…SEE the Future

Deep Depletion Technology Improves NearInfrared (700nm - 1100nm) Detection The document outlines the unique requirements of low light near-infrared detection (NIR) in scientific imaging applications. The traditional UV-NIR wavelength range detectable by silicon-based detectors, such as CCD (charge coupled devices), ranges from 200nm to 1100nm. For the purpose of this text, wavelengths above 700nm are considered to be near-infrared. For wavelengths beyond 1100nm, silicon is no longer a viable detector as it becomes transparent. In such cases, detectors made of a totally new type of material such as Indium Gallium Arsenide (InGaAs) are a preferred choice.

400

500 600 Wavelength (nanometers)

700

Figure 1: UV-VIS-NIR electromagnetic spectrum.

Over the years, interest in the far end of the wavelength spectrum has continued to grow as novel imaging techniques in biological and physical sciences make use of the advantages of working in the NIR region. These include: • • • • • •

Ability to probe materials and biological tissues at deeper depths Easy discrimination of NIR fluorescence from the tissue auto-fluorescence Increased availability of economical NIR illumination sources such as NIR lasers Development of a new class of stable, NIR fluorescent probes Increased interest in research areas such as Bose Einstein Condensate (BEC) Astronomical imaging (especially solar research) in NIR region

The detector technology has generally kept pace with increasing demands for low light level, NIR detection.

Quantum Efficiency Three types of CCD technology are widely used for scientific imaging -front-illuminated, back-illuminated and back-illuminated deep depletion. They are illustrated in the figure below.

Front

Back

Back Deep Depletion

Silicon Dioxide Thinned Silicon Incoming Light

Incoming Light

Thicker Silicon

Figure 2: Cross section of front, back and backilluminated, deep depletion CCDs.

Incoming Light

Poly-silicon Gate

Silicon Deep Depletion Technology

Silicon

Silicon page 1 of 6

In front-illuminated technology, the light enters the detector through poly-silicon gates and is absorbed in the depletion region of the silicon. Since a fraction of incoming light is lost at the poly-silicon gates, the detectors offer only a moderate quantum efficiency of around 50%* (Quantum efficiency is the fraction of incoming photons absorbed by the silicon). In back-illuminated detectors, on the other hand, the light enters directly into the silicon and as a result, the detectors offer QE upwards of 90%. In addition, the detectors also offer the highest QE in NIR for two reasons a) Longer path length for NIR photons increases the probability of photon being absorbed in silicon b) CCD surface is coated with antireflection coating that is specifically optimized for NIR wavelengths The quantum efficiency of the three types of detectors is shown below. Note that all these detectors can be coated with an optional UV conversion coating in order to increase their sensitivity in the ultraviolet. *Other front-illuminated detectors such as open-electrode and gates made of indium tin oxide offer better than typical front-illuminated sensitivity, but still below the back-illuminated sensitivity.

100

90 Back illuminated 80

CCD Quantum Efficiency (%)

…SEE the Future

Deep Depletion Technology

70 Back illuminated, deep depletion 60

50

40 Front illuminated 30

20

10

0 200

300

400

500

600

700

800

900

1000

1100

Wavelength (nm) Optional UV coating

Figure 3: Deep depletion detector technology provides the best quantum efficiency in NIR region (>700nm)

Etaloning An unfortunate drawback of back-illuminated CCD technology is that they exhibit "etaloning" - unwanted fringes caused by constructive and destructive interference patterns (see Appendix for complete description on etaloning). Unfortunately, the etaloning can not be easily removed in post-processing as it is dependent on several factors including spectral composition, degree of collimation and entrance angles of the illumination. The front-illuminated CCDs do not exhibit etaloning as the silicon is much thicker (600 µm) and the front surface is corrugated with poly-silicon gates. In fact, front-illuminated CCDs can provide an economical solution for NIR imaging applications, when the highest sensitivity is not required. The back-illuminated, deep depletion detectors are the best choice for NIR imaging as they eliminate etaloning and at the same time offer the highest sensitivity in the NIR. A series of images taken using a tungsten light source and a spectrograph shows the difference in etaloning from back-illuminated and back-illuminated, deep depletion detectors.

Deep Depletion Technology

page 2 of 6

…SEE the Future

Deep Depletion Technology

(a)

(b)

Figure 4: Series of images taken using a tungsten light source and spectrograph to show combined spectral and spatial etaloning between 650nm to 900nm (a) Images taken using back-illuminated CCD detectors clearly show etaloning. (b) back-illuminated, deep depletion CCD detectors remove the etaloning by using a thicker (~40 µm) silicon.

Dark current While deep depletion detectors offer higher NIR QE, they also suffer from higher than typical dark current rates by as much as 10 - 20 times. Typical back-illuminated detectors utilize a dark current suppressing technology called AIMO (Advanced Inverted Mode Operation) to keep the dark current as low as 0.0005 electrons/pixel/sec at -80°C. Deep depletion detectors typically use NIMO (Non-Inverted Mode Operation) to keep the diffusion of charge carriers to a minimum in the thicker depletion region. As a result, the dark current is higher. But, the latest generation of CCD cameras, such as PIXIS and VersArray cameras from Princeton Instruments, use the state-of-the-art Peltier or liquid nitrogen cooling to keep the detector temperature at -90°C or below for negligible dark current levels even in deep depletion detectors. These detectors are capable of minutes to even hours of integration times in low light level situations.

Choosing the right NIR detector From the discussions it is clear that a detector for the wavelength region >700nm should provide excellent QE and should not suffer from etaloning. In addition, care must be taken to maximize throughput by applying optimized anti-reflection coatings on the optical windows.

Deep Depletion Technology

page 3 of 6

…SEE the Future

C

Deep Depletion Technology front-illuminated CCD (F)

back-illuminated CCD (B)

Back, Deep Depletion CCD (BR)

Quantum Efficiency @ 800nm

Moderate ~40%

Good ~80%

Excellent ~92%

Dark Current

Low (1x)

Medium (1x - 1.5x)

High (10x-20x)

Etaloning

No

Yes

No or Negligible

Formats available

512 x 512 to 4096 x 4096

512 x 512 to 4096 x 4096

1024 x 1024 1340 x 1300

Table 1: back-illuminated deep depletion is the best choice for NIR (700nm - 1100nm) detection as it does not exhibit etaloning and provides the highest quantum efficiency.

NIR solutions from Princeton Instruments For over two decades, Princeton Instruments has been working with CCD vendors to provide the best NIR solutions. Notably, Princeton Instruments offers deep depletion detection technology in PIXIS and VersArray platforms. Both detectors use low noise electronics and deep cooling to provide the best low light sensitivity. PIXIS cameras offer the convenience of TE cooling and are manufactured using the state-of the art vacuum technology. As a result, the detector not only delivers deep cooling down to -90°C, but is guaranteed for lifetime.

VersArray, on the other hand, is the best choice when extended integration times and larger field of view are required. It is equipped with liquid nitrogen cooling

PIXIS: 1024BR

VersArray: 1300BR

CCD

1024 x 1024 back-illuminated, deep depletion

1340 x 1300 back-illuminated, deep depletion

Pixel Size

13 µm x 13µm

20 µm x 20 µm

CCD Area

13.3 mm x 13.3 mm

26.8 mm x 26 mm

Cooling

TE (Water/Air) -90°C

LN (-100°C to -120°C)

Readout speed

100kHz/2MHz

50kHz/100kHz/1MHz

Readout bits

16 bits

16 bits

Pixel Full well

100 ke- (typical)

250 ke- (typical)

Dark current

0.02 e-/p/sec @ -70°C

0.007 e-/p/sec @ -110°C

Table 2: Comparison of PIXIS: 1024BR and VersArray: 1300BR cameras from Princeton Instruments -two of the choices for NIR imaging.

Deep Depletion Technology

page 4 of 6

Everything in these detectors is optimized to achieve the best NIR sensitivity. 1. The CCD is made of thicker silicon, roughly double the thickness of a normal back-illuminated CCD. This contributes significantly to the absorption of NIR light, reducing the amount of light that survives a round-trip path to cause interference and increasing the QE. In addition, during the fabrication process, the CCD surface gets a NIR optimized antireflection coating to further increase the sensitivity. 2. An AR coating optimized for NIR wavelengths is applied on the vacuum window, the only window in the optical path. This is important as it recovers 3%-4% of the light loss that typically occurs at each optical surface. See Figure 4 below. 3. Deep cooling is essential for deep depletion detectors, as the starting dark current rate is higher. With deep TE cooling (PIXIS) or liquid nitrogen cooling (VersArray), the dark current is made almost negligible for the best possible signal-tonoise ratio.

Vacuum Window Transmission Data 100 90 NO AR

80 % Transmission

…SEE the Future

Deep Depletion Technology

70

UVAR

60 NIRAR

50 40 30

VISAR

20 10 0 200

400

600

800

1000

Wavelength (nm)

Figure 5: Princeton Instruments provides double sided antireflection coatings on the vacuum window that are optimized for maximum throughput in the NIR region (>700nm). Note that coatings optimized for other wavelengths are also available.

Deep Depletion Technology

page 5 of 6

…SEE the Future

Deep Depletion Technology

APPENDIX Etaloning - an introduction At NIR wavelengths, the silicon of which CCDs are made becomes increasingly transparent, causing the QE (quantum efficiency) to decline in the red. The back surface, where light enters a CCD in the back-illuminated configuration, is typically antireflection (AR) coated. Once light has passed through the body of a CCD and is about to reach the polysilicon electrodes, it encounters a sandwich of layers that generally includes silicon dioxide (refractive index 1.5). This sizeable discontinuity from the refractive index of silicon (which is 4) produces a large reflection back into the CCD. At wavelengths where silicon is transparent enough that light can traverse the thickness of the CCD several times, light bounces back and forth between the two surfaces. This sets up a standing wave pattern. Amplitude is lost at both reflective surfaces and by absorption in the body of the silicon. However, at longer wavelengths sufficient amplitude survives to cause significant constructive or destructive interference. While silicon is usually thought of as opaque, remember that a back-illuminated CCD is typically only 15 to 30 µm thick. A layer this thin can transmit a significant fraction of NIR light. For example, a back-illuminated CCD that is 17 µm thick (mechanically) would have the effective optical thickness of about 60 µm (since the refractive index of silicon in this wavelength range is 4). Thus, the round-trip optical path length between the surfaces is approximately 120 µm. At 750 nm, this would be 160 wavelengths. Therefore, there would be constructive interference at 750 nm. This pattern of interference would continue to repeat with intervals of about 5 nm.

Spatial Etaloning

Finesse

High Q Low Q

71.5

71.6

71.7

71.8

71.9

72

72.1

72.2

72.3

Optical Thickness (µm)

72.4

72.5

740

742

744

746

748

750

752

754

756

758

760

Wavelength (nm)

In addition to the spectral source of etaloning, in a thin CCD there can also be spatial etaloning. The spatial pattern arises from the incidence of monochromatic light on an etalon whose thickness is not perfectly constant. A small variation in thickness can change the local properties from constructive to destructive interference. The change required is only a halfwavelength in the round-trip path length. Since the index of silicon is 4, the change in CCD mechanical thickness required to produce this optical effect is only about 1/16 of a wavelength, or 0.05 µm at a wavelength of 800 nm. This effect can actually be used to visualize how uniform the thickness of a CCD is. If a CCD had perfectly uniform thickness, the modulation due to spatial etaloning at a given wavelength would disappear. All pixels would have the same degree of constructive or destructive interference at a given wavelength.

www.piacton.com email: [email protected] USA +1.877.4 PIACTON | Benelux +31 (347) 324989 France +33 (1) 60.86.03.65 | Germany +49 (0) 89.660.779.3 UK +44 (0) 28.38310171 | Asia/Pacific +65.6293.3130 China +86 135 0122 8135 | Japan +81.3.5639.2741

Deep Depletion Technology

page 6 of 6