Active Control of Energy Transfer in Plasmonic Nanorod

When excited with incident light, metal nanoparticles support localized surface plasmon resonances. [1][2][3] The plasmon is sensitive to changes in particle morphology and local environment, [4][5][6][7] providing dynamic optical insight into particle growth [8][9][10] and dissolution, [11][12][13] changes in local refractive index, [14][15][16] and charge [17][18][19][20] or energy transfer. [21][22][23][24] The energy available due to the plasmon's high absorption cross-section can be used in photocatalysis and photovoltaics to increase device efficiencies. 1,[25][26][27][28] Hybridization of nanoparticles with acceptors offers a way to capture the charge or energy stored by the plasmon before ultrafast relaxation of electron-hole pairs occurs. [29][30][31][32] Understanding the mechanisms of interfacial charge and energy transfer in hybrid materials is therefore crucial for future optimized device implementations. [33][34][35][36] Charge transfer (CT) is one viable mechanism for hybrid materials, but this process requires strict band alignment and is usually restricted to hard metal-inorganic interfaces. 17,31,37 More important for eventual applications is that CT requires a scavenger to avoid charge imbalance, which leads to eventual decomposition and device degradation. [38][39] On the other hand, energy transfer offers an avenue for the use of soft polymers at plasmonic interfaces while relaxing the band alignment requirement and allowing for increased processability of metal-organic interfaces. 21-23, 29, 33, 40-41 Resonance energy transfer (RET) is achieved by dipole-dipole coupling between the plasmon and acceptor, generating an electron-hole pair in the acceptor itself. The creation of an electron-hole pair in the acceptor eliminates the possibility of charge accumulation on the metal. 21,24,40,[42][43] Efficient RET occurs when the polymer acceptor is within the decay length of the plasmon's electric field and is most efficient when the spectral overlap between the plasmon donor and polymer acceptor is greatest. 22,[44][45][46][47] Here, we use gold nanorod (AuNR)-polyaniline (PANI) hybrids to demonstrate RET and its dynamic control by manipulating the degree of spectral overlap. Because of its reversible and controllable spectral modulation, PANI is a promising acceptor to couple with a plasmonic donor while providing ease of incorporation into existing platforms. 29,41,[48][49][50] Single-particle spectroscopy provides insight into changes to the scattering of single-AuNR-PANI hybrids by monitoring the plasmon resonance energy % HP��K ('&"%,'.?,&+: HVK ?,+:$7+"+:.")*./.'-."$8" .'/.4@%."(E.*(;,';",':.*.'+"+$"@7%D 4.(/7*.4.'+/B"O3"4$',+$*,';"-:(';./"+$"+:. /,';%.2)(*+,-%."V0"LP> .88,-,.'-3 HW��K -('"@." &.+.*4,'.&0"+:.*.@3"/:$?,';":$?"I71L2GI1J :3@*,&/"-('"@."7/.&"+$"-$'+*$%"+:."&3'(4,-"+7',';" $8"$)+,-(%"('&".%.-+*$',-")*$).*+,./"(+"/$8+" ,'+.*8(-./B �� J'"+:,/"/+7&30"+:."/,';%.2 )(*+,-%."V"*.)$*+/"$'"-:(';./",'"LP>",'"I71L2 GI1J":3@*,&/ (/"+:.")*$+$'(+,$'"/+(+."$8"GI1J",/" /?,+-:.&"@3")MB"9$**.%(+.&")%(/4$'"/).-+*("$8" I71L2GI1J":3@*,&/"(*."-()+7*.&",'"@$+:"(-,&,-" ('&"@(/,-"-$'&,+,$'/"7/,';"("8%7,&,-"-.%%"$'"('" ,'E.*+.&":3).*/).-+*(%"&(*D28,.%&"4,-*$/-$).B">:." ,':.*.'+"/+*7-+7*(%":.+.*$;.'.,+3"?,+:,'"('" .'/.4@%."$8"I71L2GI1J":3@*,&/",/"+:."D.3"+$" @.,';"(@%."+$"$@/.*E."("&,/+*,@7+,$'"$8"W��+:(+" -$**./)$'&/"?,+:"("*(';."$8"/).-+*(%"$E.*%()/B" P'/.4@%.".F+,'-+,$'"/).-+*( $8 I71L/ @.8$*."('&"(8+.*"I71L2GI1J":3@*,&,6(+,$' (*. /:$?'",'"X,;7*."YIB">:."@(*."I71L/ H(E.*(;." &,4.'/,$'/"$8"YZ"F"ZY"'4K :(E."('".'/.4@%. %$';,+7&,'(%"P�� $8 YB[\".]B">:."8$*4(+,$'"$8" I71L2GI1J":3@*,&/0"?,+:"("*.)*./.'+(+,E." +*('/4,//,$'".%.-+*$'"4,-*$/-$)3"H>PTK",4(;. ,'" +:.",'/.+"$8"X,;7*."YI0"*./7%+/",'"("/:,8+ ,'"+:."P��

��

to higher energy, reflected at ~1.8 eV. The AuNR-PANI hybrid extinction shift confirms the successful functionalization of the AuNRs with a PANI shell, as an increase in Eres has been seen previously by Wang and coworkers upon PANI functionalization in the oxidative solution environment. 55 The resonance in the extinction spectrum at a lower energy than the plasmon at ~1.4 eV is from PANI absorption due to transitions from the highest occupied molecular orbital (HOMO) on the benzenoid to the lowest unoccupied molecular orbital (LUMO) on the quinoid rings. [56][57] Single-particle analysis resolves inter-particle heterogeneity and reveals broadening of the Γ, which is not consistent with a simple refractive index change. Single-particle dark-field scattering (DFS) spectra of a representative AuNR and a AuNR-PANI hybrid with similar Eres reveal that the addition of a ~5 nm PANI shell drastically changes the optical response, Figure 1B. See experimental methods for a description of the single-particle measurements. The single-particle Eres and Γ are free from heterogeneous broadening that is present in the bulk measurements. With homogeneously broadened single-particle DFS spectra, changes to the Eres are directly related to changes in the local refractive index, and changes in the homogenous Γ are directly related to changes in the plasmon dephasing time. 24,[58][59] The single-particle homogeneous Γ is inversely proportional to the plasmon decay time constant, allowing change to the extent of plasmon damping to be monitored through changes to the Γ observable. 23,31,60 Single-particle distributions in Eres in the left panel of Figure 1C are consistent with previously reported changes to the AuNR-PANI hybrid measured in air, in which the addition of a 5 nm PANI shell decreased the Eres of the hybrid. [61][62] Broadening of the Γ confirms that an additional relaxation pathway is opened when AuNRs are hybridized with PANI shells (Figure 1C, right). Because the Γ is inversely proportional to the plasmon dephasing time, [31][32]51 single-particle DFS spectra can directly monitor the effects of interfacial changes on the relaxation dynamics in each AuNR. In these particles, the increase in the Γ of the AuNR-PANI hybrids when compared to bare AuNRs, Figure 1C (right), indicates the presence of an additional plasmon damping pathway. If the change in the Γ were related solely to the decrease in Eres due to refractive index changes, the Γ would decrease (rather than increase) as the AuNR Eres shifts away from interband damping contributions. The total single-particle Γ (ΓTotal) includes contributions from bulk damping (ΓBulk) intrinsic to the Au metal, and size-dependent radiation damping (ΓRad) and electron-surface scattering (ΓSurf). 60,63 Additional damping mechanisms that broaden the Γ include CT and RET and are often referred to as chemical interface damping (CID). 64 Broadening due to these additional plasmon damping pathways is represented as ΓCID here. Using the average size of the AuNRs, 14 x 41 nm, the damping contributions from ΓBulk, ΓRad, and ΓSurf were removed from the ΓTotal of the bare AuNRs and AuNR-PANI hybrids.

Damping from ΓBulk is energy-dependent due to the contributions from intraband and interband transitions intrinsic to gold and were removed using the experimental dielectric function discussed further in Figure S1. As shown in Figure S1, after the removal of the non-CID contributions the bare AuNRs have an average Γ of 16 ± 19 meV, demonstrating that these contributions alone account well for the ΓTotal. However, the average value for AuNR-PANI hybrids, after accounting for ΓBulk, ΓRad, and ΓSurf, is still 140 ± 50 meV, confirming that a large portion of the homogenous Γ must be attributed to ΓCID. In fact, we calculate a CID efficiency of 58% (Figure S1). The increase in plasmon damping after the addition of the PANI shell is therefore attributed to CID. To assign CID to RET (as opposed to CT or other mechanisms), an additional control variable is necessary, as discussed next.

( `3'(4,-"/?,+-:,'; $8"/).-+*(%"$E.*%()" @.+?..'"&$'$*"('&"(--.)+$*",/ 7/.&"(/"+:." ,4)$*+('+"-$'+*$%"E(*,(@%."('& (-:,.E.&"+:*$7;:" 4$&7%(+,$'"$8"GI1Jc/ )*$+$'(+,$'"/+(+.0"/:,8+,';" GI1J"(@/$*)+,$'"(/"+:. )$%34.*",/"+7'.&"8*$4" . 4.*(%&,'."@(/."+$".4.*(%&,'."/(%+ HX,;7*."dKB 9:(';,';"+:.")M"4$&7%(+./"GI1Jc/")*$+$'(+,$'" /+(+.0"?:,-:",'"+7*'"*./7%+/ ,'"/:,8+/",' +:."GI1J" .F+,'-+,$'"4(F,474 (/"/:$?'",'"X,;7*."dIB >:.

.F+,'-+,$'",/"&$4,'(+.&"@3"(@/$*)+,$'"('&"+:." -:(';.",'",+/"4(F,474",/"("*./7%+"$8"+:."-:(';.",'" +:."MUTU"('&"5STU"%.E.%/ (/"+:.")$%34.*",/" )*$+$'(+.&B �� >:."-:.4,-(% /+*7-+7*./",'"X,;7*." dO"&.),-+")*$+$'(+,$'"/+(+./"$8"GI1J ,'"+:." :,;:"('&"%$?")M".'E,*$'4.'+/0"*./).-+,E.%3B"�� 4$&7%(+,$'"$8"GI1J"(@/$*)+,$'")*$E,&./" &3'(4,-"-$'+*$%"$8"+:."/).-+*(%"$E.*%()"@.+?..'" +:."I71L"&$'$*"('&"GI1J"(--.)+$*"8$*"+:."('+,-,)(+.&"4$&7%(+,$'"$8"9J`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dB PE.'"+:$7;:"+:."V"&$./" '$+"*.+7*'"+$"+:."/(4."E(%7.",'"@$+:"(-,&,-" -$'&,+,$'/0"%,D.%3"&7."+$")$%34.*"*.$*;(',6(+,$'"

during the pH transitions, Figure 3B demonstrates the pH dependent dynamic damping. In acidic conditions, the intensity decreases while the Γ and Eres increase.

A statistical comparison of spectra from many single-AuNR-PANI hybrids confirms that the plasmon increases in Eres and damped in acidic pH conditions, as shown in Figure 4. The Eres of 165 single-particles increases as the pH is changed from a basic to an acidic environment, as shown by complementary cumulative distribution "+ 87'-+,$'/",'"X,;7*."ZIB">:."P�� ,'-*.(/./"(/"+:." GI1J",/")*$+$'(+.&B">:.",'-*.(/.",'"P��0"fBfZ".]0" ,/"/,4,%(*"+$"-:(';./")*.E,$7/%3"*.)$*+.&"8$*" '('$)(*+,-%./"?,+:"+:,'"GI1J"/:.%%/B �� >:." 4.-:(',/4"*./)$'/,@%."8$*"+:,/"/:,8+",/"("-:(';." ,'"+:."*.8*(-+,E.",'&.F"(//$-,(+.&"?,+:"+:."

)*$+$'(+,$'"$8"GI1JB" I"/,4,%(*"-$4)(*,/$'"$8"V�� ,'"X,;7*." ZO"/:$?/"+:(+"(/"+:.")M",/"/?,+-:.&"8*$4"YY"+$"\0" V�� ,'-*.(/./"@3"('"(E.*(;. $8"\Z"g"Ya"4.]0" ?:,-:"-(''$+"@.".F)%(,'.&"@3"("*.8*(-+,E.",'&.F" -:(';.B"U'."-$'+*,@7+$*"+$"+:."$@/.*E.&"V�� ,'-*.(/. ,/"+:(+"(/"+:."P�� $8"+:. I71L2GI1J" A correlated comparison of the changes in the Eres and Γ from individual AuNR-PANI hybrids at pH 11 and pH 3 supports that refractive index changes can account entirely for the shifts in Eres, while an additional, and at first glance, more complex, mechanism is required to explain the Γ broadening (Figure 5). Figure 5A shows the Eres for correlated AuNR-PANI hybrids in acidic and basic pH conditions.

Previous work established PANI's refractive index decreases when protonated, and the decrease in refractive index leads to an increase in the AuNR Eres. 59,66 The experimental ΔEres, 39 meV, is similar in magnitude to ΔEres in previously investigated plasmonic-PANI hybrids. 67 The PANI shell does not completely fill the AuNR sensing volume, 6.6 nm on average for the AuNR-PANI hybrid. 68 The average PANI thickness as calculated from TEM for the tips and sides is 3.4 and 5.7 nm, respectively, shown in Figure S4. The green point in Figure 5A illustrates the change in Eres from finite-difference time-domain (FDTD) simulations given in Figure S4. The simulated ΔEres, 67 meV, is slightly larger than the experimental ΔEres. One likely explanation for the difference between the calculated and measured values is that water penetration into the PANI shell also contributes to the effective refractive index sensed by the AuNR. 69 Figure 5B compares the correlated Γ for 165 AuNRs, which exhibit a more complicated relationship than seen for Eres. The Γ in Figure 5B are plotted after removing contributions from ΓBulk, ΓRad, and ΓSurf, leaving only ΓCID. ΓBulk is removed using the Eres dependent Au bulk damping, shown in Figure S3A. ΓRad of 2.3 meV is subtracted based on the average AuNR volume. ΓSurf is calculated to be 15 meV, assuming an A parameter of 0.12 determined for bare AuNRs as further explained in Figure S1. 30,63 Unlike for Eres, the Γ increases at pH 3 for 81% of the AuNR-PANI hybrids, but not for all. For a portion of single AuNR-PANI hybrids the Γ increases by 50% when the pH is changed from 3 to 11, and for other hybrids there is no change in Γ, despite consistent active switching for Eres for the same hybrids under the same conditions. This data, and the inconsistent trends, suggest an additional, nonlinear mechanism for the increased damping.

Accounting for the Eres dependent changes in the Γ for all of the AuNR-PANI hybrids can explain the apparent inconsistencies in the data shown in Figures 5A,B, and allow us to assign a mechanism of RET, as opposed to CT, for CID, as shown in Figures 5C,D. CID encompasses multiple mechanisms, including adsorbate-induced dipole scattering, CT, and RET, all of which lead to accelerated decay of the plasmon. 64 CID can contribute to a large percentage of the total plasmon damping, and which of these is the dominant mechanism can be determined based on the hybrid electronic structure and the chemical nature of the interface. 64 Damping due to dipole scattering occurs when the chemical adsorbate has a static molecular dipole moment that creates an image dipole inside the metal, a resonance independent phenomenon. CT is an interfacial decay pathway, not depending on the acceptor thickness, yet must meet band alignment energy requirements between the donor and acceptor. 37 The addition of the PANI induces plasmon damping due to CID, possibly a result of interfacial CT. CT has been shown to be independent of Eres, however, and here we see an Eres dependence on the extent of plasmon damping. 17, 70 However, the change in the Γ seen as the pH tunes the spectral overlap between the AuNR and the PANI, as shown in Figure 5B, points to RET. RET, unlike adsorbate-induced dipole scattering, occurs when the plasmon oscillation couples with the polymer transition dipole moment.

Size heterogeneity within the AuNR sample provides an internal control for RET, because corresponding shifts in Eres as a function of rod length determine whether the strongest spectral overlap between the Eres of the AuNR and PANI's absorption occurs in acidic or basic conditions (Figure 5C).

The spectral overlap determines the dipole-dipole coupling strength between the donor (in this case, AuNR) and acceptor (in this case, PANI), and thus the effective ηRET. In even the most homogeneous AuNR sample, there is a range of AuNR aspect ratios, causing different Eres values, which contributes to heterogeneous broadening in ensemble measurements. But for single-particle DFS, this range in Eres provides a set of un-broadened scattering spectra that have varying spectral overlaps with the two forms of PANI. Intrinsic sample heterogeneity therefore acts as an internal spectrometer for our analysis.

Lorentzian fits to the scattering spectra of two representative AuNRs with different Eres (black), and the extinction spectra of PANI in acidic (red) and basic (blue) conditions are shown in Figure 5C. The areas where each AuNR spectrally overlaps with the acidic PANI spectrum are shaded red, while regions "$ +:(+"$E.*%()"?,+:"+:."@(/,-"GI1J"/).-+*74 (*."/:(&.&"@%7.B >:.")7*)%.2/:(&.&"*.;,$'",'"X,;7*."a9" &.'$+./"+:. *.;,$'/ ?:.*. +:.*.",/"'$"-:(';.".F).-+.&",'"+:."/).-+*(%"$E.*%()"?:.'"+:.")M",/"/?,+-:.&B"

�� V�� V��0"V��('& V��

More details for the calculation of this region are included in Figure S5, but briefly, due to the broad absorption of PANI, AuNRs with Eres near 1.8 eV have equal amounts of spectral overlap at both pH conditions. Shorter AuNR-PANI hybrids with Eres higher than 1.8 eV have greater spectral overlap at pH 11 and thus increased RET, observed as increased ΓCID at basic conditions. In contrast, longer AuNR-PANI hybrids with Eres lower than 1.8 eV have better spectral overlap at pH 3, resulting in increased RET at acidic conditions. This switch between optimal conditions for RET as a function of AuNR size and pH implies that there should be an inversion point in ηRET around 1.8 eV.

Calculated ΔηRET, shown in Figure 5D for AuNR-hybrids with various Eres values, confirms an inversion point around 1.8 eV and strongly supports that RET is responsible for the experimentally observed increase in the Γ at pH 3. To determine if the change in the Γ is Eres dependent, the absolute percent change in the RET contribution to the Γ when switching from acidic to basic conditions is plotted.

The Eres dependent ΔηRET is given as the relative change in the Γ between pH conditions, shown in Figure 5D. Only ηRET defined as ΓCID=RET / ΓTotal (Figures S3B and S3C) is not sufficient to resolve the Eres dependence as shown in Figure S3D. The ΔηRET data is binned every 0.10 eV to clearly present the RET Eres dependence of 165 AuNRs. The red data points represent the AuNRs where the spectral overlap is greatest with the acidic conditions, corresponding to Figure 5C. The blue data points represent those with more spectral overlap in basic conditions, while the purple points represent those AuNRs with no change in the spectral overlap integral as the pH is switched. To visualize the spectral overlap integral between the AuNR and the PANI, the absolute value of the spectral overlap integral is plotted. The gray-shaded region in Figure 5D represents one standard deviation of the spectral overlap integral, calculated by integrating the area under a Lorentzian fit to a representative AuNR and the PANI absorption in both pH conditions, further detailed in Figure S5 and visualized in Figure 5C. The overlaid absolute change in spectral overlap matches well with the experimental data, identifying that ΔηRET has a strong Eres dependence as a result of the change in the spectral overlap integral. The data clearly exhibits an inversion point, with ηRET increasing where the difference in spectral overlap is the greatest, at energies both below and above 1.8 eV for acidic and basic conditions, respectively. Due to the initial Eres of the AuNR sample, 1.73 eV, used for PANI hybridization there are fewer AuNR-PANI hybrids at Eres > 1.8 eV, highlighting the importance of single-particle measurements to uncover this Eres dependent change in ΔηRET.

In conclusion, we identified RET as a new type of interaction between AuNR donors and PANI acceptors under reversible protonation of the PANI shell, only possible due to single-particle resolution.

The Eres dependent ηRET reported in this system highlights a previously unreported plasmon damping mechanism in AuNR-PANI hybrids. The dynamic tunability of ηRET by varying the PANI absorption through changes of the solution pH allows the AuNR-PANI hybrid RET to be turned on and off in situ.

For our system, shorter AuNRs with higher Eres have greater spectral overlap in basic conditions, whereas longer AuNRs with lower Eres have greater spectral overlap in acidic conditions. Single-particle measurements made it possible to map out the spectral dependence of ΔηRET when changing the pH because of variation in the Eres of the AuNRs using the same hybrid sample. Although the change in the Γ is much larger between the bare and PANI coated AuNRs (Figure 2) compared to when the pH is tuned (Figure 4), in situ Γ tuning and the dispersity in aspect ratio and hence Eres made it possible to clearly assign CID to RET in AuNR-PANI hybrids. In contrast, the inability to perform correlated DFS of the same AuNRs before and after coating together with electron microscopy hindered the accurate subtraction of the size dependent damping contributions ΓRad, and ΓSurf and therefore a clear assignment of RET.

Overall, our findings can be used to capture the energy offered by the plasmon to drive RET-based chemistry efficiently.

AuNR-PANI Hybrid Synthesis and Characterization:

Synthesis of AuNRs: AuNRs, shown in Figure S6, of 41 nm length and 14 nm width were synthesized with minor modifications as published elsewhere. 71 Briefly, seed particles were synthesized by adding 300 µL of a freshly prepared 0.01 M NaBH4 solution in a 4.7-mL-mixture of 0.1 M CTAB and 0.25 mM HAuCl4 under vigorous stirring. The solution was stirred rapidly for 2 min followed by continued slow stirring at 32 °C for 30 min. 1 L of a 0.1 M CTAB solution containing 0.25 mM HAuCl4, 0.06 mM AgNO3, 500 µL of a 0.1 M HBr solution to adjust the pH, and 0.35 mM ascorbic acid as reducing agent was prepared. 4 mL of the seeds was then added to this solution after 30 min and mixed thoroughly.

Synthesis of the PANI Shell: 876 µL aqueous SDS solution (80 mM) and 60 µL HCl (1 M) were added to 8.2 mL water. Under vigorous stirring, 2 mL of AuNR suspension (2 mM Au0, 1 mM CTAC) was added. Subsequently, 1.2 mL of aqueous aniline solution (10 mM) was added. The polymerization was initiated by adding 1.5 mL of aqueous APS solution (13.33 mM). After one hour, the stirring speed was reduced, and the reaction mixture was stirred for another 4 hours. The PANI-coated AuNRs were purified by three centrifugation-re-dispersion cycles (8500 rcf for 30 min, re-dispersion in 5 mM SDS solution). The supernatant from the first centrifugation cycle was kept for UV-vis-NIR measurements.

UV-vis-NIR Spectroscopy: Extinction spectra were acquired with a Cary 5000 spectrophotometer (Agilent Technologies Deutschland GmbH). The value of the extinction spectrum at a wavelength of 400 nm (interband transitions of Au) was used to calculate the concentration of the AuNR dispersions. 72 For the pH-dependent extinction measurements, aqueous solutions of 5 mM SDS were adjusted to pH 3 and pH 11 by adding 1 M HCl and 1 M NaOH, respectively.

Single-AuNR Sample Preparation: 600 μL (12 x 50 μL) of bare AuNRs or AuNR-PANI hybrids were spun cast onto glass slides at 3000 rpm for 60 seconds for each cycle. Bare AuNRs (14 x 41 nm) and AuNR-PANI hybrids were deposited to achieve an approximate particle density of ∼5 × 10 -5

AuNRs/mm 2 . Glass slides were oxygen plasma cleaned for 2 min directly before spin casting.

Fluidic Cell Preparation: Adhesive 0.8 mm thick silicon spacers (Grace Bio-Laboratories) were cut to create a microfluidic channel for incorporation onto the microscope. The fluidic cell was then assembled by sandwiching two 50 x 75 mm, 0.17 mm thick glass coverslips (Brain Research Lab) together using the adhesive silicon spacer with the top glass having two 1 mm diameter openings laser cut at the corners to create the flow channel. The cell was placed in a custom-made sample holder and pressured sealed. Tubing and a syringe were used to create a vacuum in the fluidic chamber allowing the solution in the cell to be changed while on the microscope.

In Situ Hyperspectral Imaging: Dark-field scattering (DFS) spectra of individual bare AuNRs and AuNR-PANI hybrids were measured using a hyperspectral dark-field microscope. 73 A Zeiss AxioObserver m1 inverted dark-field microscope body was fitted with an oil immersion condenser (numerical aperture NA=1.4) focusing light from a 100 W tungsten-halogen lamp (at 3200 K) to the sample plane. Scattering from the AuNRs and AuNR-PANI hybrids was collected using an oil immersion objective (Zeiss, PlanAchromat 63x, NA=0.7). Collected light moved through a mechanical slit (20 μm) into a spectrograph (Princeton Instruments, Acton SP2150i) with a diffraction grating (800 nm blaze wavelength, 150 lines/mm). The spectrally dispersed light was then sent to a camera (Princeton Instruments, PIXIS 400). The spectrograph and camera setup were mounted on a translation stage, allowing the accumulation of a hyperspectral data cube. Data acquisition and instrument control were achieved using NI LabVIEW software with customized sub-routines. Hyperspectral images were taken with an integration time of 5 seconds per pixel. Hyperspectral images were acquired for high and low pH conditions using sodium hydroxide and nitric acid solutions. To ensure full protonation occurred using nitric acid, a control experiment was performed with HCl, and the same spectral changes were observed (Figure S7). All AuNR and AuNR-PANI hybrid DFS spectra were fit to a single Lorentzian function