1 and 2 and coefficient μ as fitting parameter. We should note that despite the high dynamic range of our detection system (∼4 orders of magnitude), the range of laser fluencies where the Arrhenius law is applicable to our system seems to be very narrow. This is a consequence of the high background noise of Ag signals that limited the detection of Ag at low fluences, while at high fluences the signal was saturated due to silver depletion because of material evaporation during the ablation process and, to some degree, due to the nonlinearity of eq 2. The support for this hypothesis comes from the fact that the saturation of the Ag signal coincides with the plateau of the In signal where melting of ITO occurs and heat exchange parameters may change strongly. Ag signal with reasonable signal-to-noise ratio was observed at temperatures >1350 K, which explains the absence of melting plateau in comparison with the In signal. It is important to note that the temperature scale on Figure 3b is related to Ag NPs only and cannot be applied to the ITO substrate, which likely has a different temperature and is heated by a different
4. RESULTS AND DISCUSSION Laser power dependencies of ions and neutral ablation were measured at three different desorption laser wavelengths (371, 401, 421 nm). Typical results obtained at 401 nm desorption laser wavelength are shown in Figure 3a. The signals of Ag and In ions are plotted against laser fluence, which was changed with the use of neutral density (ND) filters. Both signals demonstrated strong growth starting at very similar laser fluencies 0.105 J/cm2 and 0.11 J/cm2 for Ag and In atoms, respectively (see Figure 3a). However, there are distinct differences between them. For Ag atoms, power growth is followed by the signal saturation that is connected with depletion of Ag species on the surface during ablation. For In atoms, the signal rise is followed by a plateau that eventually reverts back to a strong increase. Since In arises from the substrate and Ag from the NPs, we posit that this is connected with the onset of ITO substrate melting and subsequent surface temperature stabilization25 followed by the further temperature increase of the melted bath. It can be plausibly assumed that the evaporation of the atoms is described by the Arrhenius 9555
DOI: 10.1021/acs.jpcc.7b01061 J. Phys. Chem. C 2017, 121, 9552−9559
Article
The Journal of Physical Chemistry C
Figure 4. Typical delay dependencies of Ag post-ionized atoms signals for Ag metal and Ag NPs targets (a) and In post-ionized atoms (b).
field of view of mass-spectrometer electron optics. Finally, I(t) has the following form (see, for example, ref 31, page 68):
mechanism than NPs. Whereas NPs are directly heated by laser radiation, ITO is transparent for these wavelengths and so should not absorb radiation. Instead, it is heated due to heat flux from NPs and its temperature is defined by thermal conductivity at the interface and thermal parameters of ITO. In accordance with this analysis, NP temperatures are higher than 1000 K, and it could be assumed that those adjacent layers of the ITO substrate are also heated strongly, and the observed In signal supports this argument. The procedure described earlier cannot be used to define the ITO temperature because the enthalpy of vaporization of In from ITO is unknown. Therefore, we propose the model that is purely qualitative. The initial temperature increase of ITO is governed by heat exchange with hot NPs. However, ITO is a wide band gap semiconductor with a light absorption edge near 300−350 nm that depends on ITO stoichiometry and production parameters.27 Therefore, at elevated temperatures there should be a strong red shift of the absorption edge as is typical for most semiconductors28,29 and which results in subsequent heating of the ITO substrate by direct laser radiation adsorption rather than heat flux from already hot NPs. This process obviously has positive feedback such that above some laser fluence threshold a thermally induced optical bistability occurs.30 In this case, an induced red shift of the band gap can be expected to dominate over other heating channels of ITO. This proposed mechanism is suggested by the fact that strong visible surface damage was observed starting from laser fluencies of ∼0.8 J·cm−2, whereas in control experiments blank ITO specimens did not demonstrate any visible damage, nor was there any detectable In signal in the mass spectrum in the range of laser fluences applied. To go deeper into this problem, a different set of experiments were conducted. Ablation mass spectra were measured as a function of variable delay time τ between the desorbing laser pulse and the instrument extraction pulse in LD post ionization mode (as shown on Figure 1). In this case, the signal amplitude is defined by the number of neutral atoms that will be in the ionization volume at time equal τ. If a Maxwell distribution with temperature T(t) is used to describe desorbing atoms, the signal delay dependence I(t) can be obtained as a convolution of desorbing pulse profile P(t) with velocity distribution integrated over the ionization/extraction volume of the instrument. This volume can be approximated by a disk with thickness S and radius r0, where r0 is defined by the
⎛ M ⎞1/2 I(t ) = C1⎜ ⎟ × ⎝ 2πkBT ⎠ S + dS
∫dS ∫0
t
⎛ ⎞ z z 2M ⎟ exp⎜ − 2 2 (t − σ ) ⎝ 2(t − σ ) kBT (σ ) ⎠
⎡ ⎛ ⎞⎤ ⎛ UAg ⎞ Mr0 2 ⎟⎥ exp⎜ − × ⎢1 − exp⎜ − ⎟ dσ dz 2 ⎢⎣ ⎝ kBT (σ ) ⎠ ⎝ 2πσ kBT (σ ) ⎠⎥⎦ (3)
Here, M is the mass of desorbing species, dS is the distance between the target surface and ionization volume (see Figure 1), σ and z are variables of integration, and C1 is the constant that is proportional to the signal collection efficiency and density of ejected atoms.31 The last exponential term in eq 3 represents the desorption pulse profile P(t) that is defined by a time-dependent Arrhenius equation P(t) = exp[−UAg/kBT(t)]. To derive eq 3 it was assumed that all particles of the same kind have equal ionization probability inside the ionization volume. The only unknown function in eq 3 is the variation of the temperature over time T(t). Because the rising part of the temperature pulse could not exceed the laser pulse duration τL, which is much smaller than the time-of-flight of the atoms through ionization volume, only the trailing edge of T(t) (i.e., the cooling) is of interest. Therefore, the temperature evolution may be presented in the form T(t) = Tmax × f(t), where f(t) is a decreasing function of time. Finding the parameters of this function was the goal of this particular experiment. To validate eq 3, the signal delay dependence was measured in the same experimental conditions with the use of a pure Ag metal target (NIST standard). The problem of metal surface heating by laser pulse has a well-known solution32 and therefore the desorption pulse shape could be easily found simply by its combination with the Arrhenius equation.31 With the use of appropriate relations obtained by Bechtel32 and silver physical parameters26 we calculated the maximum surface temperature of the Ag target to be ∼1450 K, assuming a laser fluence of 10 J· cm−2 and Ag reflectivity of about 90% in the wavelength range 370−420 nm.33 This temperature was used as an initial value for the fitting procedure. Note that for a pure Ag metal target the high heat conduction of silver causes the function P(t) to decay much faster as compared to the time-of-flight of the atoms through the ionization volume. Therefore, P(t) could be 9556
DOI: 10.1021/acs.jpcc.7b01061 J. Phys. Chem. C 2017, 121, 9552−9559
Article
The Journal of Physical Chemistry C approximated by a δ-function. This makes the integration of eq 3 trivial and Tmax becomes the only fitting parameter. In Figure 4a the results of Ag ion delay measurements together with simulation curve calculated through the integration of eq 3 are demonstrated. The simulation curve was fitted to experimental data with temperature Tmax as a variable parameter. The best fitting was achieved at Tmax = 1475 ± 50 K, in good agreement with the maximal temperature estimated earlier. Figure 4a shows typical delay dependence for Ag atoms desorbed from Ag NPs and post-ionized by F2 laser radiation. The desorption laser fluence corresponds to a point on the increasing part of the Ag signal dependence in Figure 3a equal to 0.18 J cm−2, corresponding to the temperature (obtained from Figure 3b) of 1710 K. It is clear that time distribution of the Ag signal in the case of Ag NPs is much wider than those obtained for the metal target. However, the fitting of this dependence with the use of eq 3 becomes more challenging because the function P(t) could not be calculated easily. To foster discussion of this problem some simple estimates are required. Thermal diffusivity of the material is defined as a = K/cρ where K, c, and ρ represent thermal conductivity, specific heat, and mass density, respectively. With the use of tabulated parameters for silver26 one can calculate α = 0.17 × 10−3 m2/s. Therefore, for NPs of sub-10 nm size, the characteristic time of thermal equilibration will be ∼2 × 10−12 s, and therefore on the nanosecond time scale, temperature may be expected to be the same over the entire NP volume, as is commonly assumed.11 Thus, the temperature evolution of the NP is described by the equation Vcρ
dT = [Φin(t ) − Φout(t )] dt
temperature, absorption, and heated volume of ITO increase, and subsequently decrease Φout. Note that the heated ITO volume may be much larger than that of NPs because of heat propagation within ITO film and this positive feedback. As these factors affect the temperature behavior of NPs, the collective effects11 and possible change of interface conductance at the phase transition should be mentioned. Because of the many uncertainties involved, it is difficult to choose the dominant mechanism in the NP−substrate heat exchange process. However, the fact that In signal, which corresponds to ITO film desorption, is detected practically at the same laser fluencies as Ag signal suggests a very high ITO temperature that cannot be explained by heat exchange with NP only. Nevertheless, to make numerical simulations in accordance with eq 3 we need to know the temperature pulse shape f(t) and the stretched exponent f(t) = B exp[−(t/δ)β], where δ and β are fitting parameters, is a convenient approximation.13 The postulated function for f(t) has some physical basis explained as follows. In accordance with eq 4 after cessation of the laser pulse Φin(t) = 0 and only the last term on the right side of eq 4 remains, and therefore, the solution can be sought in exponential form. However, as noted by Plech et al.15 when heat conduction in ambient medium is slower than across the interface, the nonexponential decay of NP temperature may take place, which can then be approximated by a stretched exponent. The results of silver signal fitting are demonstrated in Figure 4a (dashed curve). The fitting procedure consisted of multiple iterations by varying three parameters (Tmax, δ, β) in order to minimize the least-squares residuals (LSR). The best fitting was achieved at Tmax = 1750 ± 150 K, δ = 600 ± 150 ns, and β = 0.5 ± 0.025 where the variation intervals were estimated when LSR increases by 5% from the minimum. As mentioned before, we could not estimate the ITO temperature by only using the Arrhenius equation. However, because desorbed In atom signals were detected, some data may be extracted from these delay experiments. With the use of the previously described procedure (In signal as well as fitting curve are shown in Figure 4b), optimal fitting parameters were found and their values are Tmax = 1850 ± 175 K, δ = 850 ± 150 ns, and β = 1 ± 0.1. These values are not dramatically different from those obtained for Ag desorption, and any discrepancies are likely due to experimental uncertainties except for the β value. Remarkably, significant variation of parameter β reflects the heat exchange dynamics in different media adjacent to the film/layer being heated.15 In our case, the difference between Ag and In signals is obvious: whereas temperature evolution of the Ag signal is defined by heat exchange between NPs and ITO, the In signal is governed by heat exchange between ITO and underlying glass substrate. In the previous experimental time-resolved studies of NP heating by laser pulses13,15 it was found that the characteristic time scales of temperature profiles range from pico- to subnanoseconds. This is strongly in contrast with our findings that the NP temperature time domain is orders of magnitude longer, in the range of hundreds of nanoseconds. This dramatic discrepancy is due to the rather different experimental conditions. In the case of the works cited above, NPs were completely embedded into media (liquid water), whereas in our case NPs were in contact with a solid substrate that could also be directly heated by laser radiation at elevated temperatures. As mentioned earlier, NP cooling rates are defined by the
(4)
where V is the total volume of NP. Φin(t) = 3VQabsI(t)/4R is the heat generated in NP by laser fluence F(t)14 and Φout is the outgoing heat flux due to the heat exchange with the substrate that depends on thermal parameters of ITO and glass and their thermal conductance at the interface. To get a solution of eq 4 the total heat exchange problem for the triple system consisting of NP, ITO layer, and glass substrate should be solved. Due to the complexity of the system, an analytical solution is not possible and numerical methods must be used. However, there are some unknown physical factors that strongly affect the system and make precise modeling very difficult. During the ablation process, the maximal temperature variation is far above the melting point of the NP and possibly the underlying substrates. Therefore, some of the optical and thermal constants of materials may be affected by this temperature increase and the real temperature dependencies of them are not well-known. The change of NP optical absorption due to the temperature increase is a known effect34 and could be included in the model. Another factor that strongly influences the heat exchange between NP and substrate is the temperature gradient that defines Φout. Φout ≅ −KITOA
dT dr
(5)
Here, KITO is ITO thermal conductivity and A is the contact area. At room temperature ITO is nearly transparent from nearUV to visible wavelengths, but because of the previously mentioned red-shift of the light absorption edge at elevated temperatures, the ITO substrate starts to absorb laser energy directly and is thus heated. It should be noted that this process is driven by positive feedback with increasing laser fluence: the 9557
DOI: 10.1021/acs.jpcc.7b01061 J. Phys. Chem. C 2017, 121, 9552−9559
Article
The Journal of Physical Chemistry C
exchange may cause unwanted changes in optical parameters and direct radiation absorption by an analyte, followed by its overheating and the consequent thermal destruction of biomolecules of interest. These results are also relevant to the growing use of NPs in medical applications and for nanoengineered materials, as these may also suffer ill effects at high peak temperatures during irradiation. Therefore, the kinetics of heat exchange between NPs and substrates is a problem of significance and should continue to be studied in detail.
temperature difference between the NP and substrate, which may be very small or even negative at some laser fluences. The overall picture of NP laser ablation on ITO substrates may be described as follows: when laser irradiation begins, it starts to heat Ag NPs, and by heat exchange subsequently heats the ITO film. The heated ITO starts to absorb laser energy and its temperature strongly increases because of the positive feedback mechanism mentioned previously. Within the 15 ns laser pulse, the heated ITO spot size could be estimated as l = τL·αITO ≅ 180 nm (where αITO and αglass are the calculated thermal diffusivity of ITO and glass, respectively26,27) that has heat sunk to the underlying glass substrate. After the cessation of the laser pulse, the temperature of this ITO hot spot decreases on a time scale t ∼ l2/αglass ≅ 400 ns, which is in reasonable agreement with the values obtained in our experiment. Another interesting finding of our study is that there was no difference in ion threshold formation for different laser wavelengths. In accordance with Figure 1, the absorbance of laser radiation at laser wavelengths 401 and 421 nm is approximately 2 times higher than at 371 nm. Therefore, one should expect the formation of Ag ion signal at lower intensities for 401 and 421 nm in comparison with 371 nm. In contrast with these expectations, no differences in ion threshold formation were detected, and based on our power dependence measurement (Figure 3a), this threshold was found to be 0.1 ± 0.015 J·cm−2 for all three wavelengths used. The experimentally derived NP temperature that corresponds to ion detection threshold is more than 1200 K, which exceeds the melting point of NPs.19 Therefore, before any desorption signal could be detected, NPs are heated to high temperatures and melted. Whereas the optical properties of Ag NPs have relatively small temperature dependence at temperatures below 1000 K,34 the phase transition (melting) may result in a dramatic change in optical absorption.1 If the melting point of NPs during laser irradiation is reached before the laser pulse intensity peaks, the maximum plasmon band may broaden (or may even disappear1) and absorption coefficients for all laser wavelengths become similar. This explains the lack of variation at different wavelengths in our ablation experiments.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Alexander Zinovev: 0000-0003-0072-8483 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The work was supported by the U.S. Department of Energy, Office of Science, Materials Sciences and Engineering Division. SEM images were obtained with the use of the Center for Nanoscale Materials, an Office of Science user facility, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC0206CH11357. We would like to thank the Rice University Shared Equipment Authority for the use of their instrumentation suite.
■
REFERENCES
(1) Kreibig, U.; Vollmer, M. Optical properties of metal clusters; Springer: Berlin, New York, 1995. (2) Eustis, S.; El-Sayed, M. A. Why gold nanoparticles are more precious than pretty gold: Noble metal surface plasmon resonance and its enhancement of the radiative and nonradiative properties of nanocrystals of different shapes. Chem. Soc. Rev. 2006, 35, 209−217. (3) Hoheisel, W.; Schulte, U.; Vollmer, M.; Trager, F. Metal particles on surfaces desorption, optical-spectra, and laser-induced size manipulation. Appl. Phys. A: Solids Surf. 1990, 51, 271−280. (4) Hoheisel, W.; Vollmer, M.; Trager, F. Laser-stimulated desorption of potassium atoms. Appl. Phys. A: Solids Surf. 1991, 52, 445−447. (5) Hoheisel, W.; Vollmer, M.; Trager, F. Desorption of metal atoms with laser light - mechanistic studies. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 48, 17463−17476. (6) Trim, P. J.; Snel, M. F. Small molecule MALDI MS imaging: Current technologies and future challenges. Methods 2016, 104, 127− 141. (7) Silina, Y. E.; Volmer, D. A. Nanostructured solid substrates for efficient laser desorption/ionization mass spectrometry (LDI-MS) of low molecular weight compounds. Analyst 2013, 138, 7053−7065. (8) Stolee, J. A.; Walker, B. N.; Zorba, V.; Russo, R. E.; Vertes, A. Laser-nanostructure interactions for ion production. Phys. Chem. Chem. Phys. 2012, 14, 8453−8471. (9) Novikov, A.; Caroff, M.; Della-Negra, S.; Lebeyec, Y.; Pautrat, M.; Schultz, J. A.; Tempez, A.; Wang, H. Y. J.; Jackson, S. N.; Woods, A. S. Matrix-implanted laser desorption/ionization mass spectrometry. Anal. Chem. 2004, 76, 7288−7293. (10) Tempez, A.; Ugarov, M.; Egan, T.; Schultz, J. A.; Novikov, A.; Della-Negra, S.; Lebeyec, Y.; Pautrat, M.; Caroff, M.; Smentkowski, V. S.; Wang, H. Y. J.; Jackson, S. N.; Woods, A. S. Matrix implanted laser desorption ionization (MILDI) combined with ion mobility-mass spectrometry for bio-surface analysis. J. Proteome Res. 2005, 4, 540− 545.
5. CONCLUSION The laser ablation of small Ag nanoparticles deposited by soft landing on ITO substrates was studied with laser postionization. Based on power and delay dependencies of Ag ion signals, the kinetics of Ag NP heating and cooling was studied. It was found that the desorption signal (and hence the underlying temperature pulse) of laser heated NPs lasts much longer than would be expected from previously published experiments. This observation is explained by the interaction of NPs with the ITO substrate. Being optically transparent at initial experimental conditions, the ITO film initially is heated by heat exchange with NPs, followed by increase of their optical absorption and direct heating by laser radiation. Because of poor ITO and glass substrate thermal conductivity and specific geometry factor, these hot spots cooled down much more slowly than in the case of previous experiments which used other media and geometry. Heat exchange between NPs and substrates may play a significant role in the use of NPs in MS analysis of samples deposited on SALDI chips as well as MILDI from tissue samples. The heating of surrounding media through heat 9558
DOI: 10.1021/acs.jpcc.7b01061 J. Phys. Chem. C 2017, 121, 9552−9559
Article
The Journal of Physical Chemistry C
(31) Lubman, D. M. Lasers and mass spectrometry; Oxford University Press: New York, 1990. (32) Bechtel, J. H. Heating of solid targets with laser pulses. J. Appl. Phys. 1975, 46, 1585−1593. (33) Koechner, W. Solid-state laser engineering; 5th rev. and updated ed.; Springer: Berlin; New York, 1999. (34) Hodak, J. H.; Martini, I.; Hartland, G. V. Spectroscopy and dynamics of nanometer-sized noble metal particles. J. Phys. Chem. B 1998, 102, 6958−6967.
(11) Govorov, A. O.; Richardson, H. H. Generating heat with metal nanoparticles. Nano Today 2007, 2, 30−38. (12) Govorov, A. O.; Zhang, W.; Skeini, T.; Richardson, H.; Lee, J.; Kotov, N. A. Gold nanoparticle ensembles as heaters and actuators: melting and collective plasmon resonances. Nanoscale Res. Lett. 2006, 1, 84−90. (13) Hu, M.; Hartland, G. V. Heat dissipation for Au particles in aqueous solution: Relaxation time versus size. J. Phys. Chem. B 2002, 106, 7029−7033. (14) Luk’yanchuk, B. S.; Miroshnichenko, A. E.; Tribelsky, M. I.; Kivshar, Y. S.; Khokhlov, A. R. Paradoxes in laser heating of plasmonic nanoparticles. New J. Phys. 2012, 14, 093022. (15) Plech, A.; Kotaidis, V.; Gresillon, S.; Dahmen, C.; von Plessen, G. Laser-induced heating and melting of gold nanoparticles studied by time-resolved x-ray scattering. Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 70, 195423. (16) Richardson, H. H.; Hickman, Z. N.; Govorov, A. O.; Thomas, A. C.; Zhang, W.; Kordesch, M. E. Thermooptical properties of gold nanoparticles embedded in ice: Characterization of heat generation and melting. Nano Lett. 2006, 6, 783−788. (17) Tribelsky, M. I.; Miroshnichenko, A. E.; Kivshar, Y. S.; Luk’yanchuk, B. S.; Khokhlov, A. R. Laser pulse heating of spherical metal particles. Phys. Rev. X 2011, 1, 1. (18) Yeshchenko, O. A.; Kutsevol, N. V.; Naumenko, A. P. Lightinduced heating of gold nanoparticles in colloidal solution: Dependence on detuning from surface plasmon resonance. Plasmonics 2016, 11, 345−350. (19) Ng, K. M.; Chau, S. L.; Tang, H. W.; Wei, X. G.; Lau, K. C.; Ye, F.; Ng, A. M. C. Ion-desorption efficiency and internal-energy transfer in surface-assisted laser desorption/ionization: More Implication(s) for the thermal-driven and phase-transition-driven desorption process. J. Phys. Chem. C 2015, 119, 23708−23720. (20) Ayesh, A. I.; Qamhieh, N.; Mahmoud, S. T.; Alawadhi, H. Fabrication of size-selected bimetallic nanoclusters using magnetron sputtering. J. Mater. Res. 2012, 27, 2441−2446. (21) Beik, J.; Abed, Z.; Ghoreishi, F. S.; Hosseini-Nami, S.; Mehrzadi, S.; Shakeri-Zadeh, A.; Kamrava, S. K. Nanotechnology in hyperthermia cancer therapy: From fundamental principles to advanced applications. J. Controlled Release 2016, 235, 205−221. (22) Ko, S. H.; Pan, H.; Grigoropoulos, C. P.; Luscombe, C. K.; Frechet, J. M. J.; Poulikakos, D. All-inkjet-printed flexible electronics fabrication on a polymer substrate by low-temperature high-resolution selective laser sintering of metal nanoparticles. Nanotechnology 2007, 18, 345202. (23) Zinovev, A. V.; Veryovkin, I. V.; Moore, J. F.; Pellin, M. J. Laserdriven acoustic desorption of organic molecules from back-irradiated solid foils. Anal. Chem. 2007, 79, 8232−8241. (24) Roux, A.; Muller, L.; Jackson, S. N.; Baldwin, K.; Womack, V.; Pagiazitis, J. G.; O’Rourke, J. R.; Thanos, P. K.; Balaban, C.; Schultz, J. A.; Volkow, N. D.; Woods, A. S. Chronic ethanol consumption profoundly alters regional brain ceramide and sphingomyelin content in rodents. ACS Chem. Neurosci. 2015, 6, 247−259. (25) Xu, X. F.; Grigoropoulos, C. P.; Russo, R. E. Measurement of solid-liquid interface temperature during pulsed eximer-laser melting of polycristalline silicon films. Appl. Phys. Lett. 1994, 65, 1745−1747. (26) CRC handbook of chemistry and physics; Chapman and Hall/ CRCnetBASE; Boca Raton, FL, 1999 (pp CD-ROMs). (27) Kim, H.; Gilmore, C. M.; Pique, A.; Horwitz, J. S.; Mattoussi, H.; Murata, H.; Kafafi, Z. H.; Chrisey, D. B. Electrical, optical, and structural properties of indium-tin-oxide thin films for organic lightemitting devices. J. Appl. Phys. 1999, 86, 6451−6461. (28) Beaudoin, M.; DeVries, A. J. G.; Johnson, S. R.; Laman, H.; Tiedje, T. Optical absorption edge of semi-insulating GaAs and InP at high temperatures. Appl. Phys. Lett. 1997, 70, 3540−3542. (29) Johnson, S. R.; Tiedje, T. Temperature-dependence of the Urbach edge in GaAs. J. Appl. Phys. 1995, 78, 5609−5613. (30) Kar, A. K.; Wherrett, B. S. Thermal dispersive optical bistability and absorptive bistability in bulk ZnSe. J. Opt. Soc. Am. B 1986, 3, 345−350. 9559
DOI: 10.1021/acs.jpcc.7b01061 J. Phys. Chem. C 2017, 121, 9552−9559
