Radha Setty, Technical Advisor

Introduction

Monolithic Microwave Integrated Circuits (MMICs) with no active elements such as transistors, and containing only passive elements such as resistors, capacitors, inductors, are referred to as Integrated Passive Devices (IPD). These devices do not need DC power to operate, and do not perform frequency conversion as in the case of frequency mixers or frequency multipliers.

What’s the big deal about IPDs? The short answer is they perform vital functions which active elements cannot such as filtering, equalization, balanced-to-unbalanced line conversion (or vice versa) and many more as we will describe later. IPD brings all the advantages of continuously evolving MMIC technology developed for high volume applications like computers and cell phones to passive devices which were traditionally implemented with larger, more expensive technologies such as thin film. In short IPD’s are:

  • Small – The use of lumped components at low frequency and clever layout techniques at high frequency like multiple metal layers and on-chip interconnects result in extremely small device size.
  • Repeatable – Thanks to amazing advances in MMIC manufacturing, on-chip components can be manufactured with remarkable precision.
  • Highly reliable – Intensive R&D and innovations in MMIC manufacturing have achieved low-defect or defect-free processes. MMIC manufacturing foundries set design rules based on reliability studies and manufacturing capability which results in high yield. Mini-Circuits designers follow design rules provided by the MMIC foundry to eliminate manufacturing defects.
  • Low cost – MMIC manufacturing takes place in a clean room environment using automated equipment. Foundries can manufacture a large number of wafers in a short time with high yield. Mini-Circuits designers minimize die size using lumped components and clever layouts resulting in a large number of die per wafer, and therefore lower cost per die. Packaging and testing are performed using robots and automated systems for high throughput and lower overall production cost.
  • Available in packaged or die form – Die are used by hybrid circuit manufactures utilizing chip and wire assembly. The variety of die in the market is limited, but Mini-Circuits offers the majority of its MMICs, including IPD designs, in wafer form or as singulated die for such users [7] and as a stock item for rapid prototyping and production. Packaged die enable automated pick-and-place manufacturing at the user end.

Passive components using IPD technology in Mini-Circuits’ portfolio include:

This article explains basics of IPD technology and provides an overview of performance using Mini-Circuits components as examples. Mini-Circuits’ processes for reliability testing of all IPD products will also be discussed.

A Brief History of MMIC IPD at Mini-Circuits

For many years, Mini-Circuits designed and manufactured passive components using mostly wire-wound ferrite components and microstrip lines. Continued demand for cost and size reduction eventually drove the company to explore alternative technologies such as MMIC IPD. After several decades successfully  using IPD technology for passive component design, Mini-Circuits now offers one of the widest selections of state-of-the art MMIC passives in the industry off-the-shelf.

Semiconductor Substrate

IPD technology uses low-loss semiconductor substrates upon which passive elements are formed. An earlier article in this series [1] described intrinsic semiconductors (which have no impurities, intentionally added or otherwise) whose resistivity is close to that of insulators. For example, the voltage required for 1µA current across a 3mm long Silicon crystal with a cross section of 50µm x 100 µm is 1.38kV, which is quite high. Silicon has 1.5 x 1010/cm3 of free electrons at room temperature compared to 1028/cm3 in metal. Many orders of magnitude fewer free electrons in Silicon correspond to much higher electrical resistance than in metal. Insulation resistance is typically measured at DC, but many insulators become lossy at high frequency and are therefore not suitable as substrates for microwave and millimeter-wave frequencies. Semiconductors such as High-Resistivity Silicon, however, exhibit high insulation resistance at DC and low loss at microwave frequencies. The potential of semiconductors as a low-loss substrate were first recognized when Scientists at RCA labs in Princeton, NJ published properties of 50Ω lines on Silicon, GaAs and ceramic substrates to 50 GHz as early as 1981 [2]. Table 1 shows a summary of their findings.

Table 1: Loss of 50Ω microstrip line in various substrates [2].

SubstrateSubstrate Thickness (mils)Loss (dB/cm) at 50 GHz
High Resistivity Silicon (HRS)10~0.4
GaAs10~0.25
Ceramic25~0.1

The modest increase in loss of high resistivity Silicon and GaAs relative to ceramic illustrates their usability to 50 GHz. Now that we know these materials are usable as substrates to millimeter wave frequencies, the designer’s task is to minimize the die size to realize the benefits of these substrates. This is accomplished using lumped components at low frequency and distributed elements at high frequency. Let’s take a look at lumped element circuit theory.

Lumped Elements

Extensive research has been conducted worldwide on lumped elements, summarized by Dr. Inder Bahl [3] in his book titled Lumped Elements for RF and Microwave Circuits. Research into the use of lumped elements in MMICs started as early as 1965 when RCA in New Jersey claimed to have achieved size reduction by a factor of 10 [4].

In brief, any element on a semiconductor substrate is treated as a transmission line to account for its behavior at RF and microwave frequencies (Figure 1).

Figure 1: Transmission line with lumped elements.

Series resistance (R), series inductance (L), shunt conductance (G), and shunt capacitance (C) are all defined per unit length of the line. This means that if the designer aims to realize a resistor (R), it comes with unwanted inductance (L), shunt capacitance (C) and shunt conductance (G). The same is true for relaization of other elements. For example, if a designer aims to realize a series inductor (L), other unwanted elements (R, C and G) come with it. Designers need to take these unwanted effects into consideration. This is where design software such as 2.5D (example Keysight Momemtum) and 3D (Ansys HFSS) becomes vital and the design process becomes fairly complex.

In MMICs, real estate is a premium asset, and judicious choice of lumped and distributed elements is critical to minimize the size of the chip. Lumped elements are predominantly used at frequencies below 10 GHz, which saves significant real estate. Distributed elements scale with frequency and are therefore too big to use at low frequency but become a necessity at high frequency.

Let us further review how the basic lumped elements are implemented in MMIC designs.

Resistors

ADS Equivalent Circuit
Resistor
MMIC Implementation

Figure 2: Lumped element resistor, equivalent circuit and its MMIC implementation.

Resistors are implemented by depositing lossy thin film material such as TiN (Titanium Nitride), TaN (Tantalum Nitride), and NiCr (Nichrome) on a semiconductor substrate. To minimize unwanted parasitics such as shunt capacitance and series inductance, resistor area should be kept small.

The resistance of a sheet resistor is given by:

Rm=Rs(l/w)

Where Rs is sheet resistance in ohms/square= ρ/t

  • ρ=resistivity
  • t=thickness
  • l= length
  • w=width

Inductors

ADS Equivalent Circuit
Spiral Inductor
Rectangular Inductor

Figure 3: Inductor layout and equivalent circuit.

Inductors are implemented by routing a length of transmission line in a circular or rectangular spiral form as shown in Figure 3 to minimize the real estate. At microwave or millimeter wave frequencies, interwinding capacitance will affect performance. Closed form equations can be found in literature [3,4] as a starting point for design.

MIM (Metal Insulator Metal) Capacitor

ADS Equivalent Circuit
Capacitor
MMIC Implementation

Figure 4: MIM capacitor, equivalent circuit and MMIC implementation.

Capacitors have two metal plates with a dielectric sandwiched in between. Such constructions are called MIMs (Metal-Insulator-Metal). Insulators such as Silicon Nitride are used between two metal plates. Selected dielectric materials should have lower loss at the frequency of operation. Metal plates are normally gold or copper of sufficient thickness to minimize losses.

Vias

The back plane of a MMIC is normally a metallized surface which functions as a ground plane above which all passive elements are formed on the semiconductor surface. All interconnects from passive elements to ground need to be run through metallized via holes. Vias are also used for interconnecting metal layers. Advanced IPD processes come with three metal layers and have certain benefits. Interconnecting components through additional metal layers makes it easy to arrange more elements in a smaller footprint. To increase current carrying capacity or to minimize line losses, the metal layers are interconnected with vias.

Via
ADS Equivalent Circuit

Figure 5: Via hole and equivalent circuit.

Integrated Passive Device Process Technology

The typical MMIC manufacturing process consists of a series of steps. Manufacturing passive devices using the same processes employed for active devices such as HBT or pHEMT results in small size components but is very expensive, as many of the steps in these processes are unnecessary for passive devices adding unnecessary cost. Additionally, these processes are not optimized for the electrical performance of passive components. In response to higher demand for passive MMIC components, foundries developed dedicated MMIC processes for passive devices utilizing wafers without unnecessary layers and unnecessary steps that might be required for active circuits resulting in lower cost and optimum performance. Some foundries have gone further in providing multiple metal layers of greater thickness to:

  • Reduce line losses and increase power handling capability
  • Reduce interconnect losses and improve quality factor (Q) of inductors
  • Enable smart interconnect routing between elements

Figure 4 compares cross section views of pHEMT (Fig 4a) and IPD (Figure 4b) structures. Note the significantly fewer wafer layers in the IPD compared to those in the pHEMT.

Figure 6: Typical pHEMT and IPD wafer cross section. (Not to scale).

Mini-Circuits has taken advantage of all the benefits described above to offer a series of low cost and high performance MMIC passives to customers. Below is a brief summary of some of Mini-Circuits’ IPD product lines.

Fixed Attenuators

Fixed attenuators are one of the simplest of Mini-Circuits’ passive components, but achieving a flat attenuation vs. frequency response over wide bandwidths up to millimeter wave frequencies is quite challenging. In its simplest form a fixed attenuator is a Tee, Pi or Bridged-Tee network [3], each of which is pictured in Figure 7. Complexity arises at higher frequencies due to unwanted effects of the substrate and package parasitics (packaging material, bond wire etc.). 2.5D or 3D simulators are used to design and optimize the performance while accounting for package material, geometry, thermal properties and other factors. Figure 8 shows a plastic packaged attenuator and a die designed to operate from DC to 50 GHz. Note the small size. These attenuators are used in transmit and receive chains between active and passive elements to improve return loss [5, 6] or to set gain.

Figure 7: Lumped element attenuator configurations.

QAT-Series in Plastic QFN:
2 x 2 x 1mm

QAT-Series Die Layout:
0.75 x 0.7 x 0.1mm

Figure 8: MMIC attenuators in plastic QFN and bare die formats.

Figure 9: Attenuation and VSWR of QAT series attenuators.

Figure 9 shows the attenuation and VSWR performance of QAT-series attenuators operating to 50 GHz. These are plastic packaged parts in a 2x2mm footprint saving valuable real estate and providing excellent performance. QAT-series attenuators are available from stock with attenuation values from 0 to 30 dB.

Fixed Gain Slope Equalizers

Most wideband active elements such as amplifiers exhibit gain roll-off with increasing frequency. When several such amplifiers are cascaded, this response becomes exaggerated as the gain slopes of multiple amplifiers combine. When a system demands a flat frequency response, as many wideband transceivers do, equalizers come to the rescue.  An ideal slope equalizer should exhibit decreasing insertion loss as frequency increases. When used in conjunction with an active element with negative gain slope, equalizers flatten the combined gain response.

Figure 10: Example of a lumped element equalizer circuit.

Figure 10 shows a simple lumped element equalizer circuit. Omitting capacitor C and inductor L, it is just a pi-attenuator. Series capacitor C and inductor L help to decrease the attenuation as frequency increases. Equalizer insertion loss at DC and attenuation and desired attenuation at high frequency (or attenuation slope) can be achieved by changing element values.

Figure 11: Attenuation and VSWR vs. frequency of 45 GHz equalizers.

Mini-Circuits has successfully designed wideband MMIC equalizers operating from DC to 6, 20, 28 and 45 GHz with a wide variety of fixed slope values. Figure 9 shows the performance of DC to 45 GHz equalizers with slope values ranging from 3 to 10 dB.

Reflectionless Filters

In receiver chains, incoming signal is amplified by the LNA (Low Noise Amplifier), down-converted by a mixer and further amplified by an IF amplifier. Each of these components perform their intended functions, but also introduce unintended content such as intermodulation products and spurious signals. Filters are added to the chain to remove these unwanted signals, but often create still more problems as they reflect stopband signals in the reverse path, which can wreak havoc as they in re-mix with the desired signal to create more undesired intermodulation products affecting system performance.

Mini-Circuits recognized the significance of this problem for system designers and introduced a series of lumped, a constant impedance bandpass filters or Reflectionless Filters, [9] which are patent protected [10,11,12].

Plug-In PIF-21.4
(20mmx10mmx10mm)
Insertion Loss vs. Frequency
Equivalent Circuit
Return Loss vs. Frequency

Figure 12: Constant impedance band pass filter.

IPD technology helped commercialize reflectionless filters with the optimal combination of small size, low cost, and good filter performance. For example, Mini-Circuits model XBF-24+ has about 56 lumped elements packed in a 4 x 4 x 1mm plastic package. The package, equivalent circuit and performance for this model is shown by way of example in Figure 13. Note the size reduction of 125 in volume compared to a PIF-series plug-in style filter. Even though the frequency is not same, this shows the strength of IPD for size reduction.

XBF-24+ Surface Mount
(4mmx 4mm x 1mm)
Insertion Loss vs. Frequency
Equivalent Circuit
VSWR vs. Frequency

Figure 13: Reflectionless band pass filter XBF-24+.

Reliability

Performance, size and cost reduction mean nothing if the components are not reliable. Mini-Circuits  pays tremendous attention to during design and manufacturing processes to ensure parts are reliable when operated within the specified environmental ratings. Several measures taken by the team to ensure reliability.

All passive elements need to be designed to satisfy foundry rules for maximum ratings and component ratings. Component ratings are summarized in model data sheets, see the maximum ratings below for the KAT-30+ fixed attenuator as an example.

Table 2: Maximum ratings for KAT-30+ MMIC attenuator from model datasheet.

In this example, RF input power is associated with case temperature.  Operation below these maximum ratings avoids the risk of thermal-related failures. Mini-Circuits comes up with these numbers through rigorous testing, which includes:

  1. Thermal imaging at the maximum rated operating power to ensure the semiconductor die and plastic encapsulant are below the maximum temperature ratings specified by the manufacturers.
  2. High Temperature Operating Life Tests (HTOL) to demonstrate long term reliability

Mini-Circuits’ KAT-30+, 30 dB attenuator, operating over DC to 43 GHz was tested as follows:

Thermal Imaging

Thermal imaging was performed to determine hot spot temperature at a given DC input power. These images were used to determine thermal resistance. In preparing for thermal imaging, the plastic package is chemically etched to expose the die top surface. The etched sample is mounted on a PCB substrate with proper Input/Output connections. Figure 14 shows the thermal image of KAT-30+. This image is used to derive thermal resistance ӨJC.

ӨJC= (Th-Tb)/Pd=: 217.1oC/W

Where:

Th=Hot spot temperature on the Die

Tb= Ground lead/Case Temperature and

Pd= Power Dissipation during test

 This data and calculation is used to establish HTOL (High Temperature Operating Life Test) conditions.

Figure 14: Thermal image of KAT-30+ fixed attenuator.

HTOL (High Temperature Operating Life Test)

To demonstrate the long-term reliability of KAT-30+ at high temperature, HTOL testing is conducted under the following conditions:

 Input power: 0.575W (DC)

TG (Ground Lead/Case Temperature): 125oC

Tj( Hot-spot temperature on Die)= ӨJC x Pd=249.8oC

This model underwent a life test of 5000 hours under these conditions with a sample size of 80 units and no failures were found.

Note that the ground lead temperature is 125oC during HTOL testing while the maximum rated operating temperature is 85oC (See Table 2).

This is 400C below the theoretical maximum, allowing comfortable margin for error and extending the  MTTF (Mean Time to Failure) of the product, which is shown for different hot spot temperatures in Figure 15.

Figure 15: MTTF vs. Th of KAT-30+ (Assuming Ea of 0.7) (Note 1).

Conclusion

Mini-Circuits offers a vast array of low cost, high performance IPD products to help system designers to build their product. In this article we’ve covered the basics of the IPD process, some examples of real-world IPD products in Mini-Circuits’ portfolio and Mini-Circuits’ methodology for reliability testing. Be sure to check out our other articles on MMIC technologies, and explore hundreds of MMIC models on our website!

References

  1. A Primer on RF Semiconductors (MMICs) – Mini-Circuits Blog (minicircuits.com)
  2. A. Rosen et al., “Silicon as a millimeter-wave monolithically integrated substrate-A new look,” RCA Rev. vol. 42, pp. 633-660, Dec. 1981.
  3. Inder Bahl, “Lumped Elements for RF and Microwave Circuits”, Artech House, 2003
  4. D.A. Daly et. Al, “Lumped Elements in Microwave Integrated Circuits”, IEEE Transactions on MTT, PP713-721, VOL. MTT-15, NO. 12, DECEMBER 1967
  5. Less Besser, Rowan Gilmore, “Practical RF Circuit Design for Modern Wireless Systems, Vol 1, Passive Circuits and Systems”, Artech House, 2003
  6. Fixed Attenuators Help Minimize Impedance Mismatches – Mini-Circuits Blog (minicircuits.com)
  7. Josef Buechler et.al., “Silicon High-Resistivity-Substrate Millimeter-Wave Technology”, IEEE Transactions on Electron Devices, VOL. ED-33, NO. 12, PP 2047-2052, December 1986
  8. Semiconductor Die Ordering and Packaging Information – Mini-Circuits Blog (minicircuits.com)
  9. Constant-Impedance IF Band-pass Filters Improve Circuits Performance – Mini-Circuits Blog (minicircuits.com)
  10. Reflectionless Filter Basics: A Brief History of the Genesis of Reflectionless Filters – Mini-Circuits Blog (minicircuits.com)
  11. Stabilizing Multiplier Chain Conversion Efficiency with Reflectionless Filters – Mini-Circuits Blog (minicircuits.com)
  12. Pairing Mixers with Reflectionless Filters to Improve System Performance – Mini-Circuits Blog (minicircuits.com)

Notes:

  1. The above mean time to failure (MTTF) and or Failure in Time (FIT) figures are provided only as a potential guideline and merely reflect Mini-Circuits’ opinion or recommendation.  Using MTTF/FIT calculations alone are not appropriate to predict the life expectancy of a model; other factors such as heat dissipation, specific operating current and voltages, duration of use, actual stress and signal levels, as well as other environmental conditions are just a few factors that also need to be considered.  Accordingly, the MTTF/FIT figures provided do not constitute or create a warranty, express or implied, concerning future performance or the life expectancy of the model or otherwise.